Driver Behaviour and Training, Vol. 3 (Human Factors in Road and Rail Transport)

DRIVER BEHAVIOUR AND TRAINING Human Factors in Road and Rail Transport Series Editors Dr Lisa Dorn Director of the Dr...

Author: Lisa Dorn

35 downloads 1274 Views 2MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

DRIVER BEHAVIOUR AND TRAINING

Human Factors in Road and Rail Transport Series Editors Dr Lisa Dorn Director of the Driving Research Group, Department of Human Factors, Cranﬁeld University Dr Gerald Matthews Professor of Psychology at the University of Cincinnati Dr Ian Glendon Associate Professor of Psychology at Grifﬁth University, Queensland, and is president of the Division of Trafﬁc and Transportation Psychology of the International Association of Applied Psychology

Today’s society must confront major land transport problems. The human and ﬁnancial costs of vehicle accidents are increasing, with road trafﬁc accidents predicted to become the third largest cause of death and injury across the world by 2020. Several social trends pose threats to safety, including increasing car ownership and trafﬁc congestion, the increased complexity of the human-vehicle interface, the ageing of populations in the developed world, and a possible inﬂux of young vehicle operators in the developing world. Ashgate’s ‘Human Factors in Road and Rail Transport’ series aims to make a timely contribution to these issues by focusing on the driver as a contributing causal agent in road and rail accidents. The series seeks to reﬂect the increasing demand for safe, efﬁcient and economical land-based transport by reporting on the state-of-theart science that may be applied to reduce vehicle collisions, improve the usability of vehicles and enhance the operator’s wellbeing and satisfaction. It will do so by disseminating new theoretical and empirical research from specialists in the behavioural and allied disciplines, including trafﬁc psychology, human factors and ergonomics. The series captures topics such as driver behaviour, driver training, in-vehicle technology, driver health and driver assessment. Specially commissioned works from internationally recognised experts in the ﬁeld will provide authoritative accounts of the leading approaches to this signiﬁcant real-world problem.

Driver Behaviour and Training Volume III

Edited by LISA DORN Cranﬁeld University, UK

© Lisa Dorn 2008 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise without the prior permission of the publisher. Lisa Dorn name has asserted her moral right under the Copyright, Designs and Patents Act, 1988, to be identiﬁed as the editor of this work. Published by Ashgate Publishing Limited Gower House Croft Road Aldershot Hampshire GU11 3HR England

Ashgate Publishing Company Suite 420 101 Cherry Street Burlington, VT 05401-4405 USA

Ashgate website: http://www.ashgate.com British Library Cataloguing in Publication Data International Conference on Driver Behaviour and Training (3rd : 2007 : Dublin) Driver behaviour and training. - (Human factors in road and rail transport) 1. Motor vehicle drivers - Training of - Congresses 2. Motor vehicle drivers - Attitudes - Congresses 3. Motor vehicle driving - Congresses I. Title II. Dorn, Lisa 629.2'83 Library of Congress Cataloging-in-Publication Data International Conference on Driver Behaviour and Training (1st : 2003 : Stratford-upon-Avon, England) Driver behaviour and training / edited by Lisa Dorn. p. cm. Includes bibliographical references and index. ISBN 978 0 7546 7203 6 1. Trafﬁc safety--Congresses. 2. Automobile drivers-- Congresses. 3. Automobiles--Safety appliances--Congresses. 2. Automobile driver education--Congresses. 1. Dorn, Lisa. II. Title. HE5614.I553 2003 363.12'5--dc22

2003058287

ISBN 978-0-7546-7203-6

Printed and bound in Great Britain by MPG Books Ltd. Bodmin, Cornwall.

Contents List of Figures List of Tables Preface

ix xiii xvii

Part 1 The Novice Driver Problem 1

2

3

4

5

6

7

How Do ‘Signiﬁcant Others’ Inﬂuence Young People’s Beliefs About Driving? Amanda Green and Lisa Dorn

3

Piloting a Telemetric Data Tracking System to Assess Post-training Real Driving Performance of Young Novice Drivers Robert B. Isler, Nicola J. Starkey, Peter Sheppard and Chris Yu

17

Fault Correction or Self-Assessment: Which Way Forward? Ian Edwards and Tracey Curle

31

New Elements in the Dutch Practical Driving Test: A Pilot Study Jan Vissers, Jolieke Mesken, Erik Roelofs and René Claesen

37

Personality and Attitudinal Predictors of Trafﬁc Offences Among Young Drivers: A Prospective Analysis Lisa Wundersitz and Nicholas Burns

51

Pre-driving Attitudes and Non-driving Road User Behaviours: Does the Past Predict Future Driving Behaviour? Helen N. Mann and Mark J.M. Sullman

65

Prediction of Problem Driving Risk in Novice Drivers in Ontario: Part II Outcome at Two Years Laurence Jerome and Al Segal

75

Part 2 Emotions and Driver Behaviour 8

9

A Review of Studies on Emotions and Road User Behaviour Jolieke Mesken, Marjan Hagenzieker and Talib Rothengatter

91

A Comparison of the Propensity for Angry Driving Scale and the Short Driving Anger Scale Mark J.M. Sullman

107

vi

10

11

Driver Behaviour and Training – Volume III

Aggression and Non-aggression Amongst Six Types of Drivers Évelyne F. Vallières, Pierre McDuff, Robert J. Vallerand and Jacques Bergeron

117

The Inﬂuence of Age Differences on Coping Style and Driver Behaviour Elizabeth Andrews and Stephen Westerman

129

Part 3 At Work Road Safety 12

13

14

15

16

17

18

19

20

Effects of Organisational Safety Culture on Driver Behaviours and Accident Involvement Amongst Professional Drivers Bahar Öz and Timo Lajunen

143

Stages of Change in the Australian Workplace and its Application to Driver Education Tamara Banks, Jeremy Davey and H. Biggs

155

Prospective Relationships between Physical Activity, ‘Need for Recovery’ and Driver Accidents and Absenteeism Adrian Taylor and Lisa Dorn

167

Predicting High Risk Behaviours in a Fleet Setting: Implications and Difﬁculties Utilising Behaviour Measurement Tools Jeremy Davey, James Freeman and Darren Wishart

175

Driver Celeration Behaviour in Training and Regular Driving Anders af Wåhlberg and Lennart Melin

189

A Study of Contemporary Modiﬁcations to the Manchester Driver Behaviour Questionnaire for Organisational Fleet Settings James Freeman, Jeremy Davey and Darren Wishart

201

A Comparison of Seat Belt Use Between Work Time and Free Time Driving Among Turkish Taxi Drivers Özlem Şimşekoğlu and T. Lajunen

215

A Review of Developing and Implementing Australian Fleet Safety Interventions: A Case Study Approach Update Darren Wishart, Jeremy Davey and James Freeman

227

Designing a Psychometrically Based Self-Assessment to Address Fleet Driver Risk Lisa Dorn and Julie Gandolﬁ

235

Contents

vii

Part 4 Technological Interventions, Driver Behaviour and Road Safety 21

22

23

24

25

Development of Multimedia Tests for Responsive Driving Erik Roelofs, Marieke van Onna, Reinoud Nägele, Jolieke Mesken, Maria Kuiken and Esther Cozijnsen

251

The Effect of Simulation Training on Novice Driver Accident Rates R. Wade Allen, George D. Park and Marcia L. Cook

265

Driving Experience and Simulation of Accident Scenarios Catherine Berthelon, Claudine Nachtergaële and Isabelle Aillerie

277

Investigating the Contexts in which In-Vehicle Navigation System Users Have Received and Followed Inaccurate Route Guidance Instructions Nick Forbes and Gary Burnett

291

Comparison of Novice Drivers in Austria and the Czech Republic With and Without the Use of Intelligent Speed Adaptation Christine Turetschek and Ralf Risser

311

Part 5 Human Factors and the Road Environment 26

27

28

29

30

31

What Factors are Involved in Crashes, How Do We Measure Them and What Shall We Do About Them? Frank McKenna

325

Driver Training and Assessment: Implications of the Task-Difﬁculty Homeostasis Model Ray Fuller

337

Do We Really Drive by the Seat of Our Pants? Neale Kinnear, Steve Stradling and Cynthia McVey

349

The Impact of Subjective Factors on Driver Vigilance: A Driving Simulator Study Jérémy Vrignon, Andry Rakotonirainy, Dominique Gruyer and Guillaume Saint Pierre

367

The Use of Local Case Review Panels to Determine Contributory Factors Crash Data Peter Hillard, David Logan and Brian Fildes

379

The Effectiveness of New Seat Belt Legislation in Northern Ireland A.R. Woodside, J.R. Seymour and C. Gallagher

389

viii

Driver Behaviour and Training – Volume III

Part 6 Rider Behaviour 32

33

34

Index

An Evaluation of the Portuguese Moped Rider Training Programme Patrícia António and M. Matos

399

Flow, Task Capability and Powered Two-Wheeler (PTW) Rider Training Paul Broughton

415

Understanding Inappropriate High Speed by Motorcyclists: A Qualitative Analysis Barbara Hannigan, Ray Fuller, H. Bates, Martin Gormley, Steve Stradling, Paul Broughton, Neale Kinnear and C. O’Dolan

425 443

List of Figures Figure 2.1 Figure 2.2

19

Figure 2.3 Figure 2.4 Figure 2.5 Figure 2.6

The telemetric data tracking system Mean responses and 95% conﬁdence intervals of the participants in the driver training study (N = 36) for the questions: How safe do you feel driving: 1) at night? 2) in an unfamiliar area? 3) in the city? 4) in bad weather? 5) after drinking? 6) when sleepy or tired? 7) towing a trailer? 8) an unfamiliar car? 9) when angry? 10) when being tailgated? 11) at 100 km/h? 12) at 110 km/h? 13) at 120 km/h? The crashed car of participant #1 The map function of the on-line monitoring system Mean weekly maximum speeds for participants #1 and #2 Mean weekly maximum speeds for participants #3–#7

Figure 3.1

Miller and Stacey’s driving instruction learning curve

34

Figure 5.1

Distribution of the number of trafﬁc offences recorded after questionnaire administration Flow chart of predictors of trafﬁc offences for young drivers

Figure 5.2

Figure 11.1 Figure 11.2

Figure 14.1

Schematic of coping style and major associations with driving behaviour Interactions between: (i) age group and emotion-focused coping and (ii) age group and avoidance coping, for thrillseeking. Solid regression line indicates the younger group

21 22 23 26 26

55 59

134

135

The mediating effect of ‘need for recovery’ between physical activity (total MET-mins/week) and accidents and absenteeism (after three months)

171

Figure 16.1

Measurement method example

199

Figure 19.1

Conditions inﬂuencing driver behaviour

229

Figure 21.1 Figure 21.2

Processes involved in responding to interruptions An eclectic model for the assessment of driving competence

254

Simulator conﬁgurations and deployment milieu

266

Figure 22.1

258

x

Figure 22.2 Figure 22.3 Figure 22.4 Figure 22.5 Figure 22.6 Figure 22.7 Figure 22.8 Figure 23.1. Figure 23.2 Figure 23.3 Figure 23.4 Figure 23.5 Figure 23.6 Figure 23.7 Figure 23.8

Figure 24.1 Figure 24.2 Figure 24.3 Figure 24.4

Figure 25.1 Figure 25.2 Figure 25.3 Figure 25.4 Figure 25.5 Figure 25.6 Figure 25.7 Figure 25.8

Driver Behaviour and Training – Volume III

Subject population by age and gender Time distributions for subject licensure date Accidents relative to licensure Accident rates for each simulator conﬁguration compared with previously published North American accident rates Number of subjects as a function of time beyond licensure Cumulative accident rate plots of simulator training groups as compared with rates from the literature Cumulative accident rate regression analysis trends The pedestrian shoots out from the right 2.4 seconds before the driver crosses his trajectory Average speeds and lateral positions as a function of time Average lateral positions as a function of time and driving experience Average speeds and lateral positions as a function of time Average speeds and lateral positions as a function of time Average TIVs as a function of time Average speeds and TIVs as a function of time Average lateral position as a function of time and driving experience Graph showing participants’ age distribution and gender (N = 712) Tree diagram highlighting contexts in which drivers have followed inaccurate route guidance instructions Reasons why participants have not updated the map on their navigation system (N = 498) Reasons why participants have updated the map on their navigation system (N = 374) Self-assessment and general feeling of safety on a ﬁve point scale (1 = ‘very safe’, 5 = ‘not safe at all’) Frequency of responses to the question: ‘What is a good driver in your opinion?’ Responses to the question: ‘What types of drivers endanger other road users? Responses concerning speed limits measured on a ﬁve-point-scale Awareness of ISA on a ﬁve-point scale Most frequent responses regarding advantages of ISA Most frequent responses concerning assumed disadvantages of ISA Willingness to use ISA (1 = ‘certainly not’, 5 = ‘yes, certainly’)

269 270 270 271 272 272 274

279 282 282 283 283 284 284 284

295 296 297 300

316 316 317 318 318 319 319 320

List of Figures

Figure 27.1 Figure 27.2

Figure 28.1 Figure 28.2 Figure 28.3 Figure 28.4 Figure 28.5 Figure 28.6 Figure 28.7

Figure 28.8

xi

Representation of the process of Task-Difﬁculty Homeostasis 338 Possible effects of increasing task demand (by increasing speed) to reverse a progressive decline in arousal (capability) 343 Authors’ illustration of the Task Capability Interface (TCI) Model (Fuller & Santos 2002) with inﬂuences Illustration of Task-Difﬁculty Homeostasis Illustrated summary of results from Fuller Means plot of task difﬁculty, feelings of risk and probability of collision across speed for the four road types Means plot of task difﬁculty ratings across speed for the four road types Means plot of feelings of risk ratings across speed for the four road types Magniﬁed means plot of probability of loss of control ratings across speed for the four road types (ratings re-coded into 1–7 rating scheme) Maximum speed comparison by experience level on each road type

350 351 352 357 359 360

361 362

Figure 29.1 Figure 29.2 Figure 29.3

Road condition reproduced in the driving simulator Effects of subjective factors on performance metrics Interaction of age group and gender on sigma

Figure 30.1

Case study crash site before treatment identiﬁed by the local case review panel Case study crash site after treatment identiﬁed by the local case review panel

384

Figure 31.1 Figure 31.2 Figure 31.3 Figure 31.4

Do you ensure your child is restrained? Have you placed your child without restraining them? Have you purchased a booster seat? Do you feel the legislation will impact child safety?

392 393 394 395

Figure 33.1 Figure 33.2

Linear Model of Task Difﬁculty and Flow Outcomes of the dynamic interface between task demand and capability

417

Figure 30.2

Figure 34.1 Figure 34.2

Figure 34.3

Representation of the basic process of Task-Difﬁculty Homeostasis Representation of the process of Task-Difﬁculty Homeostasis, distinguishing between proximal (clear boxes) and distal (grey boxes) determinants and inﬂuences on compliance Schematic illustration of the research process

370 373 374

384

418

426

427 431

This page intentionally left blank

List of Tables Table 2.1

Table 2.2

Table 4.1

Table 4.2

Table 5.1 Table 5.2

Table 5.3

Table 5.4

Table 7.1 Table 7.2 Table 7.3

Table 9.1 Table 9.2 Table 9.3 Table 9.4

The mean (M) weekly distance driven (Dist) in kilometres (km), number of trips (Trips) and mean speed per trip (Mean Speed) in kilometres (km/h) for seven of the eight participants. Standard Deviations (SD), minimum (Min) and maximum (Max) values are also given Weekly means of maximum speed in km/h (Max Speed), number of speeding violations per 100 km (Speeding Viol) and number of large g-forces per 100 km (G-force) for seven of the eight participants Perceived task difﬁculty of the four methods of independent driving (percentage responding that performance is ‘easy’ or ‘very easy’) Perceived task difﬁculty of the three categories of ‘productive’ special manoeuvres (percentage responding that performance is ‘easy’ or ‘very easy’) Background variables for young drivers recording and not recording a subsequent trafﬁc offence Results of a linear regression predicting kilometres driven per year, using personality and attitudinal measures as predictors (N = 179) Mean scores on selected personality and attitudinal measures for drivers recording subsequent trafﬁc offences and no subsequent trafﬁc offences (N = 208) Results of logistic regression analysis for predicting at least one subsequent trafﬁc offence, using personality and attitude measures as predictors (N = 179) Human factors predictors of problem driving events (a) Total driving events; (b) Collisions; (c) Violations Self-report predictors of problem driving events (a) Total driving incidents; (b) Collisions; (c) Violations Linear regression models of problem driving events (a)Total driving incidents; (b) Collisions; (c) Violations Alpha coefﬁcients, means and standard deviations by gender Correlations between the main variables Predicting near misses Predicting violations

24

25

45

46

55

56

57

58

80 82 83 111 112 113 113

xiv

Table 10.1 Table 10.2 Table 10.3

Table 10.4

Table 12.1 Table 12.2 Table 12.3

Driver Behaviour and Training – Volume III

Mean standardised scores on the scales deﬁning the young driver clusters Mean of reported verbal aggression, physical aggression and car using aggression Percentage of high speed driving, accidents, speeding tickets, enjoyment of driving and the mean of reported verbal and physical aggression and car-use aggression Attitudes, subjective norms, perceived control and intention to act aggressively in the intentional condition

122 123

124 125

Factor structure of the organisational safety culture scale Correlations between the study variables Regression of organisational safety culture on DBQ scales and accident involvement

150

The relationship between three levels of physical activity and ‘need for recovery’, health status, accidents and absenteeism

170

Table 15.1 Table 15.2

Alpha reliability coefﬁcients of the measurement scales Logistic regression

180 182

Table 16.1

The percentage of men and people with Swedish names and mean age and number of hours worked in 2001, for drivers in the present study sample and for the total number of active drivers at Gamla Uppsalabuss as at 30 December 2001 The means and standard deviations of the celeration variables in m/s2 during the ﬁrst (A), and second (B) runs during the training sessions, and the mean for regular driving along a route for (some of) the same drivers The N and t-values for the differences (dependent t-tests) # between different measurements of celeration behaviour. Numbers 5 and 6 were gathered after training (A and B) The Pearson correlations between driver celeration behaviour for the ﬁrst and second run during training on one hand, and for regular driving on a number of occasions; 1 before training, 5 and 6 after

Table 14.1

Table 16.2

Table 16.3

Table 16.4

Table 17.1 Table 17.2 Table 17.3 Table 17.4 Table 18.1

Alpha reliability coefﬁcients of the DBQ scale Mean scores for the DBQ factors Factor structure of the modiﬁed DBQ Logistic regression Mean response values for the reasons for not using a seat belt

147 148

193

193

194

194 205 206 208 209 218

List of Tables

Table 18.2

xv

Results of factor analysis for the items related to reasons for not using a seat belt when driving a taxi and a private car Results of factor analysis for the items of DSI Factors related to reported seat belt use frequency when driving a taxi Factors related to reported seat belt use frequency when driving a private car

222

Table 20.1 Table 20.2 Table 20.3 Table 20.4 Table 20.5 Table 20.6

Participant data Factor structure of Questionnaire A post-PCA Factor structure of Questionnaire B post-PCA Factor structure of Questionnaire C post-PCA Participant data Gender differences in FDRI factors

238 240 241 241 242 242

Table 22.1 Table 22.2

Subject population by simulator conﬁguration and gender Accident rate regression analysis

269 273

Table 23.1

Average response times as a function of driving experience and scenario; standard deviations in parentheses Singular behaviours

281 286

Speciﬁc contexts and participants’ examples of occasions where they have received and followed inaccurate route guidance instructions

298

Table 18.3 Table 18.4 Table 18.5

Table 23.2 Table 24.1

Table 26.1 Table 26.2

Table 28.1 Table 28.2 Table 28.3

Table 28.4

Table 29.1 Table 29.2

The intercorrelation of speed, close following, alcohol and violations The correlations between the two potential underlying factors (thrill and emotional outlet) and the risk factors Speed range across the different road types (clips set at 5 mph increments) Breakdown of sample by experience and gender Correlation coefﬁcients of task difﬁculty and feelings of risk ratings by speed and road type (all coefﬁcients signiﬁcant at the p < 0.001 level) Mean speeds in mph for ratings threshold and maximum speed by road type (correlation coefﬁcient between the two variables also shown: p = ns for all road types) Expected impact of low vigilance on performance metrics Impact of extraversion on performance metrics

219 220 221

331 331

354 355

358

358 371 372

xvi

Table 30.1

Driver Behaviour and Training – Volume III

Frequency of occurrence of contributory factors across the 79 crashes reviewed

385

Table 31.1

Attitudes of drivers towards seat belt legislation

394

Table 32.1 Table 32.2 Table 32.3

t-test results for public riding behaviour and trafﬁc offences 403 t-test results for rider equipment and safety accessories 404 t-test results on physical and psychological features while riding (internal risk factors) 405 Descriptive statistics for crash experience (moped/motorcycle and car) between 2000 and 2003 405 Odds ratios (95 per cent CIs) for moped/motorcycle crash experience as dependent variable 407

Table 32.4 Table 32.5

Table 33.1

The four states of ﬂow

417

Table 34.1

Broad themes discussed during the focus group

430

Preface Working in road safety is not an easy path to tread. There is a climate of poor funding, turgid political will and an almost tacit societal indifference towards the annual carnage on the world’s roads. Yet there are two important (and tenacious) professional groups without which this state of affairs would only worsen – researchers and practitioners. The International Conference in Driver Behaviour and Training unites these two groups to debate some of the latest research on how to improve road safety. In recent times, there has been an impressive increase in research in the ﬁeld of driver behaviour and training, particularly in our understanding of the effects of graduated driver licensing and the potential beneﬁts of incorporating the Goals for Driver Education model into the driver education process. In the UK, a new strategy for the training and testing of drivers is about to be announced. There are also exciting developments in the use of new technologies and innovative solutions, with more and more companies taking a keen interest to reduce the risk of work-related crashes such as with the using of simulator-based driver training for example. I introduced the conference in 2003 because there seemed to be little opportunity for researchers and practitioners to meet and debate these kinds of topics, especially in an international forum. Perhaps this is because their motivations and approaches to improving road safety differ. Practitioners are used to dealing with real world problems and ﬁnding solutions. Researchers, on the other hand, are motivated by the need to contribute to knowledge in the ﬁeld and advance their career through research funding and scientiﬁc publications. To work with practitioners is not a recognised academic activity. The major fallout here is that there is a growing scientiﬁc base of research on driver behaviour, education and training – yet custom and practice methods ﬂourish. The research ﬁndings don’t seem to be getting through at ground level. Policy makers are justiﬁably reluctant to support interventions without research evidence, leading to a slow uptake of potential solutions that could save lives. There are many methodological problems that researchers have to tackle in conducting research in the road safety ﬁeld that a practitioner may not be aware of – such as the incredibly difﬁcult process of interpreting crash data or understanding individual differences in response to interventions. In addition, collecting and analysing data is an expensive and time consuming business and research is somewhat slow and drawn out – a source of frustration for policy makers. Worse still, research just generates more questions. The complexity of the issues means that researchers may sometimes fail to communicate theories and models in a way that practitioners and policy makers can interpret and put into practice. As if I didn’t need reminding of the importance of this ﬁeld of enquiry, I lost a dear cousin, Colin Bailey, in a road trafﬁc accident only last year. His death continues to bring untold distress to my family. It also brought a renewed sense of urgency to me. If we are to succeed in designing effective road safety interventions, both

xviii

Driver Behaviour and Training – Volume III

professional groups need to collaborate and coordinate far better if we are going to make real progress across the world. This conference is just one opportunity to work together to integrate our understanding of how to improve a driver’s decision-making in the dynamic and complex process of being a road user; more importantly, which interventions deliver the most beneﬁt. I hope that the Third International Conference in Driver Behaviour and Training helps to align our motivations, maintain the momentum, extend the research foundation and consider the evidence on the best way to train and educate drivers. My gratitude goes to Cranﬁeld University for ﬁnancing the event, a2om for their sponsorship, authors and speakers for their excellent contributions. The contents of this book are a testament to a mere selection of some of the excellent research taking place across the world – but more needs to be done to get researchers and practitioners together to implement interventions that have been demonstrated to work under the scrutiny of an unbiased peer-reviewed process.

PART 1 The Novice Driver Problem

This page intentionally left blank

Chapter 1

How Do ‘Signiﬁcant Others’ Inﬂuence Young People’s Beliefs About Driving? Amanda Green and Lisa Dorn Cranﬁeld University, UK

Introduction Road safety professionals are keen to understand what causes young people to develop risky driving styles. It has long been known that we are drawn to behave in ways that are consistent with our beliefs. Given the well documented role of parents in the development of young people’s ways of thinking, one avenue of research has been to consider how parents inﬂuence their children’s future driving. Early research found a positive correlation between fathers’ and sons’ convictions (Carlson and Klein, 1970), supported by more recent studies showing associations between parents’ and offsprings’ speeding behaviour (Fleiter, Watson and Lennon, 2006; Bianchi and Summala, 2004), aggression and drink driving (Gulliver and Begg, 2004; Mulvihill, Senserrick and Haworth, 2005; Ferguson et al., 2001) and crash risk (Ferguson, Williams Chaplaine, Reinfurt and DeLeonardis, 2001). Risk taking amongst young people is also dependent on family structure, normative parental inﬂuences and social inﬂuences (Shope, Waller, Trivellore, Raghunathan and Patil, 2001; Shope, Ragunathan and Patil, 2003). The presence of other people in the car with the driver has also received some attention with both age and sex of passengers being differentially related to crash risk. For age, Regan and Mitsopoulos (2001) found a higher crash risk when teenage passengers were present, whereas when driving with an older adult or a child, young people’s crash risk was reduced. For sex, an OECD (2006) report showed that young drivers’ crash risk is signiﬁcantly increased by the presence of similarly aged passengers, particularly if both the driver and passengers are male. These two research enquiries suggest that young people’s driving behaviour may be generated through social learning processes, speciﬁcally in their experiences of the person-environment interactions of driving by ‘signiﬁcant others’ (deﬁned as parents, relatives and friends: see Bandura and Walters, 1963; Bandura, 1997; Mishel, 1999; Matthews, Deary and Whiteman, 2003), but the research tends to consider correlations between the driving behaviour of young drivers and their friends and relatives without reference to what may be the cause of this relationship. Little is known about the development of driving-speciﬁc characteristics and situational learning (Lave and Wenger, 1991) or how young drivers are inﬂuenced by signiﬁcant others.

4

Driver Behaviour and Training – Volume III

All learning is contextual and embedded within a social and physical environment (Hampson, 1988 cited by Matthews et al., 2003). It could be that driver beliefs, feelings and behaviour are a learned response according to the cues provided by others. Understanding the social processes by which these cues are internalised during adolescence and expressed when a young person starts driving will enable road safety professionals to focus educational interventions more effectively towards the development of safer driving styles. Understanding the mechanisms by which young drivers are inﬂuenced by signiﬁcant others’ behaviour and the social factors at play is an important step to identifying what underlies the development of their beliefs about driving. Social cognition is the study of how people process social information, especially its encoding, storage, retrieval and application to social situations. Social norms are developed through interactions with peers and parents and affect future decisionmaking and these processes have been studied in the context of driving (Victoir, Eertmans, Van den Bergh and Van den Brouke, 2005). Forward (2006) conducted a qualitative study to investigate drivers’ intentions to commit violations and found that ‘violators’ tended to believe that others would approve of their driving actions. They also tended to normalise their driving behaviour by using examples of others’ driving practices and arguing that their risky behaviour was therefore fairly common. The present study is an extension of Forward’s (2006) work, but adopts a social learning approach by considering the importance of observational learning and modelling of the attitudes, behaviours and emotional reactions in the driving of signiﬁcant others (Bandura, 1977). In other words, we anticipate that driving styles are learned through our observations of signiﬁcant others’ behaviours. We expect that young people may have learned both desirable or detrimental driving styles and attitudes depending on the behaviour of their role models. Further, we expect issues of self-image and identity will be of paramount importance in the way they respond and adapt to societal norms and expectations set by signiﬁcant others (Christmas, 2007). Methodologically, previous research has focused on the extent of the associations between signiﬁcant others and young people on violations, crash rates and other driving behaviours. This approach says little about the nature of young people’s driving-related experiences, nor how the driving of friends and relatives is viewed. The present study uses a qualitative analysis to explore these experiences in depth allowing young people to articulate their driving-related experiences and how they think this has affected their beliefs and behavioural intentions as a driver. The study was commissioned by a2om Ltd, a new driving academy that aims to provide a comprehensive education for learner drivers. Method An in-depth exploration of young learner drivers’ beliefs about driving using a semistructured interview method was used. This allowed a free ﬂowing discussion on the topic areas of interest using focus group methodology. This method was considered the most useful, as a number of different perspectives could be collected during the

How Do ‘Signiﬁcant Others’ Inﬂuence Young People’s Beliefs About Driving

5

interview, allowing for the viewpoints of a greater number of participants in the time allotted. A fundamental framework was used to inform the topic areas for the focus groups after a pilot study revealed its utility. The framework was constructed to elicit accounts about what young people think about the driving of signiﬁcant others, with a view to understanding how these experiences have led to the development of their beliefs about how they will drive in the future. Participants A total of 65 young people took part in 16 focus groups. The focus group sizes ranged from between three and six participants. The participants were selected from several different educational establishments across the UK. Eight schools and colleges took part from both the independent and public school sector. The schools included one mixed independent school, two mixed comprehensive schools, one all girls independent school, two all boys independent schools and two mixed state sixth form colleges. Two focus groups were conducted at each location. All participants volunteered to take part in the research. Participants were advised of the full nature of the research and advised of their right to withdraw from the study at any time. Conﬁdentiality and anonymity were also assured. Participants were aged between 17 and 19 years and included 43 males and 22 females. Twenty-nine participants had no driving experience, three had had no lessons but some driving experience, 23 were currently having driving lessons, with six of those having extra supervised practice as well, two were learning with a parent and two had passed the theory test. Procedure After introductions and ice breakers, initial questions relating to whether the participants were currently learning to drive provided an opportunity for a rapport to be established between the interviewer and interviewees. This also helped to familiarise participants with the recording equipment to help put them at ease. The focus group duration ranged between one and one and-a-half hours with a mean time of 1:04 hours. The interviews were recorded and fully transcribed. Analyses A template analysis approach to coding was utilised to reveal and organise themes relating to teenagers’ perceptions about driving that were in line with the research objectives. An initial template was constructed by systematically analysing the transcripts to identify relevant sections of text (King cited in Cassell and Symon, 2004). The pre-deﬁned codes helped to guide the analysis but these were kept to a minimum so as not to blinker identiﬁcation of further emergent themes and subthemes. Preliminary codes, along with corresponding quotations, were entered into Excel in order to manage the plethora of data generated. The data were reanalysed and a tentative template was generated. Further analysis and interpretation in relation to the project aims resulted in a simpliﬁed ﬁnal template. This consisted of two

Driver Behaviour and Training – Volume III

6

highest-order codes (themes), which were then sub-divided into lower-order codes (sub-themes). Results and discussion This study reports two main themes that help to explain how young people interpret their driving-related experiences as pre-drivers and part of fuller analyses that will be reported elsewhere. The two broad themes to emerge were ﬁrstly ‘modelling others’ driving behaviour’ and secondly ‘distancing from others’ driving behaviour’. Taking each of these themes in turn, the sub-themes will be discussed in relation to the data to derive a possible explanation for the mechanisms through which young people develop their beliefs about driving. Theme 1: modelling others’ driving behaviour The ﬁrst theme to emerge was concerned with how young people make sense of their experiences of the driving of signiﬁcant others, in particular, whether there was evidence of modelling, deﬁned as viewing the driving of signiﬁcant others as desirable, and whether this could be considered safe or not. Driving aspiration Respondents were asked, ‘Once you pass your test, what kind of driver would you like to be?’. Approximately 65 per cent of the responses referred to the desire to be patient, cautious or safe drivers. Many stated that they would want to drive in a controlled fashion and be able to handle the car in any eventuality: A controlled driver who knows the boundaries so he knows what’s coming or he knows if he is going too fast or can control his speed or is just generally aware of everything (Gp12, Independent co-education school).

But as the interviews progressed, it was clear that our participants held some contradictory beliefs with respect to speeding in particular. Modelling fathers About 20 per cent of the sample intended to imitate their fathers’ driving styles. Yet 17 teenagers described fathers as regularly exceeding the speed limit. Some of the descriptions sounded as though their father’s driving was considered somewhat heroic – hence a plausible explanation as to why their driving was considered worthy of emulating. Always in control … even though he was driving fast I would still feel safe. I feel safe when I am in a car with him (Gp3, All boys independent school).

How Do ‘Signiﬁcant Others’ Inﬂuence Young People’s Beliefs About Driving

7

You do pick up their habits. I’ve noticed that I do similar things to my dad more when my dad does something than my mum does; I have picked up some of his habits … mainly with speeding, over the limit, I drive on the limit sort of thing (Gp3, All boys independent school). Dad, he is so conﬁdent behind the wheel; he is really macho as well, he is quite macho about his car so at the same time he is quite conﬁdent. He is deﬁnitely a role model (Gp11, Independent co-education school).

Further exploration revealed that a good driver was perceived as someone who could control a car, but not necessarily within the speed limits. Driving at speed was felt to be relatively safe: Risky but safe like you’re safe but I wouldn’t be afraid to go a little bit faster. If you’re on a motorway and it’s like a clear road and I wouldn’t go absolutely crazy, as long as you’re being safe (Gp4, All Boys Independent School).

Similarly, many participants described their mothers as being ‘good drivers’; but at the same time they recalled experiences where their mothers were clearly lacking awareness or concentration and their father’s driving was held to be superior in comparison: I think my dad, well I feel more conﬁdent with my dad’s driving, although he does tend to go quite fast and he sometimes drives with his knees. My mum you have to be alert because otherwise it’s very embarrassing. She might go ‘oh my god I need to go to …’ and she just stops in the middle of the road (Gp10, All boys independent school). My mum does her make up and stuff and she is always on the phone. I don’t have to concentrate when my dad’s driving but when my mum is driving, the lights will change or something and she won’t notice because she’s sat there looking in the mirror or something (Gp10, All boys independent school).

Fathers’ driving appears to be held in particularly high esteem, with both teenage girls and boys preferring to model their father’s driving style rather than their mother’s. I think my dad is a brilliant driver and my mum’s a crap driver. It’s fun being in the car with him, it’s like being with a 20 year old driving along. He is really, really cocky but he thinks he is really good. He slows down when he sees something you know and he knows what he is doing. My mum just screams every two seconds and crashes into everything so I don’t think she should have passed her test (Gp12, Independent co-education school).

Our pre-drivers believed that conﬁdence is an important factor in driving. Those who were conﬁdent that they can take on their father’s driving style were more likely to dismiss thoughts of safety and expressed a desire to take on more challenges. Speeding sometimes if it’s like a quiet road like back lanes and stuff (Gp6, Co-education sixth form college).

8

Driver Behaviour and Training – Volume III Speeding but only by ﬁve miles an hour or something like that not like something stupid (Gp10, All boys independent school).

The general consensus was that speeding was acceptable providing there was an empty road, limits were not exceeded by too much and that they felt able to control the car. Perhaps the associations between parents’ and children’s speeding behaviour reported by Fleiter et al. (2006) and Meadows, Stradling and Lawson (1998) may be largely attributed to the modelling of fathers’ driving behaviour rather than mothers’. There was evidence of a normative framework in operation according to some of the narratives. Young people felt that their mother and father had a shared understanding of how to behave towards other drivers and they worked together to make sure other drivers understood their feelings, conﬁrming previous research (Pelsmacker and Janssens, 2007; Tabman-Ben-Ari, 2006). My dad like at the lights he gets aggressive, not road rage but he just gets aggressive with other drivers if they are going too slow … he calls them a ******. And when my mum sits in the passenger seat she sometimes, if my dad if he is going too fast, if my dad can’t stick his ﬁnger up, my mum does it for him (Gp5, Co-education sixth form college).

Friends or siblings as role models Some of the participants were inﬂuenced by the driving styles of similarly aged friends and siblings as well as parents. Some positively … One of my true best friends, who is a couple of years older, she takes me out in her car sometimes and she is really, really conﬁdent and relaxed behind the wheel and just takes everything in her stride and I would quite like to be like that (Gp1, All girls independent school).

And some negatively … My brother passed his test and then he went a cruise one night and there were three other cars as well, like his friends. We were in the second car and we went round the corner and everyone was going really fast. We went round and the ﬁrst car went into a bridge and then my brother went into him and then they all went into each other. The ﬁrst car the guy had to be cut out of it … it’s actually alright it doesn’t bother me at all (Gp12, Independent co-education school).

Night cruises (many cars congregating in one area and driving competitively for fun, usually late at night) is a worrying trend amongst young people. Young drivers often believe that positive outcomes outweigh any negative consequences to their actions (Clarke et al., 2006; Fromme, Katz and Rivet, 1997) acting for the fun of the moment and feeling invulnerable. Carmichael et al., (2005) found that teenagers who reported that their peers were involved in delinquent behaviours had lower perceptions of sanction and were more likely to violate; judging their friends’ behaviour as acceptable helps them become part of the ‘in-group’.

How Do ‘Signiﬁcant Others’ Inﬂuence Young People’s Beliefs About Driving

9

It’s great fun going in your friend’s car and yes I deﬁnitely want to do it (Gp1, All girls independent school). I would like to be safe but silly. Not racing but, it’s hard to explain I guess but just have a bit of fun, being serious but at the same time fun (Gp5, Co-education sixth form college). The friends that I have drive and they go to cruises at night where there is about 25 cars. We was driving up and there was like a slip road and one of his friends just suddenly pulled out right onto us and we had to try and swerve but we couldn’t go right because there was another car there so we had to go into a lamp post and the car was really smoking and everything (Gp5, Co-education sixth form college).

There is also some evidence of individual differences in sensation seeking. I kind of enjoy it more because it’s more fun because I’ve got a mad edge (Gp10, All boys independent school).

Drink driving experiences Analysis revealed over 40 references to drink driving experiences. Nine teenagers stated that they had one or more parents that regularly drink drive and this has inﬂuenced their future behaviour. Some experiences had not resulted in any punishment or crashes and this served to reinforce poor learning: If I had never driven before I would never have drank anything before getting into a car; I would never do it but my mum, she always drinks and drives. She doesn’t drink a lot but she’ll go over the legal amount all the time and she still drives. She hasn’t had any major crashes that were her fault so I wouldn’t mind having a bit, but I wouldn’t get drunk (Gp4, All boys independent school).

Whereas others had experienced negative outcomes: A lot of people I know, like a lot of my family have all been in like drinking accidents. My brother I think he crashed a car when he was drinking and my dad got pulled over when he was a lot younger but he had loads of people with them as well so (Gp6, Co-education sixth form college).

Narratives show that many parents suffered no serious consequences of drink driving, leading to the obvious conclusion that young people are learning that it’s safe to drink and drive. Even though most teenagers report that they would not drink and drive themselves, previous research suggests that parents’ driving records are often predictive of their children’s driving records, as cited earlier. In conclusion, we ﬁnd evidence for strong modelling effects of unsafe driving practises emanating both from parents and peers.

10

Driver Behaviour and Training – Volume III

Theme 2: distancing from others’ driving behaviour The second major theme to emerge referred to beliefs about the kinds of drivers young people want to avoid turning out like, such as those who have hesitant or reckless driving styles. In other words, the kinds of driver they want to distance themselves from. Observing others: how not to drive Several accounts suggest that participants had internalised the criticisms levied at their signiﬁcant others or had observed their driving and decided they wanted to be a more skilled driver than the example set: My dad’s quite stop start and everyone gets quite car sick with him … I want to be able to drive smoothly and right now when I brake I know I brake quite jerky and it’s not a smooth stop. So I want to be a smoother driver (Gp3, All boys independent school).

Some participants described their grandparents’ driving as ‘scary’ and felt anxious when in a car with them. Their grandparent’s driving was judged to be slow and hazardous and respondents wanted to distance themselves from becoming this kind of driver. Fourteen participants said they ﬁnd old or slow drivers frustrating and expect to feel resentful if, as a driver in the future, they are delayed by them. Sometimes you get old people on the road and they are going really slowly and people get really annoyed with them and try and over take them when they shouldn’t really over take them and then you get quite a lot of accidents (Gp10, All boys independent school).

They also distanced themselves from signiﬁcant others who had a nervous or hesitant driving style. I hate being in a car with my sister because she’s quite nervous and she’s really putting off the fact that she’s nervous about her driving which makes me nervous, is somewhat annoying. So I’d like to be quite conﬁdent and I would like to be seen, I mean my sister’s quite nervous of driving in front of other car users so I would like people to feel quite conﬁdent with me (Gp4, All boys independent school).

The distancing process is entirely consistent with the previous theme in which teenagers show a preference for calm, conﬁdent driving styles. Teenagers do not want to be associated with driving that will invoke a negative reaction from other drivers, presumably to protect their self-identify and self-esteem. Their preferred driving style will deliver a major psychological beneﬁt – allowing them to feel superior in their driving skill compared with these ‘lesser’ drivers. However, teenagers seem to draw a line between the slow, bumbling or nervous driver and those with an aggressive driving style – being equally committed to not adopting either driving style. I have a brother who drives and I ﬁnd that he can get aggressive pretty quickly. Either being impatient when trafﬁc is moving slowly or when he wants to pull out of somewhere

How Do ‘Signiﬁcant Others’ Inﬂuence Young People’s Beliefs About Driving

11

and he is there ages, waiting to go … he will sit there and just look and it’s his tone of voice like ‘hurry up, come on’ you know. Copy his driving style? No. I’d like to be a calm driver and have patience for other people and let people turn out if they want, just a patient driver really (Gp6, Co-education sixth form college).

It seems that some young people are keen to avoid the pain of being thought of as a slow, nervous or aggressive driver and instead seek the pleasure of being thought of as a skilled, conﬁdent driver; but this is largely dependent on the way they have interpreted their experiences. Our data are consistent with the ﬁndings of a recent Department for Transport study (Christmas, 2007) showing that young drivers considered good driving to be largely a function of ‘natural talent’. In this study, some young people were found to be highly conﬁdent in their driving ability, choosing to ignore caution from others with image and identity and the opinions of others being important to them. We can also conﬁrm that these beliefs and personal needs are present before they even pass their driving test according to the present data. Peer pressure inﬂuence on driving behaviour Mainstream research shows the strong effects of peer pressure on young people (Brown, 2004). With regards to driving though, many respondents considered that passengers could inﬂuence their driving, but felt they could handle these difﬁcult situations and would not concede to peer pressure: There are people who wouldn’t encourage me to do anything and then there are people I am sure would. I doubt, I don’t know it depends on the situation at the time, but I doubt I would actually put my foot down just because I was told to. I’m not likely to be taken to peer pressure so … not really a problem (Gp14, co-education college). I think it would be tempting but I wouldn’t do it because I know the risks. I would just turn around and say no. They are the passenger so it’s your choice, it’s not like they are going to jump in the front and drive (Gp6, Co-education sixth form college).

We propose that it is not socially desirable to admit to being inﬂuenced by peers as this would conﬂict with their image of being a skilled, conﬁdent driver in complete control as outlined in the modelling theme. They prefer to be seen as independent and above the inﬂuence of others. Depending on the situation like, if it’s a busy road or if it’s dark or icy then there’s more chance and I would say ‘look guys would you mind calming down because it’s difﬁcult enough to drive as it is’, but if it’s a clear enough road, I’d probably be joining in. Not like leaning over hitting people but saying things and laughing as well (Gp3, All boys independent school). You don’t mean to but I reckon that you do. Subconsciously there is something that makes someone drive faster than they would do, I reckon, than on their own (Gp3, All boys independent school).

12

Driver Behaviour and Training – Volume III

Perhaps it is the mere presence of peers that inﬂuences driving behaviour rather than any active encouragement from friends to take risks (Regan and Mitsopoulos 2001; Waylen and McKenna, 2002). Some accounts suggest that they would be too concerned for the welfare of their passengers to take risks: There is a difference in driving and something that you do yourself because if you do have a car full of people and you are putting lives at risk whereas if you are on your own it will just be your own life (Gp6, Co-education sixth form college). Because you are responsible and you have got that responsibility, especially when your friends are there, they won’t want to get into any trouble (Gp1, All girls independent school). I want a two-seater I don’t want to have to drive people and if I have an accident I will kill less people (laughs) not that I will you know but just in case (Gp1, All girls independent school). I think you have a responsibility to take care of everyone in the car and ensure that they are not a distraction … try to ignore what they are saying (Gp11, Independent co-education school).

These ﬁndings support those reported by Glendon (2005) who also found that young drivers are reluctant to put their passengers at risk. Narratives suggest that on the one hand participants consider speeding is safe, yet on the other hand are concerned for the safety of their passengers. Perhaps they reconcile this apparent inconsistency in beliefs by believing that they will be skilled enough to exceed the speed limit and still be safe. Early experiences and effects on future driving styles One salient sub-theme related to memorable early experiences: I saw, when I was nine, I was walking through, not my village but another village fairly close. I was with a friend and there was a crossroads in the village and it was on quite a steep hill. A guy on a moped came charging down a hill and tried to turn. Obviously he hadn’t been in the village much at all before; he had no clue. He just came down and skidded and went straight across, hit the curb and bounced up into a wall on the side of the road and just like lay there unconscious. My friend stayed with the guy and I legged it and called an ambulance. That completely put me off bikes (Gp9, All boys independent school).

Our data support the view that beliefs about driving are probably developed much earlier on in life than expected, perhaps because children have more exposure to driving experiences than ever before (Waylen and McKenna, 2002). Some extreme personal experiences have a lasting effect on perception of risk and leads to beliefs that will distance them from situations that could place them in danger.

How Do ‘Signiﬁcant Others’ Inﬂuence Young People’s Beliefs About Driving

13

My dad was in hospital for quite a long time, and there was loads of people there who were like paralysed for the rest of their life because someone else was drunk when they were driving and it has completely altered their life really. I wouldn’t drink and drive or get in a car when somebody had been drinking (Gp10, All boys independent school).

Conclusions Somewhere along life’s highway, young people have developed quite sophisticated views about driving and the kind of driver they would like to be before they even take to the road. Teenagers are responding and adapting to societal norms and expectations learned through their experiences of the driving of signiﬁcant others. They do this in two ways. Firstly, young people model their driving on friends and relatives that speed or drink and drive and consequently develop the view that this behaviour is safe when not done to excess. They also learn how they should respond to road users who drive slower or more cautiously. There was a strong association between skill and speed, and the general consensus of a good driver emerged as someone who can drive at speed while still maintaining control of the vehicle, with fathers being the main source of their observational learning. Secondly, they develop their driving styles through a process of distancing themselves from those with a hesitant and overly aggressive approach to driving. Their experiences tell them that this will attract criticism from other road users. The data also suggest strong indications of individual differences in the ways that experiences are interpreted by young people. Underpinning interpretation of their driving experiences is the desire to avoid the pain of criticism from others, supporting the view that self-image and identity are critical to young people. Our study shows the need to be regarded as a good driver is primarily motivated by the need to enhance self-esteem about their driving skill. Low parental monitoring has been related to risky driving behaviour, trafﬁc violations and young driver crashes (Hartos, Eitel, Haynie and Simons-Morton, 2000). Parents are in the best position to enforce driving restrictions for learners and to instil safe behaviour (Mulvihill, Senserrick and Haworth, 2005), but judging by the reports of their driving behaviour many parents are setting a poor example. Educating parents may be the ﬁrst step towards ensuring that young people avoid acting out the dangerous behaviours they have witnessed as an inexperienced driver. We consider that encouraging self-reﬂection on these formative experiences will facilitate a less distorted interpretation of their experiences and what constitutes safe driving provided this is under the guidance of a trained professional. Changing attitudes does not require deep psychological therapy; rather we only need to create doubt in the certainties that are held. Driving instructors at Cranﬁeld University are currently being trained in techniques to challenge beliefs and provide a more focused education tailored to the learners particular risk proﬁle (Dorn, 2005). A new breed of driving instructor delivering a comprehensive curriculum to educate drivers, not just in how to drive a vehicle but in their personal risks as well, has already begun with a2om.

14

Driver Behaviour and Training – Volume III

Driving instructors need some way of assessing individual differences to tailor their training. An instrument for self-assessment has been developed as part of the present research programme as a ﬁrst step to facilitate self-reﬂection on personal motives, beliefs, feelings and goals and how these individual differences might impact on future driving styles in line with the Goals for Driver Education matrix (Hatakka et al., 2002). References Bandura, A. (1977). Social Learning Theory. New York: General Learning Press. Cited at http://tip.psychology.org/bandura.html. Bandura, A. (1997). Self-Efﬁcacy: The Exercise of Control. New York: W.H. Freeman and Co. Bandura, A. and Walters, R.H. (1963). Social Learning and Personality Development. New York: Holt Rinehart and Winston. Bianchi, A. and Summala, H. (2004). ‘The “genetics” of driver behaviour: parents’ driving style predicts their children’s driving style.’ Accident Analysis and Prevention, 36, 655–9. Brown, B. (2004). ‘Adolescents’ relationships with peers.’ In R. Lerner and L. Steinberg (eds). Handbook of Adolescent Psychology. New York: Wiley and Sons. Burns, P.C. and Wilde, G.J.S. (1995). ‘Risk taking in male taxi drivers: relationships among personality, observational data and driver records.’ Personality and Individual Differences, 18, 267–78. Carlson, W.L. and Klein, D. (1970). ‘Familial vs. institutional socialization of the young trafﬁc offender.’ J. Saf. Res. 2(1), 13–25 cited in Bianchi, A. and Summala, H. (2004). ‘The “genetics” of driver behaviour: parents’ driving style predicts their children’s driving style.’ Accident Analysis and Prevention, 36, 655–9. Carmichael, S., Langton, L., Pendell, G., Tritzel J.D. and Piquero A.R. (2005). ‘Do the experiential and deterrent effect operate differently across gender?’ Journal of Criminal Justice, 33, 267–76. Cassell, C. and Symon, G. (2004). Essential Guide to Qualitative Methods in Organizational Research. London: Sage Publications. Christmas, S. (2007). ‘The good, the bad and the talented.’ Road Safety Research Report No. 74. Department for Transport. Wetherby: DfT Publications. Clarke, D.D., Ward, P., Bartle, C. and Truman, W. (2006). ‘Young driver accidents in the UK: the inﬂuence of age, experience and time of day.’ Accident Analysis and Prevention, 38, 871–8. Clarke, D.D., Ward, P. and Truman, W. (2005). ‘Voluntary risk taking and skill deﬁcits in young driver accidents in the UK.’ Accident Analysis and Prevention, 37, 523–9. Dahlen, E.R. and White, R.P. (2006). ‘The big ﬁve factors, sensation seeking and driving anger in the prediction of unsafe driving.’ Personality and Individual Differences, 41, 902–15. Dorn, L. (2005). ‘Driver Coaching.’ In L. Dorn (ed.) Driver Behaviour and Training,

How Do ‘Signiﬁcant Others’ Inﬂuence Young People’s Beliefs About Driving

15

vol. 2. Aldershot: Ashgate. Ferguson, S.A., Williams, A.F., Chaplaine, J.F., Reinfurt, D.W. and DeLeonardis, D.M. (2001). ‘Relationship of parent driving records to the driving records of their children.’ Accident Analysis and Prevention, 33, 229–34. Fleiter, J., Watson, B., Lennon, A. and Lewis, I. (2006). ‘Signiﬁcant others, who are they? – examining normative inﬂuences on speeding.’ In Proceedings 2006 Australasian Road Safety Research Policing Education Conference, Gold Coast. Forward, S.E. (2006). ‘The intention to commit driving violations: a qualitative study.’ Transportation Research Part F 9, 412–26. Fromme, K., Katz, E.C. and Rivet, K. (1997). ‘Outcome expectancies and risk-taking behaviour.’ Cognitive Therapy and Research, 21(4), 421–43. Glendon, A.I. (2005). ‘Young drivers’ attitudes towards risks arising from hazardous driving behaviours.’ In L. Dorn. (2005). Driver Behaviour and Training, vol. 2. Aldershot: Ashgate. Gulliver, P. and Begg, D. (2004). ‘Inﬂuences during adolescence on perceptions and behaviour related to alcohol use and unsafe driving as young adults.’ Accident Analysis and Prevention 36, 773–81. Hartos, J.L., Eitel, P., Haynie, D.L. and Simons-Morton, B.G. (2000). ‘Can I take the car?: relations among parenting practices and adolescent problem-driving practices.’ Journal of Adolescent Research, 15(3), 352–67. Hatakka, M., Keskinen, E., Gregersen, N.P., Glad, A., and Hernetkoski, K. (2002). ‘From control of the vehicle to personal self-control: broadening the perspectives to driver education.’ Transportation Research Part F: Trafﬁc Psychology and Behaviour, 5, 201–15. Lave, J. and Wenger, E. (1991). ‘Situated learning: legitimate peripheral participation.’ Cambridge: University of Cambridge Press. Lerner, R and Steinberg, L. (eds) (2004). Handbook of Adolescent Psychology. New York: Wiley and Sons. Matthews, G., Deary, I.J. and Whiteman, M.C. (2003). Personality Traits. Cambridge: Cambridge University Press. Meadows, M.L., Stradling, S.G. and Lawson, S. (1998). ‘The role of social deviance and violations in predicting road trafﬁc accidents in a sample of young offenders.’ British Journal of Psychology, 89, 417–31. Mishel, W. (1999). ‘Personality coherence and dispositions in a cognitive-affective personality (CAPS) approach.’ In D. Cervone and Y. Shoda (eds). The Coherence of Personality: Social Cognitive Bases of Consistency, Variability and Organisation, New York: Guildford, 37–67. Mulvihill, Senserrick and Haworth (2005). ‘Development of a model resource for parents as supervisory drivers.’ Monash University Accident Research Centre. OECD (2006). ‘Young drivers: the road to safety.’ The Transport Research Centre. Paris: OECD Publishers. Pelsmacker, P.D. and Janssens, W. (2007). ‘The effect of norms, attitudes and habits on speeding behaviour: scale development and model building and estimation.’ Accident Analysis and Prevention, 39, 6–15. Regan, M.A. and Mitsopoulos, E. (2001). ‘Understanding passenger inﬂuences on driver behaviour: implications for road safety and recommendations for counter-

16

Driver Behaviour and Training – Volume III

measure development.’ (Report No. 180) Clayton, Australia: Monash University Accident Research Centre. Shope, J.T., Ragunathan, T.E. and Patil, S.M. (2003). ‘Examining trajectories of adolescent risk factors as predictors of subsequent high-risk driving behaviour.’ Journal of Adolescent Health 32, 214–24. Shope, J.T., Waller, P.F., Ragunathan, T.E. and Patil, S.M. (2001). ‘Adolescent antecedents of high-risk driving behaviour into young adulthood: substance use and parental inﬂuences.’ Accident Analysis and Prevention, 33, 649–58. Simons-Morton, B. (2007). ‘Parent involvement in novice teen driving: rationale, evidence of effects and potential for enhancing graduated driver licensing effectiveness.’ Journal of Safety Research (forthcoming). Simons-Morton, B.G. and Hartos, J.L. (2003). ‘How well do parents manage young driver crash risks?’ Journal of Safety Research, 34, 91–7. Tabman-Ben-Ari, O. (2006). ‘Couple similarity for driving style.’ Transportation Research Part F 9, 185–93. Victoir, A., Eertmans, A., Van den Bergh, O. and Van den Brouke, S. (2005). ‘Learning to drive safely: social-cognitive responses are predictive of performance rated by novice drivers and their instructors.’ Transportation Research Part F, 8, 59–74. Waylen, A. and McKenna, F. (2002). Pre-Drivers’ Attitudes Towards Driving. Paper presentation, 67th Road Safety Congress; ROSPA.

Chapter 2

Piloting a Telemetric Data Tracking System to Assess Post-training Real Driving Performance of Young Novice Drivers Robert B. Isler, Nicola J. Starkey, Peter Sheppard1 and Chris Yu2 University of Waikato, New Zealand 1 AA Driver Education Foundation 2 SmarTrak, New Zealand Introduction Evaluating the effects of driver training interventions is a difﬁcult research affair. The ultimate goal of such interventions is to make the driver safer and therefore less likely to be involved in a road crash. A particular driver training intervention can only be considered to be effective if it can show a signiﬁcant reduction in the number of crashes for the driver or a signiﬁcant change in driver behaviour that clearly implies safer driving. Getting accurate and comprehensive crash records is difﬁcult and to measure post training behavioural driving changes based on self-reports (for example, log books) may not be accurate enough to be statistically meaningful. The majority of driver training evaluation studies in the last 30 years concluded that driver education and training contributes little to reducing crash risk or involvement for road users (pre-licence, defensive, advanced or driver improvement). And even more puzzling and paradoxical is the fact that there was no evidence that professional driver training is effective in reducing crash risk. However, failing to ﬁnd a driver training effect does not necessarily mean that it does not exist. In fact, there has been a heated scientiﬁc debate about the usefulness of the hypothesis testing procedures employed by most of these evaluation studies (Shrout, 1997). For example, the fact that statistical procedures are generally geared towards preventing type one errors (claiming an effect when there is in fact no effect) but at the same time are quite likely to lead to type two errors (failing to detect an effect when there is an effect) biases results towards non signiﬁcance. Furthermore, Crick and McKenna (1991) maintained that the lack of evidence for the beneﬁts of road safety education and training may be ascribed to a lack of methodological soundness in previous evaluations and/or to the content of the course. It is indeed interesting to note that many driver training evaluations have been published as technical reports and therefore were not subject to peer review. Often, evaluation studies have failed to use appropriate control groups and have used

18

Driver Behaviour and Training – Volume III

hypothesis testing procedures inappropriately, with very little statistical power to detect any effects. The content of the driver training courses that have been evaluated in the past tended to emphasise the teaching of vehicle control skills or alternatively, were classroom based. Since then, research has shown that increasing driver skills does not necessarily lead to more capable drivers. For example, skid training may lead to drivers over-estimating their own driving ability, without actually improving the way they manoeuvre the car (Gregersen, 1996). Furthermore, studies suggest that crash involvement is more often the result of risk taking behaviour, rather than poor driving ability (Clarke, Ward and Truman, 2005). Thus, driver training programmes which concentrate on vehicle handling skills may actually lead to increased risk taking due to learners’ inﬂated self conﬁdence and self-rated skills. Consequently, a growing consensus amongst driver training and road safety researchers is that greater emphasis should be placed on higher level cognitive functions underlying driving skills (Senserrick, 2007). Some researchers have argued further that there is an urgent need for a holistic and structured plan of education and training that addresses all goals of driver education, as outlined in the ‘Goals for Driver Education’ (GDE) model (see Engstroem, Gregerson, Hernetkostki, Keeskinen and Nyberg, 2003 for a comprehensive review on young drivers, driver education and training). At the same time there is a call for employing more sensitive and objective behavioural outcome measures, so that their accuracy can be increased and at the same time the probability for committing a type two error can be minimised. We recently conducted a large scale driver training study (Isler, Starkey, Charlton and Sheppard, 2007) in New Zealand to compare the effects of training in higher level driving skills (such as eye scanning, hazard detection and risk management) and vehicle control skills (such as manoeuvring, braking and parking) on teenagers’ real driving and risk taking behaviour, conﬁdence levels and self-rated driving skills. Thirty-six teenage drivers (across a range of ethnic and social backgrounds) on a restricted driver licence were recruited via 500 secondary schools. After the driver training camp, we installed telemetric data trackers in the vehicles of eight participants to pilot how well this technology measured post-training real driving behaviour. We tracked the driving behaviour of the participants for 32 weeks in order to evaluate if such data acquisition could help ﬁll a methodological gap in driver training evaluations. From the outset, we knew that the number of data trackers would be too small to make conclusive claims about any potential long-term effects of the driver training in our study. The idea was to test this new and promising evaluation technology and report on our ﬁndings. Method Participants From a total of 36 participants who took part in the driver training study, eight participants (four males and four females) who brought their private vehicles to the training session were selected to participate in this pilot study. They were all 16 years

Piloting a Telemetric Data Tracking System

19

old and were required to hold a current New Zealand restricted driver’s licence. This ensured that they all had some unsupervised driving experience. Their vehicles were ﬁtted with a telemetric data tracking system and their driving behaviour was monitored on-line via the internet over a 32 week period. The telemetric data tracking system The tracking system consisted of a small credit card sized global positioning module (SmarTrak Lite GPRS/GPS) ﬁtted with an accelerometer (see Figure 2.1). The system was powered by the vehicles’ battery (six volt). It took approximately 30 minutes to install the system in a vehicle. In order to obtain accurate data, the device had to be pointing forward and on a ﬂat surface. In most cases it was installed below the driver’s seat.

Figure 2.1 The telemetric data tracking system This system uses a GPS receiver and provides reliable and accurate navigational data. The software for the tracking and reporting interface via the internet was developed by SmarTrak Ltd (www.smartrak.co.nz). It allowed us to monitor, in real time, the driving performance (updated every two seconds) of the eight participants on the computer screen (see Figure 2.4 as an example of a map-based online tracking). The built-in accelerometer also provided g-force data from the vehicles. Daily, weekly and monthly reports of the driving measures for each participant could be produced and downloaded as a Windows Ofﬁce Excel spreadsheet. The following driving measures were used as dependent variables in this study: Distance driven The number of kilometres driven for each trip was recorded. Number of trips A trip started from a ‘key on’ event (starting the engine of the vehicle) to a ‘key off’ event (shutting down the engine). Mean speed per trip Every four kilometres the current speed was recorded and the mean speed for each trip was calculated. Maximum speed

The maximum speed was recorded for each trip.

20

Driver Behaviour and Training – Volume III

Speeding violation The system noted each time a participant exceeded 100 km/h (62 mph), which is the maximum speed limit for New Zealand. Lower speed limit violations (for example, driving 60 km/h on a road with a 50 km/h speed limit) were not monitored. Large g-force Each time the vehicle created a g-force (longitudinal or lateral) that was larger than 0.50, an event was triggered. The threshold setting was the same as the one used by McGehee, Raby, Carney, Lee and Reyes (2007) for their event-triggered video driver intervention trial. Negative longitudinal g-force events indicated hard braking while positive events indicated levels of acceleration that would be difﬁcult to reach without external impacts (for example, rear end collision). The system did not allow differentiation between longitudinal g-forces created by hard braking and those created by hard cornering or swerving. Results Thirty-six participants (15 females, 21 males) attended the driver training study where they were ﬁrst asked to complete a number of psychometric instruments and driver behaviour questionnaires. The data from these pre-assessments are currently being analysed. Participants were asked to rate how safe they felt driving in a variety of situations on a ﬁve point Likert scale (1 = Very Safe to 5 = Very Unsafe; adapted from Bergdahl, 2005). The responses from the eight participants in this pilot study did not differ signiﬁcantly from the responses of the other participants in the driver training camp, and therefore the results from all participants (n = 36) are presented in Figure 2.2. Most participants felt safe in the majority of driving situations, except after drinking (rated between unsafe and very unsafe), when they are sleepy or tired and when they are angry or being tailgated (rated as between ‘neither safe nor unsafe’ and ‘unsafe’). Interestingly, they felt quite safe speeding at 120 km/h even though they indicated in a different questionnaire that speeding is one of the most frequent causes of young driver crashes. We received valid telemetric driving behaviour data from six of the eight participants for the entire 32 weeks period. The data for one of the six participants (#8) was not analysed, as the tracking system did not provide the data for the variable ‘distance driven’. Two of the participants crashed during the study and the GPS system allowed us to examine their driving behaviour just before (and, in one case, during and after) the crash. Participant #1 crashed in week 19. The tracking system did not transmit any data during the crash as the power supply was disrupted, and we were not able to retrieve any data from the tracker in the crashed car (see Figure 2.3). The last data we received from the vehicle was two minutes before the crash occurred, indicating that the vehicle was travelling at 75 km/h sometime within that time period. The participant’s account of the crash was as follows:

Piloting a Telemetric Data Tracking System

Figure 2.2

21

Mean responses and 95% conﬁdence intervals of the participants in the driver training study (N = 36) for the questions: How safe do you feel driving: 1) at night? 2) in an unfamiliar area? 3) in the city? 4) in bad weather? 5) after drinking? 6) when sleepy or tired? 7) towing a trailer? 8) an unfamiliar car? 9) when angry? 10) when being tailgated? 11) at 100 km/h? 12) at 110 km/h? 13) at 120 km/h?

Hit a stationary vehicle parked half on/half off road. Was travelling at about 100 km/h when hit the vehicle. I just did not see the car – obviously lack of concentration. I was not text messaging or using phone prior to crash. I did have a passenger, though I can’t remember all that happened so I don’t know what I was doing to not see the car.

The participant suffered only some minor injuries but was shaken by the experience and decided not to drive for a while. Participant #2 crashed in week 30. She started her journey at 6.24 am, lost control on a bend at 7.22 am and swerved 180 degrees when she was hit by an oncoming car. For this incident we have a complete set of telemetric data available as the car was still functioning after the crash and power was continuously supplied to the data tracker. Figure 2.4 shows the map function of the on-line monitoring system listing the transmitted driving events on the right side of the map. The map revealed that the crash happened at 7.22 am and was preceded by a large negative g-force (–0.56),

22

Driver Behaviour and Training – Volume III

Figure 2.3 The crashed car of participant #1 probably caused by hard braking. At that time, the vehicle was travelling at 83 km/h when it swerved 180 degrees and hit an oncoming car creating a very large positive g-force (2.85). Within the same minute (7.22 am) the car was decelerated to 1 km/h. We later received the information that the crash occurred during very wet driving conditions. The vehicle of participant #3 was stolen in week ﬁve, in an early morning at 2.43 am. It seems that the vehicle was used for a ‘joy ride’ that lasted 11 minutes. Telemetric data showed that the car created seven negative large g-forces (up to –0.65), possibly indicating unsafe driving before the data ﬂow was interrupted at 2.54 am. We were later informed that the car was found burnt out in a remote parking area. Table 2.1 shows the mean weekly distance driven (in kilometres, 1 km = 0.62 miles), number of trips and mean speed per trip for seven participants. As previously mentioned, the data from participant #8 could not be analysed. The table reveals that participants #1 and #2 travelled much longer weekly distances, compared to the other participants. Participant #7 had the smallest mean weekly number of trips. In addition, it is apparent that the weekly mean speeds per trip were by far the highest for participants #1 and #2. Table 2.2 summarises the mean weekly maximum speed, number of speeding violations per 100 km and number of large g-forces per 100 km for seven of the eight participants. It shows that participants #1 and #2 had the highest mean weekly maximum speeds. The number of mean weekly speeding violations per 100 km was highest for participant #4, followed by participant #1. All participants had a great number of mean weekly large g-forces, with participants #1 and #2 having the two largest numbers. Participants #2, #5, #6 and #7 had some weeks without driving.

Piloting a Telemetric Data Tracking System

23

Figure 2.4 The map function of the on-line monitoring system Figure 2.5 shows mean weekly maximum speeds for participants #1 and #2 who crashed during the 32 week period after the driver training study. As Figure 2.5 shows, these participants had lower mean maximum speeds right after the driver training study with participant #2 keeping to the New Zealand maximum speed limit of 100 km/h for the ﬁrst six weeks before there was a substantial increase in her maximum speed in week seven, and more or less maintaining it until she crashed in week 19. Participant #1 had much higher mean weekly maximum driving speeds, which in some weeks reached up to 140 km/h. She had maximum speeds reaching 120 km/h for most weeks, except for the ﬁrst two weeks after the driver training study and the two weeks following her crash. Figure 2.6 shows the mean weekly maximum speeds for participants #3–#7. The speeds varied considerably for all participants, except for participant #5 who reached maximum speeds at around 100 km/h for most of the monitored weeks. The other participants often reached maximum speeds of up to 120 km/h with participant #6 reaching speeds close to 140 km/h (week 32). Discussion Driving behaviour research literature has identiﬁed a need for more sensitive and objective intervention outcome measures. Thus, the aim of this pilot study was to test a telemetric data tracking system to measure post-training driving behaviour of young novice drivers. Speciﬁcally, this pilot study evaluated a tool that could help close a methodological gap that seems to exist in evaluation research of driver training interventions.

The mean (M) weekly distance driven (Dist) in kilometres (km), number of trips (Trips) and mean speed per trip (Mean Speed) in kilometres (km/h) for seven of the eight participants. Standard Deviations (SD), minimum (Min) and maximum (Max) values are also given

Part

Dist (km)

Trips Max

M

SD

Min

Max

M

SD

Min

Weeks

M

SD

Max

#1

512

290

23

1317

47

14

11

74

81

5.3

60

89

1–32

#2

460

307

0

991

31

25

0

69

84

2.5

0

90

1–18

#3

206

160

5

499

43

19

14

66

51

8.6

40

63

1–6

#4

199

111

59

340

25

12

4

54

69

19.6

18

93

1–32

#5

242

827

0

827

46

32

0

168

65

14.7

0

87

1–32

#6

339

217

0

962

74

45

0

139

54

22.4

0

94

1–32

#7

110

63

0

270

33

11

0

5

69

17.3

0

84

1–32

Driver Behaviour and Training – Volume III

Min

Mean Speed (km/h)

24

Table 2.1

Table 2.2

Weekly means of maximum speed in km/h (Max Speed), number of speeding violations per 100 km (Speeding Viol) and number of large g-forces per 100 km (G-force) for seven of the eight participants

Part

Max Speed(km) SD

Min

Max

M

SD

Min

G-force (km/h)

Max

M

SD

Min

Weeks Max

#1

123

9.4

89

141

8.7

6.1

1.3

22.8

81

5.3

60

89

1–32

#2

112

9.4

97

124

1.9

3.0

0.0

10.7

84

2.5

0

90

1–18

#3

96

25.5

68

117

0.4

0.4

0.0

1.1

51

8.6

40

63

1–6

#4

98

27.5

27

126

8.9

10.1

0.0

31.4

69

19.6

18

93

1–32

#5

100

19.3

0

111

7.3

10.3

0.0

32.8

65

14.7

0

87

1–32

#6

111

32.9

0

138

3.8

5.4

0.0

23.5

54

22.4

0

94

1–32

#7

86

620

0

121

2.2

5.2

0.0

21.7

69

17.3

0

84

1–32

Piloting a Telemetric Data Tracking System

M

Speeding Violations

25

26

Driver Behaviour and Training – Volume III

Figure 2.5 Mean weekly maximum speeds for participants #1 and #2 Note: Participant #1 crashed in week 30 (C#1) but continued to drive in week 31 and 32. Participant #2 crashed in week 19 (C#2) and stopped driving.

Figure 2.6 Mean weekly maximum speeds for participants #3–#7 We received valid post-training real driver behaviour data from seven of the eight participants. Two participants, both living in rural areas, crashed their cars within the monitoring period, without being seriously injured. Their telemetric data indicated

Piloting a Telemetric Data Tracking System

27

that they were travelling longer distances, had higher average speeds and achieved higher maximum speeds than any of the other participants. It is interesting to note that road crash statistics in New Zealand indicate that young drivers in rural areas are at greater risk of being involved in a severe crash, than those who live in urban areas. Consistent with our data, these drivers normally have a higher risk exposure as they typically drive longer distances and more frequently use rural roads that allow for higher speeds than roads in urban areas. Speeding is known to be one of the most important factors of teenage crashes in New Zealand. However, our participants indicated that they felt relatively safe when speeding, even at speeds as high as 120 km/h. This is a particularly interesting ﬁnding, as most of the participants were aware that speeding is one of the most common causes of road crashes. Most participants in this study had maximum speeds reaching 120 km/h and some of them had speeds up to 140 km/h. It seems pertinent that driver training interventions should involve methods that could decrease this high risk behaviour. One of these methods could involve hazard anticipation training using video simulation, which clearly improved speed choice behaviour (McKenna, Horswill and Alexander, 2006). All participants had many large g-force events, either caused by hard braking (longitudinal g-force), and/or hard cornering/swerving (lateral g-force). Our tracking system was not able to differentiate between these events and perhaps recorded also some non risky g-forces caused by hitting a bump/pothole in the road. An event-triggered video recording system manufactured by DriveCam and used by McGehee, Raby, Carney, Lee and Reyes (2007), for their eventtriggered video driver intervention trial could help verify the cause of each large g-force. Hard braking events could have been caused by long hazard detection times of the participants, which are typically 30 per cent longer in inexperienced novice drivers compared to experienced drivers (Deery, 1999). Hazard detection times have been found to be related strongly to crash risk in young drivers and can be improved using road commentary methods or video based hazard detection training. In summary, the telemetric data tracking system used in this study seems to be a promising research tool for evaluating post-training effects by providing an objective and sensitive driver behaviour outcome measures. By using the map-based tracking function, all the recorded driver behaviour events, including crashes, could be mapped, replayed and analysed in detail on the internet. It also allowed us to create daily, weekly and monthly reports of important risk taking behaviour variables (such as speeding, average speed and large g-forces) and could also provide information on risk exposure (driving distance). In order to improve the system, an event-triggered video recording system could help verify each large g-force that was created by the monitored vehicles. It would also be beneﬁcial to record lower speeding events such as driving 60 km/h on a road with a 50 km/h speed limit, but this depends on GPS based speed limit data for all roadways being available. To fully evaluate the utility of this system and the effects of a driver training intervention, ideally the tracking device would be installed into the vehicles of the

28

Driver Behaviour and Training – Volume III

participants several months before the driver training programme, in order to obtain data based on the participants real driving behaviour. Baseline driving behaviour in experimental and control participants can then be established, so that any potential changes in the post–training driving behaviour of the experimental group can be clearly attributed to the effect of the driver training. Acknowledgments We would like to thank the Accident Compensation Corporation, the Road Safety Trust and the ITO for funding this project. Thanks to the board of the AA Driver Education Foundation and to the a2om Driving Academy (UK) who both inspired us to conduct this study. We are grateful to all people who helped us to make the ‘frontal lobe’ training camp possible. And ﬁnally, a very special thanks to all 36 participants who invested two weeks of their holidays to become safer drivers. References Bergdahl, J. (2005). ‘Sex differences in attitudes towards driving: a survey.’ The Social Sciences Journal, 42, 595–601. Clarke, D.D., Ward, P. and Truman, W. (2005). ‘Voluntary risk taking and skill deﬁcits in young driver accidents in the UK.’ Accident Analysis and Prevention, 37, 523–9. Crick, J.,and McKenna, F.P. (1991). ‘Hazard perception: can it be trained?’ In G.B. Grayson (ed.). Behavioural Research in Road Safety II. Proceedings of a seminar at Manchester University 17–18 September 1991. Crowthorne, Berkshire: Transport Research Laboratory,100–107. Deery, H.A. (1999). Hazard and risk perception among young novice drivers. Journal of Safety Research, 30 (4), 225–36. Engstrom, I., Gregersen, N.P., Hernetkostki, K., Keeskinen, E. and Nyberg, A. (2003). Young Novice Drivers, Driver Education and Training. Swedish National Road and Transport Research Institute, VTI rapport 491A 2003. Gregersen, N.P. (1996). ‘Young drivers’ overestimation of their own skill – an experiment on the relation between training strategy and skill.’ Accident Analysis and Prevention, 28 (2), 243–50. Isler, R.B., Starkey, N.J., Charlton, S. and Sheppard, P. (2007). ‘The “frontal lobe” project: a double blind, longitudinal study of the effectiveness of higher level driving skills training to enhance executive functioning in young drivers.’ Interim report to the ACC and Road Safety Trust, New Zealand. McGehee, D.V., Raby, M., Carney, C., Lee, J.D. and Reyes, M.L. (2007). ‘Extending parental mentoring using an event-triggered video intervention in rural teen drivers.’ Journal of Safety Research, 38, 215–27. McKenna, F.P., Horswill, M.S. and Alexander, J.L. (2006). ‘Does anticipation training affect drivers’ risk taking?’ Journal of Exp Psychol Appl., 12 (1): 1–10. Senserrick, T.M. (2006). ‘Reducing young driver road trauma: guidance and

Piloting a Telemetric Data Tracking System

29

optimism for the future.’ Injury Prevention, 12 (1): 56–60. Shrout, P.E. (1997). ‘Should signiﬁcance tests be banned?’ Psychological Science, 8 (1): 1–2.

This page intentionally left blank

Chapter 3

Fault Correction or Self-Assessment: Which Way Forward? Ian Edwards and Tracey Curle Alpha to Omega Motoring Ltd. (a2om), UK Introduction Many studies (for an overview see Engström et al., 2003) have identiﬁed that young novice drivers are over-represented in road crashes. Extensive research over many years has been carried out in order to identify the factors that may contribute to this phenomenon. Education has often been identiﬁed as a key counter-measure but there has been little evidence (Mayhew and Simpson, 2002) to suggest that driver education has been successful, although some studies (McKenna et al., 2006) have identiﬁed that computer-based, hazard perception training can be beneﬁcial. Recently, Hattaka et al. (1999) developed the Goals for Driver Education (GDE), a four level hierarchical approach to driver education. The top level of the matrix looks at the driver’s ‘Goals for life and skills for living’, the next level looks at ‘Goals and context of driving’, with the two lower levels covering ‘Mastery of the trafﬁc situation’ and ‘Vehicle manoeuvring’. Based on the approach outlined in the GDE, this paper will argue that the current system of developing and training driving instructors in the UK is too limited and that a wider approach is required. Assessment and training of drivers in the UK The UK’s driving test is conducted by the Driving Standards Agency (DSA) and is a fault-based assessment of a driver’s ability at the time of the test. The test asks a driver to drive under test conditions with the examiner assessing the drive for any faults committed. These faults, if serious enough, will be recorded and are subdivided into driving faults, serious faults and dangerous faults. The test candidate can make up to 15 driving faults and pass their driving test, but any serious or dangerous faults committed result in failure. This system has served the UK well since its introduction in the 1930s and has made a signiﬁcant contribution to the UK’s road safety record. However, it appears likely that the driving test informs the content of the training which is delivered by driving instructors. This fault-based system of assessment has also signiﬁcantly impacted on the way instructors are assessed by the DSA, who are also responsible for monitoring the standard of instruction. The quality assurance process for driving instructors consists of a periodic observation of a driving lesson by a specially trained DSA examiner. The examiner will mark the instructor’s performance on a number of items but most importantly three ‘core competencies’, which are:

Driver Behaviour and Training – Volume III

32

• • •

fault identiﬁcation – ability to identify a fault committed by the driver fault analysis – ability to analyse why and how the fault was committed fault correction – ability to offer strategies to correct the fault

These core competencies are marked on a one to six scale with six indicating a high level of competency. At the end of the assessment procedure the instructor will be awarded an overall grade, again on a one to six scale with six being the highest. This overall grade is strongly associated with the grades achieved on the three core competencies. The authors of this paper, based on the developmental work undertaken by the a²om academy, argue that this fault-based assessment system is too narrow and leads to restrictive and negative learning outcomes. The in-car learning environment It is widely established (for an overview see the OECD report, 2006) that the goals, context and motivation of a journey are important factors in determining how the driving task is completed. It is therefore likely that the motivation of a learner driver will have a major bearing on the type of faults committed. Edwards (2005) suggested that the true personality of a driver is rarely seen by the driving instructor during driving lessons, as the driver will be motivated to take on the social role advocated by their instructor. This is in line with Parker and Stradling (2001) who suggested that once the driving test has been passed the driver will enter an ‘Expressive Phase’. In this phase the driver will drive in a way which expresses their own personality and beliefs about driving. If this is correct, it follows that the faults identiﬁed by the instructor whilst training a driver may be limited to technical driving errors. It is therefore very unlikely that an instructor, using a fault-correction based approach to driver education, will actively engage in discussions about the driver’s own personality traits and beliefs. This argument also strengthens the case for the use of psychometric proﬁling for new drivers. Psychometric proﬁling would help the instructor to have an understanding of the learner driver’s own beliefs, thus allowing the instructor to tailor the tuition to ﬁt the individual’s own proﬁle. This argument could also be extended to cover other aspects of driver training and evaluations have shown this to be an encouraging area of research (Dorn and Garwood, 2005). The types of faults seen by driving instructors may also be limited by the context in which the driving task takes place. On a driving lesson fewer external issues are present, compared to a ‘real world’ journey. The learning environment in many ways bears little similarity to those in which the real driving task will take place. Examples of this include: there are fewer time pressures on the driver during a driving lesson, the learner driver’s goal is to drive well and not to complete a ‘real’ journey, generally there is no one else other than the instructor present and so on. In this context the faults committed are likely to be those only attributed to lack of knowledge, psychomotor skills and inexperience, rather than those associated with outside factors and inﬂuences such as peer pressure, stress, fatigue and so on. As

Fault Correction or Self-Assessment

33

these pressures are not generally present in the learning environment it is unlikely that the instructor will ever see how the driver is likely to cope with these issues later in their driving career. In order to provide a student with a complete driver education, the GDE outlines that the student must be made aware of the types of issues they will face when driving in the ‘real world’ and not simply how to avoid common errors committed whilst training. A study looking at the tuition received by 30 learner drivers delivered by four top grade instructors seemed to indicate that this holistic approach was not being delivered: The nature of the comments made by instructors indicated that instruction was carefully tailored to the situation at hand, and very little weight was placed on aspects of the driving task which the pupil did not have the opportunity to actually carry out. There is thus little scope for pupils to develop the ‘theoretical’ understanding of the driving task which may serve them well when confronting novel circumstances later in their driving career. (Groeger and Clegg, 2000).

Groeger and Clegg’s (2000) study also identiﬁed that, as the students improved and committed fewer errors, the amount of input from the tutors decreased. Whilst this approach is likely to develop the ability of the driver to control the car and integrate in trafﬁc, it is unlikely that it will address the context in which driving takes place once the driver acquires a full driving licence. In the early stages of learning to drive, a great deal of detail can be lost by the learner driver as they struggle to gain the psychomotor skills required to drive. As skill levels increase, the learner is more able to engage and absorb information on other topics related to driving. It is therefore hypothesised by the authors that the most effective time to engage in dialogue related to the higher elements of the GDE would be at the later stages of learning, once car skills have been mastered. Figure 3.1 shows the traditional approach to driver training as outlined by Miller and Stacey (2006) in the Driving Instructors’ Handbook, which is recommended reading by the Driving Standards Agency for all trainee driving instructors. The model indicates that when a driving instructor ﬁrst introduces a new topic the learner driver will require a high level of input. This input reduces over a period of time until the learner driver has reached a level of competency whereby they are able to undertake the task independently. Miller and Stacey refer to this as the learning curve and outline that You will be able to see the manipulative aspects of the driving task improving by simply watching your pupil execute them. However, prompts and verbal guidance during the early stages of learning will still be necessary. When the skill is sufﬁciently developed, hints and reassurance should be all that is required (Miller and Stacey, 2006: p. 261).

Whilst in many ways this approach seems to be intuitively correct, that is that as a driver improves there should be less need for the instructor to intervene, this approach fails to take into account the environment in which the faults are committed. As already identiﬁed, the faults committed whilst training are unlikely to be the faults committed in the ‘real world’. Therefore, instructors need to actively engage

Driver Behaviour and Training – Volume III

34

in discussions which will help the learner driver identify their own strengths and weaknesses against a ‘real world’ backdrop. This approach would require instructors to take a wider approach to driver education and would be much more in line with coaching methods. Whitmore (2006) deﬁnes coaching as ‘unlocking a person’s potential to maximise their own performance’ (8).

Prompted Practice

Transferred responsibility

Achievement

Controlled Practice

Time

Figure 3.1 Miller and Stacey’s driving instruction learning curve This coaching approach is critical in helping new drivers to understand and develop self-assessment skills, a key element contained within the GDE matrix. This approach needs to be used to address all four levels of the GDE so that, once a driver has passed the driving test, they are able to self-learn from the mistakes they commit in the ‘real world’ environment. Boud (1995) identiﬁed that ‘for effective learning of any kind to take place, learners – whoever they may be – must develop the capability of monitoring what they do and modifying their learning strategies appropriately’ (14). Boud also suggests that graduates who develop self-assessment skills will be more likely to: • • • •

wish to continue their learning know how to do so monitor their own performance expect to take full responsibility for their actions and judgements.

If the above were applied to a driving context, it could have signiﬁcant results by accelerating the ability of a newly qualiﬁed driver to learn more quickly from the experience they gain in the ﬁrst few months of post-test driving. This concept of

Fault Correction or Self-Assessment

35

learning from their own experience is critical as it is almost impossible for a learner driver to experience all of the possible driving situations they may face in the future within the pre-test learning environment. Therefore, the role of self-assessment is crucial and instructors should actively encourage the development of this attribute both in themselves and the drivers they train. Many instructors would argue that this is already being done and there could be some truth in this as instructors are assessed on their ability to use an appropriate question and answer technique. However, as this assessment is generally linked to the identiﬁcation of faults, the issues being discussed are very likely to be those associated with the two lower levels of the GDE, namely vehicle manoeuvring and mastery of trafﬁc situations, rather than the higher GDE elements. The role of the instructor One of the key ways in which self-assessment could be used by the in-car instructor is for the instructor to encourage the learner driver to think about situations they may be faced with in the future. In this way, the learner driver is given the opportunity to consider their own predisposition towards certain risk-increasing activities and to develop and rehearse possible strategies. This rehearsal could help the new driver by providing bench marks to assess performance, as well as practising the selfassessment techniques in the development of those bench marks. Conclusion The current fault-based assessment system places too much emphasis on the ability of the instructor to identify and deal with student errors, with too little emphasis on encouraging discussion of higher level knowledge and skills as outlined in the GDE. It is important to remember that fault correction does have an important role to play in the development of a driver, but it is equally important to recognise the limitations of this approach. It is doubtful that any in-car training initiative aimed at reducing the crash rate of young drivers will be successful whilst training continues to be predominantly aimed at the identiﬁcation, analysis and correction of faults, to the detriment of the development of appropriate self-assessment skills. The a²om academy is pioneering this approach by placing the development of self-assessment skills at the centre of their driver education programmes. References Boud, A. (1995). (ed. by Boud, D.). Enhancing Learning through Self-Assessment. Routledge Falmer, West Sussex, UK: 15 Dorn, L. and Garwood, L. (2005). ‘Development of a psychometric measure of bus driver behaviour’. Behavioural Research in Road Safety: 14th Seminar.

36

Driver Behaviour and Training – Volume III

Department for Transport, London. Edwards, I. (2005). Behavioural Research in Road Safety 2005: Fifteenth Seminar. DfT publications: West Yorkshire, UK. Engström, I., Gregersen, N.P., Hernetkoski, K., Keskinen, E. and Nyberg, A. (2003). Young and Novice Drivers, Driver Education and Training – Literature Review. Swedish National Road and Transport Research Institute: Sweden. Groeger, J.A., and Clegg, B.A. (2000). Practice and Instruction When Learning to Drive. Road Research Report No 14. London: HMSO. Hattaka, M., Keskinen, E., Gregersen, N., Glad, A. and Hernetkoski, K. (1999). Results of EU-Project GADGET, WP3. Parker, D. and Stradling, S. (2001). Road Safety Research Report No 17 Inﬂuencing Driver Attitudes and Behaviour. DfT publications: West Yorkshire, UK. Mayhew, D.R. and Simpson, H.M. (2002). ‘The safety value of driver education and training.’ Injury Prevention, 8: ii3–ii8. McKenna, F.P., Horswill, M.S. and Alexander, J.L. (2006). ‘Does anticipation training affect drivers’ risk taking?’ Journal of Exp Psychol Appl., 12 (1): 1–10. Miller, J. and Stacey, S. (2006). The Driving Instructor’s Handbook. Kogan Press: London: 261. OECD. (2006). Young Driver – The Road to Safety. Paris, France.

Chapter 4

New Elements in the Dutch Practical Driving Test: A Pilot Study 1

Jan Vissers,1 Jolieke Mesken,1 Erik Roelofs2 and René Claesen3 DHV Environment and Transportation, Amersfoort, The Netherlands 2 Cito National Assessment Institute, Arnhem, The Netherlands 3 CBR Driving Test Organisation, The Netherlands

Introduction In line with many other countries, novice drivers in the Netherlands are greatly overrepresented in trafﬁc collisions and fatalities, and male novice drivers even more so. While in the Netherlands the relative risk of being involved in a road accident has been decreasing over the last few years (Stipdonk, Aarts, Schoon and Wesemann, 2006), for young drivers aged between 18 and 24 this is not the case. These developments have led the Dutch Trafﬁc Department to develop a plan to revise the Dutch practical driving test. It commissioned the Driving Test Organisation (CBR) to develop and implement a new practical driving test by January 2008. The new practical driving test should, in line with the Goals for Driver Education research (GDE-matrix; see Siegrist et al, 1999) include new elements that enable the assessment of higher order skills, such as hazard perception and self-reﬂection. In 2007, pilot studies have been carried out by DHV in close cooperation with the CBR, to investigate the applicability of these new elements of the practical driving test. In this paper the results of these pilot studies (Vissers and Mesken, 2007) are described. Background of high accident risk of novice drivers Knowledge about what constitutes a safe and unsafe driver is quite extensive. A vast amount of research is available, which shows that becoming a safe driver is a very complicated matter. In two recent literature reviews (OECD, 2006; Engström, Gregersen, Hernetkoski, Keskinen and Nyberg, 2003) a lot of important correlates with unsafe driving have been documented. Compared with experienced drivers, young novice drivers have more single vehicle crashes, more often lose control of the car and more often drive too fast considering the circumstances just before the crash. They have relatively more weekend crashes and their passengers are more often of their own age group. The causes of the high accident rate of young novice drivers can be clustered around two main accident causes: ﬁrst, age-related factors and second, inadequate skills.

Driver Behaviour and Training – Volume III

38

Age-related factors The age-related factors (or ‘age cluster’) refer to the developmental phase of the young people at that particular moment. The moral, emotional and cognitive development of young women is slightly different from that of young men. Adolescents, and especially males, often rebel against existing norms, do not want to appear ‘soft’ to their friends, enjoy sensation, underestimate risks and have the feeling that they are more or less invulnerable. Here we are addressing things that are typical for youngsters, such as their lifestyle, peer group membership, the youth socialisation process and so on. These all inﬂuence attitudes, motives and the decisions which drivers make whilst driving. Although there are substantial differences between novice drivers in the extent to which these age related factors inﬂuence their driving behaviour, it is not the case that only a small core group of wild young men are completely responsible for the high accident rate. Young inexperienced female drivers also have distinctly higher crash rates than young experienced female drivers. There is often a sliding scale: the more one possesses certain personality features (for example, a greater need for excitement) and the more one belongs to a certain, more problematic subgroup, the higher the crash rate. Lack of adequate skills The lack of adequate skills is not so much a matter of poor vehicle control, but is more a lack of: • • •

being able to observe properly (observe relevant information); adequately judging trafﬁc situations (that is, capable of concentrating on aspects of the trafﬁc environment that are potentially dangerous); and predicting accurately how a particular trafﬁc situation will develop.

These skills are generally referred to as ‘higher order’ skills. Furthermore, young novice drivers often have a lack of meta-cognitive skills. This means that young novice drivers’ self-assessment skills are inadequate, leading to not being able to ﬁt their inadequate task skills (a lack of higher order skills) to the driving tasks they encounter in trafﬁc. This process of ﬁtting the task skill to the driving task is sometimes called calibration (Kuiken and Twisk, 2001). For instance, young novice drivers too rarely think ‘it is dark and raining, and as I don’t have enough experience yet, I will go by train this evening and not by car’. Calibration is not only important at this more strategic level, but also at the tactical level when driving in trafﬁc. An example of this is ‘shall I overtake this car on this slippery road on this dark evening, given my lack of experience?’ A lack of higher order skills is mainly the result of a lack of driving experience. So this cluster of causes is also referred to as the ‘experience cluster’.

New Elements in the Dutch Practical Driving Test

39

Weight of both clusters of causes Several studies investigated the weight of the ‘age’ and the ‘experience’ clusters. Most of them (among others, Maycock, Lockwood and Lester, 1991; Gregersen and Bjurulf, 1996; Sagberg, 1998) have shown that a lack of driving experience is somewhat more important than the inﬂuence of age related factors. Data from the Periodic Road Safety Surveys in the Netherlands (Barten, van Drunen, Herber, IJsselstijn and Vissers, 2005) also show that about 60 per cent of accidents can be attributed to experience related factors and about 40 per cent to age related factors. So, in general, the high accident rate of young novice drivers is more determined by a lack of driving experience than by age related factors, although there is some variation between studies in just how dominant the experience effect is. Theoretical framework: GDE-matrix The information about the factors that are related to the high accident rate of young novice drivers has been structured into a hierarchical model. A useful description of the different levels of driving behaviour was developed as part of the EU-project GADGET (Siegrist et al., 1999): the matrix of ‘Goals of Driver Education’ or GDEmatrix. The matrix is based on the assumption that the driving task can be described as a hierarchy. The idea of the hierarchical approach is that abilities and preconditions on a higher level inﬂuence the demands, decisions and behaviour on a lower level. The GDE-matrix describes four levels, starting at the basic level with ‘Vehicle manoeuvring’ (level I) and then ‘Mastery of trafﬁc situations’ (level II). The higher levels relate to ‘Goals and context of driving’ (level III; trip related) and ‘Goals for life and skills for living’ (level IV). These higher levels are less related to knowledge and skills and more to understanding, experience and self-awareness. Researchers participating in the GADGET project also emphasised the importance of providing feedback to drivers in order to address the higher level motivational and attitudinal components of driver behaviour. A key element of the GDE-matrix is self-evaluation – ‘how did I perform in this situation and what could I have done differently to reduce the risk?’ Many learner drivers in EU-countries are not trained well in these aspects of driving. Most countries are still focusing on trafﬁc rules and managing/mastering the vehicle in different trafﬁc situations, which is the historical basis for driver education and driver testing all over the world. The implication of the GDE-matrix is that focusing training and testing on the lower levels does not guarantee that drivers will drive safely once they start to drive unsupervised and are subject to level III and IV situations and goals. Indeed, focusing training and testing solely at levels I and II can have a negative effect on safety, increasing conﬁdence and encouraging drivers to explore a wider ‘performance envelope’ without making them aware of the risks that this type of behaviour imply (Vissers, 2004a). Recent work carried out by the Transport Research Laboratory (Baughan et al., 2006) illustrates the possible negative effect of concentrating on the lower level of the GDE-matrix. This study highlighted that those drivers who have the fewest

Driver Behaviour and Training – Volume III

40

lessons prior to taking their test often have the fewest faults during their driving test and have the poorest attitudes towards safe driving – and subsequently go on to have more accidents than other novice drivers. This raises the question as to whether the content of the test is correct, that is, does the driving test assess the right things in the right way? New elements for the practical driving test Based on the ﬁndings of the GDE-matrix studies, four categories of new elements for the practical driving test have been distinguished: • • • •

independent driving ‘productive’ special manoeuvres perception and management of hazards self reﬂection

Independent driving Once novice drivers have acquired their licences, they will have to drive independently. Modernisation of the driving test means a lot of emphasis on the higher levels of the GDE-matrix which deal with driver characteristics and risk awareness. Independent driving appears to be a relatively easy way to introduce elements from the higher levels into the driving test, especially level III. As part of independent driving, trip-related planning and decision skills related to both safety and environmentally friendly driving can be integrated in the practical driving test. Requiring candidates to demonstrate responsibility and independent decisionmaking in the test, goes beyond merely learning technical vehicle control and applying trafﬁc rules. In countries where driver training is not regulated and where there is no obligatory second phase, introducing this type of driving into the practical driving test means it will be incorporated into the driver training. Independent driving should be considered as a general ability. It should not be deﬁned as a separate driving task but as something related to and affecting all driving tasks. It should be seen as a tool to test other elements, to lead a candidate to situations where they have to make their own choices. Responsibility and independence should be clear and present in all parts of driver-training curriculum and consequently in the driving test. Likewise, all the different parts and elements of the driving test should enhance and encourage independent driving. Examiners will need additional training to be able to apply and identify independent driving in the practical driving test. The driving test requires cooperation between an examiner and a candidate. It is important to determine how much freedom should be given to each. There needs to be a balance. Candidates should be given enough freedom so that they feel responsible for their own driving but examiners need to maintain enough control to be able to test all they need to test. In a driving test, independent driving involves creating circumstances in which candidates shows their driving skills over a longer period

New Elements in the Dutch Practical Driving Test

41

without the help or instructions of the examiner. During this period, candidates have to ﬁnd their way and make decisions in an independent way.

Working deﬁnition of ‘independent driving’ Independent driving means that candidates make a responsible choice based on their own abilities and the requirements of the task. In the driving test, independent driving involves creating circumstances in which the candidate has to demonstrate their driving skills over a longer period without the help or instructions of the examiner. During this period, candidates have to ﬁnd their way and make decisions in an independent way.

The following forms of independent driving were tested in the pilot studies: •

•

• •

Driving towards coordination points: The candidate is asked to drive towards a certain coordination point (for example, a railway station, hospital or school). Using a navigation system: The examiner enters a destination in the navigation system and the candidate then has to follow the directions given by the system. Fixed task: The candidate is instructed to continuously take the second street on the left followed by the second street on the right. Being given a series of instructions: The examiner gives the candidate a series of instructions comparable to a situation in which someone asks for directions when they are in unfamiliar surroundings.

‘Productive’ special manoeuvres As an alternative to the reproductive performance of the special manoeuvres in the current driving test, a more independent (or productive) way of carrying out these special manoeuvres has been introduced in the Dutch driving test. The ‘productive’ special manoeuvres can be seen as a special form of independent driving in which candidates have to make their own decisions about when and how to perform which special manoeuvre. As alternative approaches to the existing special manoeuvres the following productive special manoeuvres were tested in the pilot studies: •

•

Turning manoeuvre: • independently turning the vehicle around (the candidate has to determine where and how the vehicle should be turned) Parking manoeuvre: • independently looking for a parking space in a parking lot and parking the car • independently looking for a parking space in a street in a built-up area and

Driver Behaviour and Training – Volume III

42

• • •

parking the car Stopping manoeuvre: stopping directly behind another vehicle (in such a way that the candidates can follow their route without having to reverse ﬁrst) stop and go: stopping parallel to a driveway and subsequently following the route

Perception and management of hazards In the practical driving test the decisions of the examiner are traditionally based on what they see: the decisions the candidate makes and the way in which they put these decisions into action. But whether the candidate also performs the processes that precede decision-making and vehicle handling (for example, perception of relevant information, making the right predictions and evaluations) in a proper way is not observable to the examiner. There are two possible ways to include information about the choices a candidate makes: questioning by the examiner and commentary driving by the candidate. The scope of these methods in the practical driving test is probably limited because there are difﬁculties in using verbal reports as a basis for inferring whether the right choices have been made and whether the behaviour is controlled by higher-level goals. In addition to this, formal pass/fail criteria may be difﬁcult to establish. Nevertheless it can be helpful to integrate these methods to assess adequate maximum performance of the candidate, even though the information will not be used as formal pass/ fail criteria. The following two approaches for assessing hazard perception and hazard management were investigated in the pilot studies: • •

commentary driving by the candidate questioning by the examiner

Self-reﬂection by the candidate In Sweden and Finland self-reﬂection by the candidate is already added to the driving test. Candidates have to ﬁll out a form and give an assessment of their own driving behaviour using the topics that are assessed at the driving test by the examiner. Examiners only use this self-reﬂection form after the test, when they give their ﬁnal assessment. The candidate’s self-reﬂection is used to give better feedback to the candidate about their strong and weak points. This is useful information in the case of failure (what should be focused on in driver training in order to pass the test next time) as well as success (even though the candidate passed what weaknesses they should focus on during independent driving). The examiner more or less makes an assessment of their future driving safety. This is consistent with the claim that passing the test is not an ultimate goal but just a starting point for further learning to

New Elements in the Dutch Practical Driving Test

43

become an independent, safe and responsible driver. This approach to self-reﬂection was not tested during the pilot studies.1 Instead of self-reﬂection after the driving test in the pilot studies self-reﬂection during the driving test was tested. For self-reﬂection during the test the candidate is asked to stop the car after having driven for about 15 minutes. Then the examiner asks the candidate to reﬂect on their driving performance: ‘What went well and what could you have done better?’ The examiner reacts in a neutral way and a discussion based around the candidate’s faults is presented. In the pilot studies carried out, it was decided whether higher order skills can be assessed using the new elements in the new practical driving test. In addition the practical application of the new elements was evaluated. In summary, this leads to the following research questions: 1. To what extent are the newly introduced elements in the practical driving test considered applicable? 2. To what extent do, according to driving instructors and examiners, the newly introduced elements enable the assessment of higher order skills? Method Participants About 150 candidates, who were in the ﬁnal stage of their driver training, practised with the new elements for about six weeks. After that, a Learner Interim Test2 was administered by the CBR in which the new elements were included. Instructor, candidate and examiner completed the evaluation forms. Due to incomplete forms, a total number of 109 sets were used for the analysis. The different regions in the Netherlands were represented equally in the sample. With the exception of a slight over-representation of older candidates in the study, the sample of candidates was representative for the population of CBR-candidates for the driving test. About twothirds of the candidates (65 per cent) were 18 or 19 years old, 15 per cent were between 20 and 25 years of age and the remaining 20 per cent were over 25 years of age. Background variables such as gender (49 per cent male), educational level and being employed did not differ between sample and population. Also the number of

1 The decision was already made to test self-reﬂection after the driving test (the Finnish/ Swedish model) as part of a follow-up experiment in which the most successful new elements will be integrated in a modernised driving test. 2 In the Dutch system, all learner drivers have the opportunity to take the TTT (Learner Interim Test or ‘Tussentijdse Toets’). This test usually takes place when candidates are about three quarters of their way through their driver training and gives the learner drivers a chance to show a CBR examiner how they drive. The test also gives learner drivers the chance to experience test circumstances and they can earn an exemption for the special manoeuvres if these are performed well. Following the test, the examiner will give the candidate feedback regarding what areas still need to be addressed.

44

Driver Behaviour and Training – Volume III

driving lessons (an average of 33 one-hour lessons until the moment of the Learner Interim Test) did not differ. Questionnaire Instructor, candidate and examiner completed a questionnaire immediately after the Learner Interim Test. The questionnaires were aimed to assess the effectiveness of the procedures for the new elements and included open-ended and closed questions concerning topics such as: the perceived difﬁculty of practising with the new elements, the perceived usefulness of the new elements for the practical driving test and the perceived possibility to assess higher order skills by means of the new elements. Furthermore, questions were asked about the time needed to prepare candidates for the new elements and the extent to which this was sufﬁcient. Candidates also answered questions about background variables, in order to establish whether the sample was representative for the average candidate. Procedure and data collection Each candidate was tested on one of the four new methods of independent driving and one of the three ‘productive’ special manoeuvres. As well, each candidate was tested on either hazard perception or self-reﬂection during the test. So each candidate was tested on three new elements (out of the ten being introduced) during their Learner Interim Test. Learner drivers, driving instructors and examiners completed the questionnaire directly after the Learner Interim Test. Analysis In this pilot phase of the study the emphasis was on a qualitative analysis of the data. So a descriptive analysis and a comparison between the three groups involved (candidates, driving instructors and examiners) on the questionnaire measures was conducted. Results Independent driving Independent driving is not yet a self-evident part of driver training. According to learner drivers, independent driving takes some time to master. The easiest method to demonstrate independent driving in their opinion is driving to a coordination point (for example, a railway station, hospital or school): 49 per cent evaluated this method as (very) easy to learn and perform. They evaluated the use of a navigation system (41 per cent [very] easy) and being given a series of instructions (38 per cent) as somewhat more difﬁcult. The method of continuously taking the second street on the left followed by the second street on the right was considered most to be the most difﬁcult task (only 30 per cent evaluated this task as [very] easy). (See Table 4.1.)

New Elements in the Dutch Practical Driving Test

Table 4.1

45

Perceived task difﬁculty of the four methods of independent driving (percentage responding that performance is ‘easy’ or ‘very easy’)

Driving to a coordination point Use of a navigation system Continuously taking 2nd left then 2nd right Given a series of instructions

Candidate 49%

Driving instructor 45%

Examiner 60%

41%

52%

38%

30%

24%

21%

38%

45%

40%

This is consistent with the opinion of the driving instructors. However, although the instructors agree with the learner driver that independent driving is difﬁcult, they have more conﬁdence that learner drivers can be trained to master independent driving. But to realise this, instructors believe that driver training will have to be reorganised. If independent driving is to be integrated as driver training right from the start, driving instructors expect it will only take a little extra time to learn this method. Driving instructors in our study seem very positive about the added value of independent driving for training as well as testing. According to them, independent driving gives more insight into the way the learner drivers detect important changes in trafﬁc, how they evaluate these changes, how they plan their actions, select the most appropriate action under the actual circumstances and ﬁnally carry out the selected action. When we look at the assessment of the four methods of independent driving on the Interim Learner Test, the conclusions of the examiners are in line with the opinions of the candidates and driving instructors. According to examiners, candidates had some difﬁculty carrying out the four methods of independent driving in the desired way; especially the method of continuously taking the second street on the left followed by the second street on the right, which was considered difﬁcult for candidates. Driving to a coordination point is the easiest way of independent driving: 76 per cent performed this method as required. Performance on the use of a navigation system and being given a series of instructions is somewhat more difﬁcult: in both cases 55 per cent of the candidates performed as required. On the method of continuously taking the second street on the left followed by the second street on the right candidates showed the lowest performance: 37 per cent performed as required. With the exception of the method of continuously taking the second street on the left followed by the second street on the right, the tested methods seem applicable as an assessment instrument during the driving test. Independent driving is of great value for the driving test as well as for the training of novice drivers according to instructors and examiners. By assessing independent driving during the driving test, the candidates are forced to show a higher level of driving performance. The candidates have more opportunities to show ‘productive’ driving behaviour: they have to make their own choices in trafﬁc and they are more

Driver Behaviour and Training – Volume III

46

aware of other trafﬁc participants and the trafﬁc environment. Examiners get a better insight into the way the candidate handles trafﬁc situations and they can follow the candidate’s driving-related cognitive processes in response to changes in driving situations. ‘Productive’ special manoeuvres The ‘productive’ special manoeuvres are not a problem according to the learner drivers (see Table 4.2). The parking manoeuvre is somewhat easier to perform than the turning manoeuvre and the stopping manoeuvre. Learner drivers responded that they thought it is useful to have the ‘productive’ special manoeuvres integrated into the driving test instead of the traditional ‘reproductive’ special manoeuvres. In the future when they have their driving licence they will have to make independent choices concerning what special manoeuvre to perform and where to execute it. Table 4.2

Perceived task difﬁculty of the three categories of ‘productive’ special manoeuvres (percentage responding that performance is ‘easy’ or ‘very easy’) Candidate

Driving instructor

Examiner

Turning manoeuvre

57%

69%

39%

Parking manoeuvre

71%

61%

49%

Stopping manoeuvre

60%

70%

45%

Similarly, driving instructors stated that the performance of the ‘productive’ special manoeuvres creates no problems for learner drivers. In their opinion it will take little time to teach the new methods during driver training. Again, it will be more or less a matter of reorganising the training, so no extra time will be needed to train the learner drivers the new special manoeuvres. Driving instructors stated that by introducing the ‘productive’ special manoeuvres in training and testing, independent decision-making on a more strategic level is stimulated. Learner drivers have to improve their planning and anticipation behaviour. The examiners evaluated the ‘productive’ special manoeuvres somewhat differently to the learners and instructors, believing them to be more difﬁcult, especially the turning and parking manoeuvres. According to the examiners, the ‘productive’ special manoeuvres are of great value to the driving test. As with independent driving, examiners expect that more insight is gained in the way the candidates detect important information about the trafﬁc situation, evaluate this information, plan their actions, select the most appropriate special manoeuvre under the actual circumstances and ﬁnally carry out the selected special manoeuvre correctly.

New Elements in the Dutch Practical Driving Test

47

Hazard perception skills Concerning the two methods that have been introduced to measure candidate’s hazard perception skills, the commentary driving was judged across all groups to be very difﬁcult for the candidates to perform. Although all groups agreed on the fact that commentary driving is a good method for driver training, they believed that it is not sufﬁcient as an assessment instrument. Furthermore, all three groups believed that questioning on critical emerging hazards is a useful training method. But the method is evaluated somewhat more positive as an assessment instrument. According to examiners and instructors, questioning by the examiner gives insight into the way the candidate has solved a problematic trafﬁc situation and makes clear whether they have acted deliberately or not. They add, however, that the verbal skills of candidates may be a factor for this particular method. In addition, this method is considered sensitive to subjective interpretations by the examiner. These disadvantages may threaten the reliability of the instrument. Self-reﬂection According to all groups the measurement of self-reﬂection during the driving test is not useful. In the ﬁrst place it is questionable if the candidates are able to give a reliable picture of what their strong and weak points have been during the ﬁrst ten or ﬁfteen minutes of the driving test. Besides this, examiners felt that self-reﬂection feedback (especially when things didn’t go well during the ﬁrst part of the test) can have negative consequences for performance on the remaining part of the test. Discussion The ﬁndings for the new elements introduced to measure independent driving on the driving test is that three out of the four methods tested in the pilot study are considered to be very useful: driving to a coordination point, use of a navigation system and being given a series of instructions. Independent driving is a good method to provoke more ‘productive’ driving behaviour: candidates have to take more responsibility and plan their own driving actions in advance. By introducing independent driving in the driving test, more attention is given to the third (strategic level) of the GDE-matrix. The results suggest that the examiner gets a better insight into the way the candidates detect important changes in trafﬁc, how they evaluate these changes, how they plan their actions and select the most appropriate action under the actual circumstances and ﬁnally how they carry out the selected action. This conclusion also applies to the new elements introduced to assess whether candidates are able to make an independent choice about where to perform a special manoeuvre and about which special manoeuvre to perform. All three categories of ‘productive’ special manoeuvres that have been tested are considered very useful: the turning manoeuvre, the parking manoeuvre and the stopping manoeuvre. As is the case in independent driving, more insight is gained in the way the candidates detect important information about the trafﬁc situation, evaluate this information,

Driver Behaviour and Training – Volume III

48

plan their actions, select the most appropriate special manoeuvre under the actual circumstances and ﬁnally carry out the selected special manoeuvre correctly. Two methods to assess hazard perception skills have been tested. Commentary driving is not considered a useful method for the driving test. The questioning of a critical trafﬁc situation by the examiner is considered somewhat more favourably as an assessment instrument. Questioning by the examiner gives insight into the way the candidate has solved a problematic trafﬁc situation but the reliability of this assessment instrument can be problematic. It may be useful to include questioning by the examiner in the driving test for countries with a ‘test-driven licensing system’ that relies on the test to induce drivers to take adequate training, as is the case in the Netherlands. Questioning by the examiner to assess hazard perception will mean that driving instructors will need to provide hazard perception training as well. Because of serious threats to the reliability of the instrument, formal pass/fail criteria may be difﬁcult to establish. It was decided therefore not to include results from questioning into the ﬁnal pass/failure decision. In this pilot study, a candidate’s self-reﬂection was measured during the driving test. Self-reﬂection integrated in the driving test is not a useful assessment method, according to the participants in this study. It is questionable whether reliable information is gained about the strengths and weaknesses of the candidate’s skills. Additionally, examiner’s feedback on their self-reﬂection can affect driving test performance for the rest of the test. On the basis of the results of the pilot study a new procedure for the Dutch driving test has been outlined. The main changes concern the introduction of: •

• • •

independent driving (a substantial part of the test will consist of independent driving; a random choice will be made out of the three available methods: driving to a coordination point, use of a navigation system and being given a series of instructions); ‘productive’ special manoeuvres; questioning by the examiner; self-reﬂection by the candidate (the candidate will have to ﬁll out an evaluation form before the test and the evaluation form will be used by the examiners after the driving test to explain their decision to the candidate).

In a follow-up study, the new procedure for the driving test will be tested. On the basis of the results of this study the driving test will be adapted and a ﬁnal version of the new driving test will be delivered from the beginning of 2008. References Barten, M., Drunen, R. van, Herber, N. Jsselstein, S. and Vissers, J.A.M.M. (2005). PROV 2005. Periodiek Regionaal Onderzoek Verkeersveiligheid. [Periodic Road Safety Survey.] Amersfoort, DHV. Baughan, C., Pettersen, N.P., Hendrix, M. and Keskinen, E. (2004). Towards

New Elements in the Dutch Practical Driving Test

49

European Standards for Testing. Rijswijk, CIECA. Baughan, C.J., Sexton, B., Maycock, G., Simpson, H., Quimby, A. and Chinn, L. (2005). Novice Driver Safety and the British Practical Driving Test. TRL562. Crowthorne, TRL. Engström, I, Gregersen, N.P., Hernetkoski, K., Keskinen, E. and Nyberg, A. (2003). ‘Young novice drivers, driver education and training.’ Literature review. VTIreport 491A. Linköping, VTI. Gregersen, N.P. and Bjurulf, P. (1996). ‘Young novice drivers: towards a model of their accident involvement.’ Accident Analysis and Prevention, 28, 229–41. Hatakka, M., Keskinen, E., Gregersen, N.P., Glad, A. and Hernetkoski, K. (2002). From Control of the Vehicle to Personal Self-Contro: Broadening the Perspectives to Driver Education. Transportation Research Part F, 5, 201–15. Hatakka, M., Keskinen, E., Baughan, C., Goldenbeld, Ch., Gregersen, N.P., Groot, H., Siegrist, S. Willmes-Lenz, G. and Winkelbauer, M. (2003). Basic Driver Training: New Models. Turku, University of Turku. Jonsson, H., Sundström, A. and Henriksson, W. (2003). Curriculum, Driver Education and Driver Testing: A Comparative Study of the Driver Education Systems in Some European Countries. (Educational Measurement No. 44). Umeå University (EM) Department of Educational measurement, Umeå. Kuiken, M.J. and Twisk, D.A.M. (2001). ‘Safe driving and the training of calibration.’ Literature review. Report R-2001-29. Leidschendam, SWOV. Maycock, G., Lockwood, C.R. and Lester, J.F. (1991). The Accident Liability of Car Drivers. Report 315. Crowthorne, Transport Research Laboratory. Maycock, G. and Forsyth, E. (1997). Cohort Study of Learner and Novice Drivers, Part 4: Novice Driver Accidents in Relation to Methods of Learning to Drive, Performance in the Driving Test and Self Assessed Driving Ability Behaviour. Report 275. Crowthorne, Transport Research Laboratory. McDaniel, M.A., Morgeson, F.P., Finnegan, E.B., Campion, M.A. and Braverman, E.P. (2001). ‘Use of situational judgment tests to predict job performance: a clariﬁcation of the literature’. Journal of Applied Psychology, 86, 730–40. OECD. (2006) Young Drivers: The Road to Safety. Organisation for Economic Cooperation and Development. Peräaho, M., Keskinen, E and Hatakka, M. (2003). Driver Competence in a Hierarchical Perspective: Implications for Driver Education. Turku, University of Turku/ Trafﬁc Research. Rasmussen, J. (1984). Information Processing and Hum-Machine Interaction: An Approach to Cognitive Engineering. New York/Amsterdam/London, NorthHolland. Sagberg, F. (1998). ‘Month-by-month changes in accident risk among novice drivers.’ Paper presented at the 24th International Congress of Applied Psychology. San Francisco, August 9–14. Siegrist, S. (1999). (ed.) Driver Training, Testing and Licensing: Towards Theorybased Management of Young Drivers’ Injury Risk in Road Trafﬁc. Results of the EU-project GADGET, Work Package 3. Bern, BFU. Stipdonk, H.L., Aarts, L.T., Schoon, C.C. and Wesemann, P. (2006). De essentie in de daling van het aantal verkeersdoden: Ontwikkelingen in 2004 en 2005 en nieuwe

50

Driver Behaviour and Training – Volume III

prognoses voor 2010 en 2020. SWOV Publication R-2006-4. Leidschendam: SWOV. Vissers, J.A.M.M. (2004a). Testing and Teaching of the Higher Order Skills of the GDE-Matrix. CIECA-congress, Warsaw. Vissers, J.A.M.M. (2004b). Modernisering rijexamens, Probleemanalyse beginnende bestuurders. [Modernizing driving certiﬁcation tests, problem analysis of beginning drivers.] Report number TT04-021. Veenendaal, Trafﬁc Test. Vissers, J.A.M.M. (2006). Lifelong Learning: Education and Training in Schools. CIECA-congress, Marseille. Vissers, J.A.M.M. and Mesken, J. (2007). Modernisering CBR-praktijkexamen. Resultaten deelproeven vernieuwende elementen. [Modernising the Dutch driving test. Results of the pilot studies with new elements for the practical driving test.] DHV, Amersfoort. Vlakveld, W.P. (2005). Jonge beginnende automobilisten, hun ongevalsrisico en maatregelen om dit terug te dringen. Een literatuurstudie. [Young novice car drivers, their accident risk and regulations to reduce this. A literature study.] Leidschendam, SWOV.

Chapter 5

Personality and Attitudinal Predictors of Trafﬁc Offences Among Young Drivers: A Prospective Analysis Lisa Wundersitz and Nicholas Burns University of Adelaide, Australia Introduction In Australia and other developed countries, young drivers (aged 16 to 24 years) represent only a minor proportion of the licensed driving population, yet are substantially more likely to be involved in fatal and injurious crashes than older, more experienced drivers (for example, Legge et al., 2000; Shope et al., 2001). Research suggests that around 90 per cent of crashes are, to some extent, caused by human factors or road user behaviour (Shinar, 1978). As a result, many studies have been undertaken to identify driver characteristics and behaviour associated with crash involvement. In the driving context, personality characteristics and attitudes can inﬂuence how individuals approach and behave in certain driving situations. Personality characteristics, by deﬁnition, are relatively stable over time; therefore, changing them is not an appropriate goal for young driver countermeasures. However, understanding which personality factors predict driver behaviour might assist in developing interventions and public education matched to the individual needs of young drivers. Moreover, identifying and modifying mediating factors linking personality to risky driving behaviour and crashes may be useful in changing young driver behaviour. Personality characteristics and attitudes have been found to be weakly but consistently associated with young driver crash involvement (for reviews, see Beirness, 1993; Elander et al., 1993). However, the role of personality and attitudes in crash involvement may be underestimated because crashes are relatively rare events. As a result, any differences in crash rates attributed to personality and attitudinal factors will be more difﬁcult to detect statistically (see Evans, 2004 for a discussion). Moreover, crash causation is dependent on factors other than the behaviour of a particular driver, such as environmental circumstances (for example, weather conditions), exposure (for example, annual mileage), and the behaviour of other drivers (Friedstrom et al., 1995; Struckman-Johnson et al., 1989). As crash data lacks stability and statistical power, they are not an ideal measure. A measure based on an aggregate of multiple behaviours might be more appropriate and reliable for examining the inﬂuence of personality on behaviour (for example, Epstein, 1979; Ulleberg and Rundmo, 2003). Based on this concept, personality

52

Driver Behaviour and Training – Volume III

and attitudinal measures might be expected to better predict an aggregate measure of risky driving behaviours, such as trafﬁc offences, in comparison to crashes. Driver behaviour captured in trafﬁc offence statistics is predominantly intentional and connected to the characteristics and motivations of the driver (Burg, 1970; Harrington, 1972). An increasing body of research has shown that a variety of personality characteristics and attitudes has a stronger relationship with risky driving or the propensity to commit trafﬁc offences than with crash involvement. From a review of the literature, Beirness (1993) concluded that personality factors accounted for about 10–20 per cent of the variance in crashes and up to 35 per cent of the variance in risky driving. Some of the most prominent personality factors include sensation seeking, mild social deviance, hostility, aggression, and emotional instability (for example, Jonah, 1997; Lawton et al., 1997; Miles and Johnson, 2003; Patil et al., 2006; Trimpop and Kirkcaldy, 1997). With respect to attitudes and behaviours, a risky driving style, the use of driving to reduce tension or stress and a tolerant attitude towards risky driving behaviour have all been associated with young trafﬁc offenders (for example, Baxter et al., 1990; Beirness et al., 1993; Mayer and Treat, 1977; Ulleberg and Rundmo, 2003). Despite the increasing body of research investigating relationships between personality and attitudinal factors and measures of risky driving (that is, trafﬁc offences) among young drivers, there are a number of limitations associated with this research. Firstly, most of these studies were cross-sectional or retrospective in design, whereby the relationship between personality factors and driver behaviour were measured simultaneously or after driving incidents had occurred. A prospective design is advantageous in that personality measures (especially self-reported) can be obtained before being affected by crash involvement. Secondly, these studies primarily relied on self-reported driver behaviour outcomes. Self-reported crash and trafﬁc offence data allows for the possibility of intentional or unintentional misrepresentation (Elander et al., 1993). A ﬁnal criticism is that many of the studies did not adequately consider the role of driving exposure. Generally, driving exposure varies with age (Massie et al., 1997). However, there can also be considerable variation in the level of driving exposure and travel patterns within different age groups. This is because driving exposure is not a random factor but an individual choice. Driving exposure has been found to vary among young drivers by factors such as sex and motivation for driving (Crettenden et al., 1994; Gregersen and Berg, 1994; Massie et al., 1997). Thus, while personality and attitudes may inﬂuence the way in which an individual chooses to drive, reﬂected in trafﬁc offences, it may also inﬂuence how much an individual drives (quantity of driving exposure). For example, drivers with high levels of sensation seeking might choose to drive more frequently to experience feelings of excitement or drivers with emotional problems or high levels of hostility may choose to drive more frequently to release feelings of tension or stress. The aim of the present study was to identify personality and attitudinal factors that predict subsequent trafﬁc offences, recorded in ofﬁcial driver records, among young drivers. This study also investigated whether any personality and attitudinal factors predicted different levels of driving exposure, deﬁned as self-reported kilometres

Personality and Attitudinal Predictors of Trafﬁc Offences

53

driven. This study contributes to past research on this topic by being the ﬁrst study, to the best of our knowledge, to use a prospective design and ofﬁcial records to examine the role of personality and attitudinal factors among young drivers. Method Participants The original sample consisted of 336 young drivers. A total of 128 drivers were excluded from the study because they did not consent to the release of their driver records. Consequently, the ﬁnal sample consisted of 208 young drivers (169 males, 39 females) aged 16 to 24 years (M = 18.5, SD = 1.2). Participants were recruited from the Driver Intervention Programme, a small-group discussion-based workshop for drivers aged 25 years and under who have violated the conditions of their learner’s permit or provisional licence, resulting in licence disqualiﬁcation. Thus, by deﬁnition, all participants recorded a trafﬁc offence prior to participation in the study. Participants were required to hold a current South Australian provisional driver’s licence ensuring all had some unsupervised driving experience. Participants had held a provisional licence for an average of 1.4 years (sd = 0.94) prior to questionnaire administration. Ethical approval was obtained for this study. Questionnaire Participants completed an extensive self-report questionnaire consisting of 136 items. The measures included in this questionnaire were selected for their known association with risky driving and crash involvement in the literature and primarily for the purposes of another study (see Wundersitz and Burns, 2005). The questionnaire took approximately ten to ﬁfteen minutes to complete. The ﬁrst part of the questionnaire sought information on a number of general demographic, licensing, and background variables including driving exposure (estimated number of kilometres driven). The second section incorporated 72 true or false items measuring general personality traits: assertiveness (ﬁve items: Rathus, 1973), depression (mood rather than clinical symptoms; nine items: Costello and Comrey, 1967), emotional adjustment (six items: Howarth, 1976) and sensation seeking (ten items from the Thrill and Adventure Seeking scale and seven items from the Disinhibition scale; Zuckerman, 1971). In addition, ﬁve measures of the expression of hostility or aggression were included (Buss and Durkee, 1957): assaultiveness (nine items), indirect hostility (ﬁve items), verbal hostility (nine items), irritability (eight items) and resentment (four items). A further 20 true or false items measured a variety of driving-related attitudes and behaviours: driving aggression (ten items: Parry, 1968), an attitude of competitive speed (ﬁve items: Goldstein and Mosel, 1958), driving inhibition (cautious driving when upset or angry; three items: Donovan and Marlatt, 1982) and the extent to which driving

54

Driver Behaviour and Training – Volume III

reduced tension (two items: Mayer and Treat, 1977; Pelz and Schuman, 1971). In following sections, a measure of mild social deviance (eight items: West et al., 1993), self-reported driving style or risky driving (seven items: Deery and Love, 1996) and eight separate items measuring speciﬁc driving attitudes were also incorporated, as was alcohol consumption, which is another measure of high-risk behaviour. The internal consistency of these measures has been established in other research (see Wundersitz and Hutchinson, 2006). Ofﬁcial driver records To obtain ofﬁcial trafﬁc offence records, participants provided their driver’s licence number. Driver licence numbers were used to search the DRIVERS database for trafﬁc offences detected by police on South Australian roads. DRIVERS does not include infringements from speed cameras; thus, the number of trafﬁc offences recorded was certainly an underestimate of the true number of offences. The trafﬁc offence records of participants were tracked for 12 months following questionnaire administration. It is acknowledged that some drivers were disqualiﬁed for part of this period. Statistical analysis Statistical analyses were performed to determine if young drivers recording subsequent trafﬁc offences possessed certain personality characteristics and attitudes. For univariate analyses, chi-square tests were conducted for categorical variables and independent samples t-tests were conducted for continuous variables. Note that if the assumption of normally distributed data was violated, t-tests were performed using Welch’s procedure because it does not assume equal population variances, making the t-test more robust. Cohen’s d, a standardised measure of the effect size or strength of the difference between means, was reported for t-tests with signiﬁcant results. According to Cohen’s guidelines (Cohen, 1988), an effect size of d = 0.2 represents a small effect, d = 0.5 a medium effect and d = 0.8 a large effect. Binary logistic regression was conducted for the multivariate analysis. Logistic regression does not make any assumptions about the statistical distribution of individual drivers’ trafﬁc offence frequency. Results Examination of ofﬁcial driver records showed that 38 per cent (n = 80) of young drivers were detected committing at least one trafﬁc offence during the 12 month period following questionnaire administration. The distribution of the number of trafﬁc offences is shown in Figure 5.1. Just over 14 per cent (n = 30) of young drivers recorded two or more trafﬁc offences in the following year.

Personality and Attitudinal Predictors of Trafﬁc Offences

Figure 5.1

55

Distribution of the number of trafﬁc offences recorded after questionnaire administration

Table 5.1 shows the demographic characteristics and driving exposure of young drivers with and without a recorded trafﬁc offence during the 12 month follow up period. Drivers recording a trafﬁc offence were statistically signiﬁcantly more likely to be male (43 per cent) than female (21 per cent) (χ2(1) = 6.5, p = 0.011). Trafﬁc offence status was not related to age. Driving exposure was measured in terms of the estimated number of kilometres driven per year. Young drivers recording at least one trafﬁc offence reported driving more kilometres per year than drivers without a trafﬁc offence (t(100) = 2.7, p = 0.007).

Table 5.1

Background variables for young drivers recording and not recording a subsequent trafﬁc offence

Variables Sex (%) Males Females Age (years) (sd) Kilometres driven per year (sd)

None

At least one

p-value

42.6 20.5 18.6 (1.4) 21,364.5 (19,633.8)

0.011

79.5 18.4 (1.1) 14,172.8 (12,136.8)

0.205 0.007

Driver Behaviour and Training – Volume III

56

Kilometres driven per year (driving exposure) Personality characteristics and attitudes may inﬂuence driving exposure or the number of kilometres driven per year. To investigate this possibility, a linear regression procedure was performed with kilometres driven per year as the dependent variable and all other personality and attitude measures, including sex, as predictor variables. A stepwise procedure was used for the analysis, with the level of signiﬁcance required for entry into the equation set at p < 0.05. The results of this procedure, summarised in Table 5.2, indicated that a risky driving style and the use of driving to reduce tension were the two best predictors of kilometres driven per year. However, the model featuring these two variables accounted for only 7 per cent of the variance (Adjusted R squared). Table 5.2

Results of a linear regression predicting kilometres driven per year, using personality and attitudinal measures as predictors (N = 179)

Variables in model (order of entry)

B

Adj R2

Risky driving style 494.68 0.05 Tension reduction 3129.40 0.07 Note: Final model F(178) = 7.97, p < 0.001

β

t

p-value

0.19 0.17

2.61 2.35

0.010 0.020

Trafﬁc offences To determine if young drivers recording subsequent trafﬁc offences were characterised by certain personality measures and attitudes, their mean scores on such measures were compared to drivers who did not record a trafﬁc offence (see Table 5.3). There were no statistically meaningful differences in the means of personality measures for drivers with and without trafﬁc offences. Analysis of hostility measures indicated drivers recording a trafﬁc offence had higher levels of assaultiveness (d = 0.32) but lower levels of indirect hostility (d = 0.28) than drivers without trafﬁc offences. The effect sizes indicate that these differences were small. With respect to the driving-related measures, several differences were found; drivers recording a trafﬁc offence reported higher levels of competitive speed, used driving to reduce tension, and had a riskier driving style. The corresponding effect sizes for these measures were in the small to medium range (d = 0.44, d = 0.38, d = 0.39, respectively). The attitudinal measures, speciﬁc to road safety, suggested that drivers recording a trafﬁc offence thought speeding was acceptable (d = 0.32). To determine whether any personality characteristics or attitudes predicted subsequent trafﬁc offences, all measures that differed by trafﬁc offence record in univariate analyses were entered into a logistic regression (dependent variable: at least one offence or no offences). Sex and kilometres driven per year were also

Personality and Attitudinal Predictors of Trafﬁc Offences

Table 5.3

57

Mean scores on selected personality and attitudinal measures for drivers recording subsequent trafﬁc offences and no subsequent trafﬁc offences (N = 208) At least one offence (n = 80)

No offences (n = 128)

Measure Mean SD Mean SD t-value df p-value Personality Assertiveness 7.94 1.32 7.83 1.33 0.58 206 0.563 Depression 10.48 1.84 10.05 1.81 1.62 206 0.107 Emotional 7.49 1.49 7.49 1.55 0.02 206 0.983 adjustment Sensation seeking 26.81 3.28 26.67 3.52 0.29 206 0.774 Mild social 12.29 2.96 11.80 2.86 1.19 206 0.236 deviance Hostility and aggression Assaultiveness 13.85 1.96 13.18 2.22 2.22 206 0.028 Indirect hostility 7.59 1.29 7.97 1.36 2.01 206 0.046 Verbal hostility 13.86 1.71 13.68 1.82 0.72 206 0.472 Irritability 11.44 1.71 11.25 2.01 0.69 206 0.489 Resentment 5.66 1.21 5.42 1.12 1.46 206 0.146 Driving-related Aggression 13.30 2.52 12.92 2.60 1.03 206 0.303 Competitive speed 7.74 1.62 7.00 1.72 3.08 206 0.002 Inhibition 4.25 1.15 4.55 1.13 1.88 206 0.062 Tension reduction 3.44 0.76 3.11 0.92 2.80 189 0.006 Risky driving style 20.27 6.10 17.87 6.19 2.75 206 0.007 Attitudes1 Speeding 2.93 1.29 2.52 1.29 2.18 206 0.030 acceptable Drink driving 2.45 1.56 2.59 1.72 0.62 180 0.535 acceptable Low risk of dying 1.88 1.16 1.82 1.18 0.33 206 0.744 in crash Friends don’t drive 3.14 1.28 3.13 1.18 0.07 206 0.943 safely Low likelihood of 2.35 1.20 2.44 1.31 0.48 206 0.629 being caught Lack of concern 1.25 1.61 1.17 1.34 206 0.183 for hurting others Poor driving skill 2.03 1.07 2.05 1.01 0.20 206 0.840 Low safety 2.09 1.07 2.11 1.05 0.15 206 0.885 motivation Note: For each measure, higher scores indicate higher levels of the variable, except for emotional adjustment where higher scores indicate lower levels of adjustment. 1 For each attitude measure, higher scores indicate non-safety orientated attitudes.

58

Driver Behaviour and Training – Volume III

included as predictor variables because group differences were found in univariate analyses. The results from the logistic regression, presented in Table 5.4, show that the probability of recording at least one subsequent trafﬁc offence increased with higher levels of driving to reduce tension, independent of kilometres driven. The odds ratio indicated that drivers who used driving to reduce tension were 1.5 times as likely to record a subsequent trafﬁc offence. No other personality measures entered into the logistic regression predicted trafﬁc offences. Table 5.4

Measure

Results of logistic regression analysis for predicting at least one subsequent trafﬁc offence, using personality and attitude measures as predictors (N = 179) B

SE

Wald

p-value

Tension reduction 0.42 0.19 4.80 0.028 Kilometres driven per year < 0.01 < 0.01 5.34 0.021 Note: Data for kilometres driven were missing for 29 participants.

Odds ratio

95% CI

1.52 1.00

1.05–2.21 1.00–1.00

A positive regression coefﬁcient for kilometres driven indicates that the probability of recording a trafﬁc offence increased with the greater number of kilometres driven. This model was statistically signiﬁcant (χ2(2) = 13.8, p = 0.001). Discussion Understanding which personality factors and attitudes predict driver behaviour might assist in matching interventions to the individual needs of young drivers. Thus, the primary purpose of the present study was to identify personality and attitudinal factors associated with trafﬁc offences among young drivers using ofﬁcial driver records. By using a prospective design, we attempted to minimise the effects of any crash experience on self-reported measures of individual differences. Self-reported measures provide an opportunity for drivers to give a ‘good’ or socially desirable account of themselves, particularly when the traits and behaviours assessed are undesirable (for example, hostility). Based on the ﬁndings from this study, a ﬂow chart showing the predictors of trafﬁc offences, incorporating kilometres driven per year, is depicted in Figure 5.2. The results of this study showed that trafﬁc offences, serving as a proxy for crashes, were predicted by kilometres driven per year and a driving-related behavioural measure, driving to reduce tension. The fact that relationships between variables and trafﬁc offences reported in ofﬁcial records were found is notable given that the offences reported in driver records are relatively rare and under-represent the actual number of risky driving behaviours performed (that is, they contain only the number of times a driver was detected offending).

Personality and Attitudinal Predictors of Trafﬁc Offences

59

Risky driving style

Driving to reduce tension

Kilometres driven per year

Traffic offences

Figure 5.2 Flow chart of predictors of trafﬁc offences for young drivers In contrast to previous research using self-reported driving behaviour, none of the speciﬁc personality variables were associated with trafﬁc offences. Although there are many advantages of using ofﬁcial driver records over self-reported data, there are some limitations associated with ofﬁcial records that affect their ability to detect relationships. Zylman (1972) argued that research based solely on ofﬁcial driver records may yield spurious results and, in many cases, non-signiﬁcant results because the likelihood of recording a crash or trafﬁc offence may be more dependent on local policies and practices than the driver’s proﬁciency or driving behaviour. Moreover, not all trafﬁc offences are enforced equally and this may bias the data such that some groups of drivers are over-represented (Smiley et al., 1991). An interesting ﬁnding of this study was that using driving to reduce tension predicted kilometres driven per year and trafﬁc offences (independent of kilometres driven per year). Social learning theory suggests that if an individual has not learnt sufﬁcient means of coping with tension or frustrations, driving may be used as a way of venting these feelings (Grey et al., 1989). The ﬁndings of the present study are consistent with past studies that found the use of driving to release tension was associated with trafﬁc offences and crash involvement, particularly among males (Donovan et al., 1985; Harano et al., 1975; Mayer and Treat, 1977). The use of driving to reduce tension is not a personality trait but a behavioural manifestation of such traits in the driving context that has been learned and is, therefore, more amenable to change. Consequently, it may be beneﬁcial to develop interventions or public education for young drivers that highlight the importance of using effective strategies to deal with feelings of tension or stress, other than on the road. The ﬁnding that kilometres driven per year predicted trafﬁc offences is consistent with previous research (for example, Trimpop and Kirkcaldy, 1997). High levels of driving exposure or kilometres driven has consistently been correlated with trafﬁc offences because greater driving exposure allows greater opportunity to commit, and be detected committing, a trafﬁc offence. This ﬁnding reinforces the view that kilometres driven should be included as a covariate when examining factors associated with driver behaviour. However, note that the measure of driving exposure in this study was based on self-reported estimates of kilometres driven. By

60

Driver Behaviour and Training – Volume III

nature, self-reported data are unreliable because they are subject to poor recall and misrepresentation. The association between risky driving style (that is, the manner in which one chooses to drive), the use of driving to reduce tension and kilometres driven per year suggests greater distance driven represents, to a small extent, unsafe or risky motives for driving. This notion is consistent with some previous research. For example, Gregersen and Berg (1994) found that a group of high-risk young drivers that reported more driving than other groups were characterised by an interest in cars, being ‘out and about’ and driving for extra motives other than transport. Similar to the present study, the majority of the group were male (about 80 per cent). Alternatively, young drivers’ risky driving style and resultant higher kilometres driven per year may serve purposes associated with adolescent development not examined in this study, such as opposing authority, asserting independence and impressing peers (Jessor et al., 1997). Future research could explore the nature of this interesting relationship. Nevertheless, whether kilometres driven is an expression of masculinity, an interest in cars or a claim to adulthood, the modiﬁcation of these motivations for driving might reduce the amount of kilometres driven by young drivers, resulting in crash reductions. The prospective design of this study provides an opportunity to continue following the driver records of these young drivers for two to three years to accumulate a greater number of recorded trafﬁc incidents (but bearing in mind that crash risk is not stable and varies with age). A prospective examination of the characteristics of young drivers detected for several offences or crashes would provide a convincing means of understanding the role of personality characteristics and attitudes in driver behaviour among young drivers. Consequently, information from this research would be valuable in further developing and tailoring interventions to the individual needs of young drivers. Acknowledgements I wish to thank Dr Paul Hutchinson for providing useful comments on this paper. References Baxter, J.S., Manstead, A.S.R., Stradling, S.G., Campbell, K.A., Reason, J.T. and Parker, D. (1990). ‘Social facilitation and driver behaviour.’ British Journal of Psychology, 81: 351–60. Beirness, D.J. (1993). ‘Do we really drive as we live? The role of personality factors in road crashes.’ Alcohol, Drugs and Driving, 9(3–4): 129–43. Beirness, D.J., Simpson, H.M. and Mayhew, D.R. (1993). ‘Predicting crash involvement among young drivers.’ In H.D. Utzelmann, G. Berghaus and G. Kroj (eds). Proceedings of the International Conference on Alcohol, Drugs and Trafﬁc Safety. Cologne: Verlag TUV Rhineland. Burg, A. (1970). ‘The stability of driving record over time.’ Accident Analysis and

Personality and Attitudinal Predictors of Trafﬁc Offences

61

Prevention, 2: 57–65. Buss, A.H. and Durkee, A. (1957). ‘An inventory for assessing different kinds of hostility.’ Journal of Consulting and Clinical Psychology, 21: 343–9. Cohen, J. (1988). Statistical Power Analysis for the Behavioural Sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Costello, C.G. and Comrey, A.L. (1967). ‘Scales for measuring depression and anxiety.’ Journal of Psychology, 66: 303–13. Crettenden, A.V., Yeo, E.-Y. and Drummond, A.E. (1994). Qualitative Dimensions of Young Driver Driving Exposure as a Function of Time of Day (No. CR 148). Canberra: Federal Ofﬁce of Road Safety. Deery, H.A. and Love, A.W. (1996). ‘The driving expectancy questionnaire: development, psychometric assessment and predictive utility among young drinkdrivers.’ Journal of Studies on Alcohol, 57: 193–202. Donovan, D.M. and Marlatt, G.A. (1982). ‘Personality subtypes among drivingwhile-intoxicated offenders: relationship to drinking behaviour and driving risk.’ Journal of Consulting and Clinical Psychology, 50(2): 241–9. Donovan, D.M., Queisser, H.R., Salzburg, P.M. and Umlauf, R.L. (1985). ‘Intoxicated and bad drivers: subgroups within the same population of high risk men drivers.’ Journal of Studies on Alcohol, 46: 375–82. Elander, J., West, R. and French, D. (1993). ‘Behavioral correlates of individual differences in road-trafﬁc crash risk: an examination of methods and ﬁndings.’ Psychological Bulletin, 113(2): 279–94. Epstein, S. (1979). ‘The stability of behaviour: on predicting most of the people much of the time.’ Journal of Personality and Social Psychology, 37: 1097–126. Evans, L. (2004). Trafﬁc Safety. Bloomﬁeld Hills, MI: Science Serving Society. Friedstrom, L., Ifver, J., Ingebrigtsen, S., Kulmala, R. and Thomsen, L.K. (1995). ‘Measuring the contribution of randomness, exposure, weather and daylight to the variation in road accident count.’ Accident Analysis and Prevention, 27: 1–20. Goldstein, L.G. and Mosel, J.N. (1958). ‘A factor study of drivers’ attitudes, with further study on driver aggression.’ Highway Research Board Bulletin, 172: 9–29. Gregersen, N.P. and Berg, H.Y. (1994). ‘Lifestyle and accidents among young drivers.’ Accident Analysis and Prevention, 26: 297–303. Grey, E.M., Triggs, T.J. and Haworth, N.L. (1989). Driver Aggression: The Role of Personality, Social Characteristics, Risk and Motivation. (No. CR 81). Clayton, Victoria: Monash University. Harano, R.M., Peck, R.C. and McBride, R.S. (1975). ‘The prediction of accident liability through biographical data and psychometric tests.’ Journal of Safety Research, 7: 16–52. Harrington, D.M. (1972). ‘The young driver follow-up study: an evaluation of the role of human factors in the ﬁrst four years of driving.’ Accident Analysis and Prevention, 4: 191–240. Howarth, E. (1976). ‘A psychometric investigation of Eysenck’s personality inventory.’ Journal of Personality Assessment, 40(2): 173–85. Jessor, R., Turbin, M.S. and Costa, F.M. (1997). ‘Predicting developmental change in risky driving: the transition to young adulthood.’ Applied Developmental Science,

62

Driver Behaviour and Training – Volume III

1(1): 4–16. Jonah, B. A. (1997). ‘Sensation seeking and risky driving.’ In T. Rothengatter and E. Carbonell Vaya (eds). Trafﬁc and Transport Psychology – Theory and Application. Great Britain: Pergamon: 259–67. Lawton, R., Parker, D., Stradling, S.G. and Manstead, A.S.R. (1997). ‘Predicting road trafﬁc accidents: the role of social deviance and violations.’ British Journal of Psychology, 88(2): 249–62. Legge, M., Kirov, C. and Cercarelli, R. (2000). Reported Road Crashes in Western Australia. Perth: Road Safety Council of Western Australia 2000. Massie, D.L., Green, P.E. and Campbell, K.L. (1997). ‘Crash involvement rates by driver gender and the role of average annual mileage.’ Accident Analysis and Prevention, 29(5): 675–85. Mayer, R. and Treat, J. (1977). ‘Psychological, social and cognitive characteristics of high-risk drivers: a pilot study.’ Accident Analysis and Prevention, 9(1): 1–8. Miles, D.E. and Johnson, G.L. (2003). ‘Aggressive driving behaviors: are there psychological and attitudinal predictors?’ Transport Research Part F, 6(2): 147– 61. Parry, M.H. (1968). Aggression on the Road. London: Tavistock Press. Patil, S.M., Shope, J.T., Raghunathan, T.E. and Bingham, C.R. (2006). ‘The role of personality characteristics in young adult driving.’ Trafﬁc Injury Prevention (7). Pelz, D.C. and Schuman, S.H. (1971). ‘Are young drivers really more dangerous after controlling for exposure and experience?’ Journal of Safety Research, 3(2): 68–79. Rathus, S.A. (1973). ‘A 30 item schedule for assessing assertive behavior.’ Behavior Therapy, 4: 398–406. Shinar, D. (1978). Psychology on the Road – The Human Factor in Trafﬁc Safety. New York: Wiley. Shope, J.T., Waller, P.F., Raghunathan, T.E. and Patil, S.M. (2001). ‘Adolescent antecedents of high-risk driving behaviour into young adulthood: substance use and parental inﬂuences.’ Accident Analysis and Prevention, 33(5): 649–58. Smiley, A., Hauer, E., Persaud, B., Clifford, L. and Duncan, D. (1991). Accident Potential: An Ontario Driver Records Study Summary Report. Ontario: Ministry of Transportation. Struckman-Johnson, D.L., Lund, A.K., Williams, A.F. and Osborne, D.W. (1989). ‘Comparative effects of driver improvement programs on crashes and violations.’ Accident Analysis and Prevention, 21(3): 203–15. Trimpop, R. and Kirkcaldy, B. (1997). ‘Personality predictors of driving accidents.’ Personality and Individual Differences, 23(1): 147–52. Ulleberg, P. and Rundmo, T. (2003). ‘Personality, attitudes and risk perception as predictors of risky driving behaviour among young drivers.’ Safety Science, 41: 427–43. West, R., Elander, J. and French, D. (1993). ‘Mild social deviance, type-a behaviour pattern and decision-making style as predictors of self-reported driving style and trafﬁc accident risk.’ British Journal of Psychology, 84: 207–19. Wundersitz, L.N. and Burns, N. (2005). ‘Identifying young driver subtypes: relationship to risky driving and crash involvement.’ In L. Dorn (ed.). Driver

Personality and Attitudinal Predictors of Trafﬁc Offences

63

Behaviour and Training (vol. II: 155–68). Aldershot, UK: Ashgate. Wundersitz, L.N. and Hutchinson, T.P. (2006). South Australia’s Driver Intervention Program: A Discussion of Best Practice, With Data on Characteristics of the Participants and a Selective Literature Review (No. CASR021). Adelaide, South Australia: Centre for Automotive Safety Research. Zuckerman, M. (1971). ‘Dimensions of sensation seeking.’ Journal of Consulting and Clinical Psychology, 36: 45–52. Zylman, R. (1972). ‘Drivers’ records: are they a valid measure of driving behavior?’ Accident Analysis and Prevention, 4: 333–49.

This page intentionally left blank

Chapter 6

Pre-driving Attitudes and Non-driving Road User Behaviours: Does the Past Predict Future Driving Behaviour? Helen N. Mann1 and Mark J.M. Sullman2 1 Heriot-Watt University, UK 2 Hertfordshire University, UK Introduction The number of fatalities and injuries on our roads continues to increase at an alarming rate. However, this is not just a problem for Britain and Europe, it is a global concern. It is estimated that every year 1.2 million people die in road accidents worldwide and somewhere in the region of 20–50 million people are known to be injured or permanently disabled (WHO, 2004; WHO, 2007). According to projections made by the World Health Organisation (WHO), these ﬁgures are likely to increase by about 65 per cent over the next 20 years unless preventative action is taken (WHO, 2004). Over the past four years road safety has become a priority for Government agencies around the world. In 2004, the WHO for the ﬁrst time ever selected road safety as the topic for World Health Day (WHD). The aim of this day was to raise global awareness about the magnitude, risk factors and impacts of road trafﬁc accidents. More recently the United Nations (UN) brought road safety back into the media spotlight by declaring 23–29 April 2007 Global Road Safety Week (GRSW). The focal point for the GRSW was speciﬁcally the problems that face young road users, unlike the WHD, which looked at general road safety. It is estimated that just over 1000 young people are killed every day in road trafﬁc collisions worldwide and that more than 30 per cent of all people killed and seriously injured in road accidents are under 25 years (WHO, 2007). With the number of young people involved in road trafﬁc accidents on the increase, it is unsurprising to learn that road trafﬁc injuries have been ranked as the eighth leading cause of global deaths among young people (WHO, 2007). Several factors are thought to affect the vulnerability of young pedestrians, cyclists and drivers, with some of the main issues relating to development, social inﬂuence and personality. Developmental factors can affect young road users’ safety on the road in many different ways. Firstly, children and adolescents are often smaller in stature than adults, thus drivers do not always see them; secondly, their perceptive skills have not yet fully developed. Under-developed depth perception can, for example, impair an individual’s ability to accurately judge the distance between themselves and objects in their environment and immature auditory perception can impair their

66

Driver Behaviour and Training – Volume III

ability to accurately judge from the sounds of their engines the direction and size of approaching vehicles. Finally, young road users often have a limited attention span which can mean that they are often easily distracted, ﬁnding it difﬁcult to attend to more than one problem simultaneously (WHO, 2007). These are of course important skills to master in order to navigate the road environment safely. Social factors such as peer and parental pressure can also have an impact on road user behaviour. Adolescent pedestrians and cyclists have reported that the more they go out with their friends the less often they carry out desirable road user behaviours (Elliott and Baughan, 2003). Those who report being inﬂuenced more by their parents than their peers have been shown to display more desirable behaviours on the road (Elliott and Baughan, 2003; Jessor, Turbin and Costa, 1997). For example, young drivers who report being inﬂuenced more by their parents also report less risky driving behaviour (Jessor, Turbin and Costa, 1997). Personality factors such as risk-taking (or sensation seeking) have also been linked to risky road user behaviour and poor road safety attitudes. People with a high preference for risk-taking have been found to report more negative attitudes to road safety and more frequent occurrences of risk-taking whilst driving (Iversen, 2004; Ulleberg and Rundmo, 2003). Among pre-drivers, sensation seeking and adherence to social values were associated with risky road user behaviour (West, Train, Junger, Pickering, Taylor and West, 1998). Adolescents who report high sensation seeking and low adherence to social values were those who also reported riskier behaviour on the roads. In addition to links between sensation seeking and risky driving or road user behaviour, Arnett (1994) has reported a correlation between sensation seeking and several other risky behaviours in adolescents, such as drug-taking. In social psychology it has been postulated that problem behaviours are linked to form a ‘syndrome’ of behaviours that characterise adolescence (Jessor, 1987), known as the Problem Behaviour Theory (PBT) and Arnett’s (1994) results would appear to provide supportive evidence for this theory. Support of Jessor’s theory has also been lent by the results of a study conducted in Australia by Wundersitz and Burns (2005). This research looked at personality characteristics that deﬁned highrisk young drivers and discovered that there were four different driver sub-types. The sub-type of drivers with the highest road safety risk were characterised by high levels of driving-related aggression, competitive speed, driving to reduce tension or increase personal efﬁcacy and assertiveness. The results also reported positive scores on sensation-seeking, positive attitudes towards speeding and that they engaged in other high-risk behaviours such as drinking large quantities of alcohol. Risky driving is not the only form of road user behaviour that has been linked to engagement in other forms of problem behaviour. West et al. (1998) provided evidence to suggest that there was an additional link between problem behaviour and pedestrian behaviour. These results reported the existence of a signiﬁcant relationship between engagement in problem behaviour (in the form of risk-taking) and increased pedestrian accident involvement. Having identiﬁed these links between problem behaviour, risky road user behaviour and high accident risk, the question arises as to whether or not there is any linkage contained within them. In an effort to understand the young driver problem and to aid the development of future road safety initiatives, this research aims

Pre-driving Attitudes and Non-driving Road User Behaviours

67

to establish a link between adolescents’ pre-driver road user behaviour, attitudes and future driving behaviour. It also aims to identify both pre-driver and driver characteristics that predict risky driving behaviour. Providing the research aims are met, the necessity for the designs of pre-driver interventions which create safe attitudes to both driving and general road safety will be emphasised. Method Participants This is the second in a two-part longitudinal study conducted in New Zealand. Four hundred and seventy-one students (196 males, 275 females) aged 14–16 years (mean age 15 years) took part and fully completed both parts of the study. All participants were non-drivers at Time 1, but by Time 2 there were 208 drivers. When investigating the existence of a link between pre-driver attitudes and behaviour with post-driver attitudes and behaviour, the questionnaires were ﬁltered so that only the responses of participants who were drivers by Time 2 were analysed. Out of the 208 drivers, only 195 (97 males, 98 females) had fully completed the Driver Behaviour Questionnaire (DBQ), which was the main source of information regarding their driving behaviour. Therefore, statistical analysis was conducted only on the responses of the 195 participants aged 15–16 (M = 15.1yrs, SD = 0.24). Participants were from 29 schools across New Zealand (ten South Island schools, 19 North Island schools). They were selected to participate in the study at random by teachers within their schools. Questionnaires This longitudinal study looks at individuals’ responses to two questionnaires completed approximately twelve months apart, using several different scales of measurement. The ﬁrst questionnaire consisted of three sections, including items from: the Adolescent Road User Behaviour Questionnaire (ARBQ), the Theory of Planned Behaviour applied to speeding (referred to as TPB-speeding), two of the Driver Attitudes Questionnaire (DAQ) subscales and one new DAQ subscale designed in this study to measure attitudes to seat belts. Adolescent road user behaviour was measured using the Adolescent Road User Behaviour Questionnaire (ARBQ), designed by Elliott and Baughan (2003, 2004). The scale measured three types of road user behaviour: unsafe road crossing behaviour, play and social activity on the road, and engagement in protective behaviour on the road. Responses were made on a ﬁve point Likert scale where 1 = ‘never’ and 5 = ‘very often’. To ensure that increasing scores on all three sub-scales were indicative of high risk behaviour, the items measuring protective behaviour on the road were reversed and re-labelled ‘non-engagement of protective behaviour on the road’ (that is, the higher the score, the less protective behaviour they engage in). Dangerous road users in this study were therefore those participants who scored high on each

68

Driver Behaviour and Training – Volume III

of the three subscales. The original ARBQ was composed of a 43 item scale but, for the purposes of this study, the reduced 23 item scale was used (as recommended by the authors of the study). Components of the TPB-speeding were examined to try and identify the attitudes and intentions of pre-drivers to speeding in the future. The extended TPB scale designed by Parker, Manstead and Stradling (1995) was used to measure components of the original TPB. They broadened the TPB model to include a new measure of personal norm, which included items on moral norm and anticipated regret. As participants at Time 1 were non-drivers, direct measures as opposed to beliefbased measures were used; that is, they could not be asked about control beliefs and power beliefs regarding driving behaviour. In total there were ten questions measuring 13 items (one moral norm question, two anticipated regret questions, one attitude question with four parts to it, two subjective norm questions, three perceived behavioural control questions and one measure of intention were used). Responses were made on a seven point Likert scale where 1 = ‘strongly agree’ and 7 = ‘strongly disagree’. The Driver Attitude Questionnaire (DAQ) was designed to measure attitudes to bad driving practices, such as speeding and drink-driving (Parker, Manstead, Stradling and Senior, 1998). The original DAQ consisted of four sub-sections, with ten items in each measuring attitudes towards: drink-driving, speeding, over-taking and close following. In this study there were 22 questions (nine speeding items, eight drink-driving items and ﬁve seat belt items), 17 of these measured attitudes to speeding and drink-driving (16 being taken from Parker, Manstead, Stradling and Senior’s (1998) Driver Attitude Questionnaire [DAQ]). In the DAQ there were ten items on drink-driving, however in this study only seven were chosen (three items measuring attitudes towards breath-testing and knowledge of the legal blood alcohol limit were removed because it was assumed that as participants were under the legal drinking age they may not be aware of alcohol limits and laws). A new item, ‘it’s ok to drink and drive’ was added, as it was felt that pre-drivers would be able to respond. Similarly, there were ten items on speeding in the original DAQ but only nine were chosen for this study (the item ‘I know exactly how fast I can drive and still drive safely’ was omitted because it was considered that it was not applicable to pre-drivers). As well as the two subscales measuring speeding and drink-driving, a third subscale, consisting of ﬁve items, was created to measure attitudes towards not wearing seat belts. Some of the questions used in the other two sections on speeding and drink-driving were re-worded so that they could be applied to attitudes to not wearing seat belts. Responses were made on a ﬁve point Likert scale where 1 = ‘strongly agree’ and 5 = ‘strongly disagree’. The second questionnaire consisted of three sections using eight items from the ﬁrst questionnaire, Arnett’s Inventory of Sensation Seeking (AISS) and the Driver Behaviour Questionnaire (DBQ). To establish whether attitudes and intentions changed during the intervening year’s period, as the adolescents went from being pre-drivers to drivers, eight items from the ﬁrst questionnaire were repeated in the second questionnaire. Four of the items that measured attitudes to speeding were included (responses to the four items

Pre-driving Attitudes and Non-driving Road User Behaviours

69

were added together and averaged), along with one item on speeding intention. Three items from the DAQ measuring attitude towards speeding, drink-driving and non-use of seat belts were also included. Arnett’s Inventory of Sensation Seeking (AISS) was designed to measure sensation seeking level (Arnett, 1994); the AISS scale consisted of 20 items in total, further sub-divided into two subscales, namely Novelty (the need to seek novel stimulation) and Intensity (the need to seek intense experiences). Both subscales, containing ten items each, were aggregated to ﬁnd a total score. Responses were made on a four point Likert scale where 1 = ‘does not describe me at all’ and 5 = ‘describes me very well’. High scores on both the sub-scales and on total AISS score (the combined Novelty and Intensity scores) were indicative of high sensation seekers. The DBQ was designed to measure the frequency of drivers’ self-reported behaviour on the road. There are several different versions of the scale, but the version chosen for this study consisted of 24 items measuring driving errors, lapses and violations (all violations were Highway Code violations, aggressive violations were not included). Responses were measured on a six point Likert scale where 0 = ‘never’ and 5 = ‘all the time’. Results This study focuses on the results from the second questionnaire and the data collected from those subjects in the longitudinal study who had become drivers, tests comparing responses from non-drivers and drivers at Time 2 were also conducted. The data were analysed using t-tests, ANOVAs and correlations. Regressions were also run which incorporated items from both questionnaires, in order to establish whether or not a relationship existed between self-reported driving behaviour and attitudes at Time 2 and pre-driver road user behaviour and attitudes at Time 1. The results showed that there were signiﬁcant sex differences on the second questionnaire, but there were no signiﬁcant differences according to the location of participants (North or South Island) or the area they lived in (City/town/rural). Signiﬁcant sex differences were found on both the AISS sensation seeking scale and the DBQ self-reported driving behaviour scale. Males scored signiﬁcantly higher than females on overall Sensation Seeking score (t (193) = 3.39, p < 0.001) and on Intensity (t (193) = 4.55, p < 0.001). Females, however, scored signiﬁcantly higher than males on driving errors (t (193) = –2.39, p < 0.05) and lapses (t (178) = –3.46, p < 0.001). There were no signiﬁcant sex differences regarding the eight TPB speeding items asked about in both questionnaires. Correlations that were run on items from the second questionnaire revealed that driving violations were signiﬁcantly correlated with AISS Sensation Seeking scores (r = 0.29, p < 0.001), Intensity scores (r = 0.21, p < 0.05) and Novelty scores (r = 0.26, p < 0.001). Lapses were also signiﬁcantly correlated with Novelty (r = 0.17, p < 0.05). A correlation matrix of all items from both questionnaires revealed that, with regard to the three DAQ questions about attitudes (to speeding, drink-driving and

70

Driver Behaviour and Training – Volume III

not wearing seat belts), there were correlations with Intensity on the AISS, play and social activity on the road on the ARBQ, non-engagement in protective behaviour also on the ARBQ and scores on some of the TPB-speeding items (speeding moral norm, anticipated regret, attitude and subjective norm). DAQ attitudes to speeding and drink-driving scores also correlated with scores on the DBQ violations subscale, the ARBQ unsafe road crossing subscale and the TPB speeding perceived behavioural control and speeding intention items. The correlation also showed that scores on speeding intention at Time 1 (predriving) correlated with scores on driving violations on the DBQ (r = 0.28, p < 0.001) and scores on TPB speeding intention at Time 2 (post-driving) (r = 0.30, p < 0.001). Speeding intention at Time 2 (post-driving) also correlated with scores on driving violations on the DBQ (r = 0.44, p < 0.001), driving errors on the DBQ (r = 0.21, p < 0.05), overall sensation seeking on the ARBQ (r = 0.24, p < 0.001), AISS Intensity (r = 0.20, p < 0.05) and Novelty (r = 0.20, p < 0.05). Three regressions were run to ﬁnd those questionnaire items that signiﬁcantly predicted scores on DBQ errors, lapses and violations from both Time 1 and Time 2. The best predictors for driving lapses were sex and score on the ARBQ play and social activity on the road subscale (measured at Time 1, pre-driving), which were factors that were also the best predictors of driving errors. The regression to predict driving violations, however, revealed that scores on the ARBQ subscales of unsafe road crossing behaviour and play and social activity on the road (again measured at Time 1, pre-driving) were the best predictors. A fourth regression was run on the data to look at the effect of age, sex and predriving attitudes (attitudes to drink-driving, speeding and not wearing seat belts), measured at Time 1 pre-driving, and at predicting driving violation scores at Time 2. The results showed that the only signiﬁcant predictor of driving violation scores were scores on attitudes to speeding. To see if there was a signiﬁcant difference in response between Time 1 and Time 2, paired samples t-tests were run on the data collected from the eight TPB-speeding questions that were asked in both questionnaires. The ﬁrst statistical analysis that was run focused on responses collected from all participants who completed both questionnaires regardless of whether or not they drove by Time 2 (n = 471). The results showed that there were signiﬁcant differences in responses between the two time periods. Attitudes to speeding (that is, reckless/cautious, bad/good), DAQ attitudes (attitudes to speeding, not wearing seat belts and drink-driving all being acceptable) and intentions to speed mean scores dropped signiﬁcantly from Time 1 to Time 2 (P < 0.001). Changes in responses over the two time periods for the speeding attitude items (unsafe/safe and un-enjoyable/enjoyable) were however not signiﬁcant. The second paired samples test looked at the difference between responses made at Time 1 and Time 2 according to whether or not participants drove by Time 2. The results showed that whilst mean responses for both drivers (n = 207) and non-drivers (n = 262) dropped signiﬁcantly from Time 1 to Time 2, scores on attitude to speeding being un-enjoyable/enjoyable increased signiﬁcantly for both groups.

Pre-driving Attitudes and Non-driving Road User Behaviours

71

An independent sample t-test run on each of the eight questions at Time 2 comparing responses of driving and non-driving participants did not reveal any signiﬁcant differences. Discussion The results from this study have highlighted young driver sex differences, attitude changes and links between their pre-driver attitudes and behaviours with their current driving behaviour. The research aims have thus been met and have successfully identiﬁed the need for the implementation of pre-driver interventions that create safe attitudes to both driving and general road safety. The results have conﬁrmed that young male drivers are higher sensation seekers than female drivers and seek out intense experiences, which could explain why young males are particularly over-represented in road accidents. In view of the fact that both Novelty and Intensity in sensation seeking were shown to correlate with engagement in driving violations, and that studies have reported that violations are associated with increased crash involvement, perhaps interventions need to be designed to target high sensation seekers in particular. Attitudes and speeding intentions for all participants decreased from Time 1 to Time 2, indicating that there were no signiﬁcant differences between drivers and non-drivers responses in the second questionnaire. However, although both attitudes and intention to speed decreased and became more desirable over the year period, their attitude towards speeding being enjoyable increased. These results were surprising as it was anticipated that attitudes and intentions would increase with age and driving exposure, as both driver and passenger. Speeding intention scores at both sampling points were correlated with driving violations, errors and sensation seeking. Therefore, although intention scores decreased over the course of the study, participants with the highest scores at Time 1 and Time 2 were those most likely to engage in violations and errors whilst driving. This link between pre-driver speeding intention and future engagement in driving violations again highlights the need for intentions to engage in safe driving practices to be ingrained in adolescents before they learn to drive. Pre-driver attitudes and behaviours were predictive of driving behaviour. Attitudes towards speeding, drink-driving and not wearing seat belts were correlated with unsafe road crossing, play and social activity and non-engagement in protective behaviour on the road. These results suggest that there are links between positive attitudes to driving violations and pre-driver road user behaviour. In the group who became drivers, their pre-driver attitudes towards speeding and drink-driving along with their road user behaviour, speciﬁcally unsafe road crossing behaviour, correlated with self-reported engagement in driving violations. These results were further supported by the outcomes of several regressions, which revealed that the best predictors of future engagement in driving violations were pre-driver unsafe road crossing and play and social activity on the roads. Driving lapses and errors were best predicted by engagement in play and social activity on the road as predrivers and the gender of the driver. Females, for example, reported signiﬁcantly

72

Driver Behaviour and Training – Volume III

more errors and lapses whilst driving. Interventions thus need to not only aim at changes in pre-driver attitudes but also at pre-driver behaviour. High-risk adolescent road users could thus be targeted through road safety campaigns. This study has thus provided evidence to support Jessor’s Problem Behaviour Theory, in that one form of problem behaviour, namely risky adolescent road user behaviour, was associated with another form of problem behaviour in the form of risky driving behaviour. The results also lend support to West et al.’s (1998) study, as those participants scoring high on Sensation Seeking reported riskier behaviour on the roads. Negative attitudes towards road safety were also linked to risk taking, thus supporting Iversen (2004), Ulleberg and Rundmo (2003) and Wundersitz and Burns (2005) studies. There were however a few limitations with this study. Firstly, it is not known whether or not road safety interventions were implemented in the schools that took part during the year period which could have helped towards lowering attitudes between the time points. Secondly, as the questionnaires were asking about socially undesirable behaviour it is entirely possible that participants completed the questionnaire in a more socially desirable manner or under-reported driving violations, errors and lapses. In future replications of this study intention to drink and drive and to not wear seat belts could be included to ﬁnd out whether or not they are signiﬁcant predictors of driving violations, errors or lapses. Also, it may be better to leave more than a year between sampling points, or to add a Time 3, to allow for drivers to gain more experience in an attempt to ascertain how their attitudes may or may not have changed. References Arnett, J. (1994). ‘Sensation seeking: a new conceptualization and a new scale.’ Personality and Individual Differences, 16 (2): 289–96. Elliott, M.A. and Baughan, C.J. (2003). The Behaviour of Adolescent Road Users. Department of Transport, Behavioural research in road safety, 13th Seminar, September 2003. Elliott, M.A. and Baughan, C.J. (2004). ‘Developing a self-report method for investigating adolescent road user behaviour.’ Transportation Research Part F, 7: 373–93. Iversen, H. (2004). ‘Risk-taking attitudes and risky driving behaviour.’ Transportation Research Part F, 7 (3): 135–50. Jessor, R., Turbin, M.S. and Costa, F.M. (1997). ‘Predicting developmental change in risky driving: the transmission to young adulthood.’ Applied Developmental Science, 1 (1): 4–16. Parker, D., Manstead, A.S.R. and Stradling, S.G. (1995). ‘Extending the theory of planned behaviour: the role of personal norm.’ British Journal of Social Psychology, 34: 127–37. Parker, D., Stradling, S. and Senior, V. (1998). The Development of Remedial Strategies for Driving Violations. TRL report 300. Ulleberg, P. and Rundmo, T. (2003). ‘Personality, attitudes and risk perception as

Pre-driving Attitudes and Non-driving Road User Behaviours

73

predictors of risky driving behaviour among young drivers.’ Safety Science, 41: 427–43. West, R., Train, H., Junger, M., Pickering, A., Taylor, E. and West, A. (1998). Childhood Accidents and their Relationship with Problem Behaviour. Road Safety Research Report no. 7, Department for Environment, transport regions, London. WHO (2004). World Report on Road Trafﬁc Injury Prevention: Summary. World Health Organisation: Geneva (http://www.who.int/violence_injury_prevention/publications/road_trafﬁc/world_ report/summary_en_rev.pdf) WHO (2007). Youth and Road Safety. World Health Organisation: Geneva (http:// whqlibdoc.who.int/publications/2007/9241595116_eng.pdf). Wundersitz, L. and Burns, N. (2005). ‘Identifying young driver subtypes: relationship to risky driving and crash involvement.’ In L. Dorn (ed.), Driver Behaviour and Training, Volume II. Ashgate Publishing: Aldershot.

This page intentionally left blank

Chapter 7

Prediction of Problem Driving Risk in Novice Drivers in Ontario: Part II Outcome at Two Years Laurence Jerome and Al Segal University of Western Ontario, Canada Introduction There are six million drivers in Ontario, 1.3 million or 22 per cent of whom are deemed to be problem drivers who have had previous driving offences or crashes and are at increased risk for ‘problem driving’ (Ministry of Transport of Ontario, 1991). One speciﬁc target group known to be at increased risk are younger, new drivers. New drivers are three times more likely to be killed than the average driver. They make up 15 per cent of licensed drivers and have 30 per cent of driver fatalities. In 2003, 1678 youths aged 15–24 years died as a result of injury, which represents 73 per cent of all deaths in this age group, one death in Canada every ﬁve hours. Motor vehicle crashes account for the majority (about 60 per cent) of unintentional injuries. Even more disturbing, for every youth who dies from trauma, more than ten have severe injuries often requiring one or more surgeries, prolonged hospital stays and rehabilitation. The consequences of severe trauma, especially to the brain, are often so devastating and permanent that prevention is a far better investment than late interventions and supportive care (Statistics Canada, 2003; Canadian Institute for Health Information, 2005). This is not just a problem in Canada: motor vehicle collisions continue to be the leading cause of death for 16–20-year-olds in the United States, Australia, New Zealand and most Western European countries (American Academy of Paediatrics Committees on Injury, 2006). A recent editorial opines, ‘We all appear to have become acclimatised to this public health epidemic. If 32 youths in Canada were dying each week from heart disease, inﬂuenza or meningitis, a huge outcry to stop this epidemic would be heard’ (Canadian Medical Association Journal, 2007). Statistics from ‘Drinking and Driving in Ontario Statistical Yearbook 1990’ indicates that the 19–24 year old groups are over-represented both in terms of nondrinking accidents as well as drinking related accidents. The search for ‘the holy grail’ of predicting which novice drivers will fall into the high-risk group remains elusive. Over a decade ago the then current best estimates were that two-thirds of all accidents were not predictable on the basis of current knowledge of driver characteristics and training (Ministry of Transport of Ontario, 1994). This statistic reﬂects the current state of knowledge.

76

Driver Behaviour and Training – Volume III

This study reports on a two year follow up of a group of novice drivers attending a high school driving programme who were screened at base line with on-the-road observation by experienced driving instructors who also completed a screening instrument based on the human factor literature thought to predict non-accidental trafﬁc injury and shown in part one of this study to correlate with the current gold standard of risk prediction. The outcome at two years was measured by ofﬁcial driving records listing moving violations and collisions. The ﬁndings in this study will argue that human factors measured prior to obtaining a driver’s licence will have signiﬁcant power to predict risk. Human factor research Gerald Wilde, writing in 1994 in his book Target Risk, found no statistical support that personal characteristics correlated with general accident risk. Although acknowledging some correlations with individual characteristics and collisions, he found the evidence that there is a powerful effect within speciﬁc individuals unconvincing. Whilst acknowledging demographic predictors such as age and gender, he emphasised the utility of changing the level of risk tolerance of the population as a whole as having more utility in accident prevention than concentrating on human factor research (Wilde, Robertson and Pless, 2002). However there is an extensive literature on high risk driving populations. A recent report on contributing factors to collisions stated that human factors, as opposed to vehicle and environmental factors are the predominant contributor to collisions (United States General Accounting Ofﬁce [GAO], 2003). Risky driving behaviours are found to predict collisions and moving violations (Blows, Ameratunga, Ivers, Lo and Norton, 2005; McKnight and McKnight, 2000). Such behaviours include speeding, following too close, driving under the inﬂuence of alcohol, cell phone use and not using seatbelts while driving. These behaviours cluster in young drivers and a propensity to risk-taking contributes to increased rates of unintentional injury beyond the risk due to inexperience alone (Jonah, 1986). Numerous factors including demographic variables, personality and cognitive abilities have been explored to further understand their contribution to risky driving and collisions. Young males (Turner and McClure, 2003; Williams and Shabanova, 2003) as well as older (> 65 years) drivers (Preusser, Williams, Ferguson, Ulmer and Weinstein, 1998; Williams and Shabanova, 2003; Zhang, Fraser, Lindsay, Clarke and Mao, 1998) have been consistently related to increased negative driving outcomes. The evidence shows that both educational attainment and occupational status are inversely related to motor vehicle driver collision and injury (Hasselberg and Laﬂamme, 2003; Murray, 1998). The concept of accident proneness as it relates to unintentional driving injury was ﬁrst elucidated in the psychiatric literature by Tillman and Hobbes (1949), Professors of Psychiatry at the University of Western Ontario. They described a characterological style in a group of accidentprone drivers referred by the Ministry of Transportation for recurrent accidents. The authors coined the phrase, often repeated in this literature, ‘a man drives as he lives’. Many studies have since presented evidence for associations between personality traits including risk taking, sensation seeking, impulsivity, difﬁculty in dealing with

Prediction of Problem Driving Risk

77

tension and controlling anger, substance abuse, antisocial tendencies as well as nonconformity and risky driving behaviours or collision (Deffenbacher, Deffenbacher, Lynch and Richards, 2003; Jonah, 2001; Tsuang, 1985). Psychosocial models of high risk driving including descriptions of temporary states involving high stress (Lagarde, Chastang, Gueguen, Coeuret-Pellicer, Chiron and Lafont, 2004) and Problem Behaviour Theory emphasising lifestyle factors including low parental involvement and negative peer and parental inﬂuence (Shope, Waller, Raghunathan and Patil, 2001; Shope, Raghunathan and Patil, 2003) have been related to risky driving and problem driving events. Cognitive abilities likely play a signiﬁcant role in driving risk. Inattention and distractibility, which are directly related to risky driving behaviour, are cognitive factors that have been found to account for one-fourth of collisions (Treat et al., 1977). Poor risk perception, as well as impaired capacity to deploy appropriate judgment and reasoning while driving, have also been found to play a role in risky behaviours and negative driving outcomes (McKnight and McKnight, 2000; Ryb, Dischinger, Kufera and Read, 2006). Deﬁcits in these higher order cognitive factors of executive functioning are thought to underlie risky driving behaviours. Such deﬁcits are more evident in young and older drivers and in various clinical populations, and likely contribute to the higher collision rates found in these categories of drivers (Jerome, Segal and Habinski 2006, Jerome and Segal, 2000; McKnight and McKnight, 1993; Treat et al., 1977). Normal maturational immaturities in areas of the brain underlying executive function, evident in younger ages (Blakemore and Choudhury, 2006), together with inexperience, likely contribute to increased driving risk. Method Participating sites and collaborators This study was conducted through the University of Western Ontario Department of Psychology. Dr A. Segal and Dr Laurence Jerome are Adjunct Professors at UWO in the departments of Psychology and Psychiatry. Experimental participants were recruited from students enrolled in courses offered through the Driver Training Centre at Thames Secondary School. Thames Secondary School is located in London, Ontario and is part of the Thames Valley District School Board. The participants volunteered for this study and were provided with $20 compensation. Duration of study Following the initial data collection, the participant new drivers in this study were followed for a period of two years following the initial assessment. Procedure The ofﬁcial MTO driving records for the same 66 participants from Jerome and Segal (2005) were used in this study. The ofﬁcial driving record contained a spreadsheet

78

Driver Behaviour and Training – Volume III

for each participant listing any known moving violations and collisions. Only convictions are recorded within the ofﬁcial record. Previously the driving instructors had evaluated the driving risk of each of the volunteers. The risk ratings were based on the driving instructors’ impression of the students’ observed driving ability. The instructors were asked to identify the students’ risk based upon their observational experience of the participants’ risk severity. The literature on evaluation of driving risk, suggested that individual driving instructors are reliable at predicting problem driving based on their observations of students driving. At the termination of the study at two years, all participants were to be contacted by telephone to conduct an interview designed to elicit details of moving violations, collisions and driving exposure. The Screening Instrument Computerised measures of inattention and impulsiveness The stop-signal paradigm test The stop-signal paradigm (Schachar et al., 1993) was used to assess inhibitory control; subjects were engaged in a choice reaction time (go) task and attempted to inhibit their responses to the go task when they heard a stop signal. Reaction times to the stop signal (SSRT) and to the go signal (GoRT) were used to examine inhibition and response execution respectively. The Conners’ continuous performance test (Conners, CPT- II) This computerised instrument is the most widely used commercially available test of the variables of attention and behavioural inhibition (Conners, 2000). The participant must continuously respond to non-targeted letter stimuli, but inhibit responding to infrequent visual targets. The CPT has normative data developed on both clinical and normal populations. Subjective measures of driving behaviour made by driving instructors Driving instructor risk rating (risk rating) This is a visual analogue scale completed by the driving instructor after ﬁve hours’ observation of the student’s driving. The driving instructor responded to the question ‘rate this student on your estimation of current safety, based on your observations of their current driving behaviour’. The instructor placed an X on a ten centimetre line; the further from the origin on the left, the higher the rating of risk. Driving instructor checklist Objective severity ratings of problem driving were obtained from a semi-structured behavioural observation instrument, with demonstrated reliability and validity used in an English driving study, the Driving Instructor Checklist (West and Hall, 1998). This checklist was modiﬁed for North American expression and left-hand driving. No changes to content were made. This paper and pencil instrument asked driving instructors to rate the participant

Prediction of Problem Driving Risk

79

on driving skill, safety and future ability and safety, that is, the sections included Current Risk, Safety Risk and Future Risk. Subjective self-report measures of driving behaviour and personality style Barkley adult attention scale This is an 18 item self-report questionnaire providing scores for inattention as well as hyperactivity and impulsivity. The scale items are derived from the Diagnostic and Statistical Manual of the American Psychiatric Association fourth edition with age-corrected norms (Barkley and Murphy, 1998). Jerome driving questionnaire (JDQ) This is a visual analogue scale consisting of 12 measures of the subjects’ impression of their current and future driving style over the next 12 months. The subject was asked to place an X on the line distant from the origin to indicate their subjective rating of risk. The scale provides a measure of emotional and cognitive factors thought to reﬂect underlying executive function as it relates to driving. Health and life style questionnaire (HLS) Standard information regarding health status, current medication usage and current recreational drug usage and accident history was collected. The temperament and character inventory (TCI) The TCI (Cloninger, 1996) was given to evaluate the temperamental proﬁle of impulsivity (Novelty Seeking) within the context of a broader assessment of other temperament and character traits. The TCI is a widely reference research instrument which has been shown to evaluate the temperamental characteristic of impulsivity. The TCI is computerised and presents 240 descriptive statements to which the participant responds ‘true’ (‘this statement describes me’), or ‘false’ (indicating the statement is incorrect). Normative data is available and the test is self-scoring within the software programme. There are no offensive or sexually provocative statements in the TCI. Youth risk behaviour surveillance system (YRBS) This is a widely used epidemiological survey instrument with known validated characteristics used in numerous studies evaluating health risk problem behaviours in community youth samples (Youth Risk Survey, 2001). Demographic questionnaire In addition to the above measures, standard demographic information in relation to age, gender, height, weight, grade point average in school, family composition and family occupation was collected. Telephone interview questionnaire This was a 22-item telephone questionnaire modelled after a questionnaire used by Russell Barkley (2002). A research assistant attempted to contact participants by phone and arrange a suitable time to administer the questionnaire.

80

Driver Behaviour and Training – Volume III

Results Outcome data The human factors and self-report predictors reported in Jerome and Segal (2005) were analysed with respect to the prediction of problem driving events as recorded in the ofﬁcial Ministry of Transportation of Ontario driving record taken approximately two years after licensing. Of the original sample of novice drivers (n = 66), approximately 27 per cent of students (18) had at least one problem-driving event that included either a moving violation or collision. Twenty violations and 16 collisions were found in the ofﬁcial record. There were no mortalities in the sample. Four participants had multiple collisions. These four drivers were identiﬁed by the driving instructor’s Risk Rating to be falling within the moderate to highrisk categories. A single participant had two alcohol-related violations and received a driving instructor high-risk rating. Whilst these anecdotal observations are interesting, when the entire dataset was considered, all four types of the driving instructor’s risk ratings and one additional composite risk rating (Driving Instructor Checklist plus Total Score) was not statistically signiﬁcantly associated with any measure of driving outcome. Furthermore, the driving instructor’s risk ratings whilst identifying low risk participants failed to distinguish moderate from high-risk participant outcome. A follow-up telephone survey at two years post-licensing gathered information about driving exposure, collisions and violations. Only 32 of the original 66 participants could be traced. Eight collisions and six violations were self-reported that did not appear on the ofﬁcial driving record. Again these additional problemdriving events were unrelated to the Driving Instructor’s Risk Rating. Table 7.1 Human factors predictors of problem driving events (a) Total driving incidents Predictors Gender JDQ – risk taking (b) JDQ – anger (b) JDQ – daydreaming TCI – sentimentality TCI – cooperation TCI – compassion Barkley – inattentive Barkley – total CPT – variability

0.384 0.403 0.326 0.325

Correlation p < 0.001 p < 0.001 p < 0.01 p < 0.01

–0.253 –0.293 –0.303 0.292 0.315 0.250

p < 0.04 p < 0.01 p < 0.01 p < 0.02 p < 0.01 p < 0.04

Prediction of Problem Driving Risk (b) Collisions Predictors JDQ – risk taking (a) JDQ – risk taking (b) JDQ – alertness (b) (c) Violations Predictors Gender Grade average JDQ – anger (b) JDQ – daydreaming JDQ – risk taking (b) TCI – disorderliness TCI – cooperation Barkley – inattentive Barkley – total

81

Correlation 0.282p < 0.02 0.237 p < 0.05 0.279p < 0.03

0.380 –0.291 0.364 0.477 0.383 0.275 0.283 0.369 0.368

Correlation p < 0.002 p < 0.01 p < 0.004 p < 0.0001 p < 0.002 p < 0.02 p < 0.02 p < 0.005 p < 0.005

Prediction of driving outcome Table 7.1 (a), (b) and (c) show the statistically signiﬁcant correlations of human factor predictors to driving outcome as recorded in the ofﬁcial driving record. Section (a) describes the results for the total problem driving incidents; sections (b) and (c) report on collisions and violations, respectively. Linear regression analysis was applied to the Total Driving Incidents data. The model incorporated all the identiﬁed human factor predictors, accounting for 32 per cent of the common variance (using the Adjusted R2). The linear regression model for collisions only identiﬁed the JDQ-risk taking variable, accounting for six per cent of the variance. The JDQ-daydreaming variable and male gender were related to violations, accounting for approximately 37 per cent of the common variation. Table 7.2 (a), (b) and (c) presents the statistically signiﬁcant results for the selfreport predictors of problem-driving events. Section 7.2 (a) describes the results for the total problem driving incidents; sections (b) and (c) report on collisions and violations, respectively. Linear regression analysis was applied to the Total Driving Incidents data. The model involved only the YRBS-other drugs predictor, accounting for 36 per cent of the common variance. The linear regression model for collisions only identiﬁed the HLS-head injury predictor. This accounted for roughly 19 per cent of the variance. A combination of YRBS-other drugs and HLS-other drugs used predictors accounted for 62 per cent of the violations variance.

82

Table 7.2

Driver Behaviour and Training – Volume III

Self-report predictors of problem driving events (a) Total Driving Incidents; (b) Collisions; (c) Violations

(a) Total Driving Incidents Predictors HLS – Substance use-concern YRBS – Personal safety YRBS – Marijuana use YRBS – Other drugs

Correlation 0.345 0.546 0.252 0.514

p < 0.005 p < 0.0001 p < 0.04 p < 0.0001

(b) Collisions Predictors HLS – Rx medication use HLS – Head injury HLS – Accidental poisoning YRBS – Personal safety

–0.285 0.253 0.411 0.311

Correlation p < 0.02 p < 0.04 p < 0.001 p < 0.01

0.340

Correlation p < 0.01

(c) Violations Predictors HLS – Other drug use HLS – Substance use- concern HLS – Criticism for drug use HLS – Guilt for drug use HLS – Morning alcohol use HLS – Fighting while intoxicated YRBS – Personal safety YRBS – Violence YRBS – Marijuana use YRBS – Other drugs YRBS – Physical conditioning

0.391 0.275 0.356 0.281

p < 0.001 p < 0.03 p < 0.004 p < 0.02

0.322 0.532 0.329 0.338 0.622

p < .008 p < .0001 p < .009 p < .006 p < .0001

–0.270

p < .03

Linear regression models of problem driving outcome As part of the initial stages of constructing a screening instrument for the prediction of problem driving, stepwise linear regression analysis was conducted on all statistically signiﬁcant predictors of problem-driving events. These results are shown in Table 7.3 (a), (b) and (c). As shown in this table, the Total Driving Incidents regression model involved a combination of human factor and self-report predictors accounting for 48 per cent of the variance. Similarly, a combination of human factor and self-report predictors yielded a model of collision prediction that accounted for 34 per cent of the common variance. Lastly, violations were only associated with two self-report measures of ‘other drug use’ that accounted for approximately 62 per cent of the variance.

Prediction of Problem Driving Risk

83

Table 7.3 Linear regression models of problem driving events (a) Total driving incidents Predictors YRBS – Other drugs, Gender, CPT – Variability

Adjusted R2 0.480

(b) Collisions Predictors YRBS – Personal safety, HLS – Head injury, HLS – Accidental poisoning, JDQ – Alertness (b), JDQ – Risk taking, JDQ – Risk taking (b)

Adjusted R2

0.340

(c) Violations Predictors YRBS – Other drugs, HLS – Other drug use

Adjusted R2 0.616

Discussion Limitations of this study include the small sample size. Whilst the period of exposure and the exposure dose (distance driven) was not measured directly, we feel conﬁdent that the relatively long period of observation and follow up combined with objective ofﬁcial MTO data should have captured most of the problem driving events in this group of novice drivers. Especially given current research suggesting the maximum period of collision and violation risk in novice drivers peaks within the ﬁrst months of independent driving and declines rapidly after six months (McCArrt et al., 2003, Mayhew et al., 2003). Others have indicated that the ofﬁcial record prevalence ﬁgures are likely to be underestimations, excluding unreported minor collisions and moving violations that never came to police attention. The results of our telephone interview questionnaire demonstrates this. Whilst 50 per cent of participants’ self-reported data could not be traced at two years, the available data indicated moving violations and collisions not reported in the ofﬁcial record. The under-reporting may reﬂect a delay in data entry in the ofﬁcial record, failure to be convicted or minor collisions that were not reported where ofﬁcial reporting reﬂects more ﬁnancially costly collisions. Our ﬁndings support those of Barkley (1993) and Nada Raja (1997), arguing that a combination of self-reporting and ofﬁcial driving records will likely produce the most comprehensive picture of driving outcome. The current prevalence of 27 per cent of our sample having problem driving is in keeping with the ﬁgures reported in Ontario of 25 per cent of random samples of drivers manifesting problem driving (Ministry of Transport of Ontario, 1991; 1994). Our study casts doubt on the utility of driving instructor ratings as the best measure of future risk (Donnelly et al., 1992; Dobbs, Heller et al., 1998; Dobbs and McCracken et al., 1998; West and Hall, 1998). This is the ﬁrst study to report on the predictive validity of a driving instructor’s evaluation done at the time of the initial training two years after independent driving. To our knowledge there are no published studies

84

Driver Behaviour and Training – Volume III

evaluating the contribution of both ‘objective’ measures of ecologically valid realtime driving instructor evaluations at base line along with human factor measures of impulsivity and attention after two years of independent driving experience. Our data demonstrate that for novice drivers, the driving instructors’ future predictive accuracy was ‘no better than chance’. Previous authors have argued strongly against the predictive power of human factors in identifying at-risk groups of drivers (Wilde et al., 2002). Our data demonstrate that a combination of subjective self-report from novice drivers combined with human factors measures of attention and impulsiveness and temperament proﬁling predicts a signiﬁcant percentage of the problem driving events. Interestingly, our self-report data on problem behaviour as assessed by the YRSB and HLS, unrelated to problem driving, did not support the notion of a more general factor of risky problem behaviour. The linear regression model of general problem behaviour excluded aggressive behaviours, accidental ingestion of poisonous substances, tobacco, alcohol and cannabis use, and sexual promiscuity as signiﬁcant problem behaviour predictors of problem driving. This ﬁnding supports those of Beggs et al. (1999) who argued against a general factor of risk taking behaviour predicting problem driving. The exposure to ‘other’ recreational drugs was associated with moving violations but not collisions in this relatively young population. This may reﬂect a smaller, but more behaviourally deviant sub-group in our sample, where human factors are subordinate to speciﬁc problem behavioural patterns related to substance use. Collisions, an infrequent event compared with moving violations, are complex events likely inﬂuenced by a combination of human factors, deviant problem behaviour and the unpredictable nature of the road environment. Our ﬁndings support a model of problem driving based on human factors related to inattention and impulsiveness and a second more restricted model of lifestyle factors of deviant problem behaviour arguing for a transaction between predisposing and precipitating biological and environmental factors. All the human factor instruments used in this study are normally used to measure symptoms of inattention and impulsiveness, measures of executive function in clinical populations. The fact that clinical measures of executive function have such strong predictive power in normal populations is a unique ﬁnding in this ﬁeld of study that has previously relied on instruments that assume human factors lie on a continuum. We would argue that indeed the utility of these clinical instruments measuring executive function in a normal population argues for the same continuum of severity of these human factors in both clinical and non-clinical populations. As we have previously argued deﬁcits in executive function as it relates to problem driving can be seen as lying along a continuum across a range of categorical clinical diagnoses and ages ranging from ADHD, depressive illness and dementia (Jerome and Segal, 2000). Similarly, the ﬁndings in this study of deﬁcits of impulsiveness and inattention might best be considered as continuous orthogonal variables that cut across the categorical distinction between normal and clinical populations. The ﬁndings of this study argue that the application of human factor research to predicting problem driving behaviours holds the promise of developing an instrument that could be both a sensitive and speciﬁc tool for detecting a high-risk population of young drivers. The negative ﬁndings regarding the predictive power

Prediction of Problem Driving Risk

85

of trained driving instructors, although counter-intuitive, is supported by other research ﬁndings that showed little beneﬁt of current driving instruction techniques and later driving safety. (Mayhew et al., 1998). Our ﬁndings argue that given the underlying cognitive and temperamental vulnerabilities in these high-risk normal populations, the development of screening instruments to measure future risk of problem driving in normal populations should use measures of executive function as well as self-reported data. Such instruments may be used to guide cognitive behavioural interventions for at-risk individuals to improve their attention capacities and reduce impulsiveness. Indeed, this instrument may provide a metric for assessing improvement and predicting increased safety for driving. Such instruments could well incorporate driving simulators designed to measure the critical factors found to be predictive of problem driving. Perhaps a modiﬁed approach to driving instruction for these high-risk groups emphasising cognitive and behavioural measures of driving competence might incorporate the use of such instruments to assist driving instructors, as well as ofﬁcial examiners in deciding the readiness of novice drivers to graduate to the next level, that is, independent driving. References American Academy of Pediatrics Committees on Injury, Violence and Poison Prevention and on Adolescence. ‘The teen driver.’ Paediatrics 2006, 118: 2570– 81. Barkley, R.A. and Murphy, K. (1998). A Clinical Workbook (2nd ed.). New York: Guilford Press. Barkley, R.A., Guevramont, D.C., Anastropoulos, A.D., DePaul, G.J. and Shelton, T.L. (1993). ‘Driving-related risks and outcomes of attention deﬁcit hyperactivity disorder in adolescents and young adults: a 3–5 year follow-up survey.’ Paediatrics, 92, 212–18. Barkley, R.A., Murphy, K.R., DuPaul, G.J. and Bush, T. (2002). ‘Driving in young adults with attention deﬁcit hyperactivity disorder: knowledge, performance, adverse outcomes and the role of executive functioning.’ Journal of the International Neuropsychological Society, 8, 655–72. Begg, D.J., Langley, J.D. and Williams, S.M. (1999). ‘A longitudinal study of life style factors as predictors of injuries and crashes amongst young adults.’ Accident Analysis and Prevention, 31,1–11. Blakemore, S.J. and Choudhury, S. (2006). ‘Development of the adolescent brain: implications for executive function and social cognition.’ The Journal of Child Psychology and Psychiatry, 47(3–4), 296–312. Blows, S., Ameratunga, S., Ivers, R.Q., Lo, S.K. and Norton, R. (2005). ‘Risky driving habits and motor vehicle driver injury.’ Accident Analysis and Prevention, 37(4), 619–24. Canadian Institute for Health Information. (2007). ‘National trauma registry: 2005 injury hospitalizations highlights report.’ Ottawa: The Institute. Canadian Medical Association Journal, ‘What’s killing and maiming Canadian

86

Driver Behaviour and Training – Volume III

Youth?’ CMAJ March 13th 2007 176(6), 737. Centre for Disease Control and Prevention (2001). Youth Risk Behaviour Survey (2001), Atlanta. http://www.cdc.gov/nccdphp. Cloninger, C.R. (1996). ‘Assessment of the impulsive-compulsive spectrum of behaviour by the seven-factor model of temperament and character.’ In Oldham, J.M., Hollander, E. and Skodol, A.E. (eds). Impulsivity and Compulsivity. Washington, D.C.: American Psychiatric Press, 59–96. Conners, K.A. (2000). The New Conners’ Continuous Performance Test (CPT II) Computer Programme. Toronto, ON: MHS Inc. Deffenbacher, J.L., Deffenbacher, D.M., Lynch, R.S. and Richards T.L. (2003). ‘Anger, aggression and risky behaviour: a comparison of high and low anger drivers.’ Behaviour Research and Therapy, 41(6), 701–18. Deffenbacher, J.L., Huff, M.E., Lynch, R.S., Oetting E.R. and Salvatore, N.F. (2000). ‘Characteristics and treatment of high anger drivers.’ Journal of Counselling Psychology, 47(1), 5–17. Dobbs, A.R., Heller R.B. and Schopﬂocher, D. (1998). ‘A comparative approach to identify unsafe older drivers.’ Accident Analysis and Prevention. 30(3), 363–70. Dobbs A.R., McCracken, P.N., Carstensen, B.A., Kiss, I. and Triscott., J.A.C. (1998). ‘The evaluation of competence to drive.’ Paper presented at 1998 Canadian Consensus Conference on Dementia. Donnelly, R.E., Karlinsky, H.J., Young, M.L., Ridgley, J.N. and Lamble, R.W. (1992). ‘Fitness to drive in elderly individuals with progressive cognitive impairment.’ Ministry of Transport of Ontario. Hasselberg, M. and Laﬂamme, L. (2003). ‘Socioeconomic background and road trafﬁc injuries: a study of young car drivers in Sweden.’ Trafﬁc Injury Prevention, 4, 249–54. Jerome, L. and Segal, A.U. (2000). ‘ADHD, Executive Function and Problem Driving.’ The ADHD Report, 8(2): 7–11. Jerome, L. and Segal, A.U. (2005). ‘Prediction of driving accident risk in novice drivers in Ontario: the development of a screening instrument.’ 2nd International Conference on Driver Behaviour and Training, Edinburgh, Scotland. In Lisa Dorn (ed.), Driver Behaviour and Training, vol. 2, 207–22. Jerome L., Segal A. and Habinski L. (2006). ‘What we know about ADHD and driving risk: a literature review, meta-analysis and critique.’ Journal of the Canadian Academy of Child and Adolescent Psychiatry 15(3), August 2006. Jonah, B.A. (2001). ‘Sensation seeking, risky driving and behavioural adaptation.’ Accident Analysis and Prevention, 33, 679–84. Lagarde, E., Chastang, J.F., Gueguen, A. Coeuret-Pellicer, M., Chiron, M. and Lafont, S. (2004). ‘Emotional stress and trafﬁc accidents: the impact of separation and divorce.’ Epidemiology, 15(6), 762– 6. Mayhew, D.R., Simpson, H.M. and Pak, A. (2003). ‘Changes in collision rates among novice drivers during the ﬁrst months of driving.’ Accident Analysis and Prevention, 35(5), 683–91 Mayhew, D.R., Simpson, H.M., Williams, A.D. and Ferguson, S.A. (1998). ‘Effectiveness and the role of driver education and training in a graduated licensing

Prediction of Problem Driving Risk

87

system.’ Journal of Public Health Policy, vol. 19, no.1, 51–66. McCartt, A.T., Shabanova, V.I. and Leaf, W.A. (2003). ‘Driving experience, crashes and trafﬁc citations of teenage beginning drivers.’ Accident Analysis and Prevention, 35, 311–20 McKnight, A.J. and McKnight, A.S. (1993). ‘The effect of cellular phone use upon driver attention.’ Accident and Analysis Prevention, 25, 259–65. McKnight, A.J. and McKnight, A.S. (2000). ‘The behavioural contributors to highway crashes of youthful drivers.’ Annual Proceedings – Association for the Advancement of Automotive Medicine, 44, 321–33. Ministry of Transportation of Ontario (1991). Accident Potential: An Ontario Drivers Record Study Summary Report. Ministry of Transportation of Ontario (1994). Ontario Road Safety Agenda. Murray, A. (1998). ‘The home and school background of young drivers involved in trafﬁc accidents.’ Accident Analysis and Prevention, 30(2), 169–82. Nada-Raja,S., Langley, J.D., McGee, R., Williams, S.M., Begg, D.J. and Reeder, A.I. (1997). ‘Inattentive and hyperactive behavior and driving offences in adolescence.’ Journal of the American Academy of Child and Adolescent Psychiatry, 36(4), 515–22. Preusser, D.F., Williams, A.F., Ferguson, S.A., Ulmer, R.G. and Weinstein, H.B. (1998). ‘Fatal crash risk for older drivers at intersections.’ Accident Analysis and Prevention, 30(2), 151–9. Ryb, G.E., Dischinger, P.C., Kufera, J.A. and Read, K.M. (2006). ‘Risk perception and impulsivity: association with risky behaviours and substance abuse disorders.’ Accident Analysis and Prevention, 38, 567–73. Schachar, R. J., Tannock, R., Logan, G. (1993). ‘Inhibitory control, impulsiveness and attention deﬁcit hyperactivity disorder.’ Clinical Psychology Review, vol. 13, 721–39. Shope, J.T., Raghunathan, T.E. and Patil, S.M. (2003). ‘Examining trajectories of adolescent risk factors as predictors of subsequent high-risk driving behaviour.’ Journal of Adolescent Health, 32, 214–24. Shope, J.T., Waller, P.F., Raghunathan, T.E. and Patil, S.M. (2001). ‘Adolescent antecedents of high-risk driving behaviour into young adulthood: substance use and parental inﬂuences.’ Accident Analysis and Prevention, 33, 649–58. Statistics Canada. Causes of Death, 2003. Cat. no. 84-208-XIE. www.statcan.ca/ bsolc/english/bsolc?catno=84-208-X. Tillman, W.A. and Hobbs, G.E. (1949). ‘The accident-prone automobile driver.’ The American Journal of Psychiatry, 106, 321–31. Treat, J.R., McDonald, N.S., Shinar, D., Hume, R.D., Mayer, R.E., Stansifer, R.L., et al. (1977). Tri-Level Study of the Causes of Trafﬁc Accidents, Vol. I: Causal Factor Tabulations and Assessment. (Publication Number DOT-HS-805-085). Washington, DC: U.S. Department of Transportation. Tsuang, M.T. (1985). ‘Psychiatric aspects of trafﬁc accidents‘ The American Journal of Psychiatry, 142(5), 538–46. Turner, D. and McClure, R. (2003). ‘Age and gender differences in risk taking behaviours as an explanation for high incidence of motor vehicle crashes in young

88

Driver Behaviour and Training – Volume III

males.’ Injury Control and Safety Promotion, 10(3), 123–30. United States General Accounting Ofﬁce. (2003). Research Continues on a Variety of Factors that Contribute to Motor Vehicle Crashes. Washington DC: United States General Accounting Ofﬁce. West, R. and Hall, J. (1998). Accident Liability of Novice Drivers, TRL Report 295. Crowthorne: Transport Research Laboratory. Wilde, G.J.S., Robertson, L.S. and Pless, I.B. (2002). ‘For and against: does risk homeostasis theory have implications for road safety?’ BMJ 2002, 324, 1149–52 Williams, A.F. and Shabanova, V.I. (2003). ‘Responsibility of drivers, by age and gender, for motor-vehicle crash deaths.’ J Safety Res, 34950, 527–31. Zhang, J., Fraser S., Lindsay, J., Clarke, K. and Mao, Y. (1998). ‘Age-speciﬁc patterns of factors related to fatal motor vehicle trafﬁc crashes: focus on young and elderly drivers.’ Public Health, 112, 289–95.

PART 2 Emotions and Driver Behaviour

This page intentionally left blank

Chapter 8

A Review of Studies on Emotions and Road User Behaviour 1

Jolieke Mesken,1 Marjan Hagenzieker2 and Talib Rothengatter3 DHV Environment and Transportation, Amersfoort, The Netherlands 2 SWOV Institute for Road Safety Research, Leidschendam, The Netherlands 3 University of Groningen, The Netherlands

Introduction In recent years, numerous studies have indicated the relevance of emotions for road user behaviour. Most of these studies focused on anger, although other emotions were sometimes also investigated. The studies show varying results, mostly due to inconsistencies in the use of concepts, methods and theoretical frameworks. In the present paper, a review is given of empirical studies on emotions and road user behaviour. First, studies that used emotion as an independent variable or a mediator will be reviewed. Second, studies focussing on the causes of emotions in trafﬁc will be discussed. Third, the studies will be compared with regard to methodology, concepts and deﬁnitions. Finally, the conclusions of the review will be presented. The articles discussed in this review were collected using the following databases: Library SWOV Institute for Road Safety Research, PsychLit and Online Contents from Dutch Public Library Network. The set was extended with relevant titles from the reference lists. Articles included in the set contained one of the following search terms in relation to trafﬁc (including trafﬁc, driving, road and car): Affect, Emotion, Mood, Anger, Fear and Depression. Articles that were related to driver aggression alone were left out. This was done because some studies focus on aggressive behaviour only, without making a reference to any affective concept. Examples are the series of horn-honking studies (Doob and Gross, 1968; Deaux, 1971; Ellison, Govern, Petri and Figler, 1995) and studies on the prevalence of driver aggression (Joint, 1995). Furthermore, studies on stress and workload were excluded, again as far as no reference was made to an affective concept. The concept of stress refers mainly to a mismatch between task demands and personal capabilities (Lazarus and Folkman, 1984). Negative affect may be a result, but this is not always the case. Therefore, only those studies that explicitly mention this negative affect are included. Finally, studies on depression after being involved in an accident (Post Traumatic Stress Disorder) were left out because they do not attempt to explain driving behaviour.

92

Driver Behaviour and Training – Volume III

Studies on affect as independent variable or mediator: emotion and mood effects In this section, studies that used emotion or mood as a factor inﬂuencing driving behaviour are discussed. Some of these studies have used the affective concept for building a model (for example on aggressive driving) and thus the concept is used as a mediator. The studies took different aspects of the driving task into consideration. These aspects can be categorised into effects on task performance, effects on errors and violations, effects on aggressive and risky driving and effects on general road safety and accident involvement. Task performance Several studies investigated the relationship between mood states and task performance. As early as 1967, Heimstra, Ellingstad and De Kock measured drivingrelated performance in relation to mood. A simulated driving task was constructed, which consisted of a steering wheel connected to a control element, and a pedal. Performance measures included vigilance, reaction time, tracking performance, and speed maintenance. Mood was measured prior to the task by having subjects ﬁll out a questionnaire (Mood Adjective Checklist; see Nowlis 1965). Four subscales were distinguished: anxiety, aggression, fatigue and concentration. No statistically signiﬁcant correlations were found between mood scales and performance measures. When comparing those who scored high and low on aggression, anxiety and fatigue, subjects seemed to perform worse on the driving subtasks, although the authors did not report effect sizes and p-levels. Groeger (1996) studied the effect of mood on self-rated and instructor-rated driving performance. Mood was measured before and after a driving test, by MAACL (Multiple Affect Adjective Checklist), including subscales of hostility, anxiety and depression. Both the subject and the instructor made judgements of the subject’s performance at various moments during the drive, compared to a novice driver. The subject’s judgement was related to the extent to which their mood, as measured by the MAACL, changed during the drive. That is, if subjects felt more anxious, depressed and hostile after the drive than before, they judged their performance as worse. The instructor’s judgement was only related to the change in hostility in the subject: if subjects felt more hostile after the drive, their performance was rated as worse, but this was not the case for anxiety and depression. These results suggest that all mood states inﬂuence self-evaluation, but only hostility is related to actual driving performance. A follow-up study by Stephens and Groeger (2006) investigated whether emotions affect driving behaviour in a simulator. Participants (n = 24) performed a test drive in a driving simulator. During the drive, they encountered various trafﬁc events, designed to interrupt their journey. Throughout the drive, participants were asked to give ratings of their emotions. Three emotions were considered: frustration, calmness and anger. Results showed that when drivers had to reduce speed because of trafﬁc events, they reported more anger and frustration and less calmness. Also, those who had become angry accelerated more after the impeding event.

A Review of Studies on Emotions and Road User Behaviour

93

Garrity and Demick (2001) also measured driving-related performance in relation to mood. An experienced observer evaluated the driving behaviour of 163 subjects during a test drive. A number of driving behaviours were scored on an observation form, and factor analysis on these data revealed four factors: responsiveness, manoeuvring, observation and cautiousness. Mood was measured prior to the drive by using the POMS (Proﬁle of Mood States; see Mc Nair, Lorr and Droppleman, 1992). The only factor that was related to mood was cautiousness: respondents scoring high on depression, anger and fatigue were less cautious, whereas respondents who scored high on vigour-activity were more cautious. The results of these studies suggest that anger/aggression is related to a decrease in task performance as rated by a driving instructor, to increased acceleration after an impeding event and to a less cautious driving style. The other mood states did not show consistent effects. Errors and violations Instead of focussing on general task performance, some studies took speciﬁc errors and violations into account. Two studies investigated the relation between mood and the amount of errors made in a driving course. Appel, Blomkvist, Persson and Sjöberg (1980) studied whether mood affects performance on a difﬁcult driving task. They presented a mood measure (Sjöberg, Svensson and Persson, 1978) to 55 driving school students before and after skid training. Performance on the skid training was observed by trained instructors. Several erroneous actions were scored. Correlations between performance and mood measures showed that respondents who felt more unpleasant, tired, tense and uncertain made more errors in skid training on a slippery road. Ford and Alverson-Eiland (1991) also studied errors in relation to anxiety. The anxiety levels of 107 students (11 groups of 6–12 subjects) were measured before participating in a motorcycle rider’s course, by using the State Trait Anxiety Inventory (Spielberger, Gorsuch and Lushene, 1970). Performance on the driving course was measured by the amount of errors made on a skills test, as observed by an instructor. The correlation between level of anxiety before the test ride and the amount of errors made was not signiﬁcant. Yet when correlations were calculated separately for each group, for some groups signiﬁcant correlations were shown. The authors conclude that anxiety seems to be a moderately inﬂuential factor in predicting the performance of the subjects. Stradling and Parker (1997) and Lawton, Parker, Manstead and Stradling (1997) studied the role of affect in predicting trafﬁc violations. Earlier studies (Parker, Manstead, Stradling and Reason, 1992; Parker, Reason, Manstead and Stradling, 1995) had shown that attitudes, subjective norm and perceived behavioural control together affect the intention to commit violations. Lawton et al. (1997) investigated whether the inclusion of affect in this model makes the prediction of intention to commit violations more accurate. A questionnaire with a list of 12 trafﬁc violations was presented to respondents. A factor analysis on the violation items revealed three factors (fast driving, maintaining progress and anger/hostility). Regression analysis showed that positive affective evaluations predicted all three types of violations.

94

Driver Behaviour and Training – Volume III

The study shows that anticipated affect might be a strong factor inﬂuencing trafﬁc behaviour. These results also suggest that at least for a certain group of drivers, committing violations is associated with anticipated positive affect. However, Arnett, Offer and Fine (1997) did not ﬁnd an association between positive affect and (speeding) violations. In this study, the role of several state and trait factors in driving behaviour was studied among 59 high school students who kept driving logs. Comparisons between indicated mood states were made regarding the extent to which the speed limit was exceeded. Anger was the only mood state for which a relation with driving speed was shown: respondents exceeded the speed limit to a greater degree when angry than when experiencing any other emotion. The relationship between anger and violations was also investigated by Lajunen, Parker and Stradling (1998). Their questionnaire study showed that respondents who rated their own safety orientation as high and their own perceptual-motor skills as low were less likely to report driving anger than those who rated their safety orientation as low and their perceptual motor skills as high. Also, driving anger was related to committing both Highway Code and aggressive violations. In a second study, Lajunen and Parker (2001) related driving anger to general (verbal and physical) aggressiveness on the one hand and to aggressive reactions on the other hand. Several path models were proposed, which suggested that the link between verbal aggressiveness and driving aggression was mediated by driving anger, whereas physical aggressiveness had a direct link to driver aggression. Parker, Lajunen and Summala (2002), in another questionnaire study, compared self-reported aggressive responses to driving anger by using samples from three European countries: Finland, UK and the Netherlands. Generally, those behaviours that provoke most anger also provoked the most extreme reactions. There were some differences between the countries in the amount of anger the different behaviours provoked. For example: in response to reckless driving, UK drivers reported more anger than Dutch or Finnish drivers. In sum, there is only limited evidence that a negative (tense, insecure) mood is related to errors. The results from studies connecting moods or emotions to violations are more consistent: anticipated positive affect is associated with violations, and there also seems to be a link between anger and violations. Especially the relation between anger and aggressive violations has received considerable attention. These studies are discussed separately in the next section. Aggressive and risky driving behaviour In trafﬁc research on emotion, the anger-aggression relationship has been studied most extensively. The studies by Deffenbacher and colleagues used various scales (Driving Anger Scale [DAS], Driving Anger Expression Inventory [DAX], Drivers’ Angry Thoughts Questionnaire [DATQ]) to measure the effect on aggressive and risky driving. Aggressive behaviour was deﬁned by Deffenbacher, Richards and Lynch (2004) as ‘behaviour based in anger and/or behaviour the goals of which are to harm, intimidate, threaten, dominate, retaliate upon, frustrate, or otherwise express displeasure with another driver or user of the roadway’ (p. 116). Risky behaviour is

A Review of Studies on Emotions and Road User Behaviour

95

behaviour that ‘puts the individual and/or others at increased risk for injury, crash and/or property damage’ (Deffenbacher et al., 2004: p. 116). A range of questionnaire and driving log studies showed that driving anger is associated with self-reported aggressive and risky behaviour (Deffenbacher, Lynch, Oetting and Yingling, 2001), aggressive forms of anger expression (Deffenbacher, Lynch, Deffenbacher and Oetting, 2001; Deffenbacher, Lynch, Oetting and Swaim, 2002) and angry thoughts (Deffenbacher, Petrelli, Lynch, Oetting and Swaim, 2003). These studies consistently show a pattern of high anger drivers reporting more aggressive and more risky behaviours on the road, getting angry more frequently and more intensely and reporting more crash-related outcomes (for example, loss of vehicular control, close calls, minor accidents and so on). The conclusion that driving anger is related to a range of other anger-related and driving-related variables, as was drawn from the above reported studies, was based primarily on self-reports (questionnaires and driving logs). The study reported by Deffenbacher, Deffenbacher et al. (2003) is one of few studies which did not use only self-reports but also measures of simulated driving. Subjects were 121 ﬁrst year students, either scoring high or low in driving anger. High anger drivers drove with a higher speed and shorter following distances than low anger drivers. In high impedance situations they were more likely to crash, and their state anger level increased more in high impedance situations than for low anger drivers. The authors conclude that drivers with a disposition to become angry behind the wheel are different from low driving anger drivers in terms of state anger, aggression, risky behaviours and negative driving outcomes. A second study, in which a simulator was used, was carried out by Ellison-Potter, Bell and Deffenbacher (2001). Also in this study, groups of high and low anger drivers were compared. Subjects were randomly assigned to experimental conditions: the respondents had to imagine either that they were anonymous or not, and respondents were exposed to either aggressive or neutral stimuli. They subsequently performed a test drive in a simulator, in which their aggressive driving behaviour (for example, speed, number of red lights run and collisions) was recorded. Results showed that respondents drove more aggressively when being anonymous and when exposed to aggressive stimuli. However, no main or interaction effects of driving anger were found on aggressive driving behaviour. So, whereas Deffenbacher in his studies consistently showed that high anger drivers had a propensity to report aggressive driving more frequently, Ellison-Potter et al. did not ﬁnd such a difference. The authors suggest that this may be due to the fact that no actual provocation was involved in the task: subjects were not provocated by other road users. Instead, aggressive behaviour was operationalised by average speed, number of red lights run, number of collisions and number of pedestrians hit. In fact, Deffenbacher and colleagues did not measure actual aggressive behaviour either, but used self-report measures. In two studies, Knee and Neighbors (2001) and Neighbors, Vietor and Knee (2002) also studied driving anger and aggression using questionnaires. They used selfdetermination theory to explain aggressive driving. According to self-determination theory, people may have different tendencies to regulate behaviour. They may either have a more controlled orientation (their behaviour is regulated by contingencies

96

Driver Behaviour and Training – Volume III

and pressures) or a more autonomous orientation (their behaviour is regulated by interest and choice). In the ﬁrst study (Knee et al., 2001), it was shown that people who have a more controlled orientation are more likely to behave aggressively on the road, and this link is mediated by driving anger. In the second study (Neighbors et al., 2002), in which driving logs were used, an extra component was added: not only trait motivation is important, but also how this trait affects the motivation in a speciﬁc situation (state or situational motivation). The results of the study showed that respondents high in controlled motivation report to experience more pressure and ego-defensiveness in driving situations, leading to more anger and aggression. Whereas in the previous studies the emphasis was on personal characteristics and their relation to driver aggression, Yagil (2001) focussed on cognitive processes. She proposed that drivers’ aggressive behaviour is determined by the type of attributions they make after frustrating behaviour by other drivers. Results of a questionnaire study showed that aggressive reactions were not affected by hostile attributions about the behaviour of the driver. Rather, it was general irritability and competitiveness that predicted aggression. The studies discussed in this section provide substantial support for the hypothesis that trait driving anger is related to a range of other trafﬁc related variables such as states of anger during driving and aggressive or risky driving behaviour. Besides aggressive driving behaviour, trait anger was shown to be related to Highway Code violations. Other personal characteristics and their relation to anger and aggression were discussed in this section: drivers’ tendency to regulate behaviour is related to aggression, as is drivers’ attribution of responsibility. General road safety and (near) accidents Emotions thus seem to affect a range of safety-related behaviours. Some studies also investigated whether there also exists a connection with actual road safety. Underwood, Chapman, Wright and Crundall (1999) studied both causal factors and effects of anger on self-reported driving behaviour and self-reported involvement in a (near) accident. Respondents ﬁlled in questionnaires concerning driving anger (DAS, Deffenbacher, Oetting and Lynch, 1994), driving behaviour (DBQ, Reason, Manstead, Stradling, Baxter and Campbell, 1990) and social deviance (Social Motivation Scale, West, Elander and French, 1993). Then they kept driving logs over a period of two weeks using portable micro cassette recorders. Along with other trip characteristics such as congestion, length of trip, near accidents and so on, information regarding felt anger during the trip was recorded (description of the event, intensity of anger, whether or not the anger affected driving). No straightforward relation between congestion and anger was found. Correlations were calculated between felt anger during the trip and involvement in a near-accident during the same trip. Signiﬁcant correlations were shown, but a closer examination of the driving logs revealed that felt anger was often a result of the near accident. On those occasions where anger was not a result of near accidents, it appeared that the frequency of reporting anger was related to near accidents on other occasions. In this case, felt anger was related to trait driving anger and mild social deviance.

A Review of Studies on Emotions and Road User Behaviour

97

The authors conclude that anger might be both a cause and result of near accidents, although causality remains questionable because of the correlational design. Levelt (2003b) also used driving logs to examine respondents’ own evaluation of the effects of their emotions on road safety. Causes of drivers’ emotions were equally often an event prior to the trip, an event during the trip or thoughts that occurred during the trip. Subjects rated the effects of positive emotions on average as positive for road safety, and negative emotions on average as negative for road safety. However, in some instances, fear was rated as positive for road safety. This was for example the case in situations where fast responses were needed. Bañuls, Carbonell Vaya, Casanoves and Chisvert (1996) and Carbonell Vaya, Bañuls, Chisvert, Monteagudo and Pastor (1997) developed a questionnaire to measure anxious responses to driving: the ISAT (Inventory of Situations provoking Anxiety in Trafﬁc). Four subscales were identiﬁed: situations related to selfevaluation or external evaluation, situations related to criticism and aggression, situations related to impediments and trafﬁc jams, and situations related to evaluation by the authorities. Bañuls et al. (1996) compared professional and novice drivers with regard to their responses on ISAT and their self-reported accident involvement. Results showed that for novice drivers, anxiety responses to those situations that involve some kind of evaluation of driving may be connected to increased accident risk, whereas for professional drivers the more risky situations are anxiety responses to those situations that involve delays or impediments. Carbonell et al. (1997) used only professional drivers in the sample and compared two different subgroups (taxi drivers and lorry drivers) with regard to their ISAT scores and accident involvement. For taxi drivers, anxiety responses to delays and obstacles were most clearly related to accident involvement. Also for lorry drivers, anxiety responses to impediments to completing the job and, to a lesser extent, anxiety about verbal aggression, were related to accidents. The results of these studies show that different subgroups of drivers show different relations between anxiety producing situations and accident involvement. General driving performance is thus negatively associated with negative emotions such as hostility and anxiety, although when people are asked about their own evaluation of the effects of emotions on safety, some negative emotions were evaluated as positive for trafﬁc safety as well. Conclusion The studies reviewed in this section show that emotions and moods may affect driving-related performance in a number of ways. The clearest results were shown for feelings of anger and hostility. These emotions seem to affect general task performance, but are also related to the commission of violations and to aggressive and risky driving. Furthermore, some results suggest that anxiety and feelings of tenseness are related to errors. None of the studies investigated the effects of emotions on cognitive processes while driving. This issue will be discussed further later in this chapter.

98

Driver Behaviour and Training – Volume III

Studies on antecedents of emotions and moods while driving A number of studies focussed on emotion as a dependent variable. These studies investigated the causes or antecedents of emotions while driving. Three categories can be distinguished: studies on the personality – emotion relationship, studies on the relevance of situational characteristics in causing emotions and studies on the treatment of maladaptive emotions. Personal characteristics The relations between personal characteristics and emotion in trafﬁc have been studied in terms of mood (Dorn and Matthews, 1995) and anger (Malta, Blanchard, Freidenberg, Galovski, Karl et al., 2001; Richards, Deffenbacher and Rosén, 2002). Dorn and Matthews (1995) compared two contrasting hypotheses on the mood and personality relation. One is that mood is affected by general personality traits, as measured in a study by Eysenck Personality Inventory (EPI; see Eysenck and Eysenck, 1964). The other is that mood is affected by more context speciﬁc traits, as measured in a study by the Driving Behaviour Inventory (DBI; see Gulian, Matthews, Glendon, Davis and Debney, 1989). In the study, subjects ﬁrst completed these two scales, and gave ratings of driving risk and competence. Then they were asked to perform either a passive or active driving simulator task. Afterwards, post-drive mood was measured by Uwist Mood Adjective Checklist (UMACL) which has three dimensions: Tension, Energy and Hedonic Tone. Post-drive mood was predicted better by DBI than by EPI. The subscale Dislike of Driving was the strongest predictor: subjects scoring high on this factor had higher ratings on postdrive tension and lower ratings on hedonic tone and energy. Negative post-task mood was related to negative appraisals of competence. Therefore, people high on Dislike of Driving have a disposition to make negative appraisals of their personal competence. This might cause them to make negative post-task appraisals of their performance, which leads to a negative mood. Malta et al. (2001) compared 14 aggressive and 14 non-aggressive drivers on their physiological response patterns and driving anger. The study showed that aggressive drivers, compared to controls, showed different physiological response patterns. Also, they had signiﬁcantly higher scores on driving anger and anger expression. No differences were reported between aggressive and non-aggressive drivers on state anger. Richards et al. (2002) also compared two groups on their level of driving anger: 21 ﬁrst year students scoring high and 38 ﬁrst year students scoring low on ADHD (Attention Deﬁciency Hyperactivity Disorder). Respondents ﬁlled in various scales related to driving anger (DAS, DAXI) and kept driving logs for a period of three days. Differences between the two groups were found: high ADHD subjects reported higher scores on different anger-related measures like driving anger, anger expression, aggressive driving behaviour and risky driving behaviour. Taylor and Deane (2000) and Taylor, Deane and Podd (2000) compared drivingfearful subjects (selected by advertisements) who either had or had not been involved in a motor vehicle accident. Driving fear was measured by DSQ (Driving Situations Questionnaire, Ehlers, Hofmann, Herda and Roth, 1994). The two groups did not

A Review of Studies on Emotions and Road User Behaviour

99

differ on severity of driving fear, or their general patterns of concern, although respondents that had been involved in an accident were more concerned about accident and injury than those that had not been involved in an accident. So, the two groups differ in the sense that for one group there is a clear object their fear is directed to, and the other group there is not. However this does not seem to affect fear intensity. Results thus show that several personal characteristics such as trait anger, dispositional aggression, ADHD and type of driving fear (which is in these studies treated as a personal characteristic or phobia) are related to emotions and moods in trafﬁc, although no relations with state anger were shown. Situational characteristics Parkinson (2001) investigated anger frequency in driving and non-driving contexts. Results of a questionnaire study showed that driving situations involved a higher frequency of anger than non-driving situations. Lawton and Nutter (2002) also compared anger in a driving and non-driving context and found that subjects in their study are not more likely to get angry in trafﬁc than non-trafﬁc situations. However, Lawton and Nutter only once introduced trafﬁc scenarios and asked about level of anger, whereas Parkinson asked how many times per month people got angry. Lawton and Nutter also reported a difference in the expression of anger, which is more frequent in driving than in non-driving situations. Both studies acknowledge that there are speciﬁc characteristics of the driving context, such as anonymity and lack of possibilities to communicate, which makes it different from other situations. Also, Chapman, Evans, Crundall and Underwood (2000) showed that the likelihood to act on anger depends on the context. A number of 211 drivers ﬁlled in questionnaires regarding anger in driving and non-driving contexts. Also, personal interviews were conducted. Results showed that in driving situations people report equal levels of felt anger as in non-driving situations, but in driving situations it is more likely that people react on their anger with aggression than it is in non-driving situations. So, there do seem to be differences in the nature of driving and non-driving contexts. Concepts and methods Deﬁnition of affective concept A range of different affective concepts have been used in the literature: mood, anxiety, anger, emotion, road rage, affective evaluations, fear, stress, affect and depression. Still, only ﬁve studies provide a deﬁnition of the concept. Ford and Alverson-Eiland (1991) studied the effects of anxiety on performance on a motorcycle riders’ course. They referred to Spielberger et al. (1970), deﬁning anxiety as ‘subjective feelings of tension, apprehension, nervousness and worry, as well as activation or arousal of the autonomic system’. Deffenbacher et al. (1994) refer to another study by Spielberger et al. (1983) and deﬁne driving anger as a context-speciﬁc measure of trait anger.

100

Driver Behaviour and Training – Volume III

Trait anger is deﬁned as a disposition to experience anger frequently, but still the concept of anger remains undeﬁned. Joint (1995) studied road rage, and deﬁned road rage as ‘more extreme acts of aggression, such as physical assault, that occur as a result of a disagreement between drivers’. In this case the concept is deﬁned in terms of behaviour and not in terms of affect. Stradling and Parker (1997) and Lawton et al. (1997) deﬁned affect in relation to intention to commit violations. In these studies, affect is deﬁned as extra motive, outside the more rational motives such as personal norm and behaviour intention: ‘feelings […] that an individual expects to experience while performing a particular behaviour’. Parkinson (2001) used appraisal theory (Smith and Lazarus, 1993) and deﬁned anger as appraisals of other-blame. The conclusion is that most studies do not offer a deﬁnition of the concepts they use. The deﬁnitions that are provided in the studies mentioned above are often not adequate. They sometimes refer to another undeﬁned affective concept (as in trait anger), or their deﬁnition does not seem to cover the meaning of the concept (as in road rage, which implies more than just behaviour, as mentioned in the deﬁnition). Use of a theoretical framework Most of the articles reviewed do not use a theoretical framework to build and test hypotheses about the relations between emotions and driver behaviour. However, there are a few exceptions; mainly regarding aggressive driving. Knee et al. (2001) and Neighbors et al. (2002) try to explain aggressive driving by using self determination theory (Deci and Ryan, 1985). They found that a controlled motivational orientation was related to driving anger, which in turn was related to driver aggression. Other studies have applied both the frustration-aggression hypothesis (Lajunen and Parker, 2001) and social information processing theory (Yagil, 2001). Appraisal theories have been used in studies on both stress (Matthews et al. 1996) and emotion (Levelt, 2001; Parkinson, 2001). Since the appraisal theory of emotion evolved from transactional models of stress, this is not surprising. Some of the studies refer to a theoretical framework or concept, but do not say anything about affect-driving links. Some refer only to the affective concept, for example in articles on driving related fear (Taylor and Deane, 1999). They use the theory of Rachman; ‘three pathways theory of fear acquisition’. Also, in the extensive work on driving anger that was done by Deffenbacher and colleagues since 1994, a reference is made to the concept of state-trait anger (Spielberger, 1983) but no predictions are made on the relation between anger and driving based on this theoretical concept. Dorn and Matthews (1995) refer to the temperamental approach to personality (Tellegen, 1985) and to the transactional model of driving stress (Gulian et al., 1989) and thus make contrasting hypotheses concerning the relationship between personality and mood in a driving context. However, again no predictions are made on the mood-driving relationship. Stradling and Parker (1997) used the theory of planned behaviour (Ajzen, 1985). The theory is an attitude theory, which in itself does not refer to the emotion-driving relationship, but the theory was extended with affective evaluations to be more accurate in predicting the intention to commit violations.

A Review of Studies on Emotions and Road User Behaviour

101

Measures Especially when studying the effects of emotions on driving behaviour, it would be best to have a real experiment: a design with a manipulation, a control group and random assignment to groups. However, none of the articles reviewed have actually manipulated anything, but measured the concept before and/or after a driving task. This measurement is also very diverse: about 20 different scales to measure affect have been identiﬁed. Some studies used existing scales, like (variations on) the Mood Adjective Checklist (Heimstra et al., 1967; Dorn and Matthews, 1995; Groeger, 1997); others developed scales for the purpose of the studies. One study (Malta et al., 2001) compared (self declared) aggressive drivers with non-aggressive drivers and used not only self-report scales (such as Driver’s Stress Proﬁle, see Larson 1996; Driving Anger Scale, see Deffenbacher et al., 1994) but also physiological measures such as heart rate and blood pressure. Driving related performance was in most studies measured by self-report, such as questionnaires (for example, Banũls et al., 1996; Lajunen et al., 1998) or driving logs (for example, Arnett et al., 1997; Chapman et al., 2000; Richards et al., 2002). In some studies a driving simulator was used (Deffenbacher et al., 2003; EllisonPotter et al., 2001) although the type of simulator varied as well, for example in the older study by Heimstra et al. (1967), the simulator was just a steering wheel and a tracking device. Dorn and Matthews (1995) also used a simulator but used it only to be able to measure post-task mood; no driving parameters were collected. Three studies measured actual driving performance by observation, two of which were done in the context of driver training (Appel et al., 1980; Ford and Alverson-Eiland, 1991). One study (Garrity and Demick, 2001) used driving instructors to evaluate drivers who already possessed a driving licence. Conclusion The studies that used emotion as an independent variable showed a rather inconclusive picture. Anger and hostility inﬂuenced task performance as rated by a driving instructor, but other mood states did not show such an effect. Some relations were shown between emotions or moods on the one hand and errors or violations on the other. Negative mood was related to the amount of errors made during a skid course. Another study showed, however, no straightforward results between mood and errors. Violations were shown to be related to both positive and negative affect. Drivers were shown to report more violations when they expected to experience positive affect, but also when they scored high on trait anger. These studies consistently showed a relation between anger and aggressive and risky driving behaviour. Some explanations were provided: aggressive drivers have a different attribution style and different tendencies to regulate behaviour than drivers who are not aggressive. Studies on emotion and general road safety and (near) accidents showed relations between anger and anxiety on the one hand, and self-reported involvement in a (near) accident on the other. Furthermore, when subjects were asked to evaluate the impact

102

Driver Behaviour and Training – Volume III

of emotions on their own driving behaviour, they indicated that negative emotions were negative for road safety. From the studies that used emotion as a dependent variable, it can be concluded that certain personality factors such as general aggressiveness, ADHD, motivational orientation and safety orientation, are related to driving anger. However, driving anger always refers to a trait measure, which is not always made explicit in the text of the article. Although trait driving anger is related to state driving anger, no direct relations between other personality characteristics and state anger in trafﬁc have been shown. As far as situational predictors of emotion are concerned, it is not clear whether anger in trafﬁc is more frequent than anger in other situations. The expression of anger, however, is more likely in trafﬁc than in non-trafﬁc situations (Lawton and Nutter, 2002). The studies that have been discussed vary on a number of characteristics. The affective concept that is used is different in many studies. Emotion, mood, affect and personality are often used without specifying which concept is used, and why. Anger, for example, is sometimes used as a mood, sometimes as an emotion and sometimes as a trait. The use of a theoretical framework varies as well: some studies do not use such a framework, others do. The measures that are used are in most studies selfreport measures, although some studies mention the use of a driving simulator or an instrumented car. A difﬁculty with questionnaire studies is the issue of causality. Do angry drivers take more risk, or is a certain type of driver more likely to be involved in risky situations, which might elicit anger? In the Arnett et al. (1997) study, which showed a relation between anger and exceeding the speed limit, directionality is an issue. Mood, as well as speed, was recorded after the trip. The conclusion was drawn that respondents drove faster when angry than when in another emotion; however, the conclusion might as well have been: respondents were more angry when they were speeding, than when they were not speeding. The study by Underwood et al. (1999) also shows that the direction (in this case between anger and near accidents) might be both ways. Most studies that use the Driving Anger Scale did not ﬁnd strong relations between state anger and aggression, suggesting that the link between driving anger and risky driving that are reported in these studies is more related to personality than to affect. Thus, whereas studies on determinants of emotions provided useful results, the effects of emotions in trafﬁc are still unclear. Future research on the effects of emotion on driving behaviour should pay special attention to the use of a theoretical framework, a good deﬁnition of the concept and to the direction of causality, preferably by using an experimental approach. References Ajzen, I. (1985). ‘From intentions to actions: a theory of planned behaviour.’ In J. Kuhl and J. Beckmann (eds). Action Control: From Cognition to Behaviour. Berlin, Springer. Appel, C., Blomkvist, A., Persson, L. and Sjöberg, L. (1980). ‘Mood and achievement

A Review of Studies on Emotions and Road User Behaviour

103

in a difﬁcult driving task.’ Ergonomics, 23, 605–12. Arnett, J., Offer, D. and Fine, M.A. (1997). ‘Reckless driving in adolescence: “state” and “trait” factors.’ Accident Analysis and Prevention, 29, 57–63. Bañuls Egeda, R., Carbonell Vaya, E., Casanoves, M. and Chisvert, M. (1996). ‘Different emotional responses in novice and professional drivers.’ In T. Rothengatter and E. Carbonell Vaya (eds). Trafﬁc and Transport Psychology: Theory and Application. Amsterdam, Pergamon. Carbonell Vaya, E.J., Banuls, R., Chisvert, M., Monteagudo, M.J. and Pastor, G. (1997). ‘A comparative study of anxiety responses in trafﬁc situations as predictors of accident rates in professional drivers.’ In Human Factors in Road Trafﬁc II: Trafﬁc Psychology and Engineering. Proceedings of the second seminar on human factors in road trafﬁc, Braga, Portugal, 186–92. Chapman, P., Evans, J., Crundall, D. and Underwood, G. (2000). ‘Anger and aggression in driving and non-driving contexts.’ In: Proceedings of the International Conference on Trafﬁc and Transport Psychology, Berne, Switzerland, 4–7 September 2000. Deaux, K.K. (1971). ‘Honking at the intersection: a replication and extension.’ Journal of Social Psychology, 84, 159–60. Deci, E.L. and Ryan, R.M. (1985). ‘The general causality orientations scale: selfdetermination in personality.’ Journal of Research in Personality, 19, 109–34. Deffenbacher, J.L., Deffenbacher, D.M., Lynch, R.S. and Richards, T.L. (2003). ‘Anger, aggression and risky behaviour: a comparison of high and low anger drivers.’ Behaviour Research and Therapy, 41, 701–18. Deffenbacher, J.L., Lynch, R.S., Deffenbacher, D.M. and Oetting, E.R. (2001). ‘Further evidence of reliability and validity for the driving anger expression inventory.’ Psychology Reports, 89, 535–40. Deffenbacher, J.L., Lynch, R.S., Oetting, E.R. and Swaim, R.C. (2002). ‘The driving anger expression inventory: a measure of how people express their anger on the road.’ Behaviour Research and Therapy, 40, 717–37. Deffenbacher, J.L., Lynch, R.S., Oetting, E.R. and Yingling, D.A. (2001). ‘Driving anger: correlates and a test of state-trait theory.’ Personality and Individual Differences, 31, 1321–31. Deffenbacher, J.L., Oetting, E.R. and Lynch R.S. (1994). ‘Development of a driving anger scale.’ Psychological Reports, 74, 83–91. Deffenbacher, J.L., Petrilli, R.T., Lynch, R.S., Oetting, E.R. and Swaim, R.C. (2003). ‘The driver’s angry thoughts questionnaire: a measure of angry cognitions when driving.’ Cognitive Therapy and Research, 27, 383–402. Deffenbacher, J.L., Richards, T.L. and Lynch, R.S. (2004). ‘Anger, aggression and risky behavior in high anger drivers.’ In: J.P. Morgan (ed.). Focus on aggression research. Hauppage, NY, Nova Science Publishers. Doob, A.N., and Gross, A.E. (1968). ‘Status of frustrator as an inhibitor of hornhonking responses.’ Journal of Social Psychology, 76, 213–18. Dorn, L. and Matthews, G. (1995). ‘Prediction of mood and risk appraisals from trait measures: two studies of simulated driving.’ European Journal of Personality, 9, 25–42. Ehlers, A., Hofmann, S.G., Herda, C.A. and Roth, W.T. (1994). ‘Clinical characteristics

104

Driver Behaviour and Training – Volume III

of driving phobia.’ Journal of Anxiety Disorders, 8, 323–39. Ellison, P.A., Govern, J.M., Petri, H.L. and Figler, M.H. (1995). ‘Anonymity and aggressive driving behaviour: a ﬁeld study.’ Journal of Social Behaviour and Personality, 10, 265–72. Ellison-Potter, P., Bell, P. and Deffenbacher, J.L. (2001). ‘The effects of trait driving anger, anonymity and aggressive stimuli on aggressive driving behaviour.’ Journal of Applied Social Psychology, 31, 431–43. Eysenck, H.J. and Eysenck, S.B.G. (1964). Manual of the Eysenck Personality Inventory. London, London University Press. Ford, F.H., and Alverson-Eiland, L.G. (1991). ‘The relationship between anxiety and task performance and skill acquisition in the Motorcycle Safety Foundation’s motorcycle rider course.’ In Safety Environment Future: Proceedings of the 1991 International Motorcycle Conference, Bochum, 363–79. Garrity, R.D., and Demick, J. (2001). ‘Relations among personality traits, mood states and driving behaviours.’ Journal of Adult Development, 8, 109–18. Groeger, J.A. (1997). ‘Mood and driving: is there an effect of affect?’ In T. Rothengatter and E. Carbonell Vaya (eds). Trafﬁc and Transport Psychology: Theory and Application. Amsterdam, Pergamon. Gulian, E., Matthews, G., Glendon, A.I., Davis, D.R. and Debney, L.M. (1989). ‘Dimensions of driver stress.’ Ergonomics, 32, 585–602. Heimstra, N.W., Ellingstad, V.S. and De Kock, A.R. (1967). ‘Effects of operator mood on performance in a simulated driving task.’ Perceptual and Motor Skills, 25, 729–35. Joint, M. (1995). Road Rage. London, Automobile Association. Knee, C.R., and Neighbors, C. (2001). ‘Self-determination theory as a framework for understanding road rage.’ Journal of Applied Social Psychology, 31, 889–904. Lajunen, T. and Parker, D. (2001). ‘Are aggressive people aggressive drivers? A study of the relationship between self-reported general aggressiveness, driver anger and aggressive driving.’ Accident Analysis and Prevention, 33, 243–55. Lajunen, T., Parker D. and Stradling, S. (1998). ‘Dimensions of driving anger, aggressive and Highway Code violations and their mediation by safety orientation in UK drivers.’ Transportation Research Part F: Trafﬁc Psychology and Behaviour, 1, 107–21. Larson, J.A. (1996). ‘Driver’s stress proﬁle.’ In J.A. Larson (ed.). Steering Clear of Highway Madness, Wilsonville, BookPartners, 25–8. Lawton, R. and Nutter, A. (2002). ‘A comparison of reported levels and expression of anger in everyday and driving situations.’ British Journal of Psychology, 93, 407–23. Lawton, R., Parker, D., Manstead, A.S.R. and Stradling, S.G. (1997). ‘The role of affect in predicting social behaviours: the case of road trafﬁc violations.’ Journal of Applied Social Psychology, 27, 1258–76. Lazarus, R. and Folkman, S. (1984). Stress, Appraisal and Coping. New York, Springer. Levelt, P.B.M. (2001). Emoties bij vrachtautochauffeurs. SWOV report R-2001-14. Leidschendam, SWOV Institute for Road Safety Research. Levelt, P.B.M. (2003b). Praktijkstudie naar emoties in het verkeer. SWOV report

A Review of Studies on Emotions and Road User Behaviour

105

R-2003-8. Leidschendam, SWOV Institute for Road Safety Research. Malta, L.S., Blanchard, E.B., Freidenberg, B.M., Galovski, T.E., Karl, A. and Holzapfel, S.R. (2001). ‘Psychophysiological reactivity of aggressive drivers: an exploratory study.’ Applied Psychophysiology and Biofeedback, 26, 95–116. Matthews, G., Desmond, P.A., Joyner, L., Carcary, B. and Gilliland, K.A. (1996). ‘A comprehensive questionnaire measure of driver stress and affect.’ In T. Rothengatter and E. Carbonell Vaya (eds). Trafﬁc and Transport Psychology: Theory and Application. Amsterdam, Pergamon. McNair, D., Lorr, M. and Droppleman, L. (1992). Proﬁle of mood states (3rd ed.). San Diego, Educational and Industrial Testing Service. Neighbors, C., Vietor, N.A. and Knee, C.R. (2002). ‘A motivational model of driving anger and aggression.’ Personality and Social Psychology Bulletin, 28, 324–35. Nowlis, V. (1965). ‘Research with the mood adjective checklist.’ In: S.S. Tomkins and C.E. Izard (eds). Affect, Cognition and Personality. New York, Springer. Parker, D., Lajunen, T. and Summala, H. (2002). ‘Anger and aggression in three European countries.’ Accident Analysis and Prevention, 34, 229–35. Parker, D., Manstead, A.S.R., Stradling, S.G. and Reason, J.T. (1992). ‘Determinants of intention to commit driving violations.’ Accident Analysis and Prevention, 24, 117–31. Parker, D., Reason, J.T., Manstead, A.S.R. and Stradling, S.G. (1995). ‘Driving errors, driving violations and accident involvement.’ Ergonomics, 38, 1036–48. Parkinson, B. (2001). ‘Anger on and off the road.’ British Journal of Psychology, 92, 507–26. Reason, J.T., Manstead, A.S.R., Stradling, S.G., Baxter, J.S. and Campbell, K. (1990). ‘Errors and violations on the road: a real distinction?’ Ergonomics, 33, 1315–22. Richards, T., Deffenbacher, J.L. and Rosén, L.A. (2002). ‘Driving anger and other driving-related behaviours in high and low ADHD symptom college students.’ Journal of Attentional Disorders, 6, 25–38. Sjöberg, L., Svensson, E. and Persson, L.-O. (1978). ‘The measurement of mood.’ Scandinavian Journal of Psychology, 20, 1–18. Smith, C.A. and Lazarus, R.S. (1993). ‘Appraisal components, core relational themes and the emotions.’ Cognition and Emotion, 7, 233–69. Spielberger, C.D., Gorsuch, R.L. and Lushene, R.E. (1970). STAI, Manual for the State-Trait Anxiety Inventory (‘Self-Evaluation Questionnaire’). Palo Alto, California: Consulting Psychologists Press. Spielberger, C., Jacobs, G., Russel, S. and Crane, R. (1983). ‘Assessment of anger: the state-trait anger scale.’ In J. Butcher and C. Spielberger (eds). Advances in Personality Assessment. Hillsdale, NJ, Lawrence Erlbaum Associates, Inc. Stephens, A.N. and Groeger, J.A. (2006). Do Emotional Appraisals of Trafﬁc Situations Inﬂuence Driver Behaviour? Paper presented at the Behavioural Studies Seminar, 3–4 April 2006, Bath, UK. Stradling, S.G. and Parker, D. (1997). ‘Extending the theory of planned behaviour: the role of personal norm, instrumental beliefs and affective beliefs in predicting driving violations’ In T. Rothengatter and E. Carbonell Vaya (eds). Trafﬁc and

106

Driver Behaviour and Training – Volume III

Transport Psychology: Theory and Application. Oxford, Elsevier. Taylor, J.E. and Deane, F.P. (1999). ‘Acquisition and severity of driving-related fears.’ Behaviour Research and Therapy, 37, 435–49. Taylor, J.E. and Deane, F.P. (2000). ‘Comparison and characteristics of motor vehicle accident (MVA) and non-MVA driving fears.’ Journal of Anxiety Disorders, 14, 281–98. Taylor J.E., Deane, F.P. and Podd, J.V. (2000). ‘Determining the focus of driving fears.’ Journal of Anxiety Disorders, 14, 453–70. Tellegen, A. (1985). ‘Structures of mood and personality and their relevance to assessing anxiety, with an emphasis on self-report.’ In A.H. Tuma and J.D. Maser (eds). Anxiety and the Anxiety Disorders. Hillsdale, NJ, Erlbaum. Underwood, G., Chapman, P., Wright, S. and Crundall, D. (1999). ‘Anger while driving.’ Transportation Research Part F: Trafﬁc Psychology and Behaviour, 2, 55–68. West, R.J., Elander, J. and French, D. (1993). ‘Mild social deviance, type-A behaviour pattern and decision-making style as predictors of self-reported driving style and trafﬁc accident risk.’ British Journal of Psychology, 84, 207–19. Yagil, D. (2001). ‘Interpersonal antecedents of drivers’ aggression.’ Transportation Research, Part F: Trafﬁc Psychology and Behaviour, 4, 119–31.

Chapter 9

A Comparison of the Propensity for Angry Driving Scale and the Short Driving Anger Scale Mark J.M. Sullman University of Hertfordshire, UK Introduction Driving evokes a wide range of emotions in people, including joy, frustration, anxiety, fear and anger. Anger is one of the emotions which has become increasingly researched over the last ten years. There are a number of reasons for this increase, including the fact that it is relatively common to experience this emotion while driving (Deffenbacher, Lynch and Oetting, 2002b). Furthermore, a number of studies have found that angry drivers engage more often in aggressive and dangerous driving behaviours (Dahlen, Martin, Ragan and Kuhlman, 2005; Deffenbacher, Oetting and Lynch, 1994). In fact Dahlen and Ragan (2004) went so far as to state that driving anger is one of the most inﬂuential predictors of aggressive and risky driving behaviour. Research has also found driving anger to be signiﬁcantly related to near misses (Underwood, Chapman, Wright and Crundall, 1999) and crash related conditions, such as loss of concentration, losing control of their vehicle and crash involvement (Deffenbacher, Lynch, Oetting and Yingling, 2001; Deffenbacher, Deffenbacher, Lynch and Richards, 2003; Sullman, Gras, Cunill, Planes and FontMayolas, 2007). There are a number of ways in which driving anger can be measured, with two such scales being the Driving Anger Scale (DAS) and the Propensity for Angry Driving Scale (PADS). In addition, there are two versions of the DAS, a 14 item unidimensional measure and a 33 item multidimensional measure. The 14 item version of the DAS presents 14 different situations and asks the responding driver to report the degree of anger that each situation makes them feel. In contrast, the PADS presents a situation which is likely to evoke anger and then asks the respondent to indicate how they would respond by selecting one of four potential responses, which range from mild reactions (for example, slowing down) to more extreme (for example, ramming the other car). The DAS and PADS have both been found to have good psychometric properties. Research has shown the DAS to have good internal reliability, with alpha coefﬁcients ranging from 0.80–0.92 (Deffenbacher et al., 1994; Deffenbacher, Filetti, Lynch, Dahlen and Oetting, 2002a). The alpha coefﬁcients for the PADS have also been good, ranging from 0.85–0.89 (Dahlen and Ragan, 2004; Depasquale, Geller, Clarke and Littleton, 2001). Moreover, the test–retest reliability of both scales has also

108

Driver Behaviour and Training – Volume III

been shown to be high. The PADS has been found to have a four week test–retest reliability of 0.91 (DePasquale et al., 2001), while the DAS has been shown to have a ten week test–retest reliability of 0.84 (Deffenbacher et al, 2002a). As would be expected, both scales seem to have similar relationships with descriptive variables (for example, age and gender), as well as driving behaviours and crash related conditions. For example in the two studies which have used the original 19 item version of the PADS, neither reported any age differences (Dahlen and Ragan, 2004; DePasquale et al., 2001) and only DePasquale et al. (2001) reported a gender difference. Although some research has found females score more highly on the shortened version of the DAS (Dahlen and Ragan, 2004), most research has found no gender differences (for example, Dahlen et al., 2005; Deffenbacher et al., 1994). Also, in contrast to the research using the multidimensional version of the DAS (for example, Lajunen, Parker and Stradling, 1998; Sullman, 2006), no age differences were reported in the studies using the shortened version of the scale (Dahlen and Ragan, 2004; Dahlen et al., 2005; Deffenbacher et al., 1994). However, it should be noted that the studies using the short DAS have all used samples with very narrow age ranges, whereas the two studies (mentioned above) using the longer version of the scale used samples from the general population with much broader age ranges. The DAS has also been found to be related to aggressive and risky driving behaviour (Dahlen et al., 2005; Deffenbacher et al., 2001; 2002b) and other crash related conditions such as loss of concentration, loss of control and near misses (Dahlen et al., 2005; Deffenbacher et al., 2001). In addition, although one study found a relationship between the DAS and major accidents (Deffenbacher et al., 2002b), this has not been a common ﬁnding. In contrast, the one study which has related the PADS to crash involvement found that the PADS was correlated with both major and minor crashes (Dahlen and Ragan, 2004) and other crash-related conditions, such as loss of control and receiving tickets for violating road rules (Dahlen and Ragan, 2004). Furthermore, like the DAS, the PADS has also been found to be signiﬁcantly related to aggressive and risky driving behaviour (Dahlen and Ragan, 2004). Although the PADS has been validated three times (DePasquale et al., 2001; Dahlen and Ragan, 2004; Maxwell, Grant and Lipkin, 2005), only two of these studies have used both the DAS and the PADS (Dahlen and Ragan, 2004 Maxwell et al., 2005). Furthermore, one of these two studies (Maxwell et al., 2005) modiﬁed the PADS by dropping four of the 19 items and also used a 21 item version of the DAS, rather than the 14 item version, meaning that the ﬁndings generated by that study were not comparable. Moreover, the only remaining study to compare the two scales relied solely upon psychology undergraduates as participants. This means that the participants were from a very restricted age range (medium 19) and were mainly female (75 per cent), calling into question the generalisability of these ﬁndings. This concern is highlighted further by the fact that in samples from the general population driving anger has been found to be related to both gender and age (Lajunen et al., 1998; Sullman, 2006). Therefore, it seems important that the PADS be investigated in a broader sample of drivers. The present study compared the PADS and DAS in

A Comparison of the Propensity for Angry Driving Scale

109

order to test whether the previously found relationships could be generalised to a broader sample of drivers. Method Participants Participants were randomly selected from the electoral rolls in three New Zealand cities. These electoral rolls had recently been updated due to a general election. A random number generator was used to select page numbers and was also used to select ﬁve individuals from each selected page. The names and addresses of the individuals selected were copies from the electoral roll and they were then sent a package containing a cover letter, questionnaire and reply paid envelope. All participants were required to hold a valid New Zealand licence and to have driven at least once in the last six months. If the recipient of the pack did not meet these criteria, they were asked to pass the questionnaire on to a person in the household who did and whose birthday was the closest. In total, 600 questionnaires were posted out and 225 completed questionnaires were returned, giving a moderate response rate of 37.5 per cent. Questionnaire In addition to a number of descriptive variables (for example, age, gender, annual mileage and driving speed), the fourteen item Driving Anger Scale (DAS) (Deffenbacher et al., 1994) was used to measure driving anger. Participants were instructed to imagine each of the 33 situations happening to them and to rate the amount of anger evoked by each on a six point Likert scale, which ranged from 0 = ‘not at all’ to 5 = ‘very much’. The 19 item Propensity for Angry Driving Scale (PADS) was also included (DePasquale et al., 2001). The 19 items describe driving situations that are likely to evoke anger and participants were asked to indicate how they would respond to each situation by selecting one of four choices. These choices range from mild reactions (for example, slowing down) to more extreme reactions (for example, making obscene gestures). The answers were then scored according to the procedure outlined by Depasquale et al. (2001). The eight violation items from the Driving Behaviour Questionnaire (DBQ; Reason, Manstead, Stradling, Baxter and Campbell, 1990) were used to measure risky driving behaviour. The scale asked participants to report how frequently they engaged in eight different driving violations. This was reported on a six point scale, which ranged from 0 = ‘never’ to 5 = ‘all the time’. The ten item Trait Anger Scale was also used to measure general anger (Spielberger, 1999). The TAS measures how an individual generally responds when angry and is rated on a four point scale (1 = ‘almost never’, 4 = ‘almost always’). A number of crash related conditions were measured using Deffenbacher et al.’s (2000) Driving Survey, which included: moving violations (tickets), loss of

110

Driver Behaviour and Training – Volume III

concentration, loss of control, near misses (close calls), minor accidents and major accidents in the last three months. Finally, participants were also asked to report the speed they would normally travel at on ﬁve types of roads (motorway, highway, busy main street, winding country road and a residential road). The ﬁve items were standardised and combined to form a single speed item. Results The participants were aged 17 to 89 years old (M = 44.1, SD = 15.6) and they had on average 25.8 years driving experience. More than half were female at 56.7 per cent, with 43.3 per cent male. The vast majority had not been involved in a minor (91.9 per cent) or a major crash (99.1 per cent) during the last three months. The average annual mileage was 16 069 km/year and very few (4.1 per cent) had received a ticket in the three months prior to completing the survey. Factor analysis of the DAS produced one factor which accounted for 37.3 per cent of the variance and contained all 14 items. The factor analysis of the PADS also produced one factor, but this solution only accounted for 23.3 per cent of the variance and two of the items had loadings below 0.300. However, as removing either item failed to improve the alpha coefﬁcient and to ensure comparability with previous ﬁndings, both items were retained. The PADS and DAS both showed good internal reliability, with 0.80 and 0.87, respectively. Good internal reliabilities were also found for the violations scale (0.73) and the Trait Anger Scale (0.82). As previous research has found gender differences on measures of driving anger (Dahlen and Ragan, 2004; DePasquale et al., 2001), the two driving anger scales were evaluated using a one way ANOVA. Surprisingly neither the PADS (F(1, 223) = 0.746, ns) nor the DAS (F(1, 223) = 0.492, ns) showed signiﬁcant gender effects. Furthermore, there were no gender differences for minor accidents (F(1, 219) = 0.782, ns), major accidents (F(1, 218) = 1.543, ns), tickets (F(1, 219) = 1.751, ns), loss of concentration (F(1, 211) = 0.198, ns), loss of control (F(1, 216) = 0.023, ns), involvement in near misses (F(1, 216) = 1.972) or Trait Anger (F(1, 223) = 3.519, ns). There was, however, a signiﬁcant difference for violations (F(1, 223) = 12.885, p < 0.001), with males reporting engaging more often in violations. (see Table 9.1) The PADS and DAS were correlated with a number of the background variables. Surprisingly, while the PADS was signiﬁcantly correlated with age (–0.292, p < .001) and number of years (–0.282, p < 0.001) the DAS was not. The DAS and PADS were both positively related to self-reported driving speed (DAS .183, p < 0.01; PADS .277, p < 0.001), but neither was signiﬁcantly related to annual mileage (0.124 and 0.026, respectively). The correlations between the main variables are presented in Table 9.2. The PADS was positively correlated with the DAS at 0.464, showing that although they are related they are certainly not measuring the same thing. The PADS and DAS were also both positively correlated with the TAS, violations and near misses. Interestingly the TAS was also correlated with both violations and loss of concentration.

A Comparison of the Propensity for Angry Driving Scale

111

Table 9.1 Alpha coefﬁcients, means and standard deviations by gender Measure

Alpha

PADS DAS Loss of concentration Loss of control Near misses Moving tickets Minor accidents Major accidents Violations TAS

0.80 0.87

0.73 0.82

Men M 39.76 37.43 1.85

Men SD 9.75 9.60 3.12

Women M 38.69 38.33 2.24

Women SD 8.74 9.35 8.03

0.19 1.17 0.10 0.13

0.68 2.62 0.57 0.41

0.18 0.77 0.03 0.08

0.56 1.52 0.18 0.33

0.00

0.00

0.02

0.13

2.15 16.14

0.57 3.29

1.88 17.14

0.52 4.37

PADS = Propensity for Angry Driving; DAS = Driving Anger Scale; TAS = Trait Anger Scale.

To further investigate the relationships between the driving anger scales and the two main variables they were correlated with (near misses and violations), four hierarchical multiple regressions were conducted. These were also to test whether the DAS and PADS were able to predict these two variables over and above that of the background variables and the TAS. Firstly, age, gender and annual mileage were entered in at Step 1, with the TAS being entered at Step 2. Finally the DAS and PADS were entered in at Step 3 in order to provide a direct comparison of the two. Table 9.3 shows that the descriptive variables and the TAS did not contribute signiﬁcantly to the prediction of near misses, while the addition of the PADS and DAS resulted in a signiﬁcant improvement in the prediction of near misses. Although it was only the DAS which was a signiﬁcant predictor of near misses, when the same regression was performed without the DAS, the PADS made a signiﬁcant contribution to the change in R2 (0.022, p < 0.05) and was a signiﬁcant predictor of near misses (Beta 0.177, t = 2.135, p < 0.05). Table 9.4 shows that the descriptive variables (age, sex and annual mileage) were all signiﬁcant contributors to the prediction of violations, as was the TAS. The addition of the PADS and DAS again resulted in a signiﬁcant improvement in the prediction of violations, but this time it was the PADS which was the only signiﬁcant predictor of violations. However, when the same regression was performed without the PADS the DAS made a signiﬁcant contribution to the change in R2 (0.014, p < 0.05) and was a signiﬁcant predictor of violations (Beta = 0.128, t = 1.999, p < 0.05).

112

Table 9.2 Correlations between the main variables 1

2 – 0.464*** 0.532*** 0.361*** 0.018 0.122 0.179** •0.019 •0.113 0.006

3 • 0.393*** 0.213*** 0.046 •0.006 0.233*** •0.016 •0.046 0.119

• 0.318*** 0.144* 0.084 0.082 0.066 •0.039 0.055

4

5

• 0.203** 0.145* 0.283*** 0.092 •0.061 •0.053

6

• 0.251*** 0.201** 0.015 •0.037 •0.008

PADS = Propensity for Angry Driving; DAS = Driving Anger Scale; TAS = Trait Anger Scale * p < 0.05, ** p < 0.01, *** p < 0.001

7

• 0.105 •0.004 0.024 •0.029

8

• 0.016 •0.017 •0.044

9

• 0.482*** •0.014

• 0.239***

Driver Behaviour and Training – Volume III

1. PADS 2. DAS 3. TAS 4. Violations 5. Concentration 6. Lose control 7. Near miss 8. Tickets 9. Minor crash 10. Major crash

A Comparison of the Propensity for Angry Driving Scale

113

Discussion This research found that the PADS and DAS were both signiﬁcant predictors of violations over and above that of the descriptive variables and general trait anger. This is in support of previous research which has found both scales to be signiﬁcant Table 9.3 Predicting near misses Step 1.

Variable Sex Age Mileage TAS PADS DAS

2. 3.

Beta •0.090 •0.054 •0.030 0.090 0.095 0.223

Change R2 0.010

R2 0.010

0.007 0.059**

0.017 0.077

t •1.270 •0.766 •0.419 1.203 1.098 2.867**

** p < 0.01

Table 9.4 Predicting violations Step 1.

2. 3.

Variable Sex Age Mileage TAS PONS DAS

Beta •0.173 •0.287 0.205 0.291 0.217 0.058

Change R2 0.177***

R2 0.177

0.073*** 0.043**

0.250 0.293

t •2.711** •4.572*** 3.206** 4.512*** 2.937** 0.864

** p < 0.01, *** p < 0.01

predictors of risky driving behaviour (Dahlen and Ragan, 2004; Dahlen et al., 2005; Deffenbacher et al., 2001; 2002b). However, it should be mentioned that when both driving anger scales were entered together it was only the PADS which was signiﬁcant. This was probably due to the fact that the shared variance between the two scales was attributed to the stronger related of the two, the PADS. Also in agreement with previous research was the ﬁnding that the DAS was predictive of near misses (or close calls) (Dahlen et al., 2005). However, in contrast to the one previous study to investigate this (Dahlen and Ragan, 2004), the PADS correlated both with and without the DAS in the equation was predictive of near misses. This might be due to the differences in the demographic composition of the two samples. In contrast to previous research (Dahlen et al., 2005; Deffenbacher et al., 2001), the DAS was not found to be related to crash related conditions such as loss of concentration and loss of control, while the PADS was not related to involvement in either minor or major crashes. Perhaps one reason contributing to the reduced number of signiﬁcant relationships was the relatively modest number of respondents. If the number of participants was slightly larger the number of signiﬁcant correlations may

114

Driver Behaviour and Training – Volume III

also have been larger. For example, the PADS correlated at r = 0.122 with loss of control, which may have been signiﬁcant with a larger sample size. Also along these lines the much lower number of factors related to the PADS and DAS may also be in part due to the differences in the demographic compositions of the respective studies. Most importantly the mean age here (44.1 years old) was much higher than the medium age of 19 years reported by Dahlen and Ragan (2004). As research has shown that younger drivers are more likely to engage in violations (Lajunen et al., 1998; Sullman, Meadows and Pajo, 2002), report higher levels of driving anger (Lajunen et al., 1998; Sullman, 2006) and are more frequently involved in crashes (for example, Sullman et al., 2002) it should not be surprising that more signiﬁcant relationships were found with predominantly young participants. The level of anger or hostile reactions found here using the PADS (males = 39.8, females = 38.7) appeared to be considerably lower than that reported by DePasquale et al. (2001) (overall = 50) and Dahlen and Ragan (2004) (males = 47.2, females = 43.2). The level of driving anger reported on the DAS (males = 37.4, females = 38.3) also appeared to be lower than was reported by Dahlen and Ragan (2004) (males = 43.9, females= 47.2), Dahlen et al. (2005) (males = 45.6, females = 46.2) and Deffenbacher, White and Lynch (2004) (males = 46.9, females = 45.8). However, it should be noted that in all the previously mentioned studies the participants were university students, who on average were substantially younger than the sample in the present study. Therefore, it is not surprising that the present study would report the lowest levels of anger for both the PADS and DAS. This study provides further evidence that the DAS and PADS are measuring similar, but slightly different aspects of driving anger. The differences are evident ﬁrstly in the fact that they were moderately, but not strongly correlated (r = 0.464). This was also demonstrated by the fact that they had different relationships with the descriptive variables. Nevertheless, both scales were correlated with the same crash related variables, although the PADS was more strongly related to violations and the DAS more strongly related to near misses. Furthermore, both scales were predictive of near misses and violations. This study also found the PADS was more strongly correlated to the TAS than the DAS was. This was also mildly supported by Dahlen and Ragan (2004) and is probably due to the fact that what the PADS measures is more similar to the TAS than what the DAS measures. The DAS measures how strongly each of the fourteen situations causes the driver to feel angry, while both the PADS and TAS measure reactions to anger (for example, TAS – ‘hit someone who angered you’, PADS – ‘ram someone with your car’). One potential limitation of this study is the fact that the research relied solely upon self–reported data which is subject to self-report bias. However, as all participants were assured of anonymity and conﬁdentiality there were no external pressures preventing them giving honest answers. Furthermore, while acknowledging that selfreport is not ﬂawless, there is a substantial body of research which clearly supports the accuracy of data gathered in this manner (for example, Rolls, Hall, Ingham and McDonald, 1991; Walton, 1999; West, French, Kemp and Elander, 1993). The Driving Survey measures the accident related conditions over the previous three months. Although collecting the data over this period would help by reducing

A Comparison of the Propensity for Angry Driving Scale

115

‘forgetting’, it also means very few individuals would have experienced minor or major accidents (8.1 per cent and 0.8 per cent, respectively), which are relatively rare events. Although previous research using this time period has found relationships between the PADS and accidents, it should be remembered that the previous research had a very young sample (medium 19) and young drivers are over-represented in accident statistics. Future research in this area should also collect the number of accidents over a longer period of time. In summary, although a number of the previously reported relationships between the driving anger scales and crash related conditions were replicated in this sample from the general population of drivers (rather than university students), many more were not. This research has also shown that although the PADS and DAS are measuring similar things, they are not exactly the same and should be thought of as complementary rather than competing scales. References Dahlen, E.R. and Ragan, K.M. (2004). ‘Validation of the Propensity for Angry Driving Scale.’ Journal of Safety Research, 35, 557–63. Deffenbacher, J.L., Filetti, L.B., Lynch, R.S., Dahlen, E.R. and Oetting, E.R. (2002a). ‘Cognitive-behaviour treatment of high anger drivers.’ Behaviour Research and Therapy, 40, 895–910. Deffenbacher, J.L., Huff, M.E., Lynch, R.S., Oetting, E.R. and Salvatore, N.F. (2000). ‘Characteristics and treatment of high-anger drivers.’ Journal of Counselling Psychology, 43, 131–48. Deffenbacher, J.L., Lynch, R.S., Oetting, E.R. and Swaim, R.C. (2002b). ‘The Driving Anger Expression Inventory: a measure of how people express their anger on the road.’ Behaviour Research and Therapy, 40, 717–37. Deffenbacher, J.L., Oetting, E.R. and Lynch, R.S. (1994). ‘Development of a driver anger scale.’ Psychological Reports, 74, 83–91. Deffenbacher, J.L., White, G.S. and Lynch, R.S. (2004). ‘Evaluation of two new scales assessing driving anger: the Driving Anger Expression Inventory and the Driver’s Angry Thoughts Questionnaire.’ Journal of Psychopathology and Behavioural Assessment, 26, 87–99. DePasquale, J.P., Geller, E.S., Clarke, S.W. and Littleton, L.C. (2001). ‘Measuring road rage: development of the Propensity for Angry Driving scale.’ Journal of Safety Research, 32, 1–16. Lajunen, T., Parker, D. and Stradling, S.G. (1998). ‘Dimensions of driver anger, aggressive and highway code violations and their mediation by safety orientation in UK drivers.’ Transportation Research Part F, 1, 107–21. Maxwell, J.P., Grant, S. and Lipkin, S. (2005). ‘Further validation of the propensity for angry driving scale in British drivers.’ Personality and Individual Differences, 38, 213–24. Parker, D., Lajunen, T. and Summala, H. (2002). ‘Anger and aggression among drivers in three European countries.’ Accident Analysis and Prevention, 34, 229–35. Reason, J., Manstead, A., Stradling, S., Baxter, J. and Campbell, K. (1990). ‘Errors

116

Driver Behaviour and Training – Volume III

and violations on the roads: a real distinction?’ Ergonomics, 33, 1315–32. Rolls, G.W.P., Hall, R.D., Ingham, R. and McDonald, M. (1991). Accident Risk and Behavioural Patterns of Younger Drivers. Southampton: AA Foundation for Road Safety Research. Spielberger, C.D. (1999). State-Trait Anger Expression Inventory (1st ed.). Odessa, FL: Psychological Assessment Resources. Sullman, M.J.M. (2006). ‘Driving anger amongst New Zealand drivers.’ Transportation Research Part F, 9, 173–84. Sullman, M.J.M., Gras, M.E., Cunill, M., Planes, M. and Font-Mayolas, S. (2007). ‘Driving anger in Spain.’ Personality and Individual Differences, 42, 701–13. Sullman, M.J.M., Meadows, M.L. and Pajo, K. (2002). ‘Aberrant driving behaviours amongst New Zealand truck drivers.’ Transportation Research Part F, 5, 217– 32. Underwood, G., Chapman, P., Wright, S. and Crundall, D. (1999). ‘Anger while driving.’ Transportation Research Part F, 2, 55–68. Walton, D. (1999). ‘Examining the self-enhancement bias: professional truck drivers’ perceptions of speed, safety, skill and consideration.’ Transportation Research Part F, 2, 91–113. West, R., French, D., Kemp, R. and Elander, J. (1993). ‘Direct observation of driving self-reports of driving behaviour and accident involvement.’ Ergonomics, 36, 557–67.

Chapter 10

Aggression and Non-aggression Amongst Six Types of Drivers1 Évelyne F. Vallières,1 Pierre McDuff,2 Robert J. Vallerand2 and Jacques Bergeron2 1 Télé-université, University of Québec at Montréal, Canada 2 Montréal University, Canada Introduction Aggressive driving can not only cause a lot of frustration among road users, but it can also be very dangerous, leading to violent incidents and driving accidents (Cook, Knight and Olson, 2005; Galovski and Blanchard, 2004; Galovski, Malta and Blanchard, 2006). It is a problem that road users come across frequently and, because of the diversity of aggressive driver types, it certainly represents a challenge for those concerned with road safety. Indeed, among aggressive drivers, there are those who are aggressive not only on the road but also in other aspects of their lives (Hennessy, 2005). Others, while not usually aggressive, admit to being sometimes aggressive while driving. In fact, whereas some drivers seem to always have an aggressive style of driving, others appear to become aggressive only in very speciﬁc situations or circumstances. Finally, there are those drivers who will rarely, if ever, be aggressive on the road, no matter what the situation or the circumstances they are facing. In sum, this seems to suggest that aggressive driving is multifaceted and that aggressive drivers are a heterogeneous group. In the last 30 years, aggressive drivers and their behaviours have been the focus of numerous studies. Research has shown that aggressive drivers are more often males than females, especially in cases of severe aggressive behaviours (Beck, Wang, Mitchell, 2006; Hennessy and Wiesenthal, 2001). It is also well established that aggressive drivers tend to be younger (for example, Shinar and Compton, 2004). Male drivers who hold vengeful and dangerous driving attitudes (Hennessy, 2005) and/or frequently commit driving violations (Maxwell, Grant, Lipkin, 2005), are also more aggressive on the road. Relative to situational variables, aggressive driving has been found to be more frequent at rush hours than at any other time of the day, and the presence of passengers has been associated with less aggressive driving (Shinar and Compton, 2004). Various variables related to personality and familial background have also been studied extensively. For example, in her recent studies, Malta and her 1 Authors’ note: This study was supported by Programme d’action concertée Fonds Nature et Technologies MTQ-SAAQ (project 2002-SR-86395). Québec, Canada.

118

Driver Behaviour and Training – Volume III

colleagues have found that numerous psychiatric and behavioural disorders, as well as familial antecedents such as oppositional deﬁant disorders, alcohol and substance abuse disorders, prevalence of self and familial anger problems and family conﬂicts, were associated with aggressive driving amongst young drivers (Malta, Blanchard and Freidenberg, 2005). Other dispositional measures, such as aggressiveness, narcissism or extraversion, have also been found to relate to aggressive driving (Britt and Garrity, 2006). Thus, the role of dispositional anger in fuelling aggressive behaviours has been established (Deffenbacher, Deffenbacher, Lynch and Richards, 2003; Deffenbacher, Richards and Lynch, 2004). Moreover, studies evaluating interventions aimed at reducing anger have conﬁrmed the major role played by anger in aggressive driving (Galovski and Blanchard, 2004). However, even if the role of anger as a trigger to aggressive driving has been conﬁrmed, few studies have looked at the psychological processes and cognitive variables leading to anger and aggressive driving reactions. Among the cognitive possible determinants of aggressive driving, perceived intent, which is usually implied in most deﬁnitions of aggressive driving, has rarely been studied. This could be due to the fact that driving behaviours such as risky driving are often considered aggressive, even though they may not have been performed intentionally (Maxwell et al., 2005). However, even if a risky driving manoeuvre is not done intentionally, attributional theory would predict that perceived intent of a negative encounter will lead to anger and, subsequently, to reactive aggression (Weiner, 1995; Weiner, Graham and Chandler, 1982). In other words, even if a frustrating manoeuvre was not done intentionally, the more intentional that frustrating manoeuvre will be perceived, the more angry the person subjected to it will feel and the higher the probability of an aggressive reaction. The link between perceived intent and reactive aggression has been supported in various contexts, in particular amongst aggressive children (see Crick and Dodge, 1994, for a review). These studies showed that in ambiguous situations aggressive children tend to infer greater intent and, consequently, to behave more aggressively than non-aggressive children (Dodge, 1980). This result suggests that perceived intent could be a signiﬁcant determinant of driving anger and aggressive driving. In fact, a recent study by Vallières and her colleagues (Vallières, Bergeron and Vallerand, 2005) shows support for the link between perceived intentionality, anger and reported aggressive driving reaction. However, the comparison of perceived intent among aggressive and non-aggressive drivers was not investigated. For example, in a case where aggressive and non-aggressive drivers were confronted by the same frustrating driving manoeuvre, would aggressive drivers perceive more intent than the non-aggressive ones? And if they did, would the difference be signiﬁcant only in situations that were ambiguous, such as in the Dodge study, or would it also occur in situations that were clearly intentional or clearly non-intentional? In other words, would it depend on the type of situation? Further, would aggressive drivers differ between themselves by the intensity of their reactions or by the characteristics of other drivers? The purpose of the present study was to address these questions. A ﬁrst step was to establish a typology of drivers, ranging from the most aggressive to the least aggressive, using multivariate cluster analysis. A second step was to verify if some

Aggression and Non-aggression Amongst Six Types of Drivers

119

personal and driving characteristics, driving behaviours, attitudes and beliefs could differentiate between the different types of drivers. These analyses were based on three sets of variables in two types of situations. The three sets of variables were perceived intentionality, anger and aggressive driving reactions. These variables were measured in two types of hypothetical situations in which a frustrating driving manoeuvre by another driver was done intentionally or unintentionally. Method Participants A total of 480 college and university students took part in the study (mean age = 25.6 years old, SD: 7.13). The majority of the participants (n = 276) were males (mean age = 25.8 years old, SD: 7.09), 161 were females (mean age = 25.3 years old, SD: 7.20) and gender was unknown for 43 of them. All subjects were part of one of four studies conducted on various aspects of aggressive driving and the results of which are described elsewhere. Procedure and questionnaire The questionnaires were given in class with the approval of the professor or teacher. Only those participants who had a driving licence and who voluntarily agreed to participate were given a questionnaire to ﬁll out. They were informed that the questionnaires were anonymous and were assured that their responses would be treated in a conﬁdential manner. All the questionnaires of the studies reported here contained hypothetical scenarios describing situations that one might experience while driving. There were two types of situations describing the driving manoeuvre of another driver. In some of the scenarios, the action of the other driver was clearly intentional (intentional condition), whereas in others it was clearly unintentional (unintentional condition). Some of the participants answered the questions for the two types of situation and some others for only one type of situation. When a questionnaire included more than one scenario of each type (maximum of two scenarios by type), a total score was obtained for each type of scenario. The scenarios had been selected on the basis of the results of previous pilot studies. All participants (n = 480) answered questions relative to socio-demographic variables, driving record, style of driving and love of driving. All participants were also asked to read each scenario as if the situation were really happening to them personally and to indicate on the appropriate scales the level of intentionality they attributed to the other driver’s action (one item), how angry they would feel in such a situation (one item) and their behavioural reactions (depending on the questionnaire: ﬁve to eight items: alphas: 0.68 to 0.88). Answers were rated on ﬁve point Likert scales. Cluster analyses were based on these three sets of variables in each of the conditions (intentional and unintentional).

120

Driver Behaviour and Training – Volume III

Some participants (n = 217) answered the following sub-scales of the Driving Anger Expression Inventory (DAX) (Deffenbacher, Lynch, Oetting and Swain, 2002): verbal aggression (12 items; α = 0.86), physical aggression (11 items; α = 0.89) and car using (11 items; α = 0.86). These sub-scales assess how verbally and physically aggressive one is reacting when angered while driving, and whether one uses the car to express anger. The items of the three subscales are rated on a ﬁve point scale ranging from one (‘almost never’) to ﬁve (‘almost always’). The validity of the DAX and its internal reliability has been found to be adequate (Deffenbacher, Lynch, Deffenbacher and Oetting, 2001; Deffenbacher et al., 2002). Participants (n = 183) also answered measures inspired from the TPB (Theory of Planned Behaviour: see Ajzen and Fishbein, 1980; Ajzen, 1991). The TPB is a theoretical framework that has been used in numerous studies to understand the role of some cognitive and affective variables in changing various attitudes and behaviours such as attitudes toward speeding (Parker, 2002) and the commission of driving violations (Parker, Manstead, Stradling, Reason and Baxter, 1992). Prevention programmes built on its theoretical constructs have been successful in changing risky driving behaviours (for example, Stead, Tagg, MacKintosh and Eadie, 2005). The items assessed: 1. indirect attitudes, such as behavioural beliefs and outcomes evaluation (for example, ‘To put pressure on the other driver is extremely bad/good.’ Six items each, α : 0.63 and 0.76); 2. direct attitudes (for example, ‘In this situation, the fact of tailgating is totally aggressive/harmless.’ Four items, α: 0.81); 3. indirect subjective norms, such as normative beliefs and motivation to comply (for example, ‘Most of my friends would agree with the fact that I tailgate in this situation.’ Six items each, α: 0.75 and 0.87); 4. direct subjective norms (for example, ‘In this situation, most people that are important to me would totally agree/disagree with the fact that I tailgate.’ Three items, α: 0.67); 5. indirect perceived control, such as control beliefs and perceived power (for example, ‘Police cars around the area would prevent me from tailgating.’ Five items each, α: 0.56 and 0.60); 6. perceived behavioural control (for example, ‘For me, to tailgate the other car is extremely difﬁcult/easy.’ Four items each, α: 0.76) ; and 7. behavioural intentions (for example, ‘In this situation I intend to tailgate the other car.’ Four items, α: 0.88). All measures were rated on scales ranging from –3 to +3 and were recoded on scales ranging from 1 to 7.

Aggression and Non-aggression Amongst Six Types of Drivers

121

Results Cluster proﬁles As discussed above, in order to identify possible subtypes of drivers, cluster analyses were conducted on the following three sets of variables (perceived intentionality, anger, behavioural reactions) in the two conditions (intentional, unintentional scenarios). Among the various cluster analysis strategies available, we combined Ward’s hierarchical clustering using squared Euclidean distances along a TwoStep clustering approach. Using SPSS, the TwoStep clustering approach allows the selection of the number of clusters according to a Bayesian Information criterion (BIC). Each of the variables included in the analysis were standardised in order to eliminate possible effects due to different variances. This approach suggested a six clusters solution. In addition to being theoretically meaningful, this solution was tested among selected subgroups of participants and was found sufﬁciently stable to be retained as the ﬁnal solution. As 21 participants did not ﬁt any of the six clusters, they were excluded from further analyses. As can be seen in Table 10.1, clusters one and two, which include 54 and 38 persons respectively, consist of those drivers who are the most aggressive in intentional as well as non-intentional situations. Together, these clusters represent 20 per cent of the sample, and as they consist of the most aggressive drivers we refer to them as the highly aggressive proﬁle. The most numerous clusters, three and four, include 106 and 110 participants respectively and together they represent 47 per cent of the sample. Because their mean scores on behavioural aggressive reactions are in-between those of the other groups, they are called the mildly aggressive proﬁle. There are 98 and 54 participants respectively classiﬁed in clusters ﬁve and six (33 per cent of the sample). The mean score of these drivers in both types of scenarios suggest that these drivers are rarely, if ever, aggressive and are qualiﬁed, for that reason, as the non-aggressive proﬁle. A closer look at the two groups of the aggressive proﬁle in Table 10.1 shows that they differ from each other by the intensity of anger and behavioural reaction they report, particularly in the intentional condition. As can be seen, group two reports more anger and more intense aggressive reactions than group one in this condition. In the unintentional condition, clusters one and two differ only by the level of intentionality perceived. Indeed, participants in cluster one perceive the other driver’s action as being intentional even though it is clearly unintentional. In the mildly aggressive proﬁle, both groups of drivers (three and four) differ not only in the way they perceive the other driver’s intention, but also by the intensity of their affective and behavioural reactions. In the intentionality condition, group three participants react similarly to the aggressive proﬁle in their way of perceiving intentionality and in their anger reaction. We ﬁnd a somewhat inverse pattern with group four, as the participants from that cluster react similarly to the non-aggressive proﬁle in the intentional condition. As for the non-aggressive proﬁle, even though both groups show negative mean scores on all the variables in both conditions, these scores conﬁrm that the group six participants are deﬁnitively the least aggressive.

122

Table 10.1 Mean standardised scores on the scales deﬁning the young driver clusters Clusters Scale

1. (n = 54)

Non-Intentional Scenario Perceived 0.82 intention Anger 0.81 Reaction 1.30 Demographics Age (years) 24.2 (5.2) Sex1 12.8 (% Male) Sex 8.9 (% Female)

3. (n = 106)

4. (n = 110)

5. (n = 98)

6. (n = 54)

0.34

0.45

-0.19

-0.17

-0.67

11.8**

0.38

1.25 1.93

0.75 0.29

-0.31 -0.42

-0.78 -0.51

-0.58 -0.75

72.1** 108.4**

0.71 0.77

0.10

0.08

0.34

-0.19

-1.47

43.3**

0.61

0.89 1.43

0.39 0.27

0.24 -0.06

-0.61 -0.73

-1.36 -1.04

64.7** 101.6**

0.69 0.77

25.3(7.5)

24.5(5.0)

25.4(7.6)

27.5(8.6)

26.6(7.3)

2.7*

0.03

7.8

21.6

23.8

23.4

10.6

8.9

25.0

25.6

18.5

13.1

Note: For each measure, higher scores indicate higher levels of the variable *p < 0.05, **p < 0.001 1 X2 = 3.9 ns

F

ETA

Driver Behaviour and Training – Volume III

Intentional Scenario Perceived 0.40 intention Anger 0.70 Reaction 0.84

2. (n = 38)

Aggression and Non-aggression Amongst Six Types of Drivers

123

Table 10.1 also presents the distribution of those subtypes according to two of the most used socio-demographic variables: age and gender. Analysis of variance on the mean age between our six groups reveals an overall age effect but no pairwise difference according to the Scheffe post hoc analysis. On the other hand, we found no difference according to gender, as a comparable percentage of males and females can be observed in each of the clusters. Self-reported aggression when angered In order to validate the nature of the six subtypes uncovered, we compared their mean scores on the three DAX subscales (see Table 10.2). The results obtained with the univariate analysis of variance showed important mean differences for the six groups on verbal and physical aggression and use of car. Scheffe post hoc analysis revealed that the two groups of the aggressive proﬁle are signiﬁcantly more aggressive than the other proﬁles on the three measures and that they do not differ between themselves on these same measures. Even though groups three and four of the mildly aggressive proﬁle do not signiﬁcantly differ from one another on those measures, both are signiﬁcantly more verbally aggressive than participants from the non-aggressive proﬁle. Table 10.2

Cluster DAX subscales: Variables Verbal aggression1 Physical aggression2 Car using aggression3

Mean of reported verbal aggression, physical aggression and car using aggression

Mean aggressive behaviour (SD) 1. n = 23 39.9 (6.6) 21.5 (9.8) 28.2 (9.8)

2. n = 17 39.1 (7.9) 19.5 (7.6) 29.0 (6.7)

3. n = 48 33.2 (7.4) 15.2 (5.1) 21.1 (7.0)

4. n = 58 30.5 (7.8) 14.3 (4.1) 17.0 (5.7)

5. n = 47 23.6 (7.1) 12.8 (3.8) 16.8 (6.1)

6. n = 24 25.8 (7.7) 12.5 (2.0) 13.6 (3.3)

F(5,214) 24.3 p = 0.01 12.0 p = 0.01 23.1 p = 0.01

η2 0.35 0.21 0.34

Note: Differences between groups are signiﬁcant at p < 0.01. 1 Score: 12 to 60 2 Score: 11 to 55 3 Score: 11 to 55

Self-reported driving variables Next, the subgroups were compared in terms of their driving habits, speeding, number of speeding tickets, accident(s) and their love of driving. The results (Table 10.3) show a graduated decline in percentage of those reporting frequent speeding excess according to the subtypes. In fact, while 82 per cent of the aggressive drivers

Driver Behaviour and Training – Volume III

124

reported having often exceeded the speed limit, that percentage falls progressively to 53.8 per cent for the non-aggressive groups. Another signiﬁcant difference between groups was found for speeding ticket violations, with group two participants reporting the highest percentage of speeding tickets relative to the others. Remember that this group is the one reporting the most aggressive reactions in both types of conditions. That group also reports the highest percentage of accidents and love of driving, but overall those differences between groups are not signiﬁcant. Table 10.3

Cluster Variables

Percentage of high speed driving, accidents, speeding tickets, enjoyment of driving and the mean of reported verbal and physical aggression and car-use aggression

1. n = 51

2. n = 34

Driving behaviours (%) High 82.4% speed 74.3% driving1 One and + accidents2 25.5% 36.1% Speeding tickets3 Love driving4

3. n = 102

4. n = 110

5. n = 98

6. n = 52

X2

71.6%

70.9%

64.3%

53.8%

X2 = 11.8 *

19.6%

26.4%

23.5%

25%

X2 = 4.2, ns.

41.2%

50%

41.6%

33.0%

38.1%

19.2%

X2 = 11.8 *

62.7%

73%

66.3%

58.2%

61.9%

53.1%

X2 = 5.1, n.s.

1

% of those who like to speed often % of those who had had one or more accidents in the last three years 3 % of those who had one or more speeding tickets in the last three years 4 % of those who love to drive enormously Differences between groups on high speed driving and speeding tickets were signiﬁcant at p < 0.05. In bold, adjusted standardised residual are greater than 2. 2

Attitudes, subjective norms, perceived control and behavioural intention Finally, the six groups were compared on ten subscales of the TPB (Table 10.4). Results of the analyses conducted on the direct and indirect variables related to the TPB revealed signiﬁcant differences between clusters on four of the variables: outcome evaluation, motivation to comply, direct perceived control and intention to tailgate in the intentional condition. These differences mean that some of the drivers having an aggressive proﬁle (group one) agree more than the other drivers with having behaviours such as putting pressure on another driver by tailgating in order to progress more rapidly in trafﬁc. Some of these aggressive drivers (group two) also think that they have a high control over their driving, more so than the other drivers,

Aggression and Non-aggression Amongst Six Types of Drivers

125

and (in groups one and two) when they feel they are prevented from driving the way they wish to drive, they feel more justiﬁed to drive aggressively. Table 10.4

Attitudes, subjective norms, perceived control and intention to act aggressively in the intentional condition

Cluster 3. n= 53

4. n = 21

5. n= 40

ATTITUDES, M (SD) Indirect attitudes Behavioural 5.28 beliefs (0.73) Outcomes 4.57 evaluation (1.5)

5.00 (0.63) 4.33 (1.0)

5.20 (0.93) 4.26 (1.1)

5.00 (0.63) 3.67 (1.5)

5.42 (0.84) 3.79 (1.2)

5.18 (1.1) 3.10 (1.0)

3.32 (0.78)

3.38 (0.75)

3.04 (0.63)

2.58 (0.92)

3.05 (0.85)

2.66 (0.72)

SUBJECTIVE NORMS, M (SD) Indirect subjective norms Normative 3.85 3.33 beliefs (1.0) (1.2) Motivation to 4.36 4.83 comply beliefs (1.3) (1.5)

3.86 (0.97) 5.34 (1.1)

3.33 (0.82) 5.33 (0.52)

3.63 (0.95) 5.58 (0.84)

3.27 (0.65) 5.36 (0.67)

3.90 (0.84)

2.61 (1.3)

3.88 (1.2)

3.48 (0.77)

5.33 (0.82) 4.50 (0.55) 3.62 (1.2)

5.05 (0.91) 4.31 (0.82) 4.90 (0.98)

5.63 (1.0) 4.73 (0.65) 4.00 (1.5)

1.6 ns 0.9 ns 3.3 p< 0.01

3.21 (1.4)

4.63 (1.1)

3.52 (1.4)

7.2 p< 0.00

Direct attitudes

Direct subjective norms

1. n= 23

3.86 (1.01)

3.61 (0.90)

PERCEIVED CONTROL, M (SD) Indirect behavioural control Control beliefs 4.71 5.00 4.97 (0.99) (0.63) (0.75) Perceived power 4.14 4.17 4.20 (0.66) (0.75) (0.93) Direct 5.16 4.79 4.85 perceived (1.1) (0.84) (0.84) control BEHAVIOURAL INTENTION, M (SD) Intention to 5.38 5.46 4.70 tailgate (0.76) (0.87) (0.90) Range of scores: 0–7

6. n= 22

F ()

2. n= 24

Variables

0.4 ns 2.5 p< 0.05 1.7 ns

1.1 ns 2.7 p< 0.05 2.1 p= 0.08

η2

0.02 0.13

0.09

0.06 0.14

0.11

0.09 0.05 0.16

0.30

126

Driver Behaviour and Training – Volume III

Discussion The results presented here support a classiﬁcation of drivers based on the three set of variables studied in the present research, namely perceived intentionality, reported anger and aggressive behaviours. In that sense, the present study is innovative. A classiﬁcation is uncovered based on variables that parallel a motivational sequence and are process related (Vallières et al., 2005). Indeed, the typical cluster analyses are usually based on personal characteristics and personality variables which are less amenable to change than the process related variables used in the present study. The cluster analyses show that drivers can be differentiated along a continuum from high aggression to low aggression. Furthermore, the present results reveal that aggressive drivers are not a homogeneous group and that there is more than one type of aggressive drivers. The same can be said about the mildly aggressive and the nonaggressive groups, as neither appear to be homogeneous groups. The present results also reveal that the two most aggressive groups (aggressive proﬁle) differ from each other by their cognitive interpretation of the situation and by the intensity of some of their reactions. For example, in the intentional condition, even though both groups interpret the situations in a similar way, the second group’s reactions are deﬁnitively more intense. However, in the non-intentional condition, the two groups interpret differently the situations. Indeed, if group two perceives correctly the low intentionality of the other driver’s action, group one perceives the depicted driver’s action as intentional even though it is clearly unintentional. Another surprising result relates to the comparable percentage of males and females in the different groups. That result disagrees with those found in other studies (for example, Beck, et al.,; Hennessy and Wiesenthal, 2001; Shinar and Compton, 2004) but is similar to some results obtained by Deffenbacher and his colleagues (Deffenbacher et al., 2003). These researchers suggest that there are probably more similarities than differences between male and female students in terms of aggressive driving reactions. Our results seem to agree with that statement. However, since our clusters are based on process variables (perceived intention, anger, reaction) we may also make the hypothesis that our results ensue from the type of variables used in the present cluster analyses. Our results could also mean that the actual process leading to driving aggressiveness is the same for both genders. Further studies should explore more thoroughly this hypothesis. The results presented here could be useful in planning better training for young drivers. In particular, in drawing their attention on how their perception of another driver’s action can have important impact on their own or on others affective and behavioural aggressive reactions, they may become more aware of the importance of not attributing intent to others’ driving actions thoughtlessly. In spite of the fact that the validity of the categories uncovered in the present study needs to be conﬁrmed in other research, particularly with participants other than university or college students, the ﬁndings of the present study are of interest, notably in how they suggest that drivers vary greatly by their interpretation of other drivers’ actions and by the intensity of their reactions to these interpretations. As such, driver training providers might consider this aspect in training inexperienced drivers.

Aggression and Non-aggression Amongst Six Types of Drivers

127

References Azjen, I. (1991). ‘The theory of planned behaviour.’ Organizational Behaviour and Human Decision Processes, 50, 179–211. Azjen, I., Fishbein, M. (1980). Understanding Attitudes and Predicting Social Behaviour. Englewood Cliffs, NJ: Prentice Hall. Beck, K.H., Wang, M.Q. and Mitchell, M.M. (2006). ‘Concerns, dispositions and behaviours of aggressive drivers: what do self-identiﬁed aggressive drivers believe about trafﬁc safety?’ Journal of Safety research, 37, 159–65. Britt, T.W. and Garrity, M.J. (2006). ‘Attributions and personality as predictors of the road rage response.’ British Journal of Social Psychology, 45, 127–47. Cook, L.J., Knight, S. and Olson, L.M. (2005). ‘A comparison of aggressive and DUI crashes.’ Journal of Safety Research – Trafﬁc Records Forum proceedings, 36, 491–3. Crick, N.R. and Dodge, K.A. (1994). ‘A review and reformulation of social information-processing mechanisms in children’s social adjustment.’ Psychological Bulletin, 115, 74–110. Deffenbacher, J.L., Deffenbacher, D.M., Lynch, R.S. and Richards, T.L. (2003). ‘Anger, aggression and risky behaviour: a comparison of high and low anger drivers.’ Behaviour Research and Therapy, 41, 701–718. Deffenbacher, J.L., Lynch, R.S., Deffenbacher, D.M and Oetting, E.R.,(2001). ‘Further evidence of reliability and validity for the Driving Anger Expression Inventory.’ Psychological Reports, 89, 535–40. Deffenbacher, J.L., Lynch, R.S., Oetting, E.R. and Swain, R.C. (2002). ‘The Driving Anger Expression Inventory: a measure of how people express their anger on the road.’ Behaviour Research and Therapy, 40, 717–737. Deffenbacher, J.L.,Richards, T.l. and Lynch, R.S. (2004). ‘Anger, aggression, and risky behaviour in high anger drivers.’ In J.P. Morgan (ed.). Focus on Aggression Research. Hauppauge, NY, US: Nova Science Publishers, 115–156. Dodge, K.A. (1980). ‘Social cognition and children’s aggressive behaviour.’ Child Development, 51, 162–70. Galovski, T.A. and Blanchard, E.B. (2004). ‘Road rage: a domain for psychological intervention?’ Aggression and Violent Behaviour, 9, 105–127. Galovski, T.E., Malta, L.S. and Blanchard, E.B. (2006). ‘Aggressive driving: signiﬁcance and scope of the problem.’ In Galovski, Tara E.; Malta, Loretta S.; Blanchard, Edward B. (eds). Road Rage: Assessment and Treatment of the Angry, Aggressive Driver, 3–14; Washington, DC, US: American Psychological Association, xi, 250. Hennessy, D. A. (2005). ‘Driving vengeance and wilful violations: clustering of problem driving attitudes‘ Journal of Applied Social Psychology, vol. 35(1), 61–79. Hennessy, D.A. and Wiesenthal, D.L. (2001). ‘Gender, driver aggression, and driver violence: an applied evaluation.’ Sex Roles, 44, 11/12, 661–76. Malta, L.S., Blanchard, E.B. and Freidenberg, B.M. (2005). ‘Psychiatric and behavioural problems in aggressive drivers.’ Behaviour Research and Therapy, Maxwell, J.P., Grant, S. and Lipkin, S. (2005). ‘Further validation of the propensity

128

Driver Behaviour and Training – Volume III

for angry driving scale in British drivers.’ Personality and Individual Differences, vol. 38(1), 213–224. Parker, D. (2002). ‘Changing drivers’s attitudes to speeding using the theory of planned behaviour.’ In Rutter, and Quine, L. (eds). Changing Health Behaviour. Buckingham: Open University Press, 138–52. Parker, D., Manstead, A.S.R., Stradling, S.G., Reason, J.T. and Baxter, J.S. (1992). ‘Intentions to commit driving violations: an application of the theory of planned behaviour.’ Journal of Applied Psychology, 77, 94–101. Stead, M., Tagg, S., MacKintosh, A. M. and Eadie, D. (2005). ‘Development and evaluation of a mass media theory of planned behaviour intervention to reduce speeding.’ Health Education Research, 20, 1, 36–50. Shinar, D. and Compton, R. (2004). ‘Aggressive driving: an observational study of drivers, vehicle, and situational variables.’ Accident Analysis and Prevention, 36, 429–37. Stradling, S.G. and Parker, D. (1997). ‘Extending the theory of planned behaviour: the role of personal norm, instrumental beliefs, and affective beliefs in predicting driving violations.’ In J.A. Rothengatter and E. Carbonell (eds). Trafﬁc and Transport Psychology: Theory and Application, Amsterdam: Pergamon, 367–74. Vallières, E.F., Bergeron, J. and Vallerand, R.J., (2005). Weiner, B., (1995). Judgments of Responsibility: A Foundation for a Theory of Social Conduct, New York: Guilford. Weiner, B., Graham, S. and Chandler, C.C. (1982). ‘Pity, anger, and guilt: an attributional analysis.’ Personality and Social Psychology Bulletin, 8, 226–32.

Chapter 11

The Inﬂuence of Age Differences on Coping Style and Driver Behaviour Elizabeth Andrews and Stephen Westerman University of Leeds, UK Introduction There are established age differences in driver performance (Parker, McDonald, Rabbitt and Sutcliffe, 2000). In this research we are particularly concerned with the experiences and performance of older drivers. Relative to younger drivers, older drivers are more frequently involved in speciﬁc types of accidents, such as those involving multiple-vehicles and those at complex junctions (Hakamies-Blomqvist, 1994). In large measure, age differences in driving performance may be attributable to age-related changes in physical and cognitive abilities. Older adults report problems with visual and auditory functioning that impact driving (Kline and Scialfa, 1996) and are also likely to experience decline in cognitive abilities (although see Rabbitt, 1993), including speed of information processing (Salthouse, 1992) and memory (Craik and Jennings, 1992). Impaired cognitive function has also been related to poorer driving ability in older drivers (McKnight and McKnight, 1999). Studies of age-related differences in self-report measures of driving behaviour and driver stress have produced results that are only partially consistent with these patterns of age-related changes in ability and driving performance. When considering the typology of aberrant driving behaviours developed by Reason, Manstead, Stradling, Baxter and Campbell (1990) (that is, violations, errors and lapses), a number of studies report no age effects with regard to scores for more serious errors (for example, Aberg and Rimmo, 1998; Parker, Lajunen and Stradling, 1998; Westerman and Haigney, 2000). Violations (for example, running red lights) have been consistently found to be more often reported by younger drivers (Parker et al., 1998; Westerman and Haigney, 2000). Given the described age-related decline in physical and cognitive abilities, it might be anticipated that older drivers would experience greater driving-related stress. This would be consistent with current cognitive models of stress (for example, Lazarus and Folkman, 1984) in which appraisal of coping resources relative to task performance demands is a central determinant of the individual’s stress experience. However, it has been reported that older drivers experience relatively lower levels of general driver stress (Parker et al., 2001; Matthews, Dorn and Glendon, 1991). In part this may be attributable to reduced aggression in older drivers but greater associated concentration (Matthews, Desmond, Joyner, Carcary and Gilliland, 1997) – although this latter effect was very small in research by Westerman and Haigney (2000).

130

Driver Behaviour and Training – Volume III

There are a number of potential explanations for a degree of dissociation between driving performance and self-reports of driving behaviour/stress. These include the following, non-mutually exclusive, possibilities. First, it may be that there are age differences in willingness to report certain types of driving behaviour. Somewhat contrary to this, Lajunen and Summala (2003) reported little social desirability bias on self-reported measures of poor driving behaviours (for example, errors) in a public compared to a private setting, and drivers generally report the quality of their driving as being very good (Groeger and Grande, 1996). This could result in restricted range effects when using self-reporting. In this context, it might be argued that older drivers have unrealistic assessments of their abilities and do not take into account age-related changes. However, this runs counter to the ﬁndings of Parker, Macdonald, Sutcliffe and Rabbitt (2001). Nevertheless, it is possible that older drivers are more susceptible to memory deﬁcits for driving performance. Related to this, one study reported that 14 per cent of people involved in injury-provoking accidents failed to recall the event a year later (Loftus, 1993) and memories for near-accidents are also forgotten fairly rapidly (Chapman and Underwood, 2000). The relatively lower stress levels self-reported by older drivers may result from the effects of additional driving experience (over years) that produces more advanced driving skills. Decreased mental workload and therefore an increased level of available attention resources is associated with greater driving experience (Patten, Kircher, Ostlund, Nilsson and Svenson, 2006). A related explanation of age-related performance maintenance is that through extended practice important components of the driving task are sufﬁciently automated to be protected from the effects of age-related cognitive decline (Hasher and Zacks, 1979). However, contrary to these suggestions, for some older drivers, reduced driving exposure (for example, low mileage) may equate to ‘inexperience’ and lower levels of driving ability. The ﬁnal potential inﬂuencing factor that we consider here is that older drivers adopt compensatory strategies that protect their driving performance (McKnight and McKnight, 1999). This could happen at a ‘macro’ level, such that they are more selective of when they drive and consequently the driving situations they experience (for example, not driving on busy roads or at night) and this has a positive safety effect (Hakamies-Blomqvist, 1994). However, the evidence is rather mixed with regard to the extent that drivers with cognitive and visual impairments limit their driving exposure (Stutts, 1998). Compensatory processes might also operate at a ‘micro’ level, such that older drivers adapt their driving technique to match their physical and cognitive abilities. This would enable older drivers to achieve a better match between driving demands and resources available to meet those demands. Age-related changes in coping strategy (see Gulian et al., 1989b) would be consistent with age differences in sensory, physical and cognitive abilities (see Anstey Wood, Lord and Walker, 2005; Rabbitt, 1993) and personality (see Matthews et al., 1991) and may therefore be implicated in changes in driving stress and behaviour. The concept of coping plays a particularly important role in current conceptualisations of stress. Stress is thought to result when coping resources are perceived as insufﬁcient to meet anticipated situational demands (Lazarus and Folkman, 1984). Ineffective coping strategies may lead to the experience of

The Inﬂuence of Age Differences

131

higher levels of driver stress (McCrae and Costa, 1986). This applies to drivingrelated stress (Gulian, Matthews, Glendon, Davies and Debney, 1989a). The effects of driver stress will be determined by drivers’ assessments of their availability of coping resources in relation to experienced driving demands (Matthews, Dorn and Hoyes, 1992). In this paper we examine the association between measures of driving behaviour, driver stress and driving performance, and consider the effects of age differences. In particular, we focus on the possibility that age-related differences in methods of coping impact assessments of driving behaviour, driving performance and also the experience of driving stress. Method Participants Forty-two younger drivers (22 females, 20 males) aged between 20 and 40 years (mean 31.0, SD 5.9 years) and 40 older drivers (22 females, 18 males) aged 60 years or above (mean 68.5, SD 6.3 years) were recruited from the Leeds Advanced Driving Simulator subject pool, through contact with local community, sporting and voluntary organisations, and through personal contacts within and outside the University of Leeds. Volunteers were paid to attend individual sessions lasting approximately one hour. Both younger and older samples were experienced drivers. However, as anticipated, the older sample had been driving for signiﬁcantly longer (mean 43.1, SD 10.4 years) than the younger sample (mean 11.3, SD 5.8 years) (p < 0.001). The majority of volunteers were currently driving either every day (66.7 per cent of younger and 80 per cent of older sample) or two to three days a week (23.8 per cent of younger and 15 per cent of older drivers). Most drivers reported driving between 5000 and 10 000 miles per year (20 younger and 22 older). Nine younger (21.4 per cent) and eleven older drivers (27.5 per cent) reported annual mileage less than 5000 and ﬁve younger and two older drivers reported annual mileage in excess of 15 000 miles. There were no signiﬁcant age differences among frequency of driving episodes or annual mileage (p > 0.05). Self-report measures Volunteers completed items relating to driving experience (number of years licence held, annual mileage and driving frequency), accident involvement and convictions for speeding, driving under the inﬂuence of drink/drugs and careless/dangerous driving (all within the last three years). Accidents were classiﬁed as minor (< £500 damage to vehicle/property AND no medical treatment) or major (either someone required medical treatment OR > £500 damage to vehicle/property OR both). Volunteers also completed a questionnaire pack that included the Driver Stress Inventory, the Driver Coping Questionnaire and the Driver Behaviour Questionnaire.

132

Driver Behaviour and Training – Volume III

The Driver Behaviour Questionnaire (DBQ: see Parker et al., 1995) measures driving errors (for example, underestimating the speed of an oncoming vehicle when overtaking), violations (for example, disregarding the speed limits) and lapses (for example, misreading signs and taking the wrong turn). British studies (for example, Parker et al. (2000); Parker et al. (1995a; 1995b) and in other countries (for example, Australia: see Blockley and Hartley (1995), Sweden: Aberg and Rimmo (1998)) broadly support this three factor structure. The split-half reliability of the DBQ in the current study was very good (r = 0.83). The Driver Stress Inventory (DSI: Matthews, Desmond, Joyner, Carcary and Gilliland, 1996) is a reﬁnement of the Driver Behaviour Inventory (DBI: Gulian et al., 1989a). It measures driver stress vulnerability traits relating to aggression (for example, ‘it annoys me to drive behind a slow moving vehicle’), dislike of driving (for example, ‘I ﬁnd myself worrying about my mistakes’) and alertness – redeﬁned as hazard monitoring (for example, ‘I make an effort to look for potential hazards’). It also includes two new factors relating to thrill-seeking (for example, ‘I sometimes like to frighten myself a little while driving’) and fatigue proneness (for example, ‘I become inattentive to road signs when I have to drive for several hours’). The DSI has been found to be a valid and reliable measure of driver stress (Matthews et al., 1996). The split-half reliability of the DCQ in the current study was good (r = 0.71). The Driver Coping Questionnaire (DCQ: Matthews et al., 1996) was developed to assess individual differences in coping, by asking drivers how they deal with driving speciﬁcally when it is difﬁcult or stressful and includes items concerning both explicit behaviours and internal psychological coping strategies. The DCQ measures ﬁve dimensions of coping: confrontive coping (for example, ‘showed other drivers what I thought of them’), task-focused (for example, ‘made sure I avoided reckless actions’), emotion-focused (for example, ‘blamed myself for getting too emotional’), reappraisal (for example, ‘tried to gain something worthwhile from the drive’) and avoidance (for example, ‘cheered myself up by thinking about things unrelated to the drive’). The DCQ has been found to be a valid and reliable measure of driver stress (Matthews et al., 1996). The split-half reliability of the DCQ in the current study was good (r = 0.7). Results Age differences in self-report assessments A series of t-tests were used to examine age differences in: (i) driver behaviour; (ii) driver stress and (iii) coping style. Younger drivers reported a relatively greater number of violations than older drivers, t(73.38) = –4.06, p < 0.001, but there were no age differences for errors or lapses. However, female drivers reported signiﬁcantly more lapses compared to male drivers: t(80) = –2.53, p < 0.05 (controlling for age). With regard to driver stress, younger drivers reported higher levels of aggression, t(80) = –3.62, p = 0.001, and thrill-seeking, t(80) = –3.19, p < 0.01, whilst older drivers reported higher levels of hazard-monitoring, t(80) = 4.45, p < 0.001.

The Inﬂuence of Age Differences

133

There were signiﬁcant age differences on three coping strategies. Younger drivers adopted a more confrontive coping strategy, t(74.85) = –4.57, p < 0.001, whereas older drivers placed a relatively greater emphasis on task-focused, t(79) = 2.95, p < 0.01, and reappraisal coping strategies, t(79) = 2.23, p < 0.05, to manage their driver stress. Age differences in emotion-focused and avoidance styles of coping with driver stress were non-signiﬁcant. Gender differences were noted for emotionfocused coping (women scored higher), t(79) = –2.23, p < 0.05 (associated more with older females, partial r = 0.24, p < 0.05). Age differences in accidents and speeding Accident involvement was generally low and, given skewed distributions, age differences should be interpreted with caution. However, the younger sample reported signiﬁcantly more major accidents than the older sample, t(51.173) = –2.26, p < 0.05. One older driver reported having one major accident. Seven younger drivers reported one, one driver reported two and another reported three major accidents. Eighty-ﬁve per cent of the older sample and 81 per cent of the younger sample reported no minor accidents. Five older and six younger drivers reported having one, one younger reported two and another reported four minor accidents (age differences non-signiﬁcant). There was no signiﬁcant age difference in speeding convictions (seven older and ten younger drivers reported speeding convictions). Associations between coping style and driver behaviour To examine associations between age group and coping style with respect to driving behaviours, driver stress and driving performance, a series of regression analyses was conducted. Each of the following was a dependent variable in a separate regression equation: (i) each DBQ scale; (ii) each DSI scale; (iii) minor and major accidents and (iv) speeding convictions. Given that older drivers had a greater level of driving experience than the younger sample, the effect of self-reported driving experience was ‘controlled’ by entering this ﬁrst into each of the regression equations. Following this age group and ratings for a speciﬁc coping style were entered. Finally a vector representing the interaction of age group and coping style was entered. A summary of results for the regression equations is provided in Figure 11.1. Effects of age group have been reported above (although without controlling for driving experience). Here, we focus on the effects of coping style and the interactions between age group and coping style. Reappraisal coping was negatively associated with aggression (β = –0.26, p < 0.05) and violations (β = –0.22, p < 0.05). However, there were no signiﬁcant interactive effects of age group. Task-focused coping was positively associated with hazard monitoring scores (β = 0.40, p < 0.001) and also speeding convictions (β = 0.28, p < 0.05). It was negatively associated with aggression (β = –0.42, p < 0.001), violations (β = –0.38, p < 0.001) and thrill-seeking (β = –0.22, p < 0.05). Interactive effects involving age group were not signiﬁcant.

Driver Behaviour and Training – Volume III

134

Positive Association Reappraisal

Task Focus

Confrontive

Avoidance

Negative Association

Emotion Focus

Interaction

Hazard Monitor (DSI)

Aggression (DSI)

Violations (DBQ)

Thrill Seeking (DSI)

Speeding

Minor Accidents

Major Accidents

Dislike of Driving (DSI)

Fatigue Proneness (DSI)

Figure 11.1 Schematic of coping style and major associations with driving behaviour Confrontive coping was positively associated with aggression (β = 0.65, p < 0.001), violations (β = 0.49, p < 0.001) and thrill seeking scores (β = 0.53, p < 0.001) and minor accidents (β = 0.86, p < 0.05). There were no signiﬁcant interactive effects involving age group. Avoidance coping was negatively associated with aggression scores (β = –0.23, p < 0.05) and positively associated with major accidents (β = 0.78, p < 0.05). The interaction between age group and avoidance coping was predictive of thrill-seeking behaviours (β = 0.99, p < 0.05), such that younger drivers with relatively lower avoidance coping scores were more thrill-seeking, whereas older drivers with relatively higher avoidance coping scores were relatively more thrill-seeking (see Figure 11.2). Emotion-focused coping was positively associated with dislike of driving (β = 0.72, p < 0.001) and fatigue proneness (β = 0.30, p < 0.01). The association with thrill seeking also approached signiﬁcance (p = 0.052). Emotion-focused coping was also positively associated with speeding convictions (β = 0.27, p < 0.05). There was also a signiﬁcant interaction between age group and emotion-focused coping style (β = 0.85, p < 0.05), such that younger drivers with relatively lower emotion-focused coping scores were more thrill-seeking, whereas older drivers with relatively higher emotion-focused coping scores were less thrill-seeking. Discussion This study examined age differences in driver behaviour, driver stress and driving performance. Associations between these variables were explored, and particular consideration was given to the potential inﬂuence of coping style on age differences in driving behaviour and driving stress.

The Inﬂuence of Age Differences

ii)

i)

DSI thrill seeking

135

80.00

80.00

70.00

70.00

60.00

60.00

50.00

50.00

40.00

40.00

30.00

30.00

20.00

20.00

10.00

10.00

0.00

0.00 0.00

25.00

50.00

75.00

emotion-focused coping

20.00

40.00

60.00

80.00

avoidance c oping

Figure 11.2 Interactions between: (i) age group and emotion-focused coping and (ii) age group and avoidance coping, for thrill-seeking. Solid regression line indicates the younger group Consistent with previous research (Aberg and Rimmo, 1998; Parker et al., 1995a; 1995b), younger drivers reported higher levels of violations (DBQ). They were also more aggressive and thrill-seeking in their driving (DSI). Aggression and thrillseeking are strongly associated with a confrontive coping style, with a concomitant reduction in task-focused coping for managing driver stress (Matthews et al., 1996). When considering coping styles, relative to older drivers, younger drivers adopted a more confrontive style. Confrontive coping is established as the most age-sensitive coping style and the strongest single predictor of dangerous driving behaviours; for example, violations and minor accidents (for example, Matthews et al., 1996). Results from the current study support these ﬁndings (Figure 11.1 above). Taken as a whole, this pattern of responses suggests that younger drivers are more ‘active/ external’ in their coping responses to driving situations. In contrast, older drivers reported relatively higher levels of hazard monitoring (DSI). Again, this is consistent with previous research (Matthews et al., 1996). When considered in combination with reports of relatively greater emphasis on ‘task focused’ and ‘reappraisal’ coping strategies, this suggests that older drivers adopt a more ‘passive/internal’ approach when coping with driving situations. Although age samples differed in levels of violations, there were no signiﬁcant age differences in errors or lapses (DBQ). Again, this is reasonably consistent with existing ﬁndings from samples aged between 17 and 78 years of age (Parker et al., 1995a; 1995b; Reason et al., 1990). Errors and lapses do not appear to have a strong association with accidents even after controlling for the effects of age and gender (Matthews et al., 1997). This relationship may alter with increasing age. Lapses (for example, misreading signs and taking the wrong turning off a roundabout) were the most frequently reported behaviour when the DBQ was applied to an extended age range sample of 50 to 90 year olds (Parker et al., 2000), and were associated with accident involvement. Female drivers in the current study reported high scores for lapses. Misreading signs and ‘taking the wrong turning off a roundabout’ may not

136

Driver Behaviour and Training – Volume III

be considered a serious error unless it is carried out at the expense of safe driving practice. Older drivers report low levels of driver stress and of aggression with high levels of alertness (for example, Westerman and Haigney, 2000) and hazard monitoring (for example, Matthews et al., 1996), and in this manner their accident risk is reduced. Accident rates for the sample were generally low. However, younger participants reported having a greater number of major accidents over the three-year period preceding the study. In this study older drivers reported fewer major accidents and there was no signiﬁcant difference in minor accidents. If self reported levels of deviant driving behaviours (for example, violations, accidents and convictions) were underreported due to impression management (Lajunen and Summala, 2003) or from memory loss (Loftus, 1993) this would be more likely to depress, rather than inﬂate, the effects demonstrated in the current study. Assuming that accident reports were accurate (see above) this suggests that self-reports of driving behaviour are reasonably reﬂective of ability (Parker et al., 2001). Following from the above, it may be that patterns of accident involvement are inﬂuenced by coping style to the extent that older experienced drivers adopt a more conscientious approach to driving, and younger drivers’ accident involvement may be related to increased propensity for driving violations (Parker et al., 1995a; 1995b). Driving experience gained over many years is likely to produce more advanced driving skills, and may be associated with a reduction in driver stress and mental workload (Patten et al., 2006), and therefore the possible confounding effect of driving experience was controlled in the current study. Older drivers may compensate for decline in driving and cognitive abilities both (i) ‘behaviourally’, for example, by not driving on busy roads or in rush hour trafﬁc (McKnight and McKnight, 1999) and (ii) ‘psychologically’, for example, increasing levels of task-focused coping. Previous research efforts point to robust relationships between hazard monitoring (DSI) and task-focused coping (Matthews et al., 1996). These ﬁndings were supported in the current study. However, some associations between driving behaviour/performance and coping uncovered were not anticipated (see Figure 11.1 above). For example, task-focused coping was found to be associated with speeding convictions. This ﬁnding is counterintuitive as task-focused coping may be considered a coping strategy used to reduce driver error through increased alertness and concentration on the driving task, and its association with hazard monitoring (DSI). There were two signiﬁcant interactions between coping style and age group with respect to the prediction of thrill seeking. These involved the ‘avoidance’ and ‘emotion focused’ coping styles. The nature of the interactions was very similar in both cases (see Figure 11.2 above). High scores on emotion-focused and avoidance coping scales tended to be associated with greater thrill seeking in the older group, but somewhat reduced thrill seeking in the younger group. Inappropriate or ineffective coping strategies could lead to both the experience and maintenance of higher levels of anxiety and stress while driving (Gulian et al., 1989b; McCrae and Costa, 1986). Given that older drivers report less thrill seeking the interactions (see Figure 11.2 above) are unusual. One possible explanation is that older experienced drivers do not perceive thrill seeking as risky driving behaviours (for example, Parker et al.,

The Inﬂuence of Age Differences

137

2001), thereby suffering from the generally accepted tendency of drivers to rate their quality of driving as being very good (for example, Groeger and Grande, 1996). This may have serious implications for such a safety-critical task as driving. Much of the work on driver stress and coping strategies to date has used samples of students and working adults (over 50). Indeed in Gulian et al. (1989a), the younger group were younger than 35 and the older group older than 46 years of age. The older age range in the current sample was between 60 and 80 years of age. It is possible that coping styles may interact with another variable that is related to age, for example, cognitive ability, although not necessarily the same variable for each coping style. A further possibility is that a coping style that is adaptive at one age might not be adaptive at another. Older drivers may maintain coping strategies that have become less effective or even adopt different less effective coping strategies with increasing age, and this may partly explain older driver error. The transactional model of coping developed by Gulian et al. (1989a; 1989b) and by Matthews et al. (1997) has proved to be a useful one for understanding driver stress reactions but may need to be extended to account for differences in driver stress with increasing age. Nonetheless, patterns of stress reactions broadly consistent with previous research were observed, helping to further clarify age differences in driver stress. This research has uncovered different patterns and interactions among coping styles used to manage driver stress, which may help us to explain age differences in driver behaviour and error, and therefore have implications for trafﬁc safety. Older drivers’ stress responses and associated styles of coping are a worthwhile avenue for future research. Acknowledgements The ﬁrst author has an E.S.R.C. studentship and also acknowledges ﬁnancial support for volunteer recruitment from the Institute of Psychological Sciences, University of Leeds. Thanks also to Professor Gerry Matthews for copies of DSI and DCQ scales. References Aberg, L. and Rimmo, P.-A. (1998). ‘Dimensions of aberrant driving behaviour.’ Ergonomics, 41, 39–56. Anstey, K., Wood, J., Lord, S. and Walker, J. (2005). ‘Cognitive, sensory and physical factors enabling driving safety in older adults.’ Clinical Psychology Review, 25, 45–65. Blockley, P. and Hartley, L. (1995). ‘Aberrant driving behaviours: errors and violations.’ Ergonomics, 38, 1759–71. Chapman, P. and Underwood, G. (2000). ‘Forgetting near-accidents: the roles of severity, culpability and experience in the poor recall of dangerous driving situations.’ Applied Cognitive Psychology, 14, 31–44. Craik, F. and Jennings, J. (1992). ‘Human memory.’ In F. Craik and T. Salthouse (eds). The Handbook of Aging and Cognition. Hillsdale, NJ: Lawrence Erlbaum

138

Driver Behaviour and Training – Volume III

Associates Inc. Groeger, J. and Grande, G. (1996). ‘Self-preserving assessments of skill?’ British Journal of Psychology, 97, 61–79. Gulian, E., Matthews, G., Glendon, A., Davies, D.R. and Debney, L. (1989a). ‘Dimensions of driver stress.’ Ergonomics, 32, 585–602. Gulian, E., Debney, L., Glendon, A., Davies, D.R. and Matthews, G. (1989b). ‘Coping with driver stress.’ In F. McGuigan, W. Sime and J.M. Wallace, Stress and Tension Control, Plenum Publishing, 173–86. Hakamies-Blomqvist, L. (1994). ‘Compensation in older drivers as reﬂected in their fatal accidents.’ Accident Analysis and Prevention, 26, 107–12. Hasher, L. and Zacks, R. (1979). ‘Automatic and effortful processes in memory.’ Journal of Experimental Psychology: General, 108, 356–88. Kline, D. and Scialfa, C. (1996). ‘Visual and auditory aging.’ In J. Birren and K. Schaie (eds). Handbook of the Psychology of Aging (4th ed.). San Diego, CA: Academic Press. Lajunen, T. and Summala, H. (2003). ‘Can we trust self-reports of driving? Effects of impression management on driver behaviour questionnaire responses.’ Transportation Research Part F, 6, 97–107. Lazarus, R. and Folkman, S. (1984). Stress, Appraisal and Coping. New York: Springer. Loftus, E. (1993). ‘The reality of repressed memories.’ American Psychologist, 48, 518–37. Matthews, G., Desmond, P., Joyner, L., Carcary, B. and Gilliland, K. (1996). ‘Validation of the Driver Stress inventory and Driver Coping questionnaire.’ (Personal communication with ﬁrst author, 2005.) Matthews, G., Desmond, P., Joyner, L., Carcary, B. and Gilliland, K. (1997). ‘A comprehensive measure of driver stress and affect.’ In T. Rothengatter and E. Vaya, Trafﬁc and Transport Psychology: Theory and Application. Amsterdam: Pergamon. Matthews, G., Dorn, L. and Glendon, I. (1991). ‘Personality correlates of driver stress.’ Personality and Individual Differences, 12, 535–49. Matthews, G., Dorn, L. and Hoyes, T. (1992). ‘Individual differences in driver stress and performance.’ In T. Lovesey (ed.). Contemporary Ergonomics, London: Taylor and Francis. McCrae, R. and Costa, P. (1986). ‘Personality, coping and coping effectiveness in an adult sample.’ Journal of Personality, 54, 385–405. McKnight, A.J. and McKnight, A.S. (1999). ‘Multivariate analysis of age-related driver ability and performance deﬁcits.’ Accident Analysis and Prevention, 31, 445–54. Parker, D., Lajunen, T. and Stradling, S. (1998). ‘Attitudinal predictors of aggressive driving violations.’ Transportation Research Part F, 1, 11–24. Parker, D., Macdonald, L., Rabbitt, P. and Sutcliffe, P., (2000). ‘Elderly drivers and their accidents: the aging driver questionnaire.’ Accident Analysis and Prevention, 32, 751–9. Parker, D., Macdonald, L., Sutcliffe, P. and Rabbitt, P. (2001). ‘Conﬁdence and the

The Inﬂuence of Age Differences

139

older driver.’ Ageing and Society, 21, 169–82. Parker, D., Reason, J., Manstead, A. and Stradling, S. (1995a). ‘Driving errors, driving violations and accident involvement.’ Ergonomics, 38, 1036–48. Parker, D., West, R., Stradling, S. and Manstead, A. (1995b). ‘Behavioral characteristics and involvement in different types of trafﬁc accident.’ Accident Analysis and Prevention, 27, 571–81. Patten, C., Kircher, A., Ostlund, J., Nilsson, L. and Svenson, O. (2006). ‘Driver experience and cognitive workload in different trafﬁc environments.’ Accident Analysis and Prevention, 38, 887–94. Rabbitt, P. (1993). ‘Does it all go together when it goes? The 19th Bartlett Memorial Lecture.’ Quarterly Journal of Experimental Psychologist, 46A, 385–434. Reason, J., Manstead, A., Stradling, S., Baxter, J. and Campbell, K. (1990). ‘Errors and violations on the roads: a real distinction?’ Ergonomics, 33, 1315–32. Salthouse, T. (1992). ‘Inﬂuence of processing speed on adult age differences in working memory.’ Acta Psychologica, 79, 155–70. Stutts, J. (1998). ‘Do older drivers with visual and cognitive impairments drive less?’ Journal of the American Geriatrics Society, 46, 854–61. Westerman, S.J. and Haigney, D. (2000). ‘Individual differences in driver stress, error and violation.’ Personality and Individual Differences, 29, 981–98.

This page intentionally left blank

PART 3 At Work Road Safety

This page intentionally left blank

Chapter 12

Effects of Organisational Safety Culture on Driver Behaviours and Accident Involvement Amongst Professional Drivers Bahar Öz1 and Timo Lajunen2 University of Helsinki, Finland 2 Middle East Technical University, Turkey 1

Introduction International studies have shown that professional drivers are at a high risk of trafﬁc accidents. For example, Salminen and Lähdeniemi (2002) indicated trafﬁc as the most important cause of accidental deaths at work in Finland. Similarly, Bylund, Björnstig and Larson (1997) reported the professional drivers having the highest injury rate of employed people in Sweden. Also, Caird and Kline (2004) pointed out that about four workers per day are killed in on-the-job motor vehicle crashes in the United States. Although causes of trafﬁc accidents have been studied extensively, there are only a few studies about trafﬁc accidents during working hours (Salminen and Lähdeniemi, 2002; Caird and Kline, 2004). In addition to the general factors, there are particular work-related factors inﬂuencing professional drivers’ involvement in trafﬁc accidents. Caird and Kline (2004) proposed that driver characteristics, environment, workplace, vehicle and organisation related factors inﬂuence professional drivers’ accident involvement. Different from non-professional drivers, organisational factors form an additional situational factor category for the professional drivers. The effects of organisation related factors (see Ostrom, Wilhelsen and Kaplan, 1993; Vredenburgh, 2002) and individual differences (Lajunen, 2001) on accident involvement have been demonstrated in literature. In their study about the organisational factors and on-the-job accidents of professional drivers, Caird and Kline (2004) investigated if work-related driving differed from free-time driving. One of the main differences between the nonprofessional and professional drivers is that driving is a less self-paced task for professional drivers whereas non-professional drivers can largely choose, for example, the mode of transportation, time of travel, route and target speed while driving. Hence, non-professional drivers can adjust the difﬁculty and risk level of the task whereas professional drivers’ work conditions are mostly predetermined. Among professional drivers, however, many different factors, like time schedule and long working hours, affect task demands. In addition, professional drivers have only limited inﬂuence on organisational demands and support.

144

Driver Behaviour and Training – Volume III

Organisational safety culture Organisations are thought to have different cultures (Schein, 1991). Organisational culture literature suggests that any organisation has organisational subcultures in addition to a general culture. In other words, organisational culture can be divided into components (Schein, 1991). According to Clarke (1999), talking about these small components is preferable to an investigation of the general culture since they can be measured and manipulated more easily than a general organisational culture. So far researchers have investigated organisational culture by dividing it into different components like creativity culture, motivation culture and safety culture. The key predictor of safety performance is believed to be the safety culture (Ostrom et al., 1993). Although safety culture concept is very critical for the organisations, there is no agreement about its deﬁnition or measurement among researchers (Cox and Flin, 1998). According to Lee (1993), safety culture includes shared commitment to think safely, behave safely, and trust in the safety measures put in place by the organisation. The literature has provided some concrete indicators of the existence of safety culture in a company. Wiegmann, Zhan and Von Thaden (2004) suggested that organisational commitment to safety indicates the level of safety culture because high commitment to safety by upper management provides resources for development and implementation of safety measures. If the safety culture of an organisation is well developed, the beliefs, attitudes and practices should emphasise minimising the exposure of employees to hazards. In other words, any type of application including training, selection, work schedules and use of equipment should be organised by taking employees’ safety into account. Varonen and Mattila (2000) found that a company’s attitudes to safety and its safety precautions are negatively correlated with accident rates. Similarly, Zohar (1980) suggested that employees’ perception about the company’s commitment to safety is the major indicator of safety culture. Driver behaviours In their classic study, Reason, Manstead, Stradling, Baxter and Campbell (1990) made a distinction between driver errors and violations. This differentiation provided a base for the development of the Manchester Driver Behaviour Questionnaire (DBQ; see Reason et al., 1990), which was developed for measuring aberrant driver behaviour by using self-reports. Reason and his colleagues (1990) showed that driver errors and violations are two empirically distinct types of behaviour. They deﬁned errors as ‘the failure of planned actions to achieve their intended consequences’ and violations as ‘deliberate deviations from those practices believed necessary to maintain the safe operation of a potentially hazardous system’. Unlike errors, violations were seen as deliberate behaviours, although both errors and violations are potentially dangerous and could lead to a crash. Reason et al. (1990) also found a third DBQ factor, which they named ‘slips and lapses’. This factor included attention and memory failures, which can cause embarrassment but are unlikely to have an impact on driving safety (Parker, Reason and Stradling, 1995). Since errors and violations result from different psychological processes, they should be treated differently (Reason et al., 1990).

Effects of Organisational Safety Culture on Driver Behaviours

145

Both errors and violations are negative behaviours in terms of intention or outcome. However, everyday trafﬁc also involves other behaviours in which a driver’s intention is positive. These behaviours do not depend on regulations or rules. Taking care of the trafﬁc environment or other road users, or helping them, is the main intention for these behaviours (Özkan and Lajunen, 2005). Özkan and Lajunen (2005) made an addition to the error-violation differentiation of the DBQ by introducing the concept of ‘Positive Driver Behaviours’ and developing a selfreport scale for measuring these behaviours. The main idea behind ‘Positive Driver Behaviours’ scale is that in addition to committing errors and violations, drivers also intend to help other road users and behave accordingly while driving. Driving includes both negative and positive driver behaviours at the same time. Aim of the study The aim of the present study was to investigate the effects of organisational safety culture components on driver errors, violations, positive driver behaviours, and accident involvement among professional drivers. Method Participants A total of 73 professional male drivers (38 taxi drivers and 35 cargo company drivers) volunteered to participate in the study. The participants were contacted through visits to the cargo companies and taxi stops, and they were assured of anonymity and conﬁdentiality. The mean age of the drivers was 35.18 years (SD = 7.12) and the average annual mileage was 74.08 km (SD = 73.072). The participants had an average of 13.6 years’ driving experience, and the mean number of accidents they were involved in was 1.33. Questionnaires Organisational safety culture scale An organisational safety culture scale developed for the present study was used to collect information about the drivers’ perceptions of the safety culture of the company in which they were working. The scale consisted of 15 items measuring three safety culture dimensions including ‘trafﬁc safety’, ‘general safety’ and ‘work safety’. The participants were asked to evaluate each item on a ﬁve point Likert type scale (1 = ‘strongly disagree’, 5 = ‘strongly agree’). Driver Behaviour Questionnaire (DBQ) To measure violations and errors, the DBQ with 28 items was used. Positive Driver Behaviours scale (Özkan and Lajunen, 2005) was used together with the DBQ. The Turkish translation and the factor structure of the DBQ have been validated in previous studies conducted among both professional (Sümer and Özkan, 2002) and non-professional drivers (Sümer,

146

Driver Behaviour and Training – Volume III

Lajunen and Özkan, 2002). The participants were asked to evaluate each item on a six point Likert type scale (1 = ‘never’, 6 = ‘always’). Alpha reliabilities of the scales were 0.85 for violations (12 items), 0.79 for errors (16 items) and 0.89 for positive behaviours (9 items). Background information Information related to age, sex, level of education, years of holding a full driving licence, experience, number of accidents, type of accidents and annual mileage were recorded. Results Factor structure of the organisational safety culture scale In order to determine the factor structure of the scale, factor analyses with principal axis factoring method were conducted for the 23 item scores of the organisational safety culture scale. Number of factors was determined by ‘mineigen > 1’ criterion, a screen plot and the conceptual relevance of the items to the dimension to which they should belong. As a result, three-factor solution was found to be the most interpretable one. Some items were left out from the ﬁnal solution because of low factor loadings (< 0.30), cross loadings or being conceptually irrelevant to the dimension. Remaining items were re-evaluated and the most representative ones of each dimension in terms of the content were included into the last version of the scale (see Table 12.1). The ﬁrst factor was named as ‘trafﬁc safety’. This factor included seven items, and accounted for 39.4 per cent of the variance. The second factor was named as ‘general safety’ and included three items accounting for 9.1 per cent of the variance. The third factor was named as ‘work safety’ and it included ﬁve items, which accounted for 5.4 per cent of the variance. Internal consistency reliabilities for these three factors were 0.85, 0.74 and 0.92, respectively. Correlation analyses The DBQ scales (violations, errors and positive driver behaviours) and the organisational safety culture scales (trafﬁc safety, general safety and work safety) based on factor analysis were calculated by averaging the item scores. Correlation coefﬁcients between the sum variables and descriptive statistics can be found in Table 12.2. As Table 12.2 shows, the participants perceived the organisation that they were working for as being high in trafﬁc safety, general safety and work safety. However, they tended to perceive the organisation’s general safety applications and attitude higher than that of trafﬁc safety and work safety. Drivers scored relatively low on violations and errors whereas their positive driver behaviour scores were rather high. Correlation coefﬁcients indicated that there was a positive relationship between trafﬁc safety and work safety dimensions of the organisational safety culture. The error dimension of the DBQ was negatively related to general safety dimension

Effects of Organisational Safety Culture on Driver Behaviours

147

Table 12.1 Factor structure of the organisational safety culture scale Items

Item-total correlation

Factors Factor 1 (Trafﬁc safety)

In case of a danger while working, we can get immediate help from our company. The company which I am working for follows the trafﬁc code and regulations closely and informs us about them. When hiring new employees the company which I am working for considers the safety principles. The company that I am working for takes other road users’ or customers’ complaints about the drivers into consideration. The company which I am working for emphasises safe driving more than being on time. The company which I am working for gives us training on topics like safe driving and ﬁrst aid to improve our safety as drivers. The company which I am working for makes a special effort to do the technical check of the vehicles in time. Workload in the company is organized by not taking the safety of the drivers, customers and other road users into account. When the company’s requests lead to time pressure, there is nothing wrong with not obeying the safety regulations and rules. Safety is a secondary concern for the company in urgency. We can trust that we are safe while working because of the company’s work safety related policies. I ﬁnd the work safety applications of the company satisfactory. The company which I am working for uses all the possible resources to make work safety guaranteed. In the company which I am working for there is tight control on obeying the safety related rules. In the company which I am working for, safety has the primary importance.

0.73

0.74

0.62

0.55

0.66

0.61

0.59

0.75

0.38

0.42

0.74

0.76

0.66

0.71

Factor 2 (General safety)

0.60

0.63

0.69

0.60

0.43

0.52

Factor 3 (Work safety)

0.74

0.69

0.89

0.84

0.82

0.82

0.79

0.69

0.78

0.72

Note: Factor loadings < 0.30 deleted for the sake of clarity.

148

Table 12.2 Correlations between the study variables

1. Age

8. Violations

SD

35.18

7.12

74 080

73 072

-0.13

1.33

1.56

–0.04

–0.08

3.34

1.03

–0.19

0.06

–0.18

4.31

1.24

0.02

0.20

–0.26*

–0.16

3.05

1.14

–0.18

0.04

–0.06

0.67**

–0.06

1.52

0.45

0.00

–0.12

0.07

0.11

–0.34**

1.81

0.68

–0.04

0.23

0.14

–0.22

–0.02

0.04

0.48**

1.39

0.08

0.09

0.02

–0.01

0.13

–0.10

–0.20

9. Positive driver 4.30 behaviours * p < 0.05, ** p < 0.01

1

2

3

4

5

6

7

8

0.27*

–0.17

Driver Behaviour and Training – Volume III

2. Annual km driven 3. Number of accidents 4. Trafﬁc safety 5. General safety 6. Work safety 7. Errors

Mean

Effects of Organisational Safety Culture on Driver Behaviours

149

and positively related to the work safety dimension of organisational safety culture. There was a positive relationship between errors and violations. The number of accidents drivers had been involved in had a negative relationship with the general safety dimension. As Table 12.2 indicates, positive driver behaviour dimension was not related to any of the variables. Regression analyses In order to investigate the effect of organisational safety culture on driver behaviour (errors, violations and positive driver behaviours) and accidents, four multiple regression analyses (errors, violations, positive driver behaviours and accidents as dependent variables) were conducted. In each analysis, the effects of age and mileage were controlled by forcing them into the model at the ﬁrst step. In the second step, organisational safety culture dimensions were entered into the model. Table 12.3 shows the results of regression analyses. Trafﬁc safety dimension of organisational culture was negatively related to violations and accidents. Strangely, both violations and errors were more frequent in the organisations in which workrelated safety was assessed to be high. General safety dimension was negatively related to both errors and accidents. However, none of the organisational safety culture dimensions predicted positive driver behaviours. Similarly, none of the DBQ dimensions predicted the number of accidents among professional drivers. Discussion The present study investigated the effects of organisational safety culture on driver behaviours and accident involvement among professional drivers. Results showed that safety culture predicted the frequency of errors, violations and accidents. Participants indicated that the organisation that they were working for arranged the workload by taking them, other drivers, customers, and the time pressure into account. The organisation gave priority to safety in every steps of work and used all the resources to guarantee safety in and outside the workplace. Reason (1997) suggested that two different processes are usually used to regulate factors causing accidents: production (for example, transportation of people or goods) and protection (for example, avoiding the occurrence and recurrence of accidents). Depending on the circumstances, one of these two factors is usually dominant, although the production approach is usually the dominant one because of several reasons. Firstly, production creates the resources that make protection possible. Secondly, managers are usually trained for production and are, therefore, more skilled in production than in protection. Thirdly, the quality of production can be directly evaluated whereas the quality of protection is usually reﬂected indirectly as lack of negative outcomes (such as an accident). Drivers in the present study indicated their workplace as safety oriented by emphasising that protection related applications and approach were dominant in the organisation. These applications included, for example, safety training and providing information about the trafﬁc related regulations and laws. As the results showed, protection focused applications

150

Table 12.3

Driver Behaviour and Training – Volume III

Regression of organisational safety culture on DBQ scales and accident involvement

Independent R2 Adj R2 R2 Change Variables Violations as a dependent variable 1. Age – Annual km 0.05 0.02 0.05 driven 2. Trafﬁc safety General 0.19 0.12 0.14 safety Work safety Errors as a dependent variable 1. Age Annual km 0.01 –0.02 0.01 driven 2. Trafﬁc safety General 0.22 0.16 0.21 safety Work safety Positive Driver Behaviours as a dependent variable 1. Age Annual km 0.03 –0.00 0.03 driven 2. Trafﬁc safety General 0.06 –0.02 0.03 safety Work safety Number of accidents as a dependent variable 1. Age Annual km 0.03 –0.01 0.03 driven 2. Trafﬁc safety General 0.13 0.06 0.11 safety Work safety * p < 0.05, **p < 0.01 Step

F

df

1.60

2

2.87*

5

0.46

2

3.47**

5

0.97

2

0.71

5

0.82

2

1.89

5

β

–0.03 0.21 –0.50** –0.14 0.37*

0.01 –0.12 –0.20 –0.34** 0.40**

0.15 0.10 0.16 0.08 –0.18

–0.13 –0.11 –0.30* –0.29* 0.14

Effects of Organisational Safety Culture on Driver Behaviours

151

resulted in fewer violations. If organisation’s ‘cultural message’ is that production – not people or safety – is the priority, employees might perceive organisation’s commitment to safety loose, which, in turn, may affect their safety performance (Vredenburgh, 2002). In the present study, general safety had a negative relationship to self-reported accidents and errors. This ﬁnding is in line with those of Vuuren (2000), who showed that improvement in company’s safety climate resulted in a signiﬁcant decreases in accident rate. Similarly, Kirschenbaum, Oigenblick and Goldberg (2000) reported that hazardous working conditions increased chances of being involved in an accident. Moreover, Hayes, Perander, Smecko and Trask (1998) stated that positive employee perceptions of safety are related to low accident rates. Results of the study indicated a positive relationship between high work safety (evaluated by drivers) and self-reported number of errors and violations. In other words, if the drivers found the company’s safety regulations satisfactory, company’s investment in safety sufﬁcient and safety as a priority, they reported more errors and violations. One explanation for this slightly surprising ﬁnding is that a work safety oriented company usually encourages its employees to report incidents and hazardous situations. Hence, a company’s open and blame-free atmosphere might actually make it easier for drivers to report their errors and violations. Earlier studies have shown that communication between the company and employees increases safe behaviours. Hofmann and Stetzer (1996) showed that employees working with a supervisor not mentioning safety might perceive safety as unimportant. If the involvement of management is high, that is, if the management gets involved in critical safety activities including seminars and training days, this provides good communication of safety related topics in the organisation (Wiegmann et al., 2004). Based on these studies, it can be claimed that good communication between management and the drivers increases the likelihood of reporting unwanted behaviours (errors and violations) and risky situations because drivers trust that the company is going to adopt the precautions it needs. The present study showed that the company’s safety culture is an important factor in terms of professional drivers’ trafﬁc safety. This study focused mainly on three aspects of organisational safety culture. In future studies, the effects of communication, training and reward systems on professional drivers’ safety behaviour and outcomes should be studied. Acknowledgements This study was supported by The Scientiﬁc and Technological Research Council of Turkey (TÜBİTAK Project No: 103K017) and Marie Curie Transfer of Knowledge programme (‘SAFEAST’ Project No: MTKD-CT-2004-509813).

152

Driver Behaviour and Training – Volume III

References Bylund, P.O., Björnstig, U. and Larsson, T.J. (1997). ‘Occupational road trauma and permanent medical impairment.’ Safety Science, 36(3), 187–200. Caird, K.J. and Kline, T.J. (2004). ‘The relationships between organisational and individual variables to on-the-job driver accidents and accident-free kilometers.’ Ergonomics, 47(15), 1598–1613. Clarke, S. (1999). ‘Perceptions of organisational safety: implications for the development of safety culture.’ Journal of Organisational Behaviour, 20(2), 185–98. Cox, S. and Flin, R. (1998). ‘Safety culture: philosopher’s stone or man of straw?’ Work and Stress, 12, 189–201 Lee, T.R. (1993). ‘Seeking a safety culture.’ Atom, 429, 20–3. Ostrom, L., Wilhelmsen, C. and Kaplan, B. (1993). ‘Assessing safety culture.’ Nuclear Safety, 34, 163–72. Özkan, T. and Lajunen, T. (2005). ‘A new addition to DBQ: positive driver behaviours scale.’ Transportation Research Part F, 8, 355–68. Hayes, B.E., Perander, J., Smecko, T. and Trask, J. (1998). ‘Measuring perceptions of workplace safety: development and validation of the work safety scale.’ Journal of Safety Research, 29(3), 145–61. Hofmann, D.A. and Stetzer, A. (1996). ‘A cross-level investigation of factors inﬂuencing unsafe behaviours and accidents.’ Personnel Psychology, 49, 307– 39. Kirschenbaum, A., Oigenblick, L. and Goldberg, A.I. (2000). ‘Well being, work environment and work accidents.’ Social Science and Medicine, 50, 631–9. Parker, D., Reason, J., Manstead, A. and Stradling, S. (1995). ‘Driving errors, driving violations and accident involvement.’ Ergonomics, 38, 1036–48. Reason, J., Manstead, A., Stradling, S., Baxter, J. and Campbell, K. (1990). Ergonomics, 33(10/11), 1315–32. Reason, J. (1997). Organisational Accidents. Manchester: Manchester University Press. Salminen, S. and Lähdeniemi, E. (2002). ‘Risk factors in work-related trafﬁc.’ Transportation Research Part F, 5, 77–86. Schein, E.H. (1991). Organisational Culture and Leadership (2nd ed.). San Francisco: Jossey-Bass Publishers. Sümer, N., Lajunen, T. and Özkan, T. (2002). ‘Sürücü davranışlarının kaza riskindeki rolleri: İhlaller ve hatalar.’ Trafﬁc and Road Safety International Congress, 8–12 May, Gazi University, Ankara, Turkey. Sümer, N. and Özkan, T. (2002). ‘Sürücü Davranışları, Becerileri, Bazı Kişilik Özellikleri ve Psikolojik Belirtilerin Traﬁk Kazalarındaki Rolleri.’ TürkPsikoloji Dergisi, 17(50), 1–22. Varonen, U. and Mattila, M. (2000). ‘The safety climate and its relationship to safety practices, safety of the work environment and occupational accidents in eight wood-processing companies.’ Accident Analysis and Prevention, 32, 761–9. Vredenburgh, A.G. (2002). ‘Organisational safety: which management practices are most effective in reducing employee injury rates?’ Journal of Safety Research,

Effects of Organisational Safety Culture on Driver Behaviours

153

33, 259–76. Vuuren, W. (2000). ‘Cultural inﬂuences on risks and risk management: six case studies.’ Safety Science, 34, 31–45. Wiegmann, D.A., Zhang, H., Von Thaden, T.L., Sharma, G. and Gibbons, A.M. (2004). ‘Safety culture: an integrative review.’ The International Journal of Aviation Psychology, 14(2), 117–34. Zohar, D. (1980). ‘Safety climate in industrial organisations: theoretical and applied implications.’ Journal of Applied Psychology, 65, 96–102.

This page intentionally left blank

Chapter 13

Stages of Change in the Australian Workplace and its Application to Driver Education Tamara Banks, Jeremy Davey and H. Biggs Centre for Accident Research and Road Safety, Queensland, Australia Introduction High social and ﬁnancial costs are currently being incurred by both industry and society as a result of work-related road incidents. It is estimated that worldwide 50 million people are injured and an additional 1.2 million people are killed annually in road crashes (WHO, 2004). More speciﬁcally it has been estimated that in Australia, the total cost of work-related road incidents may be in the range of $1 billion to $1.5 billion per annum (Wheatley, 1997) and that the average total insurance cost of a ﬂeet incident is approximately $28 000 with costs incurred to both the company and society (Davey and Banks, 2005). Additionally, motor vehicle incidents are the most common mechanism for Australian compensated fatalities, representing 35 per cent of all work compensated deaths (ASCC, 2006). In an attempt to reduce the frequency and severity of work-related road incidents, some organisations implement driver education. Driver education aims to improve driver knowledge, attitudes and behaviour. The term is used broadly to cover a range of instruction and learning procedures that emphasise the cognitive processes and underlying values relating to safe driving behaviour. Although many practitioners in the occupational health and safety ﬁeld assume that driver safety can be enhanced through the provision of driver safety education, research investigating this relationship has found mixed results. For example, driver education has been found to be associated with greater caution when approaching and overtaking a hazard, greater adherence to trafﬁc signals and signs, greater visual monitoring of the driving environment, fewer trafﬁc violations/ infringements, lower risk acceptance and lower accident risk (Chapman, Underwood and Roberts, 2002; Dorn and Barker, 2005; Llaneras, Swezey, Brock, Rogers and Van Cott, 1998; Lund and Williams, 1985). Comparatively, other studies have failed to ﬁnd signiﬁcant risk reductions in relation to the provision of driver education (Christensen and Glad, 1996; Katila, Keskinen, Hatakka and Laapotti, 2004; Lynam and Twisk, 1995). It is suggested that the observed contradictory ﬁndings from past studies may be partially explained by variations between participants’ readiness for change. The stages of change model, also known as the transtheoretical model of change (Prochaska, DiClemente and Norcross, 1992), is a behaviour change model that

156

Driver Behaviour and Training – Volume III

offers a framework for understanding variations in readiness for change. The model suggests that individuals require different types of support based on their position within the change process as they pass through the cyclical phases of precontemplation (not thinking of changing one’s behaviour), contemplation (seriously considering changing one’s behaviour in the near future), preparation (making plans and intending to change one’s behaviour in the very near future, may have started to make minor changes), action (engaged in changing one’s behaviour) and maintenance (working to consolidate gains from one’s changed behaviour and prevent relapse). Originally developed in the ﬁeld of psychotherapy for studying individual behaviour change, the stages of change model has strong empirical support in the area of individual health promotion. For example, research across a range of behaviours including dietary behaviour (Campbell, et al., 1994) and mammography screening (Rakowski, et al., 1998) has found that stage-matched interventions were more effective in changing behaviour than ‘one size ﬁts all’ interventions. This makes intuitive sense considering that research comparing stage distributions across a range of health-related behaviours has found that in pre-action individuals approximately 40 per cent of the population are in the precontemplation phase, 40 per cent are in the contemplation phase and only 20 per cent are in the preparation phase (Velicer et al., 1985; Laforge, Velicer, Richmond and Owen, 1999). This distribution pattern suggests that a combination of initiatives targeting awareness raising, discussing how changes could be made and making plans to change may be an effective approach to improving work-related road safety. It also may account for previous research ﬁndings that driver training is not effective. If driver training is focused as a single action-orientated initiative and has not prepared individuals by creating the conditions necessary for change, it may only be catering to the needs of approximately 20 per cent of pre-action employees. The stages of change model could be used in organisations to design driver training programs that are appropriate to the readiness for change of managers and employees. In recent years the model has started to be applied internationally to organisational change in the areas of ergonomics and health promotion (Haslam, 2002; Prochaska, Prochaska and Levesque, 2001). A case study of health and safety appraisal within an English manufacturing company identiﬁed that the Stages of Change model provided a useful framework for assessing attitudes and beliefs and assisting in recognising individual and organisational readiness to change (Barrett, Haslam, Lee and Ellis, 2005). Additionally, the model provided a framework for explaining the observed effectiveness of a cattle handling injury prevention program in New Zealand. Interviews with approximately 1500 farming personnel revealed that awareness raising methods including leaﬂets and videos were most effective in transitioning farmers from contemplation to action. Alternatively, ﬁeld days that provided farmers with an opportunity for tailored advice were more effective in transitioning farmers from action to maintenance (Slappendel, 2001 as cited in Haslam, 2002). This chapter expands upon recent organisational behaviour change research by exploring the utility of the stages of change model as a framework for understanding employee road safety behaviour change. Firstly, it is hypothesised that the stages of change model will provide a framework for identifying employee readiness to engage in work-related road safety behaviour change. Secondly, it is hypothesised

Stages of Change in the Australian Workplace

157

that the model will provide a framework for explaining employees’ perceived effectiveness of work-related road safety initiatives. Finally, this chapter suggests how practitioners can apply the stages of change framework to tailor driver training programmes to most effectively meet client needs. Method Participants To allow research ﬁndings to be generalised to a wide range of organisations, two contrasting organisations were recruited to participate in this research. Organisation A was a not-for-proﬁt state-based provider of residential and community services. It employed approximately 2000 staff and was supported by a network of approximately 500 volunteers. Organisation A operated a ﬂeet of approximately 200 vehicles. In comparison, Organisation B was a for-proﬁt national utilities provider with a workforce of approximately 35 000 employees. The organisation was jointly owned by the Australian government and private shareholders. Organisation B operated a ﬂeet of approximately 15 000 vehicles. Both organisations serviced customers in urban, rural and remote areas of Australia and therefore required their employees to operate vehicles in a range of environments. Interviews were conducted with ﬁve employees from each of the organisations. The selection of participants was a convenience sample and was ultimately determined by the employer. Efforts however were made to obtain a random and representative sample within this real-world context. Participating organisations provided access to both male and female employees from a range of roles and levels of seniority within their organisation. Participants’ roles included ﬂeet manager, occupational health and safety manager, department manager and operational employees who were required to drive as part of their work. Interviews A semi-structured interview schedule was developed to explore the utility of the stages of change model as a framework for understanding employee road safety behaviour change. Previous research suggests that it is possible to assess stage of change via individuals’ responses to a small number of questions (Haslam, 2002; Haslam and Draper, 2000). Based on adaptations from previous research (Barrett, Haslam, Lee and Ellis, 2005; Whysall, Haslam and Haslam, 2006), combinations of open and closed questions were developed. In some cases several questions were asked at each stage to elicit sufﬁcient information to identify employees’ stage of change. Interview participants were also asked to comment on any work-related road safety initiatives they were aware of that their organisation was intending to or was already engaging in. The formality and depth of interview questions was varied to suit the employees’ level of seniority and involvement in driver safety initiatives. The following core questions were asked in this order until a negative response was

158

Driver Behaviour and Training – Volume III

obtained. The last positive response indicated the employee’s stage of readiness for change. 1. Are you aware of any work-related road safety risk? (Yes = continue, No = precontemplation) 2. Are you planning to take any action to reduce work-related road safety risk in the next six months? (Yes = continue, No = precontemplation) 3. Do you have any deﬁnite plans to reduce work-related road safety risk in the next month? (Yes = continue, No = contemplation) 4. Have you taken any deliberate action to reduce work-related road safety risk? (Yes = continue, No = preparation) 5. Are you currently doing anything to maintain work-related road safety within your organisation? (Yes = maintenance, No = action). To explore whether employees’ perceived effectiveness of safety initiatives was related to their stage of change, interview participants were asked to comment on any work-related road safety initiatives they had experienced in their organisation. Several steps were taken to maximise the integrity of the interview data collected. Firstly, the interview schedule was piloted and reﬁned based on feedback from employees not participating in the main study. Secondly, interviews were conducted face-to-face in a private ofﬁce on the premises of each organisation to minimise distractions and misinterpretations of information. Thirdly, employees were interviewed individually to minimise any contamination of data arising from potential group bias. Fourthly, it was stressed that participation was voluntary and conﬁdential to encourage participants to openly report their beliefs and behaviours. Finally, consent was sought from participants for the interview to be recorded and notes to be taken during the session. All recorded data were transcribed verbatim to ensure accuracy. Analysis A three-phase approach as described by Miles and Huberman (1994) was adopted to analyse the transcribed data. Firstly, data was organised via cutting and pasting material into meaningful collections that corresponded with the interview questions. Secondly, emerging themes were identiﬁed and patterns within and between themes were explored. This phase involved summarising the data under each theme and selecting verbatim quotes to illustrate the themes. Thirdly, conclusions were drawn after interpretations of the data were veriﬁed against the interview transcripts and existing literature. Results An analysis of the interview transcripts suggests that the ﬁve core questions provided a useful starting point for classifying employees’ stage of change. However, in this research the ﬁve questions were not sufﬁcient to distinguish between adjoining stages

Stages of Change in the Australian Workplace

159

for some participants. With the inclusion of additional probing questions it was identiﬁed that the stages of change model could provide a framework for classifying employee readiness to engage in work-related road safety behaviour change. Organisation A Within Organisation A, it was identiﬁed that the observed stage of readiness varied between individuals at the management level and between the levels of managers and employees. Interviews indicated that managers (n = 3) varied between the preparation and action stages of readiness. All managers indicated an awareness of their organisation’s exposure to work-related road safety risks. Reported risks included both general road safety risks and risks that could be considered speciﬁc to their organisation. Generic risks reported included wildlife, visibility, road trains, fatigue and high exposure with reports of some employees driving up to 70 000km per annum. Organisation speciﬁc risks pertained more to the transporting of clients. For example one manager reported that some clients ‘…might just grab the handbrake or grab the wheel. We’ve had a couple of grab the wheel situations. We had a towel over the head the other day while the driver was driving. So we have a few young people that pose a risk with our cars.’ Although all managers believed that that their views of risk were shared by other managers, some felt that employees working at a lower level of the organisation may not be aware of the risks. For example when talking about road safety initiatives one manager commented, ‘I think management will embrace it, but the next couple of levels will struggle because the people I know are still saying we’re about doing this and our cars getting dinted. You know, “I was in a hurry. Doesn’t really matter…”.’ At an organisational level, managers’ comments indicated desires and intentions to enhance work-related road safety using a range of initiatives in the very near future. For example, ‘I want to have a new ﬂeet management system where we do record the data that will enable us to manage and then have appropriate consequences for their actions…’ and ‘…on my “to do” list is writing a ﬂeet policy.’ These comments would suggest that managers are currently operating in the preparation stage for some aspects of driver safety. It was also reported that some driver safety initiatives have recently been introduced in the organisation. These initiatives included, for example, consideration of safety features when selecting vehicles, monitoring vehicle servicing, ﬁtting cargo barriers in all station wagons and promoting road safety. Managers reported engaging in some work-related road safety behaviours at a personal level; for example, ‘watch other drivers is the ﬁrst thing I always do’. At an organisational level, examples include ‘making sure you haven’t got projectiles in the car and when you’re hanging clothes up, to hang them on the right rear, which I just can’t tell you how many people I keep pulling up on that. They hang them on the left rear and if you’re looking to see who’s on your left, there’s no way. It totally blocks your vision’ and ‘we have four-wheel drives in some of our residences. So we’ve had our staff in those houses do four-wheel drive classes and road safety classes as part of their induction.’ These comments would suggest that managers are currently operating in the action stage for some aspects of driver safety.

160

Driver Behaviour and Training – Volume III

Managers agreed that while some steps have been taken in the last couple of months, that there was room for improvement in the management of work-related road safety risks within their organisation. This is illustrated by one manager’s comment that ‘I think we’ve got a long way to go in terms of providing better training and information to people who are driving’. When asked about work-related road safety initiatives in their organisation, the managers reported mixed beliefs about the effectiveness of current awarenessraising initiatives. Based on previous discussions with subordinates, one manager believed that the regular road safety posters, emails and information provided on their organisation’s intranet was effective because it increased awareness without forcing employees to go to safety meetings or training. This manager commented that ‘our staff don’t talk about what they learnt at the four-wheel drive course, but you’ll go to a staff meeting and someone will say, “Hey, have you seen that thing about us (on the intranet)” – it’s like, oh my god, you read it.’ In comparison, another manager believed that the effectiveness of information bulletins to raise awareness of road safety risk was limited. This manager noted that ‘There’s been a couple of information bulletins go out on it, but I’m not actually, I can’t recall right now what was actually in it. I don’t think people stop to take a lot of notice of it.’ Managers also identiﬁed a lack of understanding within the organisation as a key barrier in trying to implement road safety initiatives. Managers felt that vehicle incidents were not currently treated as seriously as other health and safety incidents and anticipated resistance when attempting to introduce a vehicle incident monitoring database. This was illustrated through comments such as ‘it will be why do you want to make the people do that? That’s extra work, you know. All they did was reverse into a post, you know. They didn’t hurt anybody.’ Within Organisation A, the interviews indicated that the two employees were operating in the pre-contemplation stage of readiness. Unlike the managers, who recognised both general and organisation-speciﬁc road safety risks, only general road safety risks were acknowledged by the employees. Reported risks included driving long distances, mobile phone use, high volume of trafﬁc on the roads and other drivers lacking good driving ability. It was also noted that management must consider road safety risks to be a signiﬁcant issue because it was discussed at a quarterly senior meeting. Despite both employees being aware of driving risks, neither employee reported intentions to change their driving behaviour. For example one employee commented that ‘I haven’t had any damage to the car in the last seven years. So I’ve done pretty well. So I can’t see any way of improving what I do as such. That would probably change if I had an accident.’ When asked to comment on any work-related road safety initiatives they had experienced in their organisation, the employees reported the presence of many road safety posters displayed in their workplace. While they thought these were beneﬁcial, they were not considered to be sufﬁcient. For example one employee commented: you can’t just send out an e-mail, put up a poster. You really, I mean, ultimately, people should be given speciﬁc training, driver instruction for change to happen, people have to be aware of the need for change, they have to have a desire for change and they have to know – and then they have the knowledge of how to change. Then they have to have

Stages of Change in the Australian Workplace

161

the ability of how to change and then they have to have some reinforcement to make the change stay to stop them bouncing back to the old way.

Organisation B Within Organisation B the interviews indicated that managers (n = 2) and employees (n = 3) were operating in the maintenance stage of readiness. Both managers and employees indicated an awareness of their organisation’s exposure to work-related road safety risks. Reported risks included both general road safety risks and risks that could be considered speciﬁc to their organisation. Generic risks reported included wildlife, poor road conditions, fatigue, driving in isolated areas, travelling long distances and reduced concentration when answering phone calls via hands-free kits. Organisation speciﬁc risks pertained more to the nature of the work demands. For example one manager recognised that employees ‘might get called out of bed at 2 or 3 o’clock in the morning to drive long distances so sleep deprivation comes into it’. Both managers and employees reported that their organisation has continued to roll out road safety initiatives for several years. Examples of reported initiatives include a driver awareness campaign featuring stickers on cars stating that the organisation values safety, a driver proﬁling tool to identify high risk drivers, following employees up on infringement data, considering safety features when selecting vehicles, monitoring vehicle servicing, ﬁtting cargo barriers in all station wagons and enhancing awareness by featuring road safety posters, promoting road safety around holidays and including ﬂeet safety topics in the monthly ﬂeet newsletter. Managers additionally commented on plans to continue enhancing road safety behaviours by updating current initiatives and continuing to introduce new initiatives. Further evidence that Organisation B was operating in the maintenance stage was illustrated by employee comments pertaining to engaging in road safety behaviours. Reported behaviours included using cruise control to manage speeding, slowing down to allow for kangaroos potentially crossing the roads in the afternoons, taking a slower more careful approach when in the city, performing safety checks on vehicles and actively participating in the monthly health and safety meetings. When asked about work-related road safety initiatives in their organisation, both managers reported that driver safety was considered to be a health and safety issue and that they perceived current safety initiatives to be effective. For example: ‘I think it’s just a matter of improving on what we’ve already got’ and ‘they’re amongst the best that you’ll see, no doubt about that’. Alternatively, employees had mixed opinions about the effectiveness of current initiatives. For example, one employee described how they believed that the current safety meetings were working well as they provided a great opportunity to present and trouble shoot safety concerns as a team. Another employee reported that: [S]afety changes normally impact with a negative. Normally, changes will be to not drive as long or far, but increased work loads always conﬂict. All layers of management are aware of the situation about staff shortages and extra distances to travel. I believe the company does have a commitment to driver safety but is willing to overlook its own policy when it comes to a situation of resources and money.

162

Driver Behaviour and Training – Volume III

This statement would suggest that current initiatives provided limited advice on how to juggle the potentially conﬂicting requirements of production and safety. Discussion This chapter demonstrates that the stages of change model provides a useful framework for understanding employee readiness for road safety behaviour change. Additionally, perceptions pertaining to initiative effectiveness were found to vary in relation to an individuals’ stage of readiness. More speciﬁcally in Organisation A, although managers were operating within the preparation and action stages, employees were operating in the pre-contemplation stage. Managers identiﬁed employee resistance to implementing action initiatives as a potential barrier. The stages of change model would suggest that employee resistance to change may occur when safety initiatives are not targeted at an appropriate level for employees’ readiness for change. The selection of an inappropriate level may arise due to differences in readiness for change between managers and employees. When rolling out safety initiatives, managers have often previously spent much time in the contemplation and preparation phases. Understandably, managers are then ready for action and often attempt to impose action initiatives upon employees. Alternatively, employees may not have previously considered the risks of current practice or the beneﬁts of new safety initiatives. They are often not prepared for change and are therefore slow to respond or may even resist the change initiatives. Individuals in Organisation A generally believed that current initiatives aimed at increasing awareness of road safety risks were beneﬁcial. However they were not considered to be sufﬁcient. Based on the stages of change model it is suggested that safety initiatives in Organisation A should continue enhancing current understanding and commitment, but also go further in providing practical information and support to transition employees from a pre-contemplative stage towards action. In comparison, in Organisation B both managers and employees were operating in the maintenance stage of readiness. Overall, managers and employees perceived the organisation’s current safety initiatives, including ongoing advice and practical information, to be effective. However, barriers to maintaining safe driving were identiﬁed, including conﬂicts arising between reaching production targets with increased workloads while adhering to safety policies and procedures. Based on the stages of change model it is suggested that safety initiatives in Organisation B should further support employees to achieve and maintain safe driving behaviour through methods such as ongoing advice about how to resolve driving safely when faced with competing production goals, performance feedback and skills training. The organisation should also continue to monitor employees’ behaviour to identify any early signs of behaviour relapse. Based on this research it is suggested that the stages of change model could be used by practitioners to design driver behaviour and training methods that are appropriate to managers and employees’ readiness for change. Examples of how the stages of change framework could be applied to guide the design of driver education are outlined below. In the pre-contemplation stage employees would see no problem

Stages of Change in the Australian Workplace

163

with their current road safety behaviour and express no intention to change. In this stage individuals need to be persuaded that unsafe driving behaviour must be addressed. The model would suggest that attempts to impose action orientated driver education on pre-contemplative individuals may only achieve partial success as a personal understanding of the risks and a desire to engage in safe behaviours has not ﬁrst been achieved. To help transition employees from a pre-contemplative to a contemplative stage, driver education should focus on raising awareness of workrelated driving risks. In the contemplation stage employees would be aware of the risks associated with work-related driving and the need to adopt safe behaviours. Contemplative individuals would be making long term plans to reduce and manage their road safety risk. To help transition employees from a contemplative to a preparation stage, driver education should focus on providing educational material designed to reinforce their motivation to adopt safe behaviours and outline what is involved in adopting safer driving behaviours. In the preparation stage employees would be intending to take action in the very near future. To help transition employees from a preparation to an action stage, driver education should focus on providing practical information and support in learning new skills. Barriers to change should be resolved, and individuals should be encouraged to make speciﬁc plans through goal setting or contracting to foster employee commitment and ownership of safe driving behaviours. In the action stage employees would be modifying their behaviour or environment to manage work-related road safety risks. Individuals in this stage require support to achieve new safety behaviours and to maintain modiﬁed behaviours. To facilitate commitment to the modiﬁed behaviours and help transition employees from an action to a maintenance stage, driver education should focus on providing ongoing advice, feedback and skills training. In the maintenance stage employees would have been engaging in safety behaviours over a prolonged period of time. To facilitate employees remaining in the maintenance stage, driver education should focus on consolidating the gains made and preventing relapse. This can be achieved through the provision of ongoing advice, feedback and training and the monitoring of employees for early signs of behaviour relapses. It is important to note that employees may relapse to an earlier stage regardless of their current stage of change. To target relapsed employees that have failed to continue engaging in work-related road safety behaviours, driver education should support the progression back through the stages towards action and maintenance. Driver education should aim to discover the barriers that led to the employee ceasing safe practices and to motivate the employee to re-engage in safe practices through the provision of tailored information, training and feedback. This chapter builds upon previous literature pertaining to driver education by identifying how the stages of change model can be applied as a useful framework to guide the development of driver education programmes. Based on the model, it is suggested that providers of driver education for organisations could make a brief assessment during project negotiation of managers and employees readiness for change. Based on their assessment, providers could then determine the most

164

Driver Behaviour and Training – Volume III

appropriate structure and content of driver education to effectively meet the needs of their client. By adopting a stage-matched approach to driver education, providers may be able to reduce change resistance and accelerate employee movement towards the action and maintenance stage of work-related road safety behaviours. Due to the small participant sample size, further research should be conducted to validate these ﬁndings with a larger number of employees within case study organisations and across a greater range of organisations to enhance the generalisability of research ﬁndings. References Australian Safety and Compensation Council (2006). Compendium of Workers Compensation Statistics Australia 2003–04. Canberra: Commonwealth of Australia. Barrett, J.H., Haslam, R.A., Lee, K.G. and Ellis, M.J. (2005). ‘Assessing attitudes and beliefs using the stage of change paradigm: case study of health and safety appraisal within a manufacturing company.’ International Journal of Industrial Ergonomics, 35, 871–87. Campbell, M.K., DeVellis, B.M., Strecher, V.J., Ammerman, A.S., DeVillis, R.F. and Sandler, R.S. (1994). ‘Improving dietary behaviour: the effectiveness of tailored messages in primary care settings.’ American Journal of Public Health, 84, 783–7. Chapman, P., Underwood, G. and Roberts, K. (2002). ‘Visual search patterns in trained and untrained novice drivers.’ In Transportation Research Part F: Trafﬁc Psychology and Behaviour, 5(2), 157–67. Christensen, P. and Glad, A. (1996). Mandatory Course of Driving on Slippery Roads for Drivers of Heavy Vehicles: The Effect on Accidents. TOI Report 334/1996. Oslo: Transpotokonomisk institutt. Davey, J. and Banks, T. (2005). ‘Estimating the cost of work motor vehicle incidents in Australia.’ Paper presented at the Australasian Road Safety Research Policing Education Conference, Wellington, New Zealand. Dorn, L. and Barker, D. (2005). ‘The effects of driver training on simulated driving performance.’ Accident Analysis and Prevention, 37(1), 63–9. Haslam, C. and Draper, E.S. (2000). ‘Stage of change is associated with assessment of the health risks of maternal smoking among pregnant women.’ Social Science Medicine, 51, 1189–96. Haslam, R.A. (2002). ‘Targeting ergonomics interventions: learning from health promotion.’ Applied Ergonomics, 33, 241–9. Katila, A., Keskinen, E., Hatakka, M. and Laapotti, S. (2004). ‘Does increased conﬁdence among novice drivers imply a decrease in safety? The effects of skid training on slippery road accidents.’ Accident Analysis and Prevention, 36, 543– 50. Laforge, R.G., Velicer, W.F., Richmond, R.L. and Owen, N. (1999). ‘Stage distributions for ﬁve health behaviours in the USA and Australia.’ Preventative Medicine, 28, 61–74. Llaneras, R.E., Swezey, R.W., Brock, J.F., Rogers, W.C. and Van Cott, H.P. (1998).

Stages of Change in the Australian Workplace

165

‘Enhancing the safe driving performance of older commercial vehicle drivers.’ International Journal of Industrial Ergonomics, 22(3), 217–45. Lund, A.K. and Williams, A.F. (1985). ‘A review of the literature evaluating the defensive driving course.’ Accident Analysis and Prevention, 17(6), 449–60. Lynam, D. and Twisk, D, (1995). Car Driver Training and Licensing Systems in Europe. TRL Report 147. Crowthorne, UK: Transport Research Library. Lynn, P. and Lockwood, C. (1998). The Accident Liability of Company Car Drivers. TRL Report 317. Berkshire: Transport Research Laboratory. Miles, M.B. and Huberman, A.M. (1994). An Expanded Source Book: Qualitative Data Analysis. London: Sage Publications. Prochaska, J.M., Prochaska, J.O. and Levesque, D.A. (2001). ‘A trans-theoretical approach to changing organizations.’ Administration and Policy in Mental Health, 28(4), 247–61. Prochaska, J.O., DiClemente, C.C. and Norcross, J.C. (1992). ‘In Search of How People Change: Applications to Addictive Behaviours.’ American Psychologist, 47(9), 1102–14. Rakowski, W., Ehrich, B., Goldstein, M., Rimer, B., Pearlman, D., Clark, M., Valicer, W. and Woolverton, H. (1998). ‘Increasing mammography screening among women aged 40–74 by use of a stage-matched, tailored intervention.’ Preventative Medicine, 27, 748–56. Velicer, W., DiClemente, C., Prochanska, J. and Brandenburg, N. (1985). ‘Distribution of smokers by stage in three representative samples.’ Preventative Medicine, 24, 401–11. Wheatley, K. (1997). ‘An overview of issues in work-related driving.’ In Staysafe 36: Drivers as Workers, Vehicles as Workplaces: Issues in Fleet Management. (Report No. 9/51). Ninth report of the Joint Standing Committee on Road Safety of the 51st Parliament. Sydney: Parliament of New South Wales, 15–24. Whysall, Z., Haslam, C. and Haslam, R. (2006). ‘Implementing health and safety interventions in the workplace: an exploratory study.’ International Journal of Industrial Ergonomics, 36, 809–18. World Health Organization (2004). ‘Road safety: a public health issue.’ www.who. int/features/2004/road_safety/en/print.html.

This page intentionally left blank

Chapter 14

Prospective Relationships between Physical Activity, ‘Need for Recovery’ and Driver Accidents and Absenteeism Adrian Taylor1 and Lisa Dorn2 1 University of Exeter, UK 2 Cranﬁeld University, UK Introduction It is recognised that driver stress and fatigue are major contributors to at-work road trafﬁc accidents (Matthews and Desmond, 1998) and occupational health (Sluiter et al., 2003). However, it is less clear, and has been of little concern, if and how (increasingly) low levels of physical activity in western countries (Department of Health, 2004), are inﬂuencing driver stress, fatigue, health status (absenteeism) and risk of driver accident (Shephard, 1996). Taylor and Dorn (2003; 2006) provided the ﬁrst model and literature review to guide understanding of how physical (in)activity may inﬂuence risk of driver accidents. They proposed, through a number of possible pathways, that being regularly physically active may favourably inﬂuence psychological responses to stress by increasing resilience and energy, decreasing levels of fatigue and enhance sleep quality and health status. In turn these mediating factors could all contribute towards a lower risk of accidents and absenteeism. In contrast, a sedentary lifestyle may contribute to progression towards greater fatigue and less resilience for coping with the psychological demands of driving for work. Stress has been shown to reduce physical activity (Ng and Jeffery, 2003; Payne et al., 2002; Steptoe et al., 1996; Stetson et al., 1997), possibly due to a perceived need to recover from mental fatigue by resting after work. With decreasing physical activity, a downward spiral of increasing occupational fatigue (due to physical deconditioning) and stress may occur with increasing risk for driver accidents. Thus, the concept of ‘need for recovery’ would seem likely to be determined, to some degree, by general levels of physical activity, but also to have an impact on accidents and health status. In a unique study, Sluiter, van der Beek and Frings-Dresen (1997) reported that not only did less physically active coach drivers report a longer ‘need for recovery’ after work, but were more likely to have an accident. Drivers engaging in more than one weekly session of exercise had 0.78 accidents per driver, whereas less-active drivers had 1.05 accidents over a two year period. The measures of physical activity used in the study were limited as this was not a main element of the Dutch study of

168

Driver Behaviour and Training – Volume III

long-distance bus drivers. Further research is therefore needed to better understand the link between physical activity, fatigue, health status, accidents and absenteeism. The purpose of the present study was to investigate if level of activity among trainee UK bus drivers is associated with future accidents and absenteeism, using a validated measure of physical activity. A secondary purpose was to investigate if self-reported fatigue (‘need for recovery’) mediates the relationship between physical activity and both accidents and absenteeism. Methods Participants A total of 183 newly recruited bus drivers were assessed over a one year period, as part of a larger study. Recruits attended a screening session at a large bus depot in the south of England. Each driver completed a battery of pencil and paper or computerbased tests. They also undertook a detailed assessment of their performance on a bus driving simulator (reported elsewhere). Measures Self-reported physical activity was assessed using the short selfadministered version of the International Physical Activity Questionnaire (IPAQ: Craig et al., 2003). This seeks to gain information on the number of days and minutes per day in the previous seven days that the person engaged in vigorous and moderate intensity physical activity. Walking was a separate category, as was time spent sitting. The 11 item ‘need for recovery’ scale (Sluiter et al., 2003) was used to assess occupational-related fatigue. The scale includes items such as, ‘at the end of a working day I am usually feeling worn out’, ‘after a working day I am often too tired to start other activities’, with responses required on a ﬁve point scale from ‘strongly disagree’ (1) to ‘strongly agree’ (5). Self-reported health status was determined from the General Health Questionaire-12 (GHQ-12; Goldberg et al., 1997). A single item ‘Do you feel like you would like to do more physical activity than you currently do’ (with a yes/no response format), and other questions about perceived availability and preference for exercise opportunities were also asked. The number of at-fault, partly at-fault, and not at-fault bus accidents, over a subsequent three month period, were retrieved from a corporate database. Data analysis Data from the IPAQ was cleaned and scored (see guidelines at http:// www.ipaq.ki.se/), and the sample was categorised into three levels (as recommended following extensive international development work on the survey): low active (that is, < 600 MET-minutes/week of activity); moderately active (that is, 600–3000 MET-minutes/week of activity); and high active (that is, >3000 MET-minutes/week of activity). As recommended, subjects with scores of greater than 960 minutes per week of total vigorous, moderate and walking activity were excluded as outliers. Complete data (including the surveys and information on accidents and absenteeism) from 123 drivers are reported. Scores were derived for ‘need for recovery’ and GHQ12 by adding the respective items (following reverse coding as appropriate).

Prospective Relationships between Physical Activity

169

Amount of physical activity was reported for each of the three levels of activity. Analyses of variance (ANOVA) was used to compare number of accidents, absenteeism, general health (GHQ-12), and ‘need for recovery’ for the three levels of physical activity. Given the unequal distribution of the sample across the three levels of physical activity and the number of outliers excluded (that is, possible over reporting of physical activity) using the IPAQ we also conducted a series of t-tests to compare those responding ‘yes’ and ‘no’ to the question about wishing to do more physical activity than the participants were engaging in currently. Regression analysis was performed to determine the effects of physical activity on the number of accidents and absenteeism directly, and indirectly, through ‘need for recovery’. Results The sample characteristics were as follows: mean age of 37.8 (SD 10.1)(ranging from 19–60 years), 85 per cent male, 2.2 years in formal post-16 years of age education, self-reported weight/height (body mass index) of 27.3 (SD 6.5), mainly white (79 per cent white, 8 per cent black or black British, 7 per cent Asian or Asian British), mainly married (44 per cent married, 21 per cent single, 17 per cent live with partner, 9 per cent separated or divorced), and 75 per cent had smoked, with 47 per cent currently smoking at the time of the study. The latter reported smoking a mean of 104 (SD 42) cigarettes a week (calculated from per weekday and weekend day). Table 14.1 shows the descriptive data for minutes per week of vigorous and moderate intensity exercise and walking for the three activity categories. It also shows the relationship between the categorised levels of physical activity and other variables of interest. There were no signiﬁcant difference in accident rates and absenteeism over three months between the different levels of physical activity, although a trend did emerge. Low physically active drivers had approximately twice as many accidents and took almost six times as many days off work compared with high active drivers. Albeit with relatively small numbers, 8/17 (47 per cent), 22/44 (33 per cent) and 6/23 (26 per cent) of those in the low, moderate and high active groups had at least one accident, although a chi2 test failed to reveal a signiﬁcant overall effect. Surprisingly, physical activity was not related to scores on the GHQ12. In contrast, level of physical activity was associated with ‘need for recovery’, with post hoc tests revealing that the high active group had signiﬁcantly lower scores (expressed as differences and conﬁdence intervals) than both the moderate (3.6; 0.1 to 7.1) and low (6.0; 0.96 to 11.1) active drivers.

Driver Behaviour and Training – Volume III

170

Table 14.1

The relationship between three levels of physical activity and ‘need for recovery’, health status, accidents and absenteeism Low active

Vigorous minutes/week Moderate mins/week Walking Total METmins/week Total accidents after three months Absent days after three months GHQ-12 ‘Need for recovery’

Mean (SEM) N =15 5.88 (18.49)

Moderate active Mean (SEM) N = 76 73.22 (8.52)

High active Mean (SEM) N = 28 235.00 (14.40)

10.00 (28.92)

97.44 (13.33)

241.61 (22.53)

84.12 (45.47)

293.71 (20.96)

275.00 (35.43)

364.62 (139.34) 0.76 (0.20)

1935 (64.24) 0.53 (0.10)

3753.93 (108.58) 0.35 (0.17)

1.23 (0.37)

0.63 (0.17)

0.21 (0.28)

2.2 (2,122) p = 0.09

34.38 (1.20) 29.80 (1.68)

34.09 (0.55) 27.41 (0.75)

34.12 (0.96) 23.79 (1.23)

0.0 (2,113) ns 4.9 (2,116) p = 0.009

F (df) 61.8 (2,122) p < 0.001 23.21 (2,122) p < 0.001 8.9 (2,122) p < 0.001 196.7 (2,122) p < 0.001 1.2 (2,103) ns

Figure 14.1 shows a path model for the direct and indirect effects of estimated energy expenditure (from multiplying vigorous, moderate and walking mins by 8.0, 4.0 and 3.3 METs or units of energy expenditure, respectively; 1 is assumed to be equivalent to resting) from the sum of all forms of physical activity (from the IPAQ) in the past week at baseline, on accidents and absenteeism over the next three months. Conﬁrming data in Table 14.1, physical activity was not signiﬁcantly associated with either accidents or absenteeism. Physical activity was associated with ‘need for recovery’, which in turn predicted the number of accidents over three months. Mean (SEM) ‘need for recovery’ scores for those who had non versus at least one accident were 25.4 (0.7) and 28.7 (1.4)(t136 = 2.3, p < 0.05), respectively, but there were no differences in physical activity. ‘Need for recovery’ did not predict absenteeism. A series of paired t-tests revealed that those who would like to be more physically active (n = 126)(compared with those who didn’t, n = 42) reported a signiﬁcantly (corrected for unequal variances) greater mean (SEM) ‘need for recovery’, t77 = 4.0, 27.4 (0.6) v 22.7 (0.9), p < 0.001; more accidents, t93 = 2.3, 0.53 (0.08) v 0.26 (0.09), p < 0.05; and more days absent, t159 = 1.9, 0.93 (0.21) v 0.43 (0.15), p = 0.05.

Prospective Relationships between Physical Activity

171

Physical activity

-.04

-.24 **

0.08

Need for recovery .21 * Accidents in 3 months

.06 Absenteeism in 3 months

Figure 14.1 The mediating effect of ‘need for recovery’ between physical activity (total MET-mins/week) and accidents and absenteeism (after three months) Discussion The absence of a signiﬁcant difference in accident and absenteeism rates between low and high physical activity categories did not appear to be due to lack of statistical power. However, further analyses, in the on-going study, will be conducted as the number of accidents and days absent from work accumulates post-study over six and twelve months into employment. The idea that regular physical activity provides protection against accidents through greater energy, faster recovery and less workrelated fatigue received some support from the present ﬁndings. The need for recovery appears to be an important psychological variable and indicator of work-related stress and fatigue, as Sluiter and colleagues (2003) have previously reported. The study also supports previous ﬁndings that physical activity is related to ‘need for recovery’ (Sluiter et al. 1997). The prevalence of physical activity was unusually high for the present sample of professional drivers compared with other studies (Taylor and Dorn, 2006). However, the likely over-reporting by the participants did not limit the comparison between levels of activity. The use of a proxy measure of activity (that is, preference to do more activity; yes or no) helped to provide an additional insight into understanding the ﬁndings. Approximately 75 per cent of the sample expressed a desire to do more activity, and these drivers also reported much higher levels on the ‘need for recovery’ scale. They also had more accidents and absenteeism. Regular physical activity can offer greater coping resources and reduce fatigue (Puetz, O’Connor and Dishman, 2006) and single sessions can reduce cardiovascular stress reactivity to mental challenges (Hamer, Taylor and Steptoe, 2005). It would therefore seem an important behaviour to promote in the context of at work driver health. The challenge is that with occupational induced fatigue, the need to recover through physical inactivity (for example, watching tv) may be a common cognition.

172

Driver Behaviour and Training – Volume III

The effect of mood on behaviour is of increasing interest, with a need to promote moderate physical activity as something that is pleasurable and activating. Overall, the drivers in the study reported being more active than expected, with almost 90 per cent reporting that they achieved the Department of Health’s guidelines of at least 150 minutes of moderate or vigorous physical activity per week. We excluded data from 46 drivers in line with IPAQ guidelines for removing outliers, and it would seem likely that there was an overall inﬂation of physical activity reported using this survey. A simple question about desire to be more active revealed that 81 per cent would, with 91 per cent in the low active group indicating that they would like to do more physical activity. Interestingly, overall 78 per cent and 52 per cent of the sample, respectively, indicated that they would use a gym at least twice a week if one were available at the workplace, and it was free or if there was a charge of £3 per visit. Conclusions In summary, this study provides some preliminary evidence that promoting physical activity may be an important factor to consider for reducing at-work crash risk and absenteeism. It is also interesting to note that the sample of bus drivers were interested in being more physical active. It is only the second study to show a link between physically activity and risk of at work crashes, albeit a small one, with a relatively small study. ‘Need for recovery’ is further supported as an important construct that mediates the link between physical activity and accidents among professional drivers. References Craig, C.L., Marshall, A.L., Sjostrom, M., et al. (2003). ‘International physical activity questionnaire: 12-country reliability and validity.’ Medicine and Science in Sports and Exercise, vol. 35, 1381–95. de Croon, E.M., Sluiter, J.K. and Frings-Dresen, M.H. (2003). ‘Need for recovery after work predicts sickness absence: a two year prospective cohort study in truck drivers.’ Journal of Psychosomatic Research, vol. 55, 331–9. Department of Health (2004). At Least Five a Week: Evidence on the Impact of Physical Activity and its Relationship to Health. London: Dept of Health. Ekkekakis, P., Hall, E.E., VanLanduyt, L.M. and Petruzzello, S.J. (2000). ‘Walking in (affective) circles: can short walks enhance affect?’ Journal of Behavioral Medicine, vol. 23, 245–75. Goldberg, D.P., Gater, R., Sartorius, N., Ustun, T.B., Piccinelli, M., Gureje, O. and Rutter, C. (1997). ‘The validity of two versions of the GHQ in the WHO study of mental illness in general health care.’ Psychological Medicine, vol. 27, 191–7. Hamer, M., Taylor, A.H. and Steptoe, A. (2006). ‘The effect of acute aerobic exercise on blood pressure reactivity to psychological stress: a systematic review’, Biological Psychology, vol. 71, 183–90. Marcus, B.H., Rossi, J.S., Selby, V.C., Niaura, R.S. and Abrams, D.B. (1992). ‘The

Prospective Relationships between Physical Activity

173

stages and processes of exercise adoption and maintenance in a worksite sample.’ Health Psychology, vol. 11, 386–95. Puetz, T.W., O’Connor, P.J. and Dishman, R.K. (2006). ‘Effects of chronic exercise on feelings of energy and fatigue: a quantitative synthesis.’ Psychological Bulletin, vol. 132, 866–76. Shephard, R.J. (1996). ‘Worksite ﬁtness and exercise programs: a review of methodology and health impact.’ American Journal of Health Promotion, vol. 10, 436–52. Sluiter, J.K., van der Beek, A.J. and Frings-Dresen, M.H. (1997). ‘Workload of coach drivers. [Werkbelasting touringcarchauffeurs].’ Rep. No. 97-03, Amsterdam: Coronel Inst. Occup. Environ. Health, Acad. Medical Centre, 1–71. Sluiter, J.K., de Croon, E.M., Meijman, T.F. and Frings-Dresen, M.H.W. (2003). ‘Need for recovery from work-related fatigue and its role in the development and prediction of subjective health complaints.’ Occupational and Environmental Medicine, vol. 60, 62–70. Taylor, A.H. and Dorn, L. (2003). ‘The effects of exercise on stress, fatigue, sleep, health status and potential risk of at-work road trafﬁc accidents: a multi-disciplinary model.’ In L. Dorn (ed.). Proceedings of 1st International Conference on Driver Behaviour and Training, Stratford-upon-Avon, UK: Ashgate Publishers, 175–90. Taylor, A.H. and Dorn, L. (2006). ‘Effects of physical inactivity on stress, fatigue, health and risk of road trafﬁc accidents.’ Annual Review of Public Health, vol. 27, 371–91.

This page intentionally left blank

Chapter 15

Predicting High Risk Behaviours in a Fleet Setting: Implications and Difﬁculties Utilising Behaviour Measurement Tools Jeremy Davey, James Freeman and Darren Wishart Centre for Accident Research and Road Safety, Queensland (CARRS-Q), Australia Introduction Fleet and work-related motor vehicle crashes represent a substantial physical, emotional and ﬁnancial cost to the community. Previous estimations have indicated that the total cost of work-related road incidents in Australia was in the vicinity of $1.5 billion (Wheatley, 1997). More recent evidence has suggested that the average total insurance cost of a ﬂeet incident to organisations and society is approximately $28 000 (Davey and Banks, 2005), while the average cost of a fatal crash in the general Australian motoring community is estimated to be $2 million (Austroads, 2006). Furthermore, estimates of the true cost for work-related crashes suggest that hidden costs may be somewhere between 8–36 times vehicle repair/replacement costs (Murray et al., 2003). Of note is that a high proportion of work-related deaths and injuries within the overall road toll consist of work-related crashes (Murray et al., 2003; Wheatley, 1997), as work-related trafﬁc injuries have been estimated to be twice as likely to result in death or permanent disability than other workplace accidents (Wheatley, 1997). Driving assessment tools Given the tremendous burden that road crashes have on society, researchers are directing their focus towards investigating the attitudes and behaviours of general motorists, as well as determining the value of such self-reported data to predict crash involvement. Such measurement tools include: the Driving Skill Inventory (Lajunen and Summala, 1997), Driver Anger Scale (Deffenbacher, Oetting and Lynch, 1994), the Manchester Driver Behaviour Questionnaire (DBQ) (Reason et al., 1990), Driver Attitude Questionnaire (DAQ) (Parker et al., 1996) and the Safety Climate Questionnaire-MD (SCQ-MD) (Glendon and Litherland, 2001). The latter three questionnaires are proving increasingly popular in identifying the factors associated with vehicle crashes and demerit point loss among ﬂeet drivers in work settings, and will remain the focus of the present study.

176

Driver Behaviour and Training – Volume III

Firstly, in regards to the DBQ, this measurement tool has been extensively utilised within a range of driver safety research areas such as: the genetics of driving behaviour (Bianchi and Summala, 2004), age differences in driving behaviour (Dobson et al., 1999), cross cultural studies (Lajunen et al., 2003) and associations with the likelihood of being involved in an accident (Dobson et al., 1999; Parker et al., 1995; Reason et al., 1990). Such research has predominantly focused on general motorists, which has indicated that speeding violations are one of the most common factors associated with crash involvement (Parker et al., 1995). Another driving tool which is beginning to receive increasing attention within the road safety literature is the Driver Attitude Questionnaire (Parker et al., 1996). Research has begun to utilise the DAQ within a number of different applied settings such as: speed awareness training (Meadows, 2002), general driver training programs (Burgess and Webley, 2000), bicycle interventions (Anderson and Summala, 2004), as well as ﬂeet programmes (Davey et al., 2006; Wishart et al., 2006). Preliminary research indicates that the DAQ has the potential to be utilised to investigate motorists’ attitudes towards key road safety issues, such as drink driving, risky overtaking, close following and driving above the speed limit, with motorists generally reporting the most lenient attitudes towards speeding violations (Davey et al., 2006; Meadows, 2002; Wishart et al., 2006). The Safety Climate Questionnaire-Modiﬁed for Drivers (SCQ-MD) is also being utilised within road safety arenas, as researchers begin to recognise the importance of an organisation’s attitudes towards ﬂeet and road safety issues. In simple terms, ‘climate’ relates to how employees perceive the organisational culture and practice of a company (Glendon and Stanton, 2000), and it is hypothesised that this perception impacts upon the way in which workers ultimately behave at work (Wills, 2006). In regards to safety climate, a growing body of research is demonstrating a link between safety culture and a variety of outcomes, ranging from vehicle crash rates (Diaz and Cabrera, 1997; Mearns, Whitaker and Flin, 2003) to injury severity (Gillen, Baltz, Gassel, Kirsch and Vaccaro, 2002). For example, Wills, Watson and Biggs (forthcoming) investigated the driving behaviours of 323 ﬂeet employees and reported that work pressure and communication were signiﬁcantly related to driver distraction. Also, Newnam, Watson and Murray (2002) examined the self-reported driving behaviours of ﬂeet drivers and reported that the safety policies and practices within organisations had a direct impact on driving performance. Taken together, research is beginning to suggest that perceptions regarding the safety policies and practices of organisations may have a direct impact on driving outcomes. Fleets However, despite the prevalence of research currently focusing on identifying the self-reported attitudes and behaviours that inﬂuence crash involvement, relatively little research has endeavoured to examine the self-reported driving behaviours of those who drive company sponsored vehicles and/or spend long periods of time behind the wheel (Newnam et al., 2002; Newnam et al., 2004; Sullman et al., 2002; Xie and Parker, 2002). The lack of assessment tools in the Australian context appears to be a critical oversight as changes in industry/employer accountability, Occupational

Predicitng High Risk Behaviours in a Fleet Setting

177

Health and Safety (OHS) legislation, Workers Compensation legislation and public liability are requiring industry to develop better benchmarking along with more comprehensive intervention programmes related to ﬂeet safety. Currently, ﬂeet organisations cannot effectively assess current risk and thus also cannot target and develop interventions nor evaluate the effectiveness of countermeasures due to the lack of adequate measurement tools. As a result, ﬂeet companies are experiencing difﬁculties meeting their legislative requirements to reduce risk (OHS) despite compulsory third party insurance companies demanding increasingly better ﬂeet safety environments across organisations. What is presumed is that drivers of employer owned vehicles who drive for work-related purposes generally engage in a higher prevalence of aberrant driving behaviours such as speeding (Stradling, 2000), and are at greater risk of crash involvement due to their exposure to the driving environment (Newnam et al., 2002; Sullman et al., 2002). Preliminary evidence suggests that speeding is the most likely illegal behaviour to be reported by ﬂeet drivers (Davey et al., 2007; Dimmer and Parker, 1999; Wishart et al., 2006). However, further research appears necessary to determine which self-reported measurement tools are most useful within ﬂeet settings as well as what speciﬁc attitudinal and behavioural factors predict crash involvement within such settings. As a result, the present research aimed to utilise three prominent driving measurement tools to investigate the relationship between self-reported attitudes, behaviours and crash involvement. More speciﬁcally, the study aimed to: 1. examine a group of ﬂeet drivers’ attitudes and behaviours regarding road safety issues via three measurement tools (i.e., DBQ, DAQ and SCQ-MD); 2. investigate the relationship the sub-factors of the measurement tools have with self-reported crash involvement; and 3. report on the associated difﬁculties utilising behaviour measurement tools in ﬂeet settings. Method Participants A total of 4195 individuals from a large Australian company volunteered to participate in the study. There were 3642 males (88.9 per cent) and 553 females (11.1 per cent). The average age of the sample was 43.7 years (range 18–66 years). Participants were located throughout Australia in both urban and rural areas. The sample consisted of approximately equal numbers of ofﬁce workers, n = 2244 (46.8 per cent) and ﬁeld workers, n = 2264 (47.2 per cent), with n = 284 (5.9 per cent) of respondents not indicating their employment type. Examination of vehicle types revealed that the largest proportion of the sample reported driving sedans (n = 2872, 61.2 per cent), followed by station wagons (n = 1375, 28.69 per cent), vans (n = 861, 18 per cent) and ‘customer service vehicles’ (CSV) (n = 518, 11 per cent), with only a small

178

Driver Behaviour and Training – Volume III

percentage indicating usage of four-wheel drive vehicles,1 utes or heavy vehicles. The majority of driving by participants was reported to be within the city n = 1988 (42.4 per cent), or in the city and on country roads n = 1867 (39.82 per cent), with only 767 participants (16.36 per cent) reporting driving on rural roads. On average participants had held their licence for 26 years. A total of 588 participants reported being involved in a crash while driving for work in the past 12 months. Questionnaire Driver Behaviour Questionnaire (DBQ) In the current study a, modiﬁed version of the DBQ was used, which consisted of 20 items. Questions relating to lapses were omitted due to previous research indicating that this factor is not associated with crash involvement (Lawnton et al., 1997). In addition, the authors of the current paper made minor rewording or rephrasing modiﬁcations, in order to make the questionnaire more representative of Australian driving conditions. For example, references to turning ‘right’ were removed on some items, as there are instances where drivers may attempt to overtake someone who is turning left.2 Respondents were required to indicate on a six point scale (0 = ‘never’ to 5 = ‘nearly all the time’) how often they commit each of the errors (eight items), Highway Code violations (eight items) and aggressive violations (four items). Driver Attitude Questionnaire (DAQ) The DAQ is a 20 item self-report questionnaire designed to measure attitudes regarding a range of driving behaviours, which are collated to identify four factors: drink driving, close following, dangerous overtaking and speeding. Respondents are required to indicate on a six point Likert scale (0 = ‘strongly disagree’ to 5 = ‘strongly agree’) their agreement with statements regarding the appropriateness of various driving behaviours. Safety Climate Questionnaire (SCQ) A 29 item version of the SCQ was utilised in the research project. Minor modiﬁcations ensured that the questions related speciﬁcally to ‘work-related driving’. The SCQ contains ﬁve sub-factors that aim to measure perceptions towards ﬂeet safety rules, communication and support, work pressures, adequacy of ﬂeet safety procedure and management commitment. A growing body of research has demonstrated that the SCQ is a reliable tool to measure ﬂeet drivers’ attitudes towards the safety climate of an organisation (Wills et al., 2006; Wills et al., forthcoming).

1 Other than the CSV body type. 2 Previous research has demonstrated that the DBQ is robust to minor changes to some items in order to reﬂect speciﬁc cultural and environmental contexts (Blockey and Hartley, 1995; Ozkan and Lajunen, 2005; Parker et al., 2000).

Predicitng High Risk Behaviours in a Fleet Setting

179

Demographic measures A number of socio-demographic questions were included in the questionnaire to determine participants’ age, gender, driving history (for example, years experience, number of trafﬁc offences and crashes) and their weekly driving exposure (for example, type of car driven and driving hours). Procedure The vehicle insurance company provided a list of individuals who expressed interest in participating in the research. A letter of introduction, the study questionnaire and a reply paid envelope were distributed through the company’s internal mail system to the participants. In total, 1440 were mailed out and 443 were returned indicating a 30 per cent response rate. Results Structure and reliability of the questionnaires for an Australian sample The internal consistency of the DBQ, DAQ and SCQ-MD scores were examined through calculating Cronbach’s alpha reliability coefﬁcients, and are presented in Table 15.1. The SCQ factors, which speciﬁcally relate to safety, appear to exhibit the highest level of internal consistency. Similar to previous Australian research (Blockey and Hartley, 1995; Dobson et al., 1999), and on professional drivers (Sullman et al., 2002), the DBQ factors also appear to exhibit relative internal consistency. In contrast, there has been little research to determine the psychometric properties of the DAQ, and although only moderate, the alpha coefﬁcients are similar to previous research (Meadows, 2002). Table 15.1 also displays the overall mean scores for the DBQ, DAQ and SCQMD factors. Higher means on the DBQ indicate more deviant driving behaviours, while higher scores on the DAQ and SCQ-MD indicate more appropriate road safety attitudes, and positive perceptions regarding the organisation’s road safety culture, respectively. Firstly, an examination of the mean scores reveals that for the DBQ scale, participants were most likely to engage in speeding offences while at work, which was signiﬁcantly more likely compared with committing driving errors F(1, 4195) = 70.73, p <0.01 or aggressive violations F(1, 4195) = 83.42, p <0.01.3 The results indicate that speeding is the most common form of aberrant behaviour reported by the ﬂeet drivers in the current sample and, similar to previous research on professional drivers (Newnam et al., 2004; Sullman et al., 2002), speeding remains a major road safety concern (Davey et al., 2006). Secondly, an examination of participants’ attitudes (DAQ) revealed respondents were most concerned about close following (M = 4.00), however it is noted that the 3 However, it is noted that the means scores for all three DBQ factors are relatively low, which indicates participants generally reported that they did not regularly engage in the speciﬁed aberrant driving behaviours.

180

Driver Behaviour and Training – Volume III

Table 15.1 Alpha reliability coefﬁcients of the measurement scales Measurement scale

Alpha

M

SD

DBQ Errors Highway Code violations Aggressive violations

(8 items) (8 items) (4 items)

0.78 0.77 0.56

1.36 1.50 1.38

0.38 0.47 0.43

DAQ Alcohol Close following Overtaking Speeding

(5 items) (5 items) (5 items) (5 items)

0.67 0.67 0.55 0.67

3.84 4.00 3.84 3.55

0.66 0.59 0.56 0.66

SCQ Fleet safety rules Communication and support Work pressures Adequacy of procedures Management commitment

(3 items) (8 items) (8 items) (3 items) (7 items)

0.74 0.89 0.93 0.86 0.93

4.33 3.83 3.53 4.14 4.18

0.46 0.50 0.18 0.43 0.60

sample also believed it was generally unacceptable to drink and drive, speed, as well as engage in risky overtaking manoeuvres in some circumstances. The results indicate that attitudes towards drink driving are the most stringent reported by the ﬂeet drivers in the current sample and, similar to previous research (Meadows, 2002; Parker et al., 1995), speeding is the most accepted aberrant driving behaviour. However, it is noted the differences between the factors are relatively small and may therefore diminish the practical signiﬁcance of the ﬁndings. In contrast to the self-reported behaviours and attitudes, participants reported the organisation promoted positive and adequate road safety rules (M = 4.14), fostered a commitment to road safety (M = 4.18) and were able to communicate and receive support regarding road safety issues (M = 3.83). However, it is also noted that participants reported some level of work pressure (M = 3.53). These differences between the various questionnaires’ factors will be examined further in the following section. Intercorrelations between variables An examination was undertaken to determine the bi-variate relationships between the DBQ, DAQ and SCQ-MD factors as well as socio-demographic variables. While the association between the major factors and crash involvement are examined in the following logistic regression analyses, some notable bi-variate relationships are reported below. As expected, strong relationships appeared evident between the DAQ factors, with the highest correlation being between close following and risky overtaking (r =

Predicitng High Risk Behaviours in a Fleet Setting

181

0.69**). That is, those who reported an unwillingness to engage in risky overtaking manoeuvres were also unlikely to perceive close following as an acceptable driving behaviour. Similar results were also found between the DBQ factors, with the strongest bi-variate relationship identiﬁed between Highway Code and aggressive violations (r = 0.53**). With regards to bi-variate relationships between the questionnaires, signiﬁcant negative correlations were evident between all the DBQ and DAQ subfactors (for example, behaviours versus attitudes), as those who perceived aberrant driving behaviours such as speeding as serious were subsequently less likely to actually engage in such behaviours over the previous six month period (that is, r = –0.33**). Similar negative correlations were identiﬁed between the DBQ and SCQ-MD factors, as the positive work environment which provided ﬂeet safety rules, procedures and support resulted in lower levels of self-reported aberrant driving behaviour. For example, adequate ﬂeet safety rules were negatively correlated with driving errors (r = –0.21**), highway violations (r = –0.23**) and aggressive violations (r = – 0.15**). With regards to sample characteristics, a similar negative relationship was found between age and the DBQ factors, as older drivers were less likely to engage in aberrant driving behaviours as well as report positive attitudes towards road safety, as measured by the DAQ. Finally, participants who drove further distances were less likely to report positive driving attitudes as measured by the DAQ, although this did not necessarily result in a higher frequency of engagement in aberrant driving behaviours, such as speeding and aggressive violations as measured by the DBQ. However, committing a higher number of driving errors in the last six months was positively associated with self-reported work pressure (r = 0.25**). Prediction of work crashes The third part of the study aimed to examine the relationship between participants’ driving attitudes and behaviours as measured by the DAQ, DBQ, SCQ-MD and selfreported work crashes. A total of 588 participants reported being involved in a crash while driving for work in the last year. A logistic regression analysis was performed to examine the contributions of the DAQ factors (for example, overtaking, speeding, close following and alcohol), DBQ factors (for example, Highway Code violations, aggressive violations and errors), SCQ-MD factors (rules, communication, work pressure, procedures and management commitment) as well as exposure to driving (for example, kilometres driven each year and hours driving per week) to the prediction of self-reported crashes in the past 12 months. Table 15.2 depicts the variables in each model, the regression coefﬁcients, as well as the Wald and odds ratio values. Self-reported number of kilometres driven each year and hours of driving per week were entered in the ﬁrst step to examine, as well as control for, the inﬂuence of driving exposure before the inclusion of the proposed attitudinal and behavioural factors. As expected, participants who reported a higher level of driving exposure (that is, km per year) were most likely to indicate that they had been involved in a work-related crash in the past 12 months, p = 0.000.

Driver Behaviour and Training – Volume III

182

Next, the DBQ, DAQ factors and SCQ-MD factors were entered in the model to assess whether the proposed attitudes and behaviours improved the prediction of crash involvement, over and above exposure to driving (Step 2). The additional variables collectively were signiﬁcant, with a chi-square statistic of X² (12, N = 4195) = 51.59, p = 0.000. The model indicates that participants who reported a higher number of driving errors were most likely to be involved in a work-related crash (p = 0.007). Furthermore, reporting a higher level of work pressure was also predictive of crash involvement (p = 0.010). Several additional regression models were estimated to determine the sensitivity of the results. A test of the full model with all 14 variables entered together, as well as the two models entered separately conﬁrmed the same signiﬁcant predictors (for example, exposure, errors and work pressure). The inclusion of gender, age and years driving experience did not increase the predictive value of the model. Table 15.2 Logistic regression

Variables

B

SE

Step 1 Hours per week –0.21 0.13 Km per year 0.16** 0.02 Model Chi-Square 40.61** (df = 2) Step 2 0.17 –0.32 Hours per week 0.03 0.13** Km per year 0.14 0.37* Errors 0.13 0.05 Highway Code 0.13 0.13 Aggressive 0.08 –0.10 Alcohol 0.11 0.09 Close following –0.03 0.07 Speeding 0.11 –0.04 Overtaking 0.07 0.12 Fleet safety –0.11 0.08 rules 0.07 Communication –0.18* 0.10 0.04 Work pressure –0.04 0.10 Procedures Management Model Chi-Square 92.20** (df = 13) Block Chi-Square 51.59** (df = 12)

Odds ratio Exp (B)

95% CI Lower

Upper

Wald

p

1.41 42.64

0.200 0.000

0.811 1.72

0.79 1.12

1.91 1.23

0.77 23.83 7.27 0.15 1.04 0.02 0.68 0.13 0.15 2.83 1.66 6.17 0.20 0.14

0.308 0.000 0.007 0.702 0.317 0.906 0.415 0.724 0.693 0.091 0.200 0.010 0.65 0.71

0.74 1.13 1.44 1.05 1.14 0.99 1.09 0.97 0.96 1.13 0.90 0.84 1.05 0.96

0.92 1.08 1.11 0.81 0.89 0.85 0.88 0.84 0.77 0.98 0.76 0.72 0.86 0.79

1.18 1.19 1.88 1.35 1.46 1.15 1.35 1.12 1.20 1.30 1.06 0.96 1.27 1.17

Note: * p < 0.05, **p < 0.01; CI = Conﬁdence level

Predicitng High Risk Behaviours in a Fleet Setting

183

Discussion The present study aimed to examine the utility of popular self-report driving measurement tools (for example, DAQ, DBQ and SCQ-MD) to predict self-report crash involvement among a group of Australian ﬂeet drivers. In doing so, a number of issues emerged regarding conducting research in ﬂeet settings, which will also be highlighted below. Firstly, analysis of the measurement tools’ internal consistency indicated that the DBQ and SCQ-MD were moderately robust, with the results similar to previous research that has utilised the questionnaires (Blockey and Hartley, 1995; Dobson et al., 1999; Sullman et al., 2002; Wills, 2006). However, the DAQ’s internal consistency was relatively low, and as the scale has not been extensively validated within the literature, it appears that further research is necessary to determine the psychometric properties of the questionnaire, and its subsequent usefulness within ﬂeet research. Secondly, examination of the mean scores for the DBQ, DAQ and SCQ-MD factors revealed that participants generally reported positive attitudes and behaviours towards road safety. With regards to attitudes, similar to previous research (Davey et al., 2006; Meadows, 2002), respondents reported close following and drink driving as the most serious driving behaviours. Participants also reported risky overtaking practices were an additional unacceptable behaviour, while attitudes towards speeding were more lenient. This ﬁnding is consistent with research which has indicated speeding is the most common form of aberrant driving behaviour reported by motorists (Davey et al., 2007; Dimmer and Parker, 1999; Parker et al., 1995). With regards to the relationship between the measurement tools, negative associations were identiﬁed between attitudes and the corresponding behaviours. That is, participants who agreed with the seriousness of the speciﬁed aberrant driving behaviours were less likely to report engaging in such behaviours over the past six months (for example, DBQ speeding factor). Furthermore, the bi-variate correlations also provided a preliminary indication that the culture of the organisation, in particular the direction provided by the management team, is associated with driving behaviours. For example, the collected data generally indicates that the current organisation provided clear ﬂeet safety rules, appropriate communication and support as well as strong management commitment, which was negatively associated with engaging in the aberrant driving behaviours. While only preliminary, the results indicate that the ‘safety climate’4 of a ﬂeet organisation has the potential to directly impact upon the driving outcomes exhibited by employees. Despite the positive appraisal regarding the safety climate of the organisation, 588 participants reported being involved in a work-related accident in the past 12 months. With regards to the prediction of self-reported crash involvement while driving for work purposes, a number of key factors were identiﬁed. Firstly, it appears that greater exposure to the road, such as driving more kilometres per annum, increases drivers’ chances of being involved in a crash. While not surprising, the 4 Safety climate has been deﬁned as a psychological product of the behavioural and cultural ingredients within an organisation (Wills et al., forthcoming).

184

Driver Behaviour and Training – Volume III

results may provide an opportunity for ﬂeet managers to identify those at risk of crash involvement through exposure, and ensure such drivers receive appropriate interventions and supervision to reduce the likelihood of being involved in an accident. Secondly, the logistic regression analyses indicated that making a higher number of errors as well as reporting higher levels of work pressure were both predictive of work crashes. Interestingly, these two predictive variables were also correlated at a bi-variate level, as those who reported increased work pressure were also more likely to report committing a higher number of driving errors in the past six months (r = 0.25**). Further research appears necessary to determine whether there is a causal link between work pressure and committing errors, as early evidence suggests fatigue related issues are a contributor to crash involvement (Haworth et al., 2000). Implications and difﬁculties The present study forms part of a four-year programme of research conducted at CARRS-Q that involved examining ﬂeet drivers’ attitudes and behaviours from large Australian ﬂeet companies. This research project has highlighted a number of implications and difﬁculties regarding conducting research and promoting safety within ﬂeet settings. While it is noted that the following comments can only be considered subjective, the writers believe that a number of on-going issues within ﬂeet settings need to be addressed. In relation to administering tests and generally gathering self-report data, these include: 1. The predominant driving assessment tools utilised in research, such as the DBQ, DAQ and Fleet Safety Climate Survey are not conducive for administration to large scale commercial driving environments due to their length. Fleet managers and ﬂeet drivers are not willing or not able to devote the appropriate period of time necessary to accurately complete these driving assessment tools. 2. The scales are increasingly becoming antiquated as contemporary issues that inﬂuence ﬂeet drivers’ performance have not been included in assessment scales (for example, fatigue, time pressure). 3. There is a lack of modern, easily administered and user friendly measures that can be utilised for diagnostic, evaluative or appraisal purposes that speciﬁcally measure the impact of ﬂeet interventions as well as determine the assessment of associated driving risk. Conversely, with regards to generally implementing ﬂeet safety interventions, the authors note that within Australia: 1. occupational health and safety representatives are often reluctant to address

Predicitng High Risk Behaviours in a Fleet Setting

185

ﬂeet safety concerns; 2. safety interventions usually rely heavily on engineering solutions; 3. suppliers (by default) have often inﬂuenced industry responses; and 4. there is a clear lack of current behavioural interventions to improve ﬂeet safety. Taken together, changing habitual behaviours is often a complex process which should be driven by evidence-based solutions that incorporate both ofﬁcial information (for example, crash databases) and self-reported perceptions and experiences. With regards to the latter, currently only a relatively small (but growing) body of research has examined the self-reported driving attitudes and behaviours of ﬂeet drivers, despite the potential for such data to be utilised in ﬂeet interventions designed to reduce the burden of crash involvement. Limitations Some limitations should be borne in mind when interpreting the results of this study. The response rate of participants was relatively low, but consistent with previous research that has attempted to investigate ﬂeet drivers (Davey et al., 2007; Newman et al., 2002). Concerns remain regarding the reliability of the self-reported data, such as the propensity of professional drivers to provide socially desirable responses. Further research is also required to establish the reliability and validity of the scales for the Australian setting, especially the psychometric properties of the DAQ. Finally, it is also noted that a number of additional factors not examined in the current study, both personal and environmental, may inﬂuence as well as cause a vehicle crash. Despite such limitations, the results may prove to have direct implications for ﬂeet interventions, not only through monitoring the driving performance of employees and the corresponding level of perceived work pressure, but also through proactive measures to reduce the likelihood of drivers making driving errors for example, appropriate rest breaks, training and so on. Given the tremendous personal and economic cost of vehicle crashes in Australia, further research that endeavours to identify an appropriate balance between productivity and personal safety within ﬂeet settings may prove beneﬁcial at a number of levels. References Anderson, A. and Summala, H. (2004). ‘Commuter bicyclists’ self image, attitudes, behaviour and accidents.’ Paper presented at the Third International Conference on Trafﬁc and Transport Psychology, University of Helsinki. Austroads (2006). Guide to Road Safety, Part 1: Road Safety Overview. Sydney, Australia. Austroads. Clarke, S.S. and McInnes, R. (2004). ‘Unprecedented reform: the new tort law 15.’ Insurance Law Journal, 1. Bianchi, A. and Summala, H. (2004). ‘The “genetics” of driving behaviour: parents’ driving style predicts their children’s driving style.’ Accident Analysis and

186

Driver Behaviour and Training – Volume III

Prevention, 36, 655–69. Blockey, P.N. and Hartley, L.R. (1995). ‘Aberrant driving behaviour: errors and violations.’ Ergonomics, 38, 1759–71. Burgess, C. and Webley, P. (2000). Evaluating the Effectiveness of the United Kingdom’s National Driver Improvement Scheme. School of Psychology, University of Exeter. Davey, J., Banks, T. (2005). ‘Estimating the cost of work motor vehicle incidents in Australia.’ Paper presented at Policing and Education Conference New Zealand. [CD-ROM]. Davey, J., Freeman, D. and Wishart, D. (2006). ‘A study predicting crashes among a sample of ﬂeet drivers.’ Proceedings of the Road Safety Research, Policing and Education Conference, Gold Coast, Australia. [CD-ROM]. Davey, J., Wishart, D., Freeman, J. and Watson, B. (2007). ‘An application of the Driver Behaviour Questionnaire in an Australian organisational ﬂeet setting.’ Transportation Research Part F: Trafﬁc Psychology and Behaviour, 10: 11–21. Deffenbacher, J.L., Oetting, E.R. and Lynch, R.S. (1994). ‘Development of a driving anger scale.’ Psychological Reports, 74, 83–91. Diaz, R.L. and Cabrera, D.D. (1997). ‘Safety climate and attitude as evaluation measures of organizational safety.’ Accident Analysis and Prevention, 29, 643–50. Dimmer, A.R. and Parker, D. (1999). ‘The accident, attitudes and behaviour of company car drivers.’ In G.B. (ed.), Behavioural Research in Road Safety XI. Crowthorne, Berkshire: Transport Research Laboratory. Dobson, A., Brown, W., Ball, J., Powers, J. and McFadden, M. (1999). ‘Women drivers’ behaviour, socio-demographic characteristics and accidents.’ Accident Analysis and Prevention, 31, 525–35. Gillen, M., Baltz, D., Gassle, M., Kirsch, L. and Vaccaro, D. (2002). ‘Perceived safety climate, job demands, and co-worker support among union and non-union injured construction workers.’ Journal of Safety Research, 33, 33–51. Glendon, A.I. and Litherland, D.K. (2001). ‘Safety climate factors, group differences, and safety behaviour in road construction.’ Safety Science, 39, 157–88. Glendon, A.I. and Stanton, N.A. (2000). ‘Perspectives on safety culture.’ Safety Science, 34, 193–214. Haworth, N., Tingvall, C. and Kowadlo, N. (2000). Review of best practice road safety initiatives in the corporate and/or business environment (No. 166). Clayton: Monash University Accident Research Centre. Lajunen, T., Parker, D. and Summala, H. (2003). ‘The Manchester Driver Behaviour Questionnaire: a cross-cultural study.’ Accident Analysis and Prevention, 942, 1–8. Lajunen, T. and Summala, H. (1997). ‘Effects of driving experience, personality, and driver’s skill and safety orientation on speed regulation and accidents.’ In T. Rothengatter and E. Carbonell Vaya (eds.), Trafﬁc and Transport Psychology: Theory and Application. Amsterdam: Pergamon, 283–94. Lynn, P. and Lockwood, C.R. (1998). The Accident Liability of Company Car Drivers. Report No. 317. Crowthorne, Berkshire: Transport Research Laboratory. Meadows, M. (2002). ‘Speed Awareness Training.’ Paper presented at the 67th Road

Predicitng High Risk Behaviours in a Fleet Setting

187

Safety Congress, 4–6 March, Stratford, UK. Mearns, K., Whitaker, S.M. and Flin, R. (2003). ‘Safety climate, safety management practice and safety performance in offshore environments.’ Safety Science, 41, 641–80. Murray, W., Newnam, S., Watson, B., Davey, J., and Schonfeld, C. (2003). Evaluating and Improving Fleet Safety in Australia. Report for the Australian Transport Safety Bureau: Brisbane. Newnam, S., Watson, B., Murray, W. (2002). ‘A comparison of the factors inﬂuencing work-related drivers in a work and personal vehicle.’ In Proceedings of the Road Safety Policy, Education and Policing Conference, Adelaide, Australia. Newnam, S., Watson, B. and Murray, W. (2004). ‘Factors predicting intentions to speed in a work and personal vehicle. Transportation Research Part F, 7, 287– 300. Parker, D., Reason, J.T., Manstead, A. and Stradling, S.G. (1995). ‘Driving errors, driving violations and accident involvement.’ Ergonomics, 38, 1036–48. Parker, D., Stradling, S.G. and Manstead, A. (1996). ‘Modifying beliefs and attitudes to exceeding the speed limit: An intervention study based on the theory of planned behaviour.’ Journal of Applied Social Psychology, 26, 1–19. Reason, J., Manstead, A., Stradling, S., Baxter, J. and Campbell, K. (1990). ‘Errors and violations: a real distinction?’ Ergonomics, 33, 1315–32. Stradling, S.G. (2000). ‘Driving as part of your work may damage your health.’ In G.B. Grayson (ed.), Behavioural Research in Road Safety IX. Crowthorne, Birkshire: Transport Research Laboratory, 1–9. Sullman, M.J., Meadows, M. and Pajo, K.B. (2002). ‘Aberrant driving behaviours amongst New Zealand truck drivers.’ Transportation Research Part F, 5, 217– 32. Wheatley, K. (1997). ‘An overview of issues in work-related driving.’ In ‘Staysafe 36: Drivers as workers, vehicles as workplaces: Issues in ﬂeet management.’ (Report No. 9/51). Ninth Report of the Joint Standing Committee on Road Safety of the 51st Parliament. Sydney: Parliament of New South Wales. Wills, A.R. (2006). ‘An Exploration into Safety Climate and Culture, Occupational Stress, and Job Performance.’ PhD Conﬁrmation Report. CARRS-Q: Queensland University of Technology. Wills, A.R., Biggs, H.C. and Watson, B. (2006). ‘Road safety in corporate ﬂeet settings: approaches from organisational and industrial psychology.’ Paper presented at the Australasian Road Safety Research, Education and Policing Conference, Wellington, New Zealand, 14–16 November. Wills, A.R., Watson, B. and Biggs, H.C. (in press). ‘Comparing safety climate factors as predictors of work-related driving behaviour.’ Journal of Safety Research. Wishart, D., Davey, J. and Freeman, J. (2006). ‘An application of the driver attitude questionnaire to examine driving behaviours within an Australian organisational ﬂeet setting.’ Proceedings of the Road Safety Research, Policing and Education Conference, Gold Coast, Australia [CD-ROM]. Xie, C., and Parker, D. (2002). ‘A social psychological approach to driving violations in two Chinese cities.’ Transportation Research Part F, 5, 293–308.

This page intentionally left blank

Chapter 16

Driver Celeration Behaviour in Training and Regular Driving Anders af Wåhlberg and Lennart Melin Uppsala University, Sweden Introduction Testing drivers is an important undertaking in the industrialised parts of the world for the issuing of driving licences and hiring of transport personnel. Three main methods may be identiﬁed: testing knowledge of rules by written examination, evaluating attitude or personality by questionnaire or interview and testing driver behaviour (mainly knowledge of trafﬁc rules during driving). The aim of these undertakings is to decide who is a good enough driver according to whatever standard is used, usually safety. Unfortunately, there are several problems with these methods, all of which may be summarised as very weak associations with the standard of safety. Knowledge-based tests can be passed by studying hard or cheating. Also, knowledge of what the answers are to questions about rules does not make certain that drivers behave in accordance with the rules or choose the safe alternative in an actual situation. There would also seem to exist a lack of evidence for the predictive validity of such tests. Similarly, people may tend to be ‘economical with the truth’ when they respond to questionnaires; ‘socially desirable responding’ is probably a common phenomenon in trafﬁc behaviour questionnaires (for example, Lajunen, Corry, Summala and Hartley, 1997), not to mention faulty memory (Owsley et al., 1991; Joly et al., 1993; Stulginskas, Verreault and Pless, 1985; Webb, Bowman and Sanson-Fisher, 1988; Streff and Wagenaar, 1989; Hunter, Stewart, Stutts and Rodgman, 1993; West, French, Kemp and Elander, 1993; Boyce and Geller, 2002). The results may therefore be far removed from the actual behaviour as it is acted out on the road; information sifted through fallible memories, faulty understanding of items and various biases in the construction of the scales and so forth. All these problems are of course magniﬁed if interviews are used, especially unstructured ones. As a result, the predictive validity of this type of undertaking is low (see for example reviews by Johnson, 1946; Goldstein, 1964; Peck, 1993), especially for single predictors. Neither may the behaviour during driver on-the-road testing be representative for the driver’s regular behaviour; that is, the tested driver may be ‘faking good’, because he or she knows that he or she is being tested and suppresses inappropriate behaviour. As with written knowledge tests, you can pass this if you know the rules. Also, humans are used as judges, with a resulting low inter-rater reliability and added error variance. However, a few studies have actually shown that driving instructors and similar professionals can predict the safety of subsequent driving from driving

Driver Behaviour and Training – Volume III

190

performance (West and Hall, 1995a; 1995b; Hoinville, Berthoud and Mackie, 1972), although the coefﬁcients are not exactly impressive. Actually, the predictive power of all known methods of testing for trafﬁc safety is low (see reviews by Johnson, 1946; Lester, 1991). Furthermore, the results from advanced driving courses, which may be compared to driver training for licensing, do not seem to show any safety gains (Lund and Williams, 1985). This suggests that although a driver may be able to accurately perform to the satisfaction of the instructor, this does not mean that they become a safer driver. Apparently, something is wrong with the current ways of predicting safe driving in practical settings. Given the evidence referred to above, this would seem to be due to the use of humans as judges and variables that are only remotely related to safety and the actual behaviour of humans on the road. An instrument for the prediction of safe driving behaviour should therefore use behaviour variables that are: 1. possible to quantify and measure in an objective and simple way, both on testtracks and in real trafﬁc (ecologically valid measures); and 2. not understood by the tested as important, or the behaviour studied should not be possible to conceal, due to habit or limited experience or cognitive ability. This would result in a fair correspondence between the measured behaviour during testing and in regular driving. It is therefore suggested that, as an alternative or complement to the various methods described above, driver celeration behaviour should be measured and used as predictor and/or test. This concept summarises driving as the mean of speed changes that a driver undertakes during movement in the (fairly) level plane of the road (see appendix to this chapter for a mathematical deﬁnition), but also states that it is positively associated with individual accident record (af Wåhlberg, 2006a; submitted). Empirically, celeration behaviour has, so far, been shown to be associated with accidents (af Wåhlberg, 2000; 2004; 2006d; forthcoming) for bus drivers, an association which is independent of speed (af Wåhlberg, 2006c) and increases strongly with aggregation of data (af Wåhlberg, 2007a) and to have fair psychometric properties (af Wåhlberg, 2003; 2007a; 2007b). However, similar variables have been used by other researchers (Lajunen and Summala, 1997; Lajunen et al., 1997;1 Quimby et al., 1999) as accident predictors, although with worse results. Furthermore, despite large differences in celeration behaviour found as a result of training of bus drivers, the drivers could apparently not change their driving behaviour into full accordance with the instructors’ advice. Instead, their initial driving style was still evident after instruction (af Wåhlberg, 2002). However, this does not necessarily mean that results for this variable in the training situation is a good measure of regular behaviour, as it could be argued that celeration behaviour would be easy to ‘fake good’. There are any number of driver training courses and tests available, and they may for the present purpose be regarded as fairly similar, in the sense that they create a 1

The two studies by Lajunen seem to use the same set of data.

Driver Celebration Behaviour in Training and Regular Driving

191

social situation where the pupil is under the scrutiny of an experienced driver and teacher, and thus probably tries to look good in terms of whatever is taught. One such driver training is EcoDriving (and its close relative Heavy EcoDriving), where a fuel-efﬁcient driving style is taught (see af Wåhlberg, forthcoming). This type of training is of particular methodological interest for two reasons: objective data (fuel consumption) are used as the main evaluation tool and the efﬁciency of the teaching is tested by comparing consumption on two drives over the same route, where the ﬁrst round is a test of the regular driving behaviour of the trainee so that the instructor can tailor instruction to remedy the speciﬁc faults the individual makes. Thereafter, training commences and the same route is driven again. Fuel consumption is measured for both these drives and compared. Usually, the driver burns about ten per cent less fuel on the second round, without using more time. This type of driver training therefore presents a good opportunity to study the differences and similarities between a driver’s regular behaviour and that displayed in a type of testing situation, given that data on regular behaviour can be measured. Summing up, the goal of the present study was to investigate whether driver celeration behaviour was similar (in terms of relative standing between individuals, that is, correlation) during training and regular driving, before and after training had taken place, using instruction in Heavy EcoDriving compared with regular driving en route for bus drivers, as the intervention. Method General Data were collected as part of a project evaluating the (short- and long-term) effects of training in fuel-efﬁcient driving at a bus company (af Wåhlberg, 2006b; forthcoming). One of the papers from this work analysed the changes in acceleration and deceleration behaviour and fuel consumption during the training in Heavy EcoDriving, that is, between the ﬁrst and second drive (af Wåhlberg, 2002). For the present study, these data (recalculated as longitudinal celeration behaviour) were related to celeration data from the same drivers driving under regular working conditions, unaware of being measured. Celeration behaviour measurements The principle used for celeration measurement is fairly simple: changes in speed are measured during driving and the absolute mean during movement (standstills are edited out) for each driver is calculated (see for example af Wåhlberg 2006c). The basic theory behind celeration behaviour uses all changes (lateral and longitudinal) in speed for the calculation of the driver’s mean behaviour (af Wåhlberg, 2006a; submitted). However, longitudinal celeration behaviour has been found to contribute almost all the variance to the resultant (af Wåhlberg, 2004). Therefore, only longitudinal acceleration was used in the present study (see the formula in the

192

Driver Behaviour and Training – Volume III

appendix to this chapter). This also simpliﬁed technical matters, as it was possible to use the speedometer signal as the source of data; see further in the next section. Technical equipment The equipment used for measurement was a vehicle computer that collected the data from the signal to the speedometer. This signal is an electric pulse that is emitted every time a magnet passes another magnet in the rear axle, yielding 6900 pulses per kilometre for the model of the bus used. With the help of an internal clock in the computer, speed and speed changes were calculated (with 10 and 2.5 Hz, respectively, see the appendix to this chapter), storing these data in a separate memory. Both units were hidden under the dashboard, and all measurements unknown to the drivers. Five buses were thus equipped and running in regular trafﬁc on one of the busiest routes in Uppsala, Sweden. One of these was also used during some of the training sessions, yielding the presently used data. On the other hand, during training it was quite obvious that fuel consumption was measured (by a different system), as there was a display/interface on the dashboard showing continuous consumption. Subjects A small group of drivers used one of the celeration measurement equipment buses during training, and could thus be measured. Selection was fairly random, in the sense that no speciﬁc principle was used for allotting the drivers to training for any day. Instead, this was due to the drivers being available, either during their free time, or when someone else could replace them during regular working hours. The mean values on some available basic variables for the sample used in the present study and the population of drivers at the company are shown in Table 16.1. It can be seen that the samples are fairly similar to the population; although an unrepresentative sample would probably only mean a restriction of variation in celeration behaviour and an under-estimation of the true associations. The data for regular driving were gathered continuously over several years, but were ordered into ‘samples’ for analysis, numbered consecutively over time. The samples and data used in the present paper are the same as in other papers (for example, af Wåhlberg, 2003; 2006c) and there is therefore a discontinuity in the numbering; that is, the present data on regular driving were gathered more than six months after the start of these measurements. However, the number of subjects in each analysis is restricted by how many are present in both measurements when different ones are compared. Procedure Drivers were required to drive along a pre-determined route2 for the training in Heavy EcoDriving. (First run, A) with the instruction to ‘drive as usual’ (en route 2 The route was about 11 km. It could differ somewhat between drivers, but not between drives.

Driver Celebration Behaviour in Training and Regular Driving

Table 16.1

The percentage of men and people with Swedish names and mean age and number of hours worked in 2001, for drivers in the present study sample and for the total number of active drivers at Gamla Uppsalabuss as at 30 December 2001 N

Sample Population

193

29 414

Sex 93.1% 89.4%

Ethnicity 51.7% 57.0%

Age 50.2 45.4

Hours worked 1373.1 1214.9

as a bus driver). During this drive, the instructor would assess the behaviour of the driver and target some wasteful habits for training, which proceeded thereafter. When the driver had been instructed and had tried out the new behaviours, the route was driven again and the fuel consumption and driving times were compared between runs. Usually, changes in fuel consumption of about 15 per cent were found (af Wåhlberg, 2002). For the regular driving (with passengers), the same bus route was used for all subjects, with the distance used for measurement being a multiple of the total from terminus to terminus, about 12 km (most drivers were measured by driving this stretch back and forth repeatedly). Results In Table 16.2 the means and standard deviations of the celeration variable in the two training sessions are presented, while in Table 16.3 the Ns and signiﬁcances for the comparisons between each sample can be seen. Note that the means used for calculations are not for all drivers (shown in Table 16.2), but most often the largest possible sub-part. Thereafter, these variables were correlated (see Table 16.4). It can be seen from the means in Table 16.2 that drivers tended to drive in a less forceful manner during the ﬁrst training drive as compared to their regular driving (although this might Table 16.2

The means and standard deviations of the celeration variables in m/s2 during the ﬁrst (A), and second (B) runs during the training sessions, and the mean for regular driving along a route for (some of) the same drivers Variable A, N = 27 B, N = 23 Regular 1, N = 19 Regular 5, N = 27 Regular 6, N = 25

Celeration behaviour 0.463/0.068 0.494/0.046 0.552/0.085 0.548/0.074 0.537/0.066

in part be due to the driving environment being different). However, despite this difference, the celeration values of drive A did correlate with regular driving. The

Driver Behaviour and Training – Volume III

194

Table 16.3

The N and t-values for the differences (dependent t-tests) between different measurements of celeration behaviour. Numbers 5 and 6 were gathered after training (A and B)

Measurement B Regular 1 Statistic N t N t First run (A) 23 –1.483 19 –4.658*** Second run (B) 16 –2.364* Regular 1 Regular 5 * p < 0.05, ** p < 0.01, *** p < 0.001

Regular 5 N t 27 –5.741*** 23 –2.909** 19 –0.579

Regular 6 N t 25 –4.342*** 21 –2.580* 15 0.475 21 0.870

second drive, however, did not correlate with regular driving, and only weakly and non-signiﬁcantly with the ﬁrst drive (0.33, p > 0.10, N = 23). Note that some of the regular driving (2 and 3) was actually undertaken after training, but still correlated with the ﬁrst drive. This is in agreement with the analysis of the long-term stability of celeration behaviour; the Heavy EcoDriving training does not seem to have had any long-term impact (af Wåhlberg, forthcoming). The accident records for the bus drivers were also available. No signiﬁcant correlations were found between incidents for a two-year period and the celeration variable. Discussion The present data lend support to the hypothesis that driver celeration behaviour is habitual and hard to suppress, even in a situation where this would be expected to happen. Also, even when the absolute levels of celeration behaviour do change Table 16.4

The Pearson correlations between driver celeration behaviour for the ﬁrst and second run during training on one hand, and for regular driving on a number of occasions; 1 before training, 5 and 6 after

Celeration behaviour First run (A) Second run (B) * p < 0.05

Regular 1, N = 16 0.52* 0.11

Regular 5, N = 23 0.47* –0.06

Regular 6, N = 21 0.24 0.16

substantially, as shown by the t-tests, the individual relative level tends to remain the same, as shown by the correlations. In other words, there was a fairly constant displacement of behaviour. This ﬁnding might be interpreted as the lingering of a habitual driving style despite the strong incentive and possibility of openness to change. The non-signiﬁcant associations between the second drive and all other measurements was probably due in part to the restriction of range (the standard deviation was lower), and an expected result of training.

Driver Celebration Behaviour in Training and Regular Driving

195

There are several consequences for these ﬁndings; ﬁrst it may be noted that driver celeration behaviour has again been found to be a semi-stable individual differences variable, this time over situations, which lends further credibility to it as a psychological measure of driver behaviour. Furthermore, it would even seem viable to use this measure for testing and screening purposes. This statement could of course be disputed on the grounds that the correlations found here, and the stability over time in regular driving (af Wåhlberg, 2003), are not that impressive. However, this can be counteracted by larger amounts of data; aggregation over measurements will stabilise the individual driver value and yield better predictions (af Wåhlberg, 2007a). This could be achieved by constant monitoring of the drivers’ behaviour (this is of course only viable in professional populations), a direction taken in Japan (Ministry of Transport, 1998), making the validity actually close to 1. The advantages of such surveillance would also include fuel savings, as deceleration values are related to fuel consumption (af Wåhlberg, 2002). The technical systems for this type of monitoring are available, and are becoming common in Europe, although they are not used in the suggested way. It was evident that the drivers had very much lower celeration values during training, as compared to regular driving. There are three possible reasons for this, all of which may have contributed: a different road layout (Ericsson, 2000), fewer mandatory stops and intentional suppressing of regular behaviour. However, the cause of these differences is not as important in the present situation as the ﬁnding that despite the changes there remained some predictability to the drivers’ behaviour before instruction had commenced. However, a few further limitations to these conclusions should also be pointed out. For example, it would seem probable that the driving style of inexperienced drivers would not be as habitual (automatic) as that of the drivers tested here. Also, the sample studied in the present paper was small, although fairly representative for the population on the available demographic variables. However, even if this is not so, so far no indications have been found that this would in any way threaten the generalisability of the results to the rest of the drivers employed by the bus company and similar drivers. The logic behind this statement is that celeration behaviour would seem to be only weakly correlated with variables like sex, age and experience (af Wåhlberg, 2000; 2003; 2004). Also, the present study concerns similarities in behaviour over environments, and no reasons are apparent that one group of bus drivers would be more or less inclined to behave in a stable way than any other, although this research is still in its infancy, and important discoveries of its properties may still be made. Still, the present study suffers methodologically from the very small sample, and replications of greater sample size are necessary. However, there is also a need to enlarge the measurement per person, not on one occasion, but during several sessions, as there seems to be a lot of intra-person variability between days (af Wåhlberg, 2003). This restriction is important; it is not claimed here that it is possible to achieve good predictive validity (of accidents) from a single measurement of celeration behaviour. Another restriction is the unknown generalisability of the results between driver groups. So far, all research that has tested predictions from the driver celeration

Driver Behaviour and Training – Volume III

196

behaviour theory has used bus drivers in a city environment as subjects, and although the theory is proposed to be valid for all types of driving, this has not been tested. The only available research that is closely related did use car drivers as subjects, but the results were not clear (Lajunen and Summala, 1997; Lajunen et al., 1997;3 Quimby et al., 1999). All results and conclusions in this paper should be seen as provisional. The discussion and arguments, on the other hand, would seem to be very valid; there is a great need for an objective instrument to measure driver behaviour. The practical uses of such an instrument would be manifold; as a feedback instrument during driver training, as selection tool for companies and departments, and as a surveillance package for the monitoring of daily driver behaviour. During driver training, an apparatus, which could give exact and objective feedback, could enhance the instruction given, and facilitate comparisons with other drivers. It may be noted in this context that the association of the celeration measure with trafﬁc density is rather low (af Wåhlberg, 2003; 2007b). The uses as a selection tool would therefore be fairly straightforward; a prospective employee or licence holder would drive a pre-determined route (or rather several different routes on different days), and only those below a certain cut-off value would be accepted. Finally, by monitoring daily behaviour of drivers in transport ﬂeets, any sign of dangerous behaviour could be instantly countered with feedback, training and, in the most severe cases, different work assignments. It is therefore concluded from this study, and the others referred to, that celeration behaviour has some prospect of becoming very useful for road safety when it comes to deterring the dangerous driver. Acknowledgements The present study was part of a project ﬁnanced by the Swedish Road Administration (grant AL90AB 2001:7758) and Gamla Uppsalabuss. Hanna Skagerström carried out all the basic data transfers and calculations and administration, while Johan Göthe (Drivec AB) constructed the technical equipment and the mathematical formulas. As usual, the personnel at Gamla Uppsalabuss assisted in various ways. Thanks to Mats Hammarström at Uppsalabuss for supplying data. The assistance of Axelssons Traﬁkskolor (Axelsson’s Driving School) in the gathering of training data is gratefully acknowledged. References Boyce, T. and Geller, E. (2002). ‘An instrumented vehicle assessment of problem behaviour and driving style: do younger males really take more risks?’ Accident Analysis and Prevention 34, 51–64. Ericsson, E. (2000). ‘Variability in urban driving patterns.’ Transportation Research

3

The two studies by Lajunen seem to use the same set of data.

Driver Celebration Behaviour in Training and Regular Driving

197

Part D, 5, 337–54. Hoinville, G., Berthoud, R. and Mackie, A. (1972). A Study of Accident Rates Among Motorists who Passed or Failed an Advanced Driving Test. TRRL Report 499. Crowthorne: Transport and Road Research Laboratory. Hunter, W., Stewart, J., Stutts, J. and Rodgman, E. (1993). ‘Observed and selfreported seat belt wearing as related to prior trafﬁc accidents and convictions.’ Accident Analysis and Prevention, 25, 545–54. Johnson, H. (1946). ‘The detection and treatment of accident-prone drivers.’ Psychological Bulletin, 43, 489–532. Joly, P., Joly, M.-F., Desjardins, D., Messier, S., Maag, U., Ghadirian, P. and LabergeNadeau, C. (1993). ‘Exposure for different licence categories through a phone survey: validity and feasibility studies.’ Accident Analysis and Prevention, 25, 529–36. Lajunen, T., Corry, A., Summala, H. and Hartley, L. (1997). ‘Impression management and self-deception in trafﬁc behaviour inventories.’ Personality and Individual Differences, 22, 341–53. Lajunen, T., Karola, J. and Summala, H. (1997). ‘Speed and acceleration as measures of driving style in young male drivers.’ Perceptual and Motor Skills, 85, 3–16. Lajunen, T. and Summala, H. (1997). ‘Effects of driving experience, personality, driver’s skill and safety orientation on speed regulation and accidents.’ In T. Rothengatter and E. Carbonell Vaya (eds). Trafﬁc and Transport Psychology, Theory and Application, Amsterdam: Pergamon, 283–94. Lund, A. and Williams, A. (1985). ‘A review of the literature evaluating the defensive driving course.’ Accident Analysis and Prevention, 17, 449–60. Ministry of Transport (1998). ‘MOT Report: Trafﬁc accident prevention plan for commercial vehicle using digital tachograph.’ JSAE Review, 19, 87–8. Owsley, C., Ball, K., Sloane, M., Roenker, D. and Bruni, J. (1991). ‘Visual/cognitive correlates of vehicle accidents in older drivers.’ Psychology and Aging, 6, 403–15. Peck, R. (1993). ‘The identiﬁcation of multiple accident correlates in high risk drivers with speciﬁc emphasis on the role of age, experience and prior trafﬁc violation frequency.’ Alcohol, Drugs and Driving, 9, 145–66. Quimby, A., Maycock, G., Palmer, C. and Grayson, G. (1999). Drivers’ Speed Choice: An In-Depth Study. TRL Report 326. Crowthorne: Transport Research Laboratory. Streff, F. and Wagenaar, A. (1989). ‘Are there really shortcuts? Estimating seat belt use with self-report measures.’ Accident Analysis and Prevention, 21, 509–516. Stulginskas, J., Verreault, R. and Pless, I. (1985). ‘A comparison of observed and reported restraint use by children and adults.’ Accident Analysis and Prevention, 17, 381–6. Webb, G., Bowman, J. and Sanson-Fisher, R. (1988). ‘Studies of child safety restraint use in motor vehicles: some methodological considerations.’ Accident Analysis and Prevention, 20, 109–15. West, R., French, D., Kemp, R. and Elander, J. (1993). ‘Direct observation of driving, self reports of driver behaviour and accident involvement.’ Ergonomics, 36, 557–67. West, R. and Hall, J. (1995a). ‘Predicting accident risk in novice drivers.’ In G.

198

Driver Behaviour and Training – Volume III

B. Grayson (ed.), Behavioural Research in Road Safety V, Transport Research Laboratory, Crowthorne, 97–105. West, R. and Hall, J. (1995b). Accident Liability of Novice Drivers. TRL Report 295. Crowthorne: Transport Research Laboratory. af Wåhlberg, A. (2000). ‘The relation of acceleration force to trafﬁc accident frequency: a pilot study.’ Transportation Research Part F: Trafﬁc Psychology and Behaviour, 3, 29–38. af Wåhlberg, A. (2002). ‘Fuel efﬁcient driving training – state of the art and quantiﬁcation of effects.’ E141 Proceedings of Soric’02. <www.psyk.uu.se/ hemsidor/busdriver/index.htm>. af Wåhlberg, A. (2003). ‘Stability and correlates of driver acceleration behaviour.’ In L. Dorn (ed.), Driver Behaviour and Training. First International Conference on Driver Behaviour and Training. Stratford-upon-Avon, 11–12 November, 2003, 45–54. af Wåhlberg, A. (2004). ‘The stability of driver acceleration behaviour and a replication of its relation to bus accidents.’ Accident Analysis and Prevention, 36, 83–92. af Wåhlberg, A. (2006a). The Prediction of Trafﬁc Accident Involvement From Driving Behaviour. Doctoral thesis, Uppsala University, Sweden. af Wåhlberg, A. (2006b). ‘Short-term effects of training in economical driving; passenger comfort and driver acceleration behaviour.’ International Journal of Industrial Ergonomics 36, 151–63. af Wåhlberg, A. (2006c). ‘Speed choice versus celeration behaviour as trafﬁc accident predictor.’ Journal of Safety Research, 37, 43–51. af Wåhlberg, A. (2006d), ‘Driver celeration behaviour and the prediction of trafﬁc accidents.’ International Journal of Occupational Safety and Ergonomics, 12, 281–96. af Wåhlberg, A. (2007a), ‘Aggregation of driver celeration behaviour data: effects on stability and accident prediction.’ Safety Science, 45, 487–500. af Wåhlberg, A. (2007b), ‘Effects of passengers on bus driver celeration behaviour and incident prediction.’ Journal of Safety Research, 38, 9–15. af Wåhlberg, A. (forthcoming). ‘Long term effects of training in economical driving: fuel consumption, accidents, driver acceleration behaviour and technical feedback.’ International Journal of Industrial Ergonomics. af Wåhlberg, A. (submitted). ‘Driver celeration behaviour and accidents – an analysis.’ www.psyk.uu.se/hemsidor/busdriver.

Driver Celebration Behaviour in Training and Regular Driving

199

Appendix: Measurement method The speed signal is tapped from the vehicle’s speedometer with a frequency of 10Hz, an interval of 100 ms. During this interval, the number of complete puls-cycles is counted, and speed and acceleration are calculated with the formulas shown in Figure 16.1. Example

Puls 0

Cycletime T0

Puls 1

Cycletime T1

Puls 2

Cycletime T2

Measurement interval 100ms

Figure 16.1 Measurement method example Example t = T0 + T1 + T2 n=3 Speed calculation formula v = ((n / t)/w)*1000 where v = speed (m/s) n = number of pulses from the speedometer t = time for n pulses to accumulate (seconds) w = pulses per kilometre Acceleration calculation formula a(4n) = ((v(4n) + v(4n+1) + v(4n+2) + v(4n+3)) / 4 – (v(4n) + v(4n–1) + v(4n–2) + v(4n–3))/4) / 2.5 where: a = acceleration (m/s2) v = speed (m/s) n = number of acceleration measurement points

This page intentionally left blank

Chapter 17

A Study of Contemporary Modiﬁcations to the Manchester Driver Behaviour Questionnaire for Organisational Fleet Settings James Freeman, Jeremy Davey and Darren Wishart Centre for Accident Research and Road Safety, Queensland (CARRS-Q), Australia Introduction DBQ and the present driving context The Manchester Driver Behaviour Questionnaire (DBQ) is increasingly becoming one of the most prominent measurement scales to examine self-reported driving behaviours (Lajunen and Summala, 2003). For example, the DBQ has been extensively utilised in a range of driver safety research areas, such as: age differences in driving behaviour (Dobson et al., 1999), the genetics of driving behaviour (Bianchi and Summala, 2004), cross cultural studies (Lajunen et al., 2003), as well as factors contributing to accident involvement (Dobson et al., 1999; Parker et al., 1995b) and demerit point loss (Davey et al., 2007). Furthermore, the versatility of the DBQ has also been demonstrated via the utilisation of the instrument in a number of countries, including China (Xie and Parker, 200), Australia (Davey et al., 2006; Dobson et al., 1999; Newnam, Watson and Murray, 2004), New Zealand (Sullman, Meadows and Pajo, 2002), Finland (Bianchi and Summala, 2004) and the United Kingdom (Parker et al., 1995a; Parker et al., 2000). The popularity of the DBQ to assess current driving performance is also reﬂected in the considerable evolution of the scale since its inception. The original DBQ was developed by Reason et al. (1990) and focused on two distinct driving behaviours that were identiﬁed as errors and violations. Errors were believed to consist of actions that are not planned (for example, mistakes and misjudgements), while violations were considered to be deliberate deviations from safe driving behaviours (for example, speeding). However, an additional factor referred to as ‘slips and lapses’ was also developed that focused on attention and memory failures, which were traditionally not considered to affect overall road safety. Speciﬁcally, such behaviours were associated with attention and memory problems, while errors include more serious mistakes such as failures of observation and misjudgement (Lajunen and Summala, 2003). The original DBQ scale has undergone further modiﬁcation by Lawton et al. (1997), incorporating additional items to assess other factors that have been proposed

202

Driver Behaviour and Training – Volume III

to contribute to driving violations. For example, aggressive violation items have been included in the questionnaire and focus on an interpersonal aggressive component, such as showing or exhibiting frustration. However, ‘ordinary’ violations remained within the scale and consist of aberrant driving behaviours that do not have an aggressive aim, for example, speeding behaviours. Taken together, currently the scale distinguishes between two forms of violations: Highway Code violations (for example, speeding and running red lights) and interpersonal aggressive violations (for example, chasing another motorist when angry and sounding one’s horn). A closer examination of the deﬁnitions reveal that Highway Code violations focus on gaining an advantage such as speeding and engaging in risky overtaking manoeuvres, while aggressive violations are more hostile in nature and are usually directed towards other motorists. In addition to the considerable level of modiﬁcation of DBQ items, there has been a high level of variation within the literature regarding the number of factors identiﬁed from using the DBQ. Firstly, some earlier research conﬁrmed the original three factors of errors, violations and lapses (Adberg and Rimmo, 1998; Blockey and Hartley, 1995; Parker et al., 1995a). For example, Aberg and Rimmo (1998) identiﬁed inattention and inexperience error factors from a large group of Swedish drivers, but overall found the same factor structure. In contrast, there has been evidence of four factors reported by Sullman et al. (2002) that focused on errors, lapses, aggressive violations and ordinary violations. Similarly, Lajunen et al. (2003) identiﬁed four factors with a group of UK drivers, and Mesken et al. (2002) reported four factors (errors, lapses, speeding and interpersonal violations) when examining the driving behaviours of Finnish motorists. In addition to the different number of factors identiﬁed, research has generally reported differences in factor structure, as speciﬁc items often load on different factors depending on the driving context (Davey et al., 2006), which ultimately inﬂuences the naming and interpretation of each factor. Despite such variability, previous applied research has demonstrated that the DBQ is robust to minor changes to some items that have been made to suit speciﬁc cultural and environmental contexts (Blockey and Hartley, 1995; Davey et al., 2007; Ozkan and Lajunen, 2005; Parker et al., 2000). As highlighted above, the DBQ has been utilised in a number of motorised countries and has thus been translated and modiﬁed to tailor a vast array of driving situations. Professional drivers and ﬂeet safety Despite the tremendous amount of research that has utilised the DBQ to investigate general motorists’ driving behaviours, there is currently only a small (but expanding) body of knowledge regarding the self-reported driving behaviours of those who drive on public roads for professional reasons (Davey et al., 2006; Newnam et al., 2002; Newnam et al., 2004; Sullman et al., 2002; Xie and Parker, 2002). That is, relatively little research has endeavoured to examine the driving behaviours of those who drive company sponsored vehicles and/or spend long periods of time behind the wheel (Davey et al., 2006; Newnam et al., 2004), despite this group being at a greater level of risk to accident involvement (Newnam et al., 2002; Sullman et al., 2002), either through increased exposure to the road or as a result of time pressures

A Study of Contemporary Modiﬁcations

203

and other distractions (Stradling et al., 2000). This lack of research is of particular concern as early estimates suggest work related road incidents cost approximately AUD$1.5 billion (Wheatley, 1997), with the hidden costs somewhere between 3 and 36 times vehicle repair/replacement costs (Murray et al., 2003). Similar to the above, the small amount of ﬂeet-based research that has utilised the DBQ has also reported a high level of factor structure variability for the measurement tool. For example, Xie and Parker (2002) examined the driving behaviours of professional drivers and identiﬁed four factors for example, violations, lapses and errors. In contrast, Sullman et al. (2002) utilised the DBQ to examine factors associated with crash involvement and reported four factors, while Dimmer and Parker (1999) focused on company car drivers and reported a six factor DBQ structure. One of the few Australian studies by Davey et al. (2006) utilised the DBQ to examine the behaviours of a group of ﬂeet drivers and reported a traditional three factor solution of errors, aggressive and speeding violations, although it is noted that a greater number of traditional items considered to be speeding violations actually loaded on the aggressive violation factor. That is, the aggressive violations factor consisted of a mixture of emotion-oriented responses to driving situations and traditional Highway Code violations. Contemporary DBQ modiﬁcations When considering that the work vehicle may be increasingly becoming an extension of the ofﬁce (for example, taking phone calls), the process of multitasking while driving and time pressures placed on drivers may yet prove to have a direct impact on driving performance (Davey et al., 2006). As a result, there appears to be an opportunity to identify additional contemporary factors that may inﬂuence professional driving performance, such as fatigue, time pressure and multitasking (for example, driving and eating and/or mobile phone use) and determine what impact, if any, such issues have on driving performance. Therefore, the present research aims to extend the traditional DBQ by including contemporary items to the traditional 20 item measurement tool and investigate the utility of the additional items in predicting aberrant driving behaviours. More speciﬁcally, the study endeavoured to: • •

examine the factor structure of the DBQ after inclusion of contemporary ﬂeet driving items; and investigate the relationship the modiﬁed DBQ has with self-reported crash involvement and trafﬁc offences.

Method Participants A total of 443 individuals, who were all employees of a large insurance company in Australia, volunteered to participate in the study. There were 345 (78 per cent) males and 98 (22 per cent) females. The average age of the sample was 44 years (range

204

Driver Behaviour and Training – Volume III

18–68 years). Participants were located throughout Australia in both urban and rural areas. The largest proportion of vehicles driven by participants were reported to be for tool of trade (56 per cent), although vehicles were also salary sacriﬁced (43 per cent), and a small proportion were leased or the participant’s own vehicle (1 per cent). Vehicles were reported to be sedans (85 per cent), four wheel drives (12 per cent) or other (3 per cent). The majority of driving by participants was reported to be within the city (46 per cent), or in the city and on country roads (40 per cent). On average participants had held their licence for 26 years (range 5–48 years), and had been driving a work vehicle for approximately ﬁve years (range 1–33 years), with the largest proportion driving between 11 and 20 hours per week (43 per cent), and between 30 000–40 000 km per year. Materials Driver Behaviour Questionnaire (DBQ) A modiﬁed version of the DBQ, consisting of 35 items was used. Similar to previous research, questions relating to lapses were omitted due to earlier evidence that has indicated this factor is not associated with crash involvement (Lawnton et al., 1997; Stradling, personal communication, 2003). In addition to the traditional 20 items incorporated with the DBQ, the authors of the current paper also included another 15 items that focused speciﬁcally on contemporary ﬂeet safety issues such as fatigue, tiredness and multitasking. These items were derived from the implementation of focus groups with ﬂeet drivers from a number of large Australian ﬂeet organisations which facilitated the identiﬁcation of key themes proposed to inﬂuence driving performance such as fatigue, tiredness, multitasking and general distraction. Some of the added items were ‘drive while under time pressure’, ‘eat a meal while driving for work’ and ‘drive while using a mobile phone’. Respondents were required to indicate on a six point scale (0 = ‘never’ to 5 = ‘nearly all the time’) how often they commit each of the errors (eight items), Highway Code violations (eight items), aggressive items (four items), as well as the additional 15 items. The complete list of the additional questions are: 1. Drive while tired 2. Have difﬁculty driving because of tiredness or fatigue 3. Find yourself nodding off while driving for work 4. Drive while under time pressure 5. Find yourself driving on autopilot 6. Save time during the day by driving quicker between jobs 7. Eat a meal while driving for work 8. Find your attention being distracted from the road 9. Drive home from work after a long day 10. Not wear your seatbelt

A Study of Contemporary Modiﬁcations

205

11. Remove your seatbelt for some reason while driving 12. Do paperwork or other administration work while driving 13. Drive while using a ‘hand-held’ mobile phone 14. Drive while using a ‘hands-free’ mobile phone 15. Do paperwork or admin while driving Demographic measures A number of socio-demographic questions were included in the questionnaire to determine participants’ age, gender, driving history (for example, years experience, number of trafﬁc offences and crashes) and their weekly driving exposure (for example, type of car driven, driving hours). Procedure The vehicle insurance company provided a list of individuals who expressed interest in participating in the research. A letter of introduction, the study questionnaire and a reply paid envelope was distributed through the company’s internal mail system to the participants. In total 1440 were mailed to ﬂeet drivers and 443 were returned, which indicates a 30 per cent response rate. Results Factor structure and reliability of the Driver Behaviour Questionnaire for an Australian sample The internal consistency of the original DBQ factor scores was examined through calculating Cronbach’s alpha reliability coefﬁcients, which are presented in Table 17.1. Similar to previous Australian research (Blockey and Hartley, 1995; Dobson et al., 1999) and research on professional drivers (Sullman et al., 2002), the factors appear to exhibit relative internal consistency. Examination of the scores reveals that the items traditionally associated with Highway Code violations indicate the highest reliability coefﬁcients (0.80) while aggressive violations, which consisted of only four items, had the lowest reliability (0.60). Table 17.1 Alpha reliability coefﬁcients of the DBQ scale

Errors (8 items) Highway Code violations (8 items) Aggressive violations (4 items)

Current

Sullman et al.

0.77 0.80 0.60

0.71 0.62 0.57

Sample (2002)

206

Driver Behaviour and Training – Volume III

Self-reported frequent driving behaviours Table 17.2 reports the overall mean scores for the three factors, revealing that participants reported a similar frequency for each of the driving categories, although further analyses indicated Highway Code violations occurred signiﬁcantly more frequently than errors F(1, 443) = 80.73, p < 0.01 as well as aggressive violations F(1, 433) = 94.42, p < 0.01. The means are higher than previous research that has focused on college students (Bianchi and Summala, 2004) elderly drivers (Parker et al., 2000) and professional drivers (Sullman et al., 2002; Xie and Parker, 2002), indicating that the current sample engaged in, or at least reported, a higher level of aberrant driving behaviours.1 In addition, Table 17.2 reports the mean and standard deviation scores for the four highest ranked items from the complete 35 item questionnaire, which were: ‘drive while under time pressure’ (M = 2.79, SD = 1.20); ‘drive while tired’ (M = 2.68, SD = 1.10); ‘exceed the speed limit on a highway’ (M = 2.62, SD = 0.93) and ‘ﬁnd your attention being distracted from the road’ (M = 2.26, SD = 0.83). The results indicate that while speeding remains one of the most common forms of aberrant behaviour reported by the ﬂeet drivers (Newnam et al., 2004; Sullman et al., 2002), drivers are also at risk of driving while fatigued, tired or while distracted. Table 17.2 Mean scores for the DBQ factors

Errors (8 items) Highway Code violations (8 items) Aggressive violations (4 items) Highest Ranked Items 1. Drive while under time pressure 2. Drive while tired 3. Exceed the speed limit on a highway 4. Find your attention being distracted from the road

M 1.61 1.70 1.53

SD 0.37 0.58 0.48

2.79 2.68 2.62 2.26

1.20 1.10 0.93 0.83

Factor analysis was administered on the complete 35 item questionnaire. Principal components analysis with oblique rotation was implemented to determine the factor structure of the DBQ, which revealed a three factor solution that accounted for 49 per cent of the total variance. The ﬁrst factor accounted for approximately 31 per cent of the total variance and contained nine items relating to a combination of Highway Code violations and some aggressive violations. Firstly, seven Highway Code violation items loaded on the factor, with the ﬁrst ﬁve speeding items identiﬁed 1 However, it is noted that the DBQ questionnaire utilised in the current study most likely varies slightly on the wording of some items compared to previous DBQ research, which should be borne in mind when making comparisons with previous research.

A Study of Contemporary Modiﬁcations

207

as the strongest contributors to the factor (for example, race away from trafﬁc lights, drive especially close, speed on residential road and so on). Secondly, similar to previous research (Davey et al., 2006), some aggressive items also loaded on the speeding factor. However, this factor was labelled Highway Code violations as the predominant theme to collectively emerge from the items focuses on speeding behaviours. The second factor accounted for ten per cent of the total variance and contained nine of the new items that centred on a combination of fatigue and distraction issues, such as driving while tired, nodding off while driving and driving on autopilot. It is noted that some additional items that may be perceived as being associated with multitasking and time pressure were also evident within this factor, such as eating a meal while driving and saving time during the day by driving quicker between jobs. While it was originally anticipated that ‘fatigue’ and ‘multi-tasking’ items would be represented in distinct factors, this did not occur in the current sample of drivers. Rather at this stage, the factor was labelled ‘fatigue’ as the largest proportion of items relate to symptoms that result from this experience, although it is noted that closer analytic scrutiny could produce a different interpretation. Finally, the third factor accounted for approximately eight per cent of the overall variance and comprised of 11 items, the majority associated with traditional error items; for example, miss a stop or give way sign, nearly hit a car and skid while breaking. However it is also noted that one traditional aggressive item also loaded on the factor (for example, become angered by another driver and give chase) and three new items, two of which focused on non-seat belt wearing and one on multitasking. All items and factors for the modiﬁed DBQ are reported in Table 17.3. Reliability and intercorrelations of the modiﬁed Driver Behaviour Questionnaire The internal consistency of the modiﬁed DBQ scale scores were examined through calculating Cronbach’s alpha reliability coefﬁcients. The resulting analysis indicated internal consistency of: 0.82 = speeding/aggression, 0.76 = fatigue, 0.69 = errors. In addition, bivariate analysis indicated that the strongest relationship was between speeding and errors (r = 0.58**), followed by fatigue and errors (r = 0.52**) and then fatigue and speeding (r = 0.50**). Interestingly, there was only a moderate bivariate relationship between fatigue and hours driven per week (r = 0.20**) or number of kilometres driver per year (r = 0.24**). Prediction of offences The ﬁnal part of the study aimed to examine the utility of the modiﬁed Driver Behaviour Questionnaire to predict self-reported work crashes as well as demerit point loss. Due to the relatively small number of participants who reported a workrelated crash in the last 12 months (N = 48), it was not possible to implement regression analyses and thus the following analyses focus on predicting work-related driving infringements (N = 73). A logistic regression analysis was performed to examine the contributions of the three factors (for example, speeding/aggression, fatigue/tiredness and errors) as well as driving exposure (for example, kilometres

208

Driver Behaviour and Training – Volume III

Table 17.3 Factor structure of the modiﬁed DBQ Description

F1

Race away from trafﬁc lights to beat car beside you Drive especially close to the car in front to signal drive faster Disregard speed limit on a residential road Disregard speed limit on a highway or freeway Stay in a closing lane and force your way into another Become angered by another driver and show anger Cross junction knowing the trafﬁc lights have already turned Become impatient by slow driver and overtake on inside Sound your horn to indicate your annoyance at another driver

0.73 0.64 0.62 0.62 0.57 0.55 0.54 0.52 0.41

Drive while tired Have difﬁculty driving because of tiredness or fatigue Find yourself nodding off while driving for work Drive while under time pressure Find yourself driving on autopilot Save time during the day by driving quicker between jobs Eat a meal while driving for work Find your attention being distracted from the road Drive home from work after a long day Nearly hit a cyclist while turning Not wear your seatbelt Remove your seatbelt for some reason while driving Become angered by another driver and give chase Skid while breaking or cornering on a slippery road When overtaking, underestimate speed of oncoming vehicle Attempt to overtake someone you had not noticed turning Miss stop or give way sign Pull out of junction so far that you disrupt trafﬁc Nearly hit another car while queuing to enter a main road Do paperwork or other administration work while driving Fail to notice pedestrian crossing in path

0.41

F2

F3

0.79 0.79 0.66 0.64 0.63 0.61 0.54 0.54 0.44 0.63 0.63 0.60 0.58 0.55 0.52 0.52 0.49 0.48 0.45 0.41 0.40

Note: Five questions did not load. They were: (1) Have one or two alcoholic drinks before driving for work; (2) Fail to check rear-view mirror; (3) Drive while using a ‘hands-free’ mobile phone; (4) Drive while using a ‘hand-held’ phone and (5) Do paperwork or admin while driving.

A Study of Contemporary Modiﬁcations

209

driven each year and hours driving per week) to the prediction of self-reported infringements in the past 12 months. Table 17.4 depicts the variables in each model, the regression coefﬁcients, as well as the Wald and odds ratio values. Self-reported number of kilometres driven each year and hours of driving per week were entered in the ﬁrst step to examine, as well as control for, the inﬂuence of driving exposure before the inclusion of the DBQ factors. As expected, the number of kilometres driven per year was predictive of incurring demerit point loss (p = 0.000) as those who drive longer distances are at a greater risk. Table 17.4 Logistic regression Variables

B

SE

Wald

p

Odds ratio Exp (B)

95% CI Lower

Upper

Step 1 Hours per week –0.148 0.18 Km per year 0.36** 0.10 Model Chi-Square 16.65** (df = 2)

0.58 14.18

0.447 0.000

0.87 1.42

0.61 1.19

1.24 1.72

Step 2 Hours per week Km per year Speeding/agress Errors Tiredness/ fatigue

1.02 11.47 1.09 0.01 4.69

0.312 0.001 0.924 0.702 0.030

0.87 1.39 1.05 1.05 1.56

0.57 1.15 0.39 0.81 1.046

1.20 1.70 2.92 1.35 2.431

–0.19 0.34** 0.27 0.05 0.466

0.19 0.09 0.26 0.52 0.215

Model Chi-Square 25.49** (df = 5) Block Chi-Square 12.89** (df = 3)

Next, the three modiﬁed DBQ factors were entered in the model to assess whether the proposed behaviours improved the prediction of demerit point loss over and above exposure to driving (Step 2). The additional variables collectively were signiﬁcant, with a chi-square statistic of X² (4, 3 = 443) = 10.89, p = 0.005, as was the fatigue variable. The model indicates that as participants become more tired and/or lose concentration, the corresponding likelihood of engaging in infringements that results in demerit point loss increases (p = 0.030). Several additional regression models were estimated to determine the sensitivity of the results. A test of the full model with all six predictors entered together, as well as the two models entered separately, conﬁrmed the same signiﬁcant predictors (for example, exposure and fatigue). Forward and backward stepwise regression identiﬁed the same predictors. Inclusion of gender, age and years driving experience did not increase the predictive value of the model.

210

Driver Behaviour and Training – Volume III

Discussion The chapter aims to explore possible contemporary changes to the DBQ for utilisation within organisational ﬂeet settings. More speciﬁcally, the chapter reports on an attempt to add additional contemporary driving behaviour items (for example, fatigue and multitasking) to the traditional DBQ in order to increase the utility of the measurement tool to examine and predict aberrant driving behaviours. At present, the DBQ has become increasingly popular as a measurement tool to investigate motorists’ self-reported driving behaviours (Lajunen et al., 2003; Parker et al., 1995b) as well as determine what driving behaviours are directly related to increased risk of crash involvement (Parker et al., 1995). However, within the professional driving setting, only a small body of research has begun to examine the driving behaviours of professional drivers (Newnam et al., 2004; Willis et al., 2004) and there has been little examination of necessary measurement tools to accurately capture their driving experiences and perceptions. Firstly, reliability analysis of the original DBQ indicated coefﬁcients that were relatively robust and similar to both the small amount of previous Australian research (Blockey and Hartley, 1995; Dobson et al., 1999) and recent ﬂeet safety ﬁndings (Sullman et al., 2002). Encouragingly, despite the subtle alterations to the DBQ to reﬂect Australian driving conditions, the reliability of the scale appears acceptable. Secondly, examination of the overall mean scores for the original DBQ factors revealed similar scores between the constructs, although it appears that the current sample were most likely to engage in Highway Code violations. This ﬁnding is consistent with previous research that has indicated speeding to be the most frequently reported aberrant driving behaviour on public roads (Dimmer and Parker, 1999; Lajunen et al., 2003). Furthermore, given the time pressures often placed on professional drivers, it may not be surprising that speeding violations are the most common form of aberrant behaviour both exhibited and reported by ﬂeet drivers. However, once the additional items were analysed that focused on fatigue, tiredness and multitasking, it became evident that participants in the current sample were most likely to report driving while under time pressure as well as driving while tired, followed by exceeding the speed limit on the highway. While only preliminary, the results indicate that although speeding remains one of the most common forms of aberrant behaviour reported by the ﬂeet drivers (Newnam et al., 2004; Sullman et al., 2002), drivers are also at risk of driving while fatigued, tired or while distracted. A series of factor analytic techniques were implemented to assist with the interpretation of the scale scores. Both exploratory and oblique rotations produced ﬁve factor models. While some previous studies have reported ﬁve factor structures (Parker et al., 2000), the present factors were difﬁcult to interpret, which resulted in a three factor solution being sought. This endeavour proved fruitful as three factors emerged that generally consisted of errors, highway violations/aggressive violations and the fatigue/tiredness factor. The three factor model was relatively consistent with previous research that has found distinctions between the different aberrant driving behaviours (Lajunen and Summala, 2003; Sullman et al., 2002). Driving error was the clearest factor to interpret and appeared to be associated with failures of observation and judgement, while general highway violations were characterised

A Study of Contemporary Modiﬁcations

211

by items that were a combination of traditional speeding behaviours as well as some aggressive acts. It is noteworthy that some of the highway violations that loaded on this factor may be interpreted as aggressive violations, especially for experienced professional drivers. For example, while driving especially close to a car in front of you to indicate for them to drive faster and crossing a junction knowing that the lights have already turned against you have traditionally been considered to be highway violations, they may also constitute an aggressive behaviour or at least indicate some level of frustration. Thus, behaviours traditionally viewed as highway violations may be classiﬁed as aggressive and aberrant, or at least, may originate from emotions associated with frustration. However, it is also noted that a temporality issue may be evident as aggressive violations in any current context may result in speeding violations (for example, highway violations) within a matter of moments. The third factor consisted purely of the additional items that focused on themes associated with tiredness, fatigue, loss of concentration and distraction. Taken together, the themes can be considered quite broad and thus identifying a clearly deﬁnable title for the factor proved difﬁcult. As highlighted previously, while it was originally anticipated that fatigue and multitasking items would be represented in distinct factors, this did not occur in the current sample of drivers. Nevertheless, the writers believe that an overall component of the factor is driving while fatigued, as some form of relationship may be identiﬁed between many of the individual items and the subsequent condition of being tired and at risk of losing concentration. However, ﬁve additional questions did not load on the factors, and of particular note was that of mobile phone use. While the two items relating to ‘hands-free’ and ‘hands-held’ phone use were not clearly interpretable in the current factor structure, further research appears to be needed to determine whether this ﬁnding is speciﬁc to the current sample or if phone use is not a contemporary ﬂeet driving issue that does not impact on driving outcomes. The corresponding calculation of Cronbach’s alpha reliability coefﬁcients for the new factors (speeding/aggression, errors and fatigue) revealed higher reliability scores than for the traditional item loading structure calculated before the factor analytic technique.2 However, the item-loading characteristics of the current study may be inﬂuenced by a number of additional issues, such as the speciﬁc characteristics of the sample. In addition, the lack of research into ﬂeet drivers combined with the difﬁculties interpreting the factor structure within the current study may indicate that individuals who drive for work, especially ﬂeet drivers, are a special population who experience and exhibit different driving behaviours to the general motoring population. Given that the factor structure of the DBQ has varied considerably (for example, three to six factors) in different countries and different settings, situational and cultural factors need to be taken into account when utilising the DBQ (Lajunen and Summala, 2003). Finally, with regards to the prediction of self-reported driving offences and crashes, only a small proportion of the sample reported being in a crash within the last year, which contributed to difﬁculties identifying factors associated with the 2 Although this is to be expected, as Cronbach’s alpha usually increases with the inclusion of additional items.

212

Driver Behaviour and Training – Volume III

event. Subsequently, an examination of self-reported driving violations through step-wise logistic regression analysis revealed that both exposure to the road and reporting symptoms of fatigue (for example, tiredness and loss of concentration) were predictive of incurring driving violations. Firstly, exposure to the road was expected to be a signiﬁcant predictor given that increasing driving distances is likely to increase the probability of deliberately or unintentionally making driving errors which may lead to demerit point loss. Secondly, fatigue was also identiﬁed as a predictor of demerit point loss and is of particular importance. Not only was driving while fatigued and driving while tired the most commonly reported behaviours by the sample, but this factor also predicted demerit point loss over and above exposure to the road. Given that feeling fatigued and making driving errors may be considered one of the most likely methods to incur infringement notices, it may not be surprising that driving under pressure and feeling tired are predictive of ﬁnes. Interestingly, there was only a moderate bivariate relationship between fatigue and exposure to the road, although it is possible this anomaly is speciﬁc to the current sample. Future research that identiﬁes the particular reason for motorists’ demerit point loss (for example, errors versus deliberate acts) may provide for a more reﬁned analysis to determine the speciﬁc contribution of fatigue to driving infringements and even crash involvement. Despite this, the current study provides preliminary evidence that driving under pressure and the associated feelings and behaviours of fatigue may warrant further investigation within ﬂeet settings. A number of limitations should be taken into account when interpreting the results of this study. The response rate of participants was relatively low, but consistent with previous research utilising the DBQ scale in Australia (Dobson et al., 1999). Similar to research in this area, concerns remain regarding the reliability of the selfreported behaviour, such as the propensity of professional drivers to provide socially desirable responses. Questions also remain about the representativeness of the sample, as participants were mainly corporate ﬂeet drivers (for example, involved in insurance sales) and such driving styles may not be easily transferable to other ﬂeet driving populations. In summary, further research is required to establish the predictive utility of including additional items in the DBQ which are ﬂeet speciﬁc, such as fatigue and multitasking. The present study has provided some additional preliminary evidence that modifying the DBQ to suit applied settings can produce favourable results with regards to identifying the factors that inﬂuence the driving task. References Aberg, L. and Rimmo, P. (1998). ‘Dimensions of aberrant driver behaviour.’ Ergonomics, 41, 39–56. Bianchi, A. and Summala, H. (2004). ‘The “genetics” of driving behaviour: parents’ driving style predicts their children’s driving style.’ Accident Analysis and Prevention, 36, 655–69. Blockey, P.N. and Hartley, L.R. (1995). ‘Aberrant driving behaviour: errors and

A Study of Contemporary Modiﬁcations

213

violations.’ Ergonomics, 38, 1759–71. Davey, J., Freeman, D. and Wishart, D. (2006). ‘A study predicting crashes among a sample of ﬂeet drivers.’ Proceedings of the Road Safety Research, Policing and Education Conference, Gold Coast, Australia [CD-ROM]. Davey, J., Wishart, D., Freeman, J. and Watson, B. (2007). ‘An application of the Driver Behaviour Questionnaire in an Australian organisational ﬂeet setting.’ Transportation Research Part F: Trafﬁc Psychology and Behaviour, 10: 11–21. Dimmer, A.R. and Parker, D. (1999). ‘The accidents, attitudes and behaviour of company car drivers.’ In G.B. Grayson (ed.), Behavioural Research in Road Safety IX, Crowthorne: Transport Research Laboratory. Dobson, A., Brown, W., Ball, J., Powers, J. and McFadden, M. (1999). ‘Women drivers’ behaviour, socio-demographic characteristics and accidents.’ Accident Analysis and Prevention, 31, 525–35. Lajunen, T., Parker, D. and Summala, H. (2003). ‘The Manchester Driver Behaviour Questionnaire: a cross-cultural study.’ Accident Analysis and Prevention, 942, 1–8. Lajunen, T. and Summala, H. (2003). ‘Can we trust self-reports of driving? Effects of impression management on driver behaviour questionnaire responses.’ Transportation Research Part F, 6, 97–107. Lawton, R., Parker, D., Stradling, S. and Manstead, A. (1997). ‘The role of affect in predicting social behaviours: the case of road trafﬁc violations.’ Journal of Applied Social Psychology, 27, 1258–76. Mesken, T., Lajunen , T. and Summala, H. (2002). ‘Interpersonal violations and their relation to accident involvement in Finland.’ Ergonomics, 45 (7), 469–483. Newnam, S., Watson, B. and Murray, W. (2002). ‘A comparison of the factors inﬂuencing the safety of work-related drivers in work and personal vehicles.’ Proceedings of the Road Safety Research, Policing and Education Conference, Adelaide [CD-ROM]. Newnam, S., Watson, B. and Murray, W. (2004). ‘Factors predicting intentions to speed in a work and personal vehicle.’ Transportation Research Part F, 7, 287– 300. Ozkan, T. and Lajunen, T. (2005). ‘A new addition to DBQ: positive driver behaviours scale.’ Transportation Research Part F, 8, 355–68. Parker, D., McDonald, L., Rabbitt, P. and Sutcliffe, P. (2000). ‘Elderly drivers and their accidents: the aging driver questionnaire.’ Accident Analysis and Prevention, 32, 751–9. Parker, D., Reason, J.T., Manstead, A. and Stradling, S.G. (1995a). ‘Driving errors, driving violations and accident involvement.’ Ergonomics, 38, 1036–48. Parker, D., West, R.J., Stradling, S.G. and Manstead, A.R. (1995b). ‘Behavioural characteristics and involvement in different types of trafﬁc accident.’ Accident Analysis and Prevention, 27(4), 571–81. Reason, J., Manstead, A., Stradling, S., Baxter, J. and Campbell, K. (1990). ‘Errors and violations: a real distinction?’ Ergonomics, 33, 1315–32. Shinar, D. ‘Aggressive driving: the contribution of the drivers and the situation.’ Transportation Research Part F, 1, 137–60. Stradling, S.G., Meadows, M.L. and Beatty, S. (2000). ‘Driving as part of your work

214

Driver Behaviour and Training – Volume III

may damage your health.’ In G.B. Crayson (ed.), Behavioural Research in Road Safety IX, Crowthorne: Transport Research Laboratory. Sullman, M.J., Meadows, M. and Pajo, K.B. (2002). ‘Aberrant driving behaviours amongst New Zealand truck drivers.’ Transportation Research Part F, 5, 217– 232. Wheatley, K. (1997). ‘An overview of issues in work-related driving.’ In Staysafe 36: Drivers as Workers, Vehicles as Workplaces: Issues in Fleet Management. (Report No. 9/51). Ninth Report of the Joint Standing Committee on Road Safety of the 51st Parliament. Sydney: Parliament of New South Wales. Wills, A., Watson, B. and Biggs, B. (2004). ‘The relative inﬂuence of ﬂeet safety climate on work-related driver safety.’ Proceedings of the Road Safety Research, Policing and Education Conference, Perth [CD-ROM]. Xie, C. and Parker, D. (2002). ‘A social psychological approach to driving violations in two Chinese cities.’ Transportation Research Part F, 5, 293–308.

Chapter 18

A Comparison of Seat Belt Use Between Work Time and Free Time Driving Among Turkish Taxi Drivers Özlem Şimşekoğlu1 and T. Lajunen2 1 University of Helsinki, Finland 2 Middle East Technical University, Turkey Introduction International research has demonstrated that using a seat belt reduces considerably the number of car occupants injured or killed in trafﬁc accidents (Elvik and Vaa, 2004; Evans, 1986). According to the results of a meta-analysis on effects of seat belt use, using a seat belt decreases the probability of being killed by about 40–50 per cent for drivers and front seat passengers, and by about 25 per cent for passengers in the back seats (Elvik and Vaa, 2004). Similarly, Evans (1986) reported that nearly 41 per cent reduction in fatalities would occur if all the front seat occupants in the US were to use lap/shoulder belts. Despite the proven effectiveness of seat belts, however, a large number of car occupants do not use seat belts in Turkey, which has been seen as one of the main reasons for low trafﬁc safety in Turkey (SWE ROAD, 2001). Low seat belt use rate, especially among professional drivers, is a serious trafﬁc safety problem in Turkey. For example, almost no taxi drivers, who are not required by the trafﬁc code to use a seat belt, or front seat passengers of taxis, use a seat belt (T.C. Emniyet Genel Müdürlüğü, Traﬁk Hizmetleri Başkanlığı, 1999). Therefore, in order to improve driver and passenger safety, seat belt use should increase among taxi drivers and passengers. Lack of strict seat belt legislation, enforcement and frequently getting in and out of their vehicles are common reasons why taxi drivers do not use seat belts (Ferguson, Wells, Williams and Feldman, 1999; Fernandez, Park and Olshaker, 2005). In the United States, for example, considerably higher seat belt use rate (74 per cent) among taxi drivers in the District of Columbia compared with Maryland (20 per cent) and Virginia (38 per cent) was explained by the primary seat belt law, which applies to all drivers in the District of Columbia (Ferguson et al., 1999). Similarly, stricter seat belt legislation, which also applies to taxi drivers, was believed to increase the low seat belt use rate (6.8 per cent) among taxi drivers in Boston (Fernandez, Park and Olshaker, 2005). Also, taxi passengers use seat belts infrequently due to belt inaccessibility and short trips (Welkon and Reisinger, 1977; Ferguson et al., 1999). The present study investigated mainly taxi drivers’ self-reported seat belt use frequencies and reasons for not using a seat belt when driving a taxi (for example,

Driver Behaviour and Training – Volume III

216

during work hours) and a private car (for example, during free time). Although there are studies that investigated seat belt use among taxi drivers, we did not come across a study investigating whether seat belt use among taxi drivers changed between work time and free time driving. The main aims of the study were to investigate: •

• •

whether reported seat belt use frequencies and degree of agreement with the reasons for not using a seat belt differs according to the driver status (taxi driver, private car driver); factors related to reported seat belt use frequency when driving a taxi and a private car; and speciﬁc reasons for not using a seat belt when driving a taxi.

Method Participants One hundred and twenty-two male Turkish taxi drivers aged between 24 and 61 years (M = 39.8, SD = 9.1) participated in the study. Most of them were high school graduates (40.2 per cent), followed by elementary school graduates (27 per cent), primary school graduates (23.8 per cent) and university graduates (9 per cent). The mean number of years spent working as a taxi driver was 11.7 years (SD = 9.2). All the taxi drivers were working for different private taxi companies in Ankara. Questionnaire Taxi drivers completed the questionnaire at taxi stations while they were off work or waiting for passengers. The questionnaire included the following topics: demographic and driving related information (for example, age, education and total kilometres driven last year), seat belt use frequency in different travelling conditions, reasons for not using a seat belt, attitudes towards seat belt use, and the Driving Skills Inventory (DSI; Lajunen and Summala, 1995). Eleven seat belt use frequency items were asked for measuring the frequency of seat belt use in different travelling conditions (for example, city trips, outside city trips) on a ﬁve point Likert scale (1 = ‘never’, 5 = ‘always’). For measuring the reasons for not using a seat belt, participants were given 27 possible reasons and asked to indicate their degree of agreement with the items on a ﬁve point Likert scale (1 = ‘completely disagree’, 5 = ‘completely agree’). These items were chosen by the authors from a pool of reasons for not using a seat belt suggested by colleagues and students on a trafﬁc psychology course. Among the 27 items, six items were only applicable to either taxi driving or private car driving, whereas the remaining 21 items were applicable to both taxi and private car driving. The questionnaire included 12 attitude items to which the respondents answered by using a semantic differential scale (1 = ‘not safe at all’, 7 = ‘very safe’). Lastly, the DSI included a ﬁve point response scale (1 = ‘absolutely weak’, 5 = ‘absolutely strong’) to measure self-reported driving skill. The participants were asked to rate items related to seat belt use frequency and reasons for not using a seat belt thinking

A Comparison of Seat Belt Use

217

of the times both when they were driving a taxi (at work) and driving a private car (in free time). Statistical analysis Factor analyses were conducted for the items asking reasons for not using a seat belt when driving a taxi and a private car, and for the DSI. Two hierarchical multiple regression analyses were also performed on the data to investigate the factors related to the reported general seat belt use frequency of the participants when driving a taxi and a private car. After entering the demographic and some driving-related variables in the ﬁrst step, other variables were entered to the model using the stepwise selection method. Before the analyses, data were checked for multicollinearity and singularity. No sign of multicollinearity or singularity were found. Lastly, paired sample t-tests were conducted to compare the responses participants gave to some items both as a taxi driver and as a private car driver. Results Reported seat belt use frequencies and reasons for not using a seat belt The majority (78.7 per cent) of the participants reported that they never use a seat belt when driving a taxi, while about 21 per cent reported that they use a seat belt seldom or sometimes. However, 14 per cent of the participants reported that they never use a seat belt and 30 per cent of them reported that they always use a seat belt when driving a private car. Mean response values for the given reasons for not using a seat belt when driving a taxi and private car are given in Table 18.1. Discomfort followed by not having the habit and driving short distances was among the mostly agreed items when driving both a taxi and private car. No obligation by law to use a seat belt in city trafﬁc was the second mostly agreed to item for the times when driving a taxi. Also, answers given to an open-ended question asking for other reasons for not using a seat belt as a taxi driver revealed that about 41 per cent of the respondents reported fear of not being able to defend themselves in case of an attack and robbery by the passengers. Factor analyses Principal axis factoring analysis with varimax rotation was applied to items related to reasons for not using a seat belt when driving a taxi and a private car and to the items of the DSI. Three factors emerged from the items related to reasons for not using a seat belt both when driving a taxi and private car. For the times when driving a taxi, the ﬁrst factor was labelled as ‘situational factors’, the second factor was called ‘over trust of driving skills and experience and under estimation of accident probability’ and the third factor was labelled as ‘following the example of the other taxi drivers’. For the times when driving a private car, the ﬁrst factor was labelled as ‘situational factors’, the second factor was labelled as ‘over trust of driving skills and experience

218

Driver Behaviour and Training – Volume III

Table 18.1 Mean response values for the reasons for not using a seat belt

Reasons

1 2 3 4 5 6 7 8 9 10 11 12 13 14

Mean response value for taxi driving*

Mean response value for private car driving* 1.42 2.11 2.61 1.39 2.08 1.55 2.58 1.88 1.22 1.75 1.71 2.11 2.31 1.85

Not believing in the effectiveness of seat belt 1.48 Fear of getting stuck in the car in an accident 2.12 Discomfort 3.50 Waste of time 1.80 No need 2.67 Low accident probability 1.62 No habit 2.86 Forgetting 1.75 Risk taking 1.20 Trust in driving experience 1.78 Trust in driving skills 1.77 Underestimation of a serious accident on city roads 2.27 Underestimation of a serious accident on short trips 2.43 Underestimation of a serious accident in good 1.93 weather 15 Underestimation of a serious accident on good roads 1.87 16 Underestimation of a serious accident on day trips 1.87 17 Driving with low speed 2.28 18 Driving a short distance 2.62 19 Airbag in the car 1.40 20 Belief in destiny 1.40 21 No obligation to use seat belt in city trafﬁc 3.44 22 Following the example of other taxi drivers who do 1.76 not use seat belt 23 Social disapproval for seat bet use from other taxi 1.52 drivers 24 Fear of being perceived as a bad driver by passengers 1.34 25 Fear of being perceived as a bad driver by family members 26 Following the example of family members who do not use a seat belt 27 Following the example of close friends who do not use 1.38 a seat belt *1 = Completely disagree, 2 = Disagree, 3 = Neither agree nor disagree, 4 = Agree, 5 = Completely agree

1.83 1.78 2.19 2.48 1.42 1.39

1.16 1.20 1.29

A Comparison of Seat Belt Use

Table 18.2

Results of the factor analysis for the items related to reasons for not using a seat belt when driving a taxi and a private car When driving a taxi

3 5 6 7 9 10 11 12 13 14 15 16 17 18 19 22

23

26

27

Item Discomfort No need Low accident probability No habit Risk taking Trust in driving experience Trust in driving skills Underestimation of a serious accident on city roads Underestimation of a serious accident on short trips Underestimation of a serious accident in good weather Underestimation of a serious accident on good roads Underestimation of a serious accident on day trips Driving with a low speed Driving a short distance Airbag in the car Following the example of other taxi drivers who do not use seat belt Social disapproval for seat belt use from other taxi drivers Following the example of family members who do not use seat belt Following the example of close friends who do not use seat belt Initial Eigenvalue Explained variance (%) α/r

F1

F2

F3

When driving a private car F1 F2 F3 0.69 0.65

0.49 0.68 0.51 0.91 0.89

0.62 0.56

0.57

0.62

0.65

0.69

0.72

0.78

0.69

0.73

0.78

0.84

0.68 0.58 0.45

0.60 0.50 0.47 0.59

0.67

0.70

0.60

7.42 16.71

1.81 13.81

0.91

0.83

1.69 6.95

0.42** **p < 0.01

219

7.63 18.20

1.99 12.98

1.63 11.04

0.91

0.75

0.75

Driver Behaviour and Training – Volume III

220

and following the example of close others’ and the last factor was labelled as ‘not needing a seat belt, discomfort and no habit’. Items with their item loadings for the factors and internal consistency coefﬁcient (α) or correlation coefﬁcient (r) (in the case of only two items) of the sub-scales can be seen in Table 18.2. Similar to earlier research ﬁndings, two factors (driving skills and safety skills) emerged from the DSI. The DSI items with their item loadings on the two factors and internal consistency coefﬁcient (α) of the sub-scales can be seen in Table 18.3. Sum variables were calculated on the basis of factor analysis results by averaging the item scores. Table 18.3 Results of factor analysis for the items of DSI Item 1 2 3 4 5 6 7 8 9 10 11 12 13 14 16 17 18 19 20

F1

F2

Fluent driving Perceiving hazards in trafﬁc Driving behind a slow car without getting impatient Managing the car through a skid Predicting trafﬁc situation ahead Knowing how to act in particular trafﬁc situations Fluent lane-changing in heavy trafﬁc Making ﬁrm decisions Staying calm in irritating situations Controlling the vehicle Keeping a sufﬁcient following distance Adjusting your speed to the conditions Make a hill start on a steep incline Overtaking Conforming to the speed limits Avoiding unnecessary risks Tolerating other driver errors calmly Obeying the trafﬁc lights carefully Reverse parking into a narrow gap

0.54 0.48

0.58

Initial Eigenvalue Explained variance (%)

4.98 20.25

2.69 11.64

α

0.84

0.73

0.45 0.56 0.48 0.52 0.59 0.61 0.50 0.59 0.46 0.43 0.62 0.77 0.69 0.47 0.37 0.66

Multiple regressions analyses Results of the multiple regression analysis for predicting taxi drivers’ seat belt use frequency by the study variables when driving a taxi are shown in Table 18.4. The number of accidents the drivers had experienced in last three years, driving skills and situational reasons (for example, short trips) were negatively related to the reported seat belt use frequency, whereas penalty score, safety skills and following

A Comparison of Seat Belt Use

221

the example of other taxi drivers were positively related to the reported seat belt use frequency of the participants. Table 18.4

Factors related to reported seat belt use frequency when driving a taxi

Step 1

Variable Age

R2 0.04

Change in R2 0.04

F 0.53

df 8

β –0.10

Primary school graduated

–0.15

Elementary school graduated

–0.11

University graduated

–0.12

Total km driven last year

0.10

Years working as a taxi driver

0.06

Number of accidents in last three years

–0.21*

Penalty score

0.21*

2

Driving skills

0.09

0.05

1.12

9

–0.34**

3

Safety skills

0.14

0.05

1.73

10

0.30**

4

Situational reasons

0.17

0.03

2.02*

11

–0.24*

5

Following the example of other taxi drivers

0.21

0.04

2.33*

12

0.21*

Results of the multiple regression analysis for predicting taxi drivers’ seat belt use frequency by the study variables when driving a private car are shown in Table 18.5. The factor of not needing to use a seat belt, discomfort and not having the habit of seat belt use, and total attitude towards seat belt use were negatively related to self-reported seat belt use frequency; whereas safety skills were positively related to the reported seat belt use frequency of the participants.

Driver Behaviour and Training – Volume III

222

Table 18.5

Step

1

Factors related to reported seat belt use frequency when driving a private car Variable

Age

R2

0.16

Change in R2

F

0.16

2.97**

df

7

Primary school graduated

β

0.12 0.07

Elementary school graduated

–0.002

University graduated

0.02

Total km driven last year

–0.15

Number of accidents in last three years

0.06

Penalty score

2

Not needing to use seat belt, discomfort and no habit

0.40

0.24

8.93***

8

0.09 –0.56***

3

Safety skills

0.43

0.03

9.21***

9

0.23**

4

Total attitude

0.46

0.03

9.21**

10

–0.19**

Paired sample tests Results of the paired sample t-tests for comparing the reported seat belt use frequencies between work driving and free time driving showed that participants reported signiﬁcantly higher seat belt use frequency when driving a private car than a taxi in all travelling conditions. T-test values for the different travelling conditions are as follows: all trips in general (t(121) = –14.74, p < 0.001), city trips (t(121) = –10.58, p < 0.001), outside city trips (t(121) = –4.58, p < 0.001), short trips (t(121) = –8.30, p < 0.001), long trips (t(121) = –7.99, p < 0.001), night trips (t(121) = –10.37, p < 0.001), day trips (t(121) = –9.96, p < 0.001), trips in bad weather conditions (t(121) = –10.41, p < 0.001), trips in good weather conditions (t(121) =–9.61, p < 0.001), trips in good roads (t(121) = –9.77, p < 0.001) and trips in bad roads (t(121) = –9.90, p < 0.001). Also, results of the paired sample t-tests for comparing reasons for not using a seat belt between work driving and free time driving showed that

A Comparison of Seat Belt Use

223

participants agreed signiﬁcantly more with discomfort (t(121) = 6.69, p < 0.001), waste of time due to seat belt use (t(121) =3.81, p < 0.001), no need to use a seat belt (t(121) = 4.84, p < 0.001), not having the habit (t(121) = 3.22, p < 0.001) and underestimation of accident probability in urban roads (t(121) = 2.30, p < 0.05) and short distances (t(121) = 2.13, p < 0.05) when driving a taxi than driving a private car. Discussion The present study mainly investigated whether reported seat belt use frequencies and degree of agreement with the given reasons for not using a seat belt differ between the times when driving a taxi and private car in a sample of Turkish taxi drivers. Participants reported signiﬁcantly higher seat belt use frequency in all travelling conditions for the times when they were driving a private car than a taxi. This ﬁnding clearly shows that taxi drivers change their seat belt use behaviour according to their driver status (taxi driver versus car driver). Also, this ﬁnding implies that there should be reasons speciﬁc to taxi driving that explain why the same drivers use a seat belt less while driving a taxi than a private car. After discomfort, no obligation to use a seat belt in city trafﬁc was the second mostly agreed reason for not using a seat belt when driving a taxi. Similar to the previous ﬁndings (Ferguson et al., 1999; Fernandez, Park and Olshaker, 2005), this ﬁnding emphasises the urgent need for seat belt laws to also apply to taxi drivers to increase seat belt use among them. In addition, fear of not being able to defend themselves in case of an attack and robbery by the passengers was an important motivation for not using a seat belt when driving a taxi. This ﬁnding supports a previous study emphasising the importance of workplace hazards in predicting health outcomes and safety behaviour among taxi drivers (Machin and Souza, 2004). Thus, no obligation to use a seat belt in city trafﬁc and fear of being attacked or robbed by the passengers appear as two speciﬁc reasons for not using a seat belt when driving a taxi. When driving a taxi, driving skills were negatively related to reported seat belt use frequency, while safety skills were positively related to reported seat belt frequency. These results are in line with previous ﬁndings indicating that driving skills were positively associated with number of penalties, accidents and level of speed, while safety skills were negatively associated with these variables (Lajunen, Corry, Summala and Hartley, 1998; Özkan, Lajunen, Chliaoutakis, Parker and Summala, 2006; Sümer, 2002). Also, driving experience was positively associated with driver’s conﬁdence in his/her driving skills (Lajunen and Summala, 1995). Thus, taxi drivers’ overconﬁdence in their driving skills related to their high annual mileage and exposure may partly explain their infrequent seat belt use. The number of accidents in the last three years was negatively related to reported seat belt use frequency when driving a taxi. A low number of accidents, which can be accepted as a sign of safe driving, can explain the negative relationship between number of accidents and reported seat belt use frequency. For the times when driving a private car, the factor of not needing to use a seat belt, discomfort and not in the habit was the strongest predictor of seat belt use,

224

Driver Behaviour and Training – Volume III

similar to previous ﬁndings (Şimşekoğlu and Lajunen, 2006). Interestingly, attitude towards seat belt use was negatively related to seat belt use indicating that as the participants had a more positive attitude towards seat belt use, their reported seat belt use frequency decreased. This ﬁnding is in line with the previous ﬁndings indicating that having positive attitudes, beliefs and intentions about using a seat belt were not strong predictors of actual seat belt use all the time (Chliaoutakis et al., 2000; Knapper, Cropley and More, 1976; Loo, 1984). The present ﬁndings have some practical implications for the interventions aiming at increasing seat belt use among taxi drivers. Using a seat belt is crucial for taxi drivers for two main reasons. First of all, taxi drivers are at a higher risk of being involved in a car accident due to their high driving exposure. Secondly, travelling by taxi is one of the most common public means of transportation among Turks. Therefore, taxi drivers should feel responsible for the safety of their passengers by using a seat belt and requiring their passengers to use a seat belt. The present ﬁndings clearly show that strict seat belt laws with high enforcement are needed to increase seat belt use among taxi drivers. Also, in taxis, the presence and availability of seat belts should be controlled by the vehicle inspection agencies or the police. Besides the laws, seat belt campaigns aiming at increasing seat belt use among taxi drivers should be more common. These campaigns should emphasise the safety beneﬁts of seat belt use in all travelling conditions and include messages to taxi drivers about being a responsible and safe driver who is concerned for public safety. Acknowledgements This study has been supported by the Graduate School of Psychology in Finland and Marie Curie Transfer of Knowledge programme (‘SAFEAST’ Project No: MTKDCT-2004-509813). References Chliaoutakis, E.J., Gnardellis, C., Drakou, I., Darviri, C. and Sboukis, V. (2000), ‘Modelling the factors related to the seatbelt use by the young drivers of Athens.’ Accident Analysis and Prevention, 32, 815–25. Elvik, R. and Vaa, T. (2004). The Handbook of Road Safety Measures. Amsterdam: Elsevier. Evans, L. (1986). ‘The effectiveness of safety belt in preventing fatalities.’ Accident Analysis and Prevention, 18, 229–41. Ferguson, S.A., Wells, J.K., Williams, A.F. and Feldman, A.F. (1999). ‘Belt use rates among taxicab drivers in a jurisdiction with license points for nonuse.’ Journal of Safety Research, 30, 87–91. Fernandez, W.G., Park, J.L. and Olshaker, J. (2005). ‘An observational study of safety belt use among taxi drivers in Boston.’ Annals of Emergency Medicine, 45, 626–9. Knapper, K.C., Cropley, J.A. and Moore, J.R. (1976). ‘Attitudinal factors in the nonuse of seat belts.’ Accident Analysis and Prevention, 8, 241–6. Lajunen, T., Corry, A., Summala, H. and Hartley, L. (1998). ‘Cross-cultural

A Comparison of Seat Belt Use

225

differences in drivers’ self-assessment in their perceptual motor and safety skills: Australians and Finns.’ Personality and Individual Differences, 24, 539–50. Lajunen, T. and Summala, H. (1995). ‘Driver experience, personality, and skill and safety-motive dimensions in drivers’ self assessments.’ Personality and Individual Differences, 19, 307–18. Loo, R. (1984). ‘Correlates of reported attitudes towards and use of seat belts.’ Accident Analysis and Prevention, 16, 417–21. Machin, M.A. and De Souza, J.M.D. (2004). ‘Predicting health outcomes and safety behaviour in taxi drivers.’ Transportation Research Part F, 7, 257–70. Özkan, T., Lajunen, T., Chliaoutakis, J.E., Parker, D. and Summala, H. (2006). ‘Cross-cultural differences in driving skills: a comparison of six countries.’ Accident Analysis and Prevention, 38, 1011–18. Sümer, N. and Özkan, T. (2002). ‘The role of driver behavior, skills and personality traits in trafﬁc accidents.’ Turkish Journal of Pyschology, 17(50), 1–25. SWE ROAD (2001). Türkiye için ulusal traﬁk güvenliği programı. Taslak Nihai Rapor. Ankara. Şimşekoğlu, Ö. and Lajunen, T. (2006). ‘Why Turks do not use seat belts?’ An interview study. Submitted. T.C. Emniyet Genel Müdürlüğü, Traﬁk Hizmetleri Başkanlığı. (1999). Ülkemizde emniyet kemeri kullanımı. Ankara: Traﬁk Araştırma Merkezi Müdürlüğü Yayınları. Welkon, C. and Reisinger, K.S. (1977). ‘The phantom taxi seat belt.’ Public Health Briefs, 67, 1091–2.

This page intentionally left blank

Chapter 19

A Review of Developing and Implementing Australian Fleet Safety Interventions: A Case Study Approach Update Darren Wishart, Jeremy Davey and James Freeman Centre for Accident Research and Road Safety, Queensland (CARRS-Q), Australia Introduction There is relatively little research in Australia or overseas devoted to the area of ﬂeet and work-related road safety, and this is being increasingly addressed. The recent focus is in part due to an awareness of workplace health and safety issues and the overall impact that ﬂeet-related safety issues have on business effectiveness and road safety (Downs et al., 1999; Haworth et al., 2000). Historically, costs associated with work-related vehicle crashes have more often than not been calculated in terms of vehicle damage or write off costs. Murray et al. (2003) suggest that the direct cost of crashes in terms of repairs is only the tip of the iceberg. In recent years, changes in industry/employer accountability, business processes, occupational health and safety, workers’ compensation legislation, insurance and third party coverage, and a generally more litigious environment, require industry to develop better benchmarking along with more comprehensive programmes to improve ﬂeet safety. There is currently only a small amount of work in this area and estimates of the true cost for work-related crashes suggest that hidden costs may be somewhere between 8–36 times vehicle repair or replacement costs (Murray et al., 2003). Based solely on workers compensation data, estimates of costs to Australian industry for workrelated crashes have been in the vicinity of $400–$500 million per year (Wheatley, 1997). Furthermore, a recent estimate of the average cost to society for a fatal crash is approximately $2 million (Austroads, 2006) and the average total insurance cost of a ﬂeet incident to organisations and society is approximately $28 000 (Davey and Banks, 2005). Previous research has highlighted work-related road safety as an area that requires further attention with a focus on developing research informed interventions aimed at improving road safety outcomes and, in turn, offering huge ﬁnancial savings to industry and the community (Bibbings, 1997; Murray et al., 2003; Haworth et al., 2000; Staysafe, 1997). In Australia, road crashes are the most common cause of workrelated injury, death and absence from work (Howarth et al., 2000). Work-related trafﬁc injuries are about twice as likely to result in death or permanent disability

228

Driver Behaviour and Training – Volume III

than other workplace accidents (Wheatley, 1997) and account for up to 26 per cent of work-related fatalities in Australia and 13 per cent of the national road toll (Murray et al., 2003). There is an obvious and growing need for industry, government and the community to allocate resources and build knowledge and expertise in this area. Historically, in terms of exploring and implementing ﬂeet safety interventions, industry has often taken a ‘silver bullet’ approach aimed at developing and implementing a single countermeasure or intervention strategy to encompass and address all ﬂeet-related road safety issues. This approach is often reactive, rather than proactive which aims to not only reduce similar incidents but also is aimed at improving behaviour. One shortcoming with a reactive approach is that often the single implemented countermeasure results in only a short-term ﬁx and does not address the underlying contributing behavioural factors relating to the crash. Thus the organisation embarks on a cyclical process similar to a dog chasing its tail and may not demonstrate signiﬁcant improvement in their ﬂeet safety records over time. More recently one of the facilitators of progress in ﬂeet safety has been the occupational health and safety domain (OHS). OHS and chain of responsibility (COR) legislation has helped to create further awareness of an organisation’s responsibility to ensure safe work practice. Industry, as a means of trying to address OHS responsibilities in ﬂeet safety, adopts what they consider to be a best practice approach. Historically, best practice to improving ﬂeet safety has often meant any practice or type of intervention being implemented. This can result in countermeasures and intervention strategies that have not been previously evaluated or organisations not implementing a thorough and empirical evaluation process. Furthermore, the silver bullet approach is no longer used in other areas of road safety, as research would suggest that intervention approaches need to be proactive and multi-dimensional. For instance, strategies and interventions to reduce the incidence of drink driving often involve not only law enforcement and random breath testing, but also incorporate advertising and awareness campaigns, rehabilitation programmes and technological interventions such as alcohol interlock devices. However, the current state of ﬂeet safety has many organisations not addressing the work-related road safety issue as comprehensively as other work-related safety risk issues within their workplace. For example, organisations often allocate more safety-related resources to lower exposure and lower workplace risk processes, in contrast to the high exposure and high risk of work-related driving. In attempting to satisfy legislative needs of OHS, organisations plan the development of work-related road safety intervention strategies, although the reality within the majority of organisations is that they often struggle to implement such interventions. The failure to effectively implement ﬂeet safety interventions often stems from a lack of management commitment and support, and general underresourcing. Thus there is an immense discrepancy between what organisations plan to do and what is actually undertaken in addressing work-related road safety risks and initiatives.

A Review of Developing and Implementing Australian Fleet Safety Interventions

229

Factors of inﬂuence in ﬂeet driver behaviour A proactive multi-dimensional approach to ﬂeet safety is required to help address the many factors that inﬂuence ﬂeet driver behaviour. The following ﬁgure provides an indication of the numerous conditions inﬂuencing driver behaviour and subsequently ﬂeet driver behaviour (Lonero and Clinton, 1998). Historically, ﬂeet safety initiatives, in part due to ﬂeet safety coming from an asset management perspective, have taken on a one size ﬁts all approach. This approach has often been lacking in addressing the varied inﬂuences underlying ﬂeet driver behaviour often resulting in only shortterm ﬂeet safety improvement.

Figure 19.1 Conditions inﬂuencing driver behaviour Source: Lonero and Clinton, 1998.

Case study research Research conducted by CARRS-Q with a variety of industry ﬂeets reveal similar patterns emerging across ﬂeets in relation to causal and contributing factors to crashes, data recording and reporting issues, types of crashes and the types of vehicles involved. Throughout a number of large diverse vehicle ﬂeets the most common types of crashes accounting for the vast majority of ﬂeet incidents are represented by: • reversing • rear enders • road conditions • loss of control

230

• • •

Driver Behaviour and Training – Volume III

animal-related incidents damage whilst parked accumulated damage

Interestingly, these crash categories appear to be a reﬂection of a combination of a blameworthy and asset management approach to crashes. Categorisation in this manner does not provide any insight into the perceptions, attitudes, safety climate and organisational culture contributing to crashes through the inﬂuence on human behaviour. In contrast, transport authorities’ recording of crashes indicate a broader range of contributing factors to crashes which encompasses driver and road conditions. For example, Queensland Transport (2001) lists factors contributing to crashes such as: • • • • • • • • • • • • •

disobeying road rules alcohol/drugs speed inexperience inattention age fatigue other driver conditions negligence rain/wet road road conditions vehicle defects street lighting

These two approaches to recording crashes demonstrate the different genres of approaches to ﬂeet safety within organisations; one being asset management and the second having more of a human behaviour interface. Each method of recording crashes provides different types of information that can be used to inform organisational objectives and interventions. The asset management approach is the most widely used approach to inform interventions. Whilst this may often result in short-term ﬁnancial gain, it does not supply the information necessary for large-scale behavioural interventions and workplace culture change. The alternate approach used in other domains focuses more on driver behaviour and road conditions. These two approaches to data collection which inform interventions are reactive in that the core data collection occurs post crash. What is needed is a data collection approach that centres on driver behaviour and subsequently inﬂuences safer workrelated driving. The majority of current approaches in the workplace, while helpful to an organisation in some sense, do not provide the information necessary to implement targeted interventions designed to address the speciﬁc behavioural, attitudinal and cultural inﬂuences impacting on work-related road safety. In addition, the current reactive data collection approaches also do not provide an effective manner in which to empirically evaluate ﬂeet safety interventions and initiatives that are implemented.

A Review of Developing and Implementing Australian Fleet Safety Interventions

231

For example, the collected data are often more reﬂective of insurance recording mechanisms which incorporate the process of drivers ‘attributing blame’ elsewhere rather than the objective identiﬁcation of the factors contributing to the crash. Historical approach One of the historical approaches to ﬂeet safety focuses on ‘behind the wheel’ driver training and education. Although many of these programmes are to teach road users the skills necessary for the successful operation of a vehicle on our roads, caution needs to be exercised to ensure that the distinction between performance and behaviour is recognised, as is the difference between what road users are capable of doing and what they actually do. Performance levels of road users can often be linked to the skills and demands of certain road situations, whereas road user behaviour is often inﬂuenced by cultural, personality, attitudinal and motivational factors (Parker, Lajunen and Stradling, 1998). This suggests that high levels of skill or proﬁciency in a task does not necessarily translate into better behaviour. There is also a common misunderstanding that improving road user skills will automatically improve road user behaviour, which in turn is expected to result in improved road safety. Furthermore, increased skill proﬁciency needs to be complimented by organisational processes and procedures that support safe driving behaviour. For example, although training may provide the skills and possible awareness to drive safely (for example, to not speed), organisational processes and work tasks may create time pressure demands that compromise safe driving operations. Driver training and education programmes involving a strong practical component, such as that the development of vehicle control skills may inadvertently create an inﬂated belief in one’s own driving ability, which in turn may lead to an increase in aggressive driving behaviour (Katila, Keskinen, Hatakka and Laapotti, 2003). In order to improve ﬂeet safety, organisations need to adopt a broader perspective and develop initiatives targeted at the underlying cultural issues further inﬂuencing ﬂeet safety, along with adopting the necessary supportive organisational processes that facilitate safe driving. Cultural approach Recent research conducted across various vehicle ﬂeet settings suggests there is a strong inﬂuence on work-related driving behaviours by an organisation’s safety climate (Wills, 2003). Safety climate can be expressed as an employee’s psychological perceptions of safety culture and practice (Hayes et al., 2002). These perceptions are developed from the employee’s continual observation of other work colleagues’ safety practice. These observations in turn inﬂuence employee behaviour in relation as to what are considered accepted levels of safety required to perform work-related tasks (Varonen and Mattila, 2000). An example of the inﬂuence that organisational culture and safety climate can have on performance can be demonstrated through the practice of speeding. There is a strong focus on road safety and educational campaigns highlighting the dangers

Driver Behaviour and Training – Volume III

232

of speeding and the need for drivers to obey speed limits, yet enforcement data demonstrate that speeding still frequently occurs. Organisational culture within a ﬂeet setting may dictate that it is more important to attend an appointment on time or complete a ‘necessary’ task urgently, than it is to be late or leave a task incomplete. In this instance the employee may compromise their safety and the safety of others by driving above the speed limit in order to ‘make up time’ or ‘deliver the goods’. Needs analysis Organisations embarking on a programme of improving ﬂeet safety often undertake a needs analysis investigating what is currently being done in relation to addressing ﬂeet safety issues. This process often involves investigation into areas such as: • • • • • • • •

organisational process interventions reporting recording policy recruitment interventions evaluation

The results of a needs analysis are often used by organisations to assist in identifying areas for improvement and to ensure that appropriate processes, mechanisms and structure are adequately in place to support change and intervention strategies. However, the information provided by the needs analysis often exposes deﬁciencies in processes, reporting, recording and policy mechanisms without actually informing the design of behavioural-based intervention strategies. Future ﬂeet safety research and the subsequent development of intervention programmes must address the inﬂuences on behaviour to achieve long-term improvements in ﬂeet safety. Fleet safety research has previously been lacking in developing research-based and informed intervention strategies directed at behaviours, attitudes, intentions, perceptions, organisational culture and safety climate. It is with this in mind that current research should be directed at addressing a number of domains that inﬂuence behaviour. The results obtained from baseline measures in these domains should guide the development and implementation of targeted interventions aimed at high risk sectors and behaviours in an operational ﬂeet environment. Identiﬁed baseline measures Organisations need to gather baseline measures from a number of areas that current research has identiﬁed as inﬂuencing the design, development and implementation of appropriate and targeted intervention strategies. These can include:

A Review of Developing and Implementing Australian Fleet Safety Interventions

• • • • • • • • •

233

driver attitudes road safety knowledge behavioural intentions perceptions risk taking sensation seeking crash records driver history safety climate.

Current research undertaken by CARRS-Q is examining the development of targeted intervention strategies tailored toward speciﬁc issues identiﬁed from baseline measures in the above-mentioned areas. The results obtained from these baseline measures are used to assist organisations in making informed choices regarding the implementation of countermeasures. High risk areas of vehicle ﬂeets can be identiﬁed from baseline measures not only in terms of vehicle types and geographical location, but also in relation to inﬂuences of human behaviour, perceptions, attitudes, personality traits, beliefs, safety climate and organisational culture. Once identiﬁed, these high risk sectors assist the design and implementation of appropriate intervention strategies. As the implementation of intervention strategies and their subsequent results often takes time, a further advantage of appropriate baseline measures is that any countermeasures and interventions implemented can be evaluated against changes across a wide variety of performance indicators. For example, an intervention strategy may not demonstrate initial improvements in crash rates but may demonstrate improvements in cultural inﬂuences of behaviour and attitudes, which in both the short and longer term can lead to improvements in vehicle ﬂeet safety. Conclusion In summary, this chapter has highlighted some of the major issues that are inﬂuencing ﬂeet safety intervention development and implementation in Australia. The importance of undertaking a baseline measurement approach along with accurately measuring driving behaviour has been identiﬁed as a crucial element of the development and implementation of work-related road safety initiatives. It is necessary for future ﬂeet safety improvements that organisations and researchers work collaboratively to ensure that ﬂeet intervention strategies are research based and aimed at developing targeted interventions toward the numerous high risk sectors and inﬂuences on ﬂeet driver behaviour. However, it remains of concern that organisations are reluctant to adequately resource and implement ﬂeet safety interventions that have been tailored to reduce their speciﬁc work-related road safety risks. Despite such difﬁculties, continued efforts to develop, implement and evaluate effective ﬂeet safety interventions can only contribute to the reduction in the burden of work-related road trauma.

234

Driver Behaviour and Training – Volume III

References Austroads (2006). Guide to Road Safety, Part 1: Road Safety Overview. Sydney, Australia. Bibbings, R. (1997). ‘Occupational road risk: towards a management approach.’ Journal of the Institution of Occupational Safety and Health, 1, 61–75. Downs, C.G., Keigan, M., Maycock, G. and Grayson, G.B. (1999). The Safety of Fleet Car Drivers: A Review. No. 390. Crowthorne: Transport Research Laboratory. Haworth, N., Tingvall, C. and Kowadlo, N. (2000). Review of Best Practice Road Safety Initiatives in the Corporate and/or Business Environment. No. 166. Clayton: Monash University Accident Research Centre. Hayes, B.C., Bartle, S.A., and Major, D.A. (2002). ‘Climate for opportunity: a conceptual model.’ Human Resource Management Review, 12, 445–468. Joint Standing Committee on Road Safety (1997). Staysafe 36: Drivers as Workers, Vehicles as Workplaces: Issues in Fleet Management. Ninth report of the Joint Standing Committee on Road Safety of the 51st Parliament. (Report No. 9/51): Parliament of New South Wales. Katila, A., Keskinen, E., Hatakka, M. and Lapotti, S. (2003). ‘Does increased conﬁdence among novice drivers imply a decrease in safety: the effects of skid training on slippery road accidents.’ Accident Analysis and Prevention, 36(4), 543–50. Lonero L.P. and Clinton K.M. (1998). Changing Road User Behaviour: What Works, What Doesn’t. Toronto: PDE Publications. Murray, W., Newnam, S., Watson, B., Davey, J. and Schonfeld, C. (2003). Evaluating and Improving Fleet Safety in Australia. ACT: Australian Transport Safety Bureau. Parker, D., Lajunen, T. and Stradling, S. (1998). ‘Attitudinal predictors of interpersonally aggressive violations on the road.’ Transportation Research Part F, 1, 11–24. Queensland Transport (2001). Road Trafﬁc Crashes in Queensland. Brisbane: Queensland Government. Varonen, U. and Mattila, M. (2000). ‘The safety climate and its relationship to safety practices, safety of the work environment and occupational accidents in eight wood-processing companies.’ Accident Analysis and Prevention, 32, 761–9. Wheatley, K. (1997). ‘An overview of issues in work-related driving.’ In Staysafe 36: Drivers as Workers, Vehicles as Workplaces: Issues in Fleet Management. (Report No. 9/51). Ninth report of the Joint Standing Committee on Road Safety of the 51st Parliament. Sydney: Parliament of New South Wales. Wills, A. (2003). ‘Work-related driver behaviour and intentions: the inﬂuence of ﬂeet safety climate.’ Submitted honours thesis, Queensland University of Technology, Queensland Australia.

Chapter 20

Designing a Psychometrically Based Self-Assessment to Address Fleet Driver Risk Lisa Dorn and Julie Gandolﬁ Cranﬁeld University, UK Introduction Drivers of company vehicles are more likely to be involved in crashes than private motorists, even when exposure has been accounted for (Chapman, Roberts and Underwood, 2000; Broughton et al., 2003). As many as one in three road deaths may involve someone driving while at work. This group includes not just those who drive for a living – bus, coach and goods vehicle drivers – but also company car drivers driving to or from a meeting or appointment. This latter group may not see driving as the key activity of their working day. It is, however, the time when they are most at risk while at work. There are many possible reasons why this might be the case (for a review, see Lancaster and Ward, 2002). Research has suggested that at-work drivers are more vulnerable to fatigue brought about by driving for long periods of time (Horne and Reyner, 1995; Maycock, 1996). Other studies suggest that driving under time pressure to meet schedules is one of the main contributors to at-work crashes (AdamsGuppy and Guppy, 1995; Downs, Keigan, Maycock and Grayson, 1999). Several large Health and Safety Executive surveys have indicated that stress and related conditions formed the second most commonly reported group of work-related illhealth conditions affecting 420 000 people in Great Britain in 2005–2006 (www. hse.gov.uk/statistics/). It is certainly the case that company car drivers are more likely to engage in risky driving behaviours compared with that of private motorists (Downs et al., 1999; Broughton et al., 2003; Chapman, Roberts and Underwood, 2000; Parker, West, Stradling and Manstead, 1995; Stradling, 2000). The increased risk of work-related crashes may well be due to a combination of both driver stress and fatigue caused by the pressure to meet tight deadlines and schedules whilst at work. In a sense, the car has become a place of work in which their response to stress affects the way they drive. According to the Transactional Model of Driver Stress, stress outcomes are generated by the driver’s cognitive appraisals that the demands of the task stretch or exceed the driver’s capabilities and coping resources (Gulian et al., 1989). Several distinct cognitive processes contribute to an evaluation of this kind, including appraisal of task demands and personal competence, and selection and

236

Driver Behaviour and Training – Volume III

implementation of coping strategies. Two instruments have been developed with the Transactional Model as their theoretical underpinning called the Driver Stress Inventory (DSI) and the Driver Coping Questionnaire (DCQ). The DSI (Matthews et al., 1996) measures ﬁve dimensions of vulnerability to driver stress: aggression items, concerning feelings of anger and frustration and willingness to engage in risky behaviours such as tailgating and frequent overtaking; dislike of driving, associated with anxiety about driving and lack of enjoyment and conﬁdence; hazard monitoring, related to active search for danger and threats, even when driving seems routine; fatigue-proneness describes vulnerability to physical and mental fatigue symptoms; and thrill-seeking relates to enjoyment of risk and danger. The DSI attained good psychometric properties (Matthews et al., 1997) with reliable scales distinct from standard personality traits. The DSI predicts driver behaviour criteria in British, American and Japanese samples (Matthews et al., 1997; Matthews, 2002). Aggression and thrill-seeking are associated with increased rates of accident involvement, convictions for trafﬁc offences and self-reported violations and errors. Dislike of driving relates to increased risk of driving errors accompanied by fewer speeding convictions. The three dimensions most predictive of stress states are dislike of driving, aggression and fatigue proneness; they are similarly correlated with state dimensions in both simulator and ﬁeld studies (Matthews et al., 1998; Matthews, 2001; 2002). Dislike of driving is associated with both distress and worry, mediated by task-related cognitive interference and correlated with negative selfappraisal. In combination, these processes create negative moods and anxiety and have a detrimental effect on task performance (Matthews, 2001; Dorn and Matthews, 1995). Aggression and fatigue-proneness also tend to relate to poorer mood, but aggression is most strongly correlated with anger and fatigue-proneness with scales linked to low task engagement and fatigue. The DCQ (Matthews et al., 1996) was developed to assess individual differences in coping, measuring ﬁve dimensions including: confrontive coping (for example, ‘showed other drivers what I thought of them’), task-focused (for example, ‘made sure I avoided reckless actions’), emotion-focused (for example, ‘blamed myself for getting too emotional’), reappraisal (for example, ‘tried to gain something worthwhile from the drive’) and avoidance (for example, ‘cheered myself up by thinking about things unrelated to the drive’). Drivers high in confrontive coping tend to make unsafe decisions (Matthews, 2001) and it is the strongest single predictor of tailgating (Matthews, 2005), as well as violations and minor accidents (for example, Matthews et al., 1996). On the other hand, task-focused coping is a more adaptive strategy, such as deliberately controlling speed, and is more likely to be used by drivers high in hazard monitoring. So how do companies manage their at-work road risk? Mostly employees are given one-to-one skills-based driver training but behavioural responses to stress and fatigue whilst driving is not addressed with this particular intervention. Driver education is increasingly focusing on raising awareness about risk and getting drivers to evaluate their abilities (Sundstrom, 2005) and personal strengths and weaknesses (Vissers et al, 2007). Educating drivers needs to take a far broader approach and consider self evaluation methods and self-analysis (Hatakka et al, 2002) so that they can understand their personal risk better. To illustrate, Gregersen et al (1996)

Designing a Psychometrically Based Self-Assessment

237

has found stronger evidence of a safety beneﬁt amongst company car drivers using group discussion techniques compared with driver training. When it comes to educating at work drivers, we suggest that the main focus should be on how to manage driver stress and fatigue, but a driver trainer needs a way of understanding an individual’s vulnerability to driver stress and what kind of coping strategies they might employ. Coaching can then be delivered speciﬁc to their needs. Self-assessments of driver behaviour have several advantages then. Individuals can report their beliefs, feeling and behaviours in a way that the driver trainer cannot be privy to. Self-assessments direct attention towards areas for improvement. Any instrument for this purpose must be capable of determining whether the individual is claiming competence in their driving, when they may not possess it. When people provide a self-assessment they know that their responses are going to impact on them in some way. Hence, drivers may respond with what they think is an appropriate answer, rather than how they really behave. It is well known that drivers tend to rate themselves as better drivers than their peers, particularly young male drivers. However, in a review of studies, Shrauger and Osberg (1981) found that self-assessments were as good if not better than external ratings and scores. We therefore considered it important in the development of a self-assessment instrument for company car drivers to incorporate a measure of socially desirable responding, using an adapted version of the Driver Social Desirability Scale (DSDS: Lajunen and Summala, 1997). The DSDS measures two factors called impression management and self deception. The impression management scale refers to a deliberate tendency to give favourable self descriptions. The self deception scale is deﬁned as a positively biased but subjectively honest self description. In an earlier version of the DSI, impression management was negatively related to self-reported accidents, penalties, overtaking frequency, speeding and aggression and positively related to rule compliance. Self deception was correlated positively with control in trafﬁc. Hitherto, we have been unable to ﬁnd studies that have considered driver stress and coping amongst company car drivers, nor whether they are prone to socially desirable responding. This paper reports the results of two studies. Study 1 aims to report the results of a factor analysis of at-work drivers’ responses to the three instruments as potentially useful predictors of work-related driver behaviour. This will be achieved by investigating whether the factor structure of the instruments are relevant for this particular group with the aim of incorporating these instruments into one measure if this is found to be the case. Study 2 reports a validity analysis of the reﬁned measure by assessing age and sex differences in driver stress, coping and socially desirable responding. The relationship between violation and crash history and responses to the scales will also be presented to determine whether there is evidence that the reﬁned instrument is predictive of an external measure of at risk behaviour. Finally, preliminary analyses of the validity of self-assessment using two internal criteria, impression management and self deception, will be reported.

238

Driver Behaviour and Training – Volume III

Method: Study 1 Participants Three hundred and thirty-four company car drivers completed the ﬁrst part of this study. All participants were employed by companies as sales representatives. Age ranged from 18 to 74 years (Mean = 41.26, SD = 10.95). One hundred and three participants were female and 230 were male. Descriptive statistics for demographic and situational data are provided in Table 20.1. Table 20.1 Participant data

Age Mileage Hours driving per week Years licence held Points on licence Number of collisions in previous three years Number of weeks of post-licensed driver training received

Mean 41.26 17 840 7.54 18.94 0.06 0.2

SD 10.95 13 060 8.573 9.64 1.43 0.496

3.92

1.53

Procedure Three separate questionnaires were administered at one sitting taking about 20 to 40 minutes to complete online and participants were informed that their data would be conﬁdential. Questionnaire A consisted of all 48 items of the DSI and 40 newly constructed items based on in-depth interviews and a literature review of the risks of driving for work. Questionnaire B consisted of 34 of the original 35 DCQ items. The item not included was ‘Thought about the consequences of having an accident’, based on responses from a pilot study in which at-work drivers considered this item to be not appropriate. Questionnaire C incorporated 12 items based on the DSDS. Treatment of results There are two alternative analyses that could be used to analyse the questionnaire data – principal components analysis (PCA) and factor analysis (FA). The techniques are very similar, as they both aim to produce a set of linear combinations of items that represent the underlying constructs within a data set (Pallant, 2001). However, they are different in the way they do this. PCA uses all the variance in the variables as it produces the linear combinations, while FA estimates factors using a mathematical model which only analyses shared variance (Tabachnick and Fidell, 1996). Guadagnoli and Velicer (1988) reviewed the literature on PCA and Factor Analysis (FA), and reported that the solutions created by PCA and FA were largely very similar. Critics cite the increased error found in PCA to argue in favour of

Designing a Psychometrically Based Self-Assessment

239

allegedly more rigorous FA methods, but in this study the data relate to a questionnaire comprised of well-established factors measuring driver behaviour, therefore it is not necessary to estimate factors based on a mathematical model – instead, the focus is on identifying whether the existing variables group together in the same way for company car drivers as they do for commuter drivers. PCA establishes the linear components that exist within the data and how each variable contributes to a component or factor (Field, 2000). It produces a set of correlations between each item and the factor onto which it loads, which enable analysis of the extent to which an item contributes to the factor. Once a number of factors have been established, interpretation is assisted through rotation of factors. This does not alter the underlying constructs, but it presents the loadings in a more interpretable manner. Oblique rotation is used when factors are highly correlated. When the items are plotted on a graph it allows the axes to move closer together or further away, so that the factor represented by the axis can intersect the items when keeping the axes at 90 degrees would not. Orthogonal rotation keeps the axes in the same position but allows them to rotate until clusters of items have been intersected sufﬁciently to form factors. When factors are uncorrelated, the clusters are less likely to plot close to one another so keeping the axes perpendicular to one another is suitable, and produces a more interpretable factor solution. An orthogonal rotation (Varimax) was chosen for this analysis, although in the development of the original DSI an oblique rotation was used. Field (2000) suggests running the analysis using both kinds of rotation and choosing the most appropriate solution. Both procedures were carried out, and minimal differences between the two factor structures were found. When oblique and orthogonal rotations create a very similar structure, choosing the orthogonal rotation is supported by Kline (1994), who states: ‘Varimax is an excellent method of reaching orthogonal simple structure, and in many cases oblique solutions are virtually identical.’ Shaffer and Sinnett (1964) state that, ‘agreement between two such diverse methods of rotation would increase conﬁdence in the simple structure solution’ and go on to explain that in their factor analysis of ratings of personality theories, such similarity in factor structure despite different rotations provides a simple structure which is easily interpreted. Field (2000) states that an oblique rotation should only be used if there are good reasons to suppose that the underlying factors are related in theoretical terms. In the development of the original DSI, there was evidence that the factors of its forerunner, the Driver Behaviour Inventory (DBI: see Gulian, Matthews, Glendon, Davies and Debney, 1989) were highly correlated for commuter drivers. However, it is reasonable to suppose that the underlying mechanisms behind company car driving are quite different from those of commuters and the same relationships between factors should not be assumed. Results PCA used for this analysis has three main advantages: ﬁrstly to understand the structure of a group of variables; secondly to divide variables such as questionnaire items into groups relating to different constructs; and thirdly to reduce a large data

Driver Behaviour and Training – Volume III

240

set to a reasonable size without losing the essence of the data. PCA was used in the development of the original DSI (Matthews et al., 1996; 1997) and is therefore deemed appropriate for this analysis. Analyses revealed that the value produced by the Kaiser-Meyer-Olkin test of sampling adequacy (Kaiser, 1970) for the PCA exceeded 0.6 and thereby indicated that the sample was large enough to provide reliable PCA results (KMO = 0.858). Bartlett’s test of sphericity was also signiﬁcant (Sig. = < 0.0005) which indicated that the data set was suitable for PCA (Bartlett, 1954). The purpose of the PCA analyses was to test whether the original factor structure for these measures was reproduced in our data (Gandolﬁ and Dorn, 2005; Garwood and Dorn, 2005). Varimax rotation was used to provide a simple structure with easily interpretable factors. Questionnaire A The PCA for Questionnaire A showed that six factors explained 46.31 per cent of the variance. Of the six factors that emerged, ﬁve were largely replicated factors from the DSI – aggression, thrill-seeking and hazard monitoring were replicated. Fatigueproneness reversed to form fatigue resistance consistent with a factor analysis of the DSI for police drivers (Gandolﬁ and Dorn, 2005) and dislike of driving reversed to form the enjoyment of driving factor, consistent with analyses of previous DSI responses for both police drivers and bus drivers (Gandolﬁ and Dorn, 2005; Garwood and Dorn, 2005). An additional factor also emerged and was labelled work-related risk. This factor describes the extent to which company car drivers report taking risks when under time pressure. The work-related risk factor is similar to the risk inevitability factor identiﬁed in a sample of police drivers (Gandolﬁ and Dorn, 2005). An internal consistency analysis was conducted using Cronbach’s alpha coefﬁcient to establish the internal reliability of each of the six scales shown in Table 20.2. The PCA and the internal consistency analysis removed 38 of the 88 items. Table 20.2 Factor structure of Questionnaire A post-PCA

Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6

Number of items 12 items 11 items 7 items 9 items 6 items 5 items

Cronbach’s alpha 0.896 0.874 0.863 0.763 0.857 0.641

Factor label Thrill-seeking Fatigue resistance Enjoyment of driving Work-related risk Hazard monitoring Aggression

Questionnaire B The PCA of Questionnaire B revealed that ﬁve factors explained 50.44 per cent of the variance. The ﬁve factors of the original DCQ were replicated – task focused, avoidance, reappraisal, emotion focused and confrontive coping. These factors were

Designing a Psychometrically Based Self-Assessment

241

also subjected to internal consistency analysis and presented in Table 20.3. The PCA and internal consistency analysis removed nine of the 34 original items. Table 20.3 Factor structure of Questionnaire B post-PCA

Factor 1 Factor 2 Factor 3 Factor 4 Factor 5

Number of items 8 items 5 items 7 items 5 items 5 items

Cronbach’s alpha 0.860 0.847 0.756 0.699 0.727

Factor label Task focus coping Confrontive coping Avoidance coping Emotional coping Reappraisal coping

Questionnaire C The PCA of the socially desirable responding items showed that two factors explained 55.60 per cent of the variance – and labelled impression management and driver conﬁdence. The PCA and internal consistency analysis removed two of the 12 items. The Cronbach’s alphas are shown in Table 20.4. Table 20.4

Factor 1 Factor 2

Factor structure of Questionnaire C post-PCA Number of items 6 4

Cronbach’s alpha 0.840 0.852

Factor name Impression management Driver conﬁdence

Method: Study 2 Given the ﬁndings of the PCA reported in Study 1, there was a ﬁrm rationale to incorporate the remaining items to form a new psychometric instrument to measure company car driver behaviour for self-assessment and training purposes. We named this instrument the ﬂeet driver risk index (FDRI). As part of the development of the FDRI, an investigation of its validity was undertaken using a fresh sample. Participants and procedure A sample of 1089 completed the new online ﬂeet driver risk index containing 90 items. Participants ranged in age from 20 to 71 years (mean = 41.05, sd = 9.28). Eight hundred and ﬁfty-two of the participants were male and 237 were female. All participants were employed by several UK companies as sales representatives. Further demographic and situational information is provided in Table 20.5.

242

Driver Behaviour and Training – Volume III

Table 20.5 Participant data

Age Mileage Hours driving per week Years licence held Points on licence Number of collisions in previous three years Number of weeks of post-licensed driver training received

Mean 41.05 23 245 13.60 18.64 0.75 0.36

SD 9.28 15 361 12.63 8.72 1.52 0.73

3.72

1.72

Table 20.6 Gender differences in FDRI factors Factor Aggression Thrill-seeking Enjoyment of driving Hazard monitoring Driver conﬁdence

F 0.423 7.954 2.622 6.833 1.668

Sig. 0.001 0.000 0.005 0.000 0.000

Male (mean) 37.20 23.38 68.83 84.71 66.68

Male (sd) 17.02 16.71 14.91 13.17 19.39

Female (mean) 41.41 17.93 65.73 80.58 60.88

Female (sd) 17.70 14.56 15.97 14.24 19.46

Results and discussion Gender differences in FDRI factors Firstly, t-tests were conducted to establish gender differences in the FDRI factor. (See Table 20.6) Higher levels of aggression reported by female company car drivers may be due to role conﬂict with professional women having to cope with the demands of working in a competitive sales environment as well as family life. Male drivers’ higher mean scores for thrill-seeking, enjoyment of driving and hazard monitoring were consistent with previous ﬁndings for the DSI (Matthews et al., 1996) and males’ higher mean score for driver conﬁdence was also consistent with expectations (Lajunen and Summala, 1997). Age, experience and driver stress and coping Given there were signiﬁcant correlations between age and the factor scores, partial correlations controlling for age was conducted. The ﬁndings revealed that, consistent with previous research on commuter drivers (Matthews et al., 1997), aggression was strongly correlated with thrill-seeking (r = 0.37, p < 0.001) and confrontive coping (r = 0.46, p < 0. 001).

Designing a Psychometrically Based Self-Assessment

243

When controlling for age, experience was signiﬁcantly correlated with thrillseeking (r = 0.10, p < 0.001) and reappraisal coping (r = 0.13, p < 0.001), suggesting that more experienced company car drivers report higher levels of thrill-seeking. Perhaps one of the reasons that more experienced ﬂeet drivers are over-represented in crashes is due to a desire to seek excitement and danger whilst driving for work. Previous research has found that thrill-seeking is associated with increased rates of accident involvement, trafﬁc offences and self-reported violations and errors amongst commuter drivers (Matthews et al., 1996; 1997). Number of hours driving Results showed that number of hours spent driving per week was positively correlated with enjoyment of driving (r = 0.14, p < 0.001), hazard monitoring (r = 0.15, p < 0.001), fatigue resistance (r = 0.18, p < 0.001) and negatively correlated with aggression (r = –0.15, p < 0.001). It appears that the more driving undertaken in a week, the more a company car driver reports driving enjoyment, looking out for hazards, resisting the negative inﬂuences of fatigue and responding in a less hostile manner to other road users. However, hours spent driving was also positively correlated with work-related risk (r = 0.24, p < 0.001), driver conﬁdence (r = 0.09, p < 0.01) and reappraisal coping (r = 0.20, p < 0.001) indicating a greater tendency to take risks when under pressure, report higher levels of conﬁdence in driving skills and adopt ineffective coping strategies. Work related risk The new factor to emerge as a result of the PCA – work-related risk – was found to be strongly correlated with aggression (r = 0.34, p < 0.001), thrill-seeking (r = 0.47, p < 0.001), confrontive coping (r = 0.40, p < 0.001) and emotion focused coping (r = 0.36, p < 0.001), and signiﬁcantly lower levels of fatigue resistance (r = –0.39, p < 0.001) and task focus coping (r = – 0.30, p < 0.001). These results show that at-work drivers who believe that taking risks under pressure is an inevitable part of their job report greater levels of driver stress, employ ineffective coping strategies and are more vulnerable to fatigue. Socially desirable responding Impression management was found to be strongly correlated with driver conﬁdence consistent with expectations (Lajunen et al., 1997). High scores for impression management were also associated with task focused coping (r = 0.37, p < 0.001) and negatively correlated with confrontive coping (r = –0.40, p < 0.001) and thrill-seeking (r = –0.44, p < 0.001) suggesting that impression management scores are associated with a ‘safer’ proﬁle and more effective driver coping strategies supporting its use as a measure of socially desirable responding as part of the FDRI. Driver conﬁdence (a reﬁned measure of self deception from the original DSDS) was found to be strongly correlated with task focused coping, enjoyment of driving, hazard monitoring and fatigue resistance. Strong negative correlations were found

244

Driver Behaviour and Training – Volume III

between driver conﬁdence and confrontive coping and emotion focused coping. In other words, more conﬁdent drivers are reporting their behaviour in a more socially desirable way as with impression management, but perhaps through self-deception mechanisms rather than the more deliberate deception practised by high impression management scorers. We then conducted more powerful statistical tests to determine whether drivers who exhibit socially desirable responding tendencies also report safer driving behaviours and coping strategies. Participants were grouped according to whether they scored low, medium or high for impression management and for driver conﬁdence based on 33rd and 66th percentiles as cut-off points. Signiﬁcant differences between groups were found for all factors in the expected direction. High impression management scorers display a consistently ‘safer’ proﬁle compared with low scorers except driver conﬁdence and avoidance coping, which were both signiﬁcantly higher for high IM scorers. Similarly, high driver conﬁdence scorers display a consistently ‘safer’ proﬁle compared with low scorers. Full analyses and discussion is beyond the scope of this paper but will be carried out as part of the ongoing development of the FDRI. Violations Penalty points on a driver’s licence is an indication of behaviours that increase the chances of being involved in a crash and are a useful criteria for validating a self reported measure of driver behaviour. We investigated whether violation history measured in terms of number of penalty points on their driver’s licence was associated with the FDRI scales. As found in previous studies, the number of licence points correlated positively with thrill-seeking (r = 0.08, p < 0.01) and we also found a positive correlation with work-related risk (r = 0.16, p < 0.001). We explored this relationship further using ANOVA. Results revealed signiﬁcant differences for work-related risk according to violation history. Drivers with seven or more penalty points reported higher levels of risk taking due to time pressure compared with drivers with fewer penalty points (F = 9.15, p < 0.001). These ﬁndings support the use of this factor as an important determinant of at-work road risk. The negative correlation found between impression management (r = –0.12, p < 0.001) and violation history could be explained by the tendency for high scorers not to report their penalty points in the same way as they seem to report more socially desirable driving behaviours. We explored this relationship further using ANOVA and found that drivers with a higher number of penalty points reported signiﬁcantly lower levels of impression management compared with drivers with none or few. Crash involvement Using crash involvement frequency as a criterion value for validity purposes, we grouped participants according to crash history and used ANOVA to determine whether high and low accident involved drivers differed according to their FDRI scores. The results show that drivers with two or more crashes reported higher levels of aggression (F = 2.67, p < 0.05) compared with those with one or no accidents

Designing a Psychometrically Based Self-Assessment

245

– consistent with previous research on commuter drivers. Drivers with three or more crashes reported higher levels of work-related risk (F = 5.73, p < 0.001) and reappraisal coping (F = 3.72, p < 0.001) compared with drivers with two crashes or less. It appears that with multiple accidents involving company cars, drivers report increased levels of aggression and take risks in response to pressure at work and the demands of driving. They also report looking on the bright side when the demands of driving are difﬁcult, which appears to be an ineffective coping strategy in comparison to other more safety-enhancing strategies such as task-focused coping. Conclusions We ﬁnd support for previous research using the DSI, DCQ and DSDS and provide further evidence for a speciﬁc company car driver behavioural style in response to the demands of driving for work. Factor analysis of the three instruments revealed that the factor structures remained largely similar to that of drivers who do not drive for work, and support their use for understanding ﬂeet driver behaviour, stress and coping. The PCA helped to reﬁne the instrument, allowing several items to be discarded across all scales to create the ﬂeet driver risk index. The reversals of dislike of driving to form enjoyment of driving and fatigue proneness to form fatigue resistance suggest that company car driving is more similar to the kind of driving behaviour reported by professional drivers than commuter drivers (Gandolﬁ and Dorn, 2005; Garwood and Dorn, 2005). It makes intuitive sense that company car drivers report that driving is generally enjoyable, otherwise they would be unlikely to work in a capacity in which it is a large part of their working day and, like the police, company car drivers report a greater resilience to fatigue. A new factor emerged, work-related risk, which went on to show important relationships with other factors and driver behaviours. We found support for the hypothesis that company car drivers experience signiﬁcant levels of driver stress and utilise ineffective strategies such as confrontive coping. The transactional analysis of aggression, thrill-seeking and work-related risk suggests the need to direct interventions both towards negative appraisals of other drivers and motivations to take risk, replacing confrontive coping with more safetyfocused coping skills. Indeed, training of self-management skills in coping appears to be an effective treatment for driver anger (Deffenbacher et al., 2002). Both impression management and driver conﬁdence were found to be important determinants of more positive responses to other scales, conﬁrming their utility for inclusion in assessing company car driver behaviour. Perhaps drivers who lack competence fail to recognise they are incompetent (Kruger and Dunning, 1999). It has been found that when individuals are aware that their self-assessments are being validated by external measures there is an improvement in accuracy (Mabe and West, 1982). It is also the case that self-assessment is a skill that needs to be developed (Falchikov and Boud, 1989). Further research will investigate the role of impression management and driver conﬁdence in self-assessment of company car driver behaviour and how to moderate its effects.

246

Driver Behaviour and Training – Volume III

Our data suggest that educating drivers to manage the stress of driving for work should be targeted towards the speciﬁc vulnerabilities of the individual driver, using the FDRI as a self-assessment in the ﬁrst instance. To achieve this, driver trainers must have sufﬁcient competence and coaching skills to be able to judge whether a driver has achieved their training goals. Work at Cranﬁeld is ongoing to develop professional short and long courses to develop driving instructors’ competencies in how to manage behavioural, emotional and attitudinal risk. References Adams-Guppy, J.R. and Guppy, A. (1995). ‘Speeding in relation to perceptions of risk, utility and driving style by British company drivers.’ Ergonomics, 38, 2525– 35. Bartlett, M.S. (1954). ‘A note on the multiplying factors for various chi square approximations.’ Journal of the Royal Statistical Society, 16 (Series B), 296–8. Broughton, J., Baughan, C.J., Pearce, L., Smith, L. and Buckle, G. (2003). WorkRelated Road Accidents. TRL report 582. Crowthorne: Transport Research Laboratory. Chapman, P., Roberts, K. and Underwood, G. (2000). ‘A study of the accidents and behaviours of company car drivers.’ In G. B. Grayson (ed.). Behavioural Research in Road Safety X. Crowthorne: Transport Research Laboratory. Cooper, M.D. (2000). ‘Towards a model of safety culture.’ Safety Science, 36, 111– 36. Deffenbacher, J.L., Filetti, L.B., Lynch, R.S., Dahlen, E.R. and Oetting, E.R. (2002). ‘Cognitive-behavioral treatment of high anger drivers’. Behaviour Research and Therapy, 40, 895–910. Dorn, L. (2005). ‘Professional driver training and driver stress: effects on simulated driving performance.’ In G. Underwood (ed.), Trafﬁc and Transport Psychology, Elsevier. Dorn, L. and Garwood, L. (2005). ‘Development of a psychometric measure of bus driver behaviour.’ Behavioural Research in Road Safety: 14th Seminar. Department for Transport, London. Dorn, L. and Matthews, G. (1995). ‘Prediction of mood and risk appraisals from trait measures: two studies of simulated driving.’ European Journal of Personality, 9(1), 25–42. Downs, C.G., Keigan, M., Maycock, G. and Grayson, G.B. (1999). The Safety of Fleet Car Drivers: A Review. TRL Report 390. Crowthorne, Berkshire: Transport Research Laboratory. Falchikov, N. and Boud, D. (1989). ‘Student self-assessment in higher education: a meta-analysis.’ Review of Educational Research, 59(4), 395–430. Field, A. (2000). Discovering Statistics Using SPSS (2nd ed.), London: Sage. Flin, R., Mearns, K., O’ Connor, P. and Bryden, R. (2000). ‘Measuring safety climate: identifying the common features.’ Safety Science, 34, 177–92. Gandolﬁ, J. and Dorn, L. (2005). ‘Development of the Police Driver Risk Index.’ in

Designing a Psychometrically Based Self-Assessment

247

L. Dorn (ed.), Driver Behaviour and Training, vol. II. Aldershot: Ashgate. Garwood, L. and Dorn, L. (2005). ‘Stress vulnerability and choice of coping strategies in UK bus drivers.’ In L. Dorn (ed.), Driver Behaviour and Training, vol. II. Aldershot: Ashgate. Glendon, A.I., Dorn, L., Matthews, G., Gulian, E., Davies, D.R. and Debney, L.M. (1993). ‘Reliability of the Driver Behaviour Inventory.’ Ergonomics, 36, 719– 26. Gregersen, N.P., Brehmer, B. and Moren, B. (1996). ‘Road safety improvement in large companies: An experimental comparison of different measures.’ Accident Analysis and Prevention, 28, 3, 297–306. Grifﬁths, M. (1997). ‘Selecting safe vehicles: issues in vehicle crash safety.’ In Staysafe 36: Drivers as workers, vehicles as workplaces: Issues in ﬂeet management. Report No. 9/51. Ninth report of the Joint Standing Committee on Road Safety of the 51st Parliament. Sydney: Parliament of New South Wales. Guadagnoli, E. and Velicer, W.F. (1988). ‘Relation of sample size to the stability of component patterns.’ Psychological Bulletin, 103(2), 265–75. Gulian, E., Matthews, G., Glendon, A.I., Davies, D.R. and Debney, L.M. (1989a). ‘Dimensions of driver stress.’ Ergonomics, 32, 585–602. Hatakka, M., Keskinen, E., Gregersen, N.P., Glad, A. and Hernetkoski, K. (2002). ‘From control of the vehicle to personal self-control; broadening the perspectives to driver education.’ Transportation Research Part F, 5, 201–15. Haworth, N., Tingvall, C. and Kowadlo, N. (2000). Review of Best Practice Road Safety Initiatives in the Corporate and/or Business Environment. Report No. 166. Accident Research Centre, Monash University, Victoria. Horne, J.A. and Reyner, L.A. (1995). ‘Sleep related vehicle accidents.’ British Medical Journal, 310, 565–7. Kaiser, H. (1970). ‘A second generation Little Jiffy.’ Psychometrika, 35, 401–415. Kline, P. (1994). An Easy Guide to Factor Analysis, London: Routledge. Kruger, J. and Dunning, D. (1999). ‘Unskilled and unaware of it: how difﬁculties in recognizing one’s own incompetence lead to inﬂated self assessments.’ Journal of Personality and Social Psychology, 77(2), 221–32. Lajunen, T. Corry A, Summala, H. and Hartley, L. (1997). ‘Impression management and self-deception in trafﬁc behaviour inventories.’ Personality and Individual Differences, 22(3) 341–53(13). Lancaster, R. and Ward, R. (2002). The Contribution of Individual Factors to Driving Behaviour: Implications for Managing Work-Related Road Safety. Research Report. HSE Contract Research Report: HSE Books. Mabe, P.A. and West, S.G. (1982). ‘Validity of self-evaluation of ability: A review and meta-analysis.’ Journal of Applied Psychology, 67, 280–96. Matthews, G. (2001). ‘A transactional model of driver stress.’ In P.A. Hancock and P.A. Desmond (eds), Stress, Workload and Fatigue, Mahwah, NJ: Lawrence Erlbaum. Matthews, G. (2002). ‘Towards a transactional ergonomics for driver stress and fatigue.’ Theoretical Issues in Ergonomics Science, 3, 195–211. Matthews, G., Dorn, L., Hoyes, T.W., Davies, D.R., Glendon, A.I., and Taylor, R.G. (1998), ‘Driver stress and performance on a driving simulator.’ Human Factors,

248

Driver Behaviour and Training – Volume III

40, 136–49. Matthews, G., Emo, K.A. and Funke G.J. (2005). ‘The transactional model of driver stress and fatigue and its implications for driver training.’ In L. Dorn (ed.), Driver Behaviour and Training, vol. II, Aldershot: Ashgate. Matthews, G., Desmond, P.A., Joyner, L.A., Carcary, B. and Gilliland, K. (1996). ‘Validation of the Driver Stress Inventory and Driver Coping Questionnaire.’ Paper presented at the International Conference on Trafﬁc and Transport Psychology, Spain. Matthews, G., Desmond, P.A., Joyner, L.A., Carcary, B. and Gilliland, K. (1997). ‘A comprehensive questionnaire measure of driver stress and affect.’ In T. Rothengatter and E. Carbonell Vaya, (eds), Trafﬁc and Transport Psychology: Theory and Application, Amsterdam: Pergamon. Matthews, G., Dorn, L. and Glendon, A.I. (1991). ‘Personality correlates of driver stress.’ Personality and Individual Differences, 12, 6, 535–49. Matthews, G., Tsuda, A., Xin, G. and Ozeki, Y. (1999). ‘Individual differences in driver stress vulnerability in a Japanese sample.’ Ergonomics, 42(3), 401–15. Maycock, G. (1996). ‘Sleepiness and driving: the experience of UK car drivers.’ Accident Analysis and Prevention, 29(4), 453–62. Mitchell, R., Driscoll, T. and Healey, S. (2004). ‘Work-related road fatalities in Australia.’ Accident Analysis and Prevention, 36, 851–60. Pallant, J. (2001). SPSS Survival Manual, Maidenhead: Open University Press. Parker, D., West, R., Stradling, S. and Manstead, A.S.R. (1995). ‘Driving errors, driving violations and accident involvement.’ Ergonomics, 38, 1036–48. Shaffer, J.P. and Sinnett, E.R. (1964). ‘A factor analysis of Hall and Lindsey’s ratings of personality theories.’ Acta Psychologica, 22, 135–44. Shrauger, S.J. and Osberg, T.M. (1981). ‘The relative accuracy of self-prediction and judgement by others in psychological assessment.’ Psychological Bulletin, 90, 322–51. Stradling, S.G. (2000). ‘Driving as part of your work may damage your health.’ In G.B. Grayson (ed.), Behavioural Research in Road Safety IX. Crowthorne, Berkshire: Transport Research Laboratory, 1–9. Sundstrom, A. (2005). ‘Self-assessment of knowledge and abilities: a literature review.’ Report No. 54, UMEA Universitet, Sweden. Tabachnick, B.G. and Fidell, L.S. (1996). Using Multivariate Statistics (4th ed.), MA: Pearson Education. Vissers, J., Mesken, J., Roelofs, E. and Claesen, R. (2007). ‘New elements in the Dutch practical driving test: a pilot study.’ In L. Dorn (ed.), Driver Behaviour and Training, vol. III. Aldershot: Ashgate.

PART 4 Technological Interventions, Driver Behaviour and Road Safety

This page intentionally left blank

Chapter 21

Development of Multimedia Tests for Responsive Driving Erik Roelofs,1 Marieke van Onna,1 Reinoud Nägele,2 Jolieke Mesken,2 Maria Kuiken2 and Esther Cozijnsen2 1 Cito, The Netherlands 2 DHV Environment and Transportation, The Netherlands Introduction Throughout Europe, increased attention is being given to road safety issues. Policy makers are looking for new ways to address road fatalities, in particular amongst novice drivers. Like in other countries, Dutch trafﬁc safety policy is aimed at fostering safe trafﬁc participation by means of stating trafﬁc rules, maintaining the law, trafﬁc education and communicative campaigns (Vissers, 2004). In various European countries a two-phased driver training programme, including a pre-licence and post-licence phase, is now under consideration. In addition, initial driver training programmes are being further improved by using professional instructional lesson designs aimed at systematic teaching of driving tasks. On top of that, initiatives for permanent trafﬁc education have been developed. Trafﬁc education is considered to be a life-long process. These developments are inﬂuenced by a changing conception of driving and driver training. Until recently the driver’s task was conceived of as a set of elementary driving tasks pertaining to vehicle control and applying trafﬁc rules. Nowadays driving is considered as a broad domain of competence, in which the driver is expected to make independent decisions, taking into account the task environment, his or her interests and those of other trafﬁc participants. A recently issued broad taxonomy of driving competence is known as the GDE matrix (Goals for Driver Education). It is used for developing curricula for driver training at both pre- and post-licence stages. The matrix stresses the overriding signiﬁcance of the higher levels of driver behaviour with regard to accidents and the need for drivers to possess not only knowledge and skills, but also risk awareness and self-evaluation capacities at multiple levels (Hatakka, Keskinen, Gregersen, Glad and Hernetkoski, 2002; Hatakka et al., 2003). This shift towards competence-based learning induces new forms of driving assessment, as is the case for many forms of (vocational) education (Dierick and Dochy, 2001). Until recently, congruent with traditional views on driver training, assessment instruments took the form of rather isolated testing of knowledge about trafﬁc rules in theory tests and driving skills in predominantly examiner-led road tests. As in other countries, in the Netherlands it is increasingly advocated to additionally assess higher order aspects of driving as represented in the GDE-matrix:

252

Driver Behaviour and Training – Volume III

risk tolerance; reﬂection on own driving behaviour; and hazard perception (Vissers, 2004). In 2007, Cito, a leading Dutch testing and assessment company, formed a consortium with several other Dutch organisations: consultancy and engineering group DHV, specialised in sustainable environmental planning, including transport; the Foundation for Scientiﬁc Research of Trafﬁc Safety (SWOV); and TNO, a unit for Human Factors, whose mission is to apply scientiﬁc knowledge with the aim of strengthening the innovative power of industry and government. This consortium, named Cito Drive, is aimed at developing instruments for the assessment of driving competence in initial and post-licence driver training. The current chapter discusses the development of computer-based tests to assess the competence of drivers in perceiving and detecting relevant cues in a dynamic driving environment and in selecting an adequate response to changes in the complex task environment. In the next section we discuss some theoretical notions and research ﬁndings on essential cognitive aspects of the driving tasks. After that, a general model of competent performance is discussed and applied to the assessment of driving competence, followed by a section on the construction of the instruments for the assessment of responsive driving. The chapter closes with a short discussion. Perceiving the right cues and making the right decisions: change detection and action selection It can be expected that safe drivers are good at detecting and perceiving relevant cues in their environment at an early stage. Instead of continually responding to stimuli, a competent driver is able to anticipate situations. The degree to which a driver can operate in advance of real time events is a hallmark of experience and competence (Drummond, 1990). Driving is a complex task in a dynamic environment and it would be impossible to see and attend to all of the relevant cues in the environment while driving. The ability and capacity of drivers to attend to and consciously process information is limited. Through training and experience drivers apparently can develop ways to compensate for these cognitive limitations. Several very important dimensions can be distinguished (Drummond, 1990): •

•

•

Competent drivers develop more valid expectations about how events can turn out and become better in estimating the likelihood of occurrence of possible scenarios. Competent drivers develop optimal searching-strategies for relevant cues so they only need to recognise and process information that is essential for the driving task in speciﬁc situations. Competent drivers develop more efﬁcient ways of processing information when they operate on a more automated level. Automation leads, for example, to chunking of relevant information into larger, more holistic and complex units and to reﬁnement of the schematas or programmes that direct the task performance.

Development of Multimedia Tests for Responsive Driving

253

Many research ﬁndings support the relationship between insufﬁcient perceptualcognitive competences and the involvement in accidents. Underestimation of the complexity and demands of the task in speciﬁc situations (inadequate detection and perception skills) is often identiﬁed as a very important factor related to the accident proneness of young drivers (Engström et al., 2003, Joly et al., 1997; Gregersen, 1996; Gregersen and Bjurulf, 1996; Preusser, Ferguson and Williams, 1998; McGwin and Brown, 1999; Twisk, 1998a; 1998b; Summala, 1987). Differences in detection and perception skills between novice and experienced drivers are also reported. Experienced drivers are better and faster in detecting relevant cues than novice drivers (McKenna and Crick, 1991; 1994). Results of different studies show that with experience, drivers are more able to integrate information quickly and consider hazardousness as a holistic attribute of the driving environment (Deery, 1999). Other research has shown that search strategies of experienced drivers were found to be more ﬂexible and they had a wider horizontal search pattern (Falkmer and Gregersen, in Engström et al., 2003). Novice drivers also ﬁxate more on stationary objects, whereas experienced drivers ﬁxate more on moving objects (Mayhew and Simpson, 1996). Novice drivers tend to be less ﬂexible in their scanning behaviour and have a stronger ﬁxation on cues that are irrelevant in these speciﬁc complex situations (Chapman, Underwood and Roberts, in Engström et al., 2003). In sum, research suggests that novice drivers perceive task situations less holistically and detect relevant cues less quickly and efﬁciently than experienced drivers. This seems to stem, at least in part, from novice drivers adopting less efﬁcient information gathering strategies (Deery, 1999). Groeger’s four facet model of driving behaviour There are many models of driving behaviour that address the role of cognitive competence that underlie driver performance and responding to changes in a complex task environment. Most of them describe or explain the driving task from one perspective: either from a perspective of risk or threat, from a performance perspective or from a social psychological perspective. In a recently developed and validated model of driving behaviour, an attempt is made to integrate these factors (Mesken, 2006). Groeger identiﬁes in his model (see Figure 21.1) four main facets in describing the cognitive processes that are involved in driving and in responding to changes in driving situations (Groeger, 2000; Grayson et al., 2003). The four facets explain how drivers adapt to changes in the task environment: an event may happen that implies an interruption of the drivers’ goals, the driver evaluates whether this event has consequences for the future and for future goals, and depending on that, he decides on action planning and implementation (Mesken, 2006, 31).

The ﬁrst process, implied goal interruption, detects changes which imply some discontinuity in currently active goals. In other words, to start the process of a response to a change, some sort of discrepancy must be detected between the environment as is and the expected environment. Discrepancies are especially relevant when they are predictive of being endangered (Grayson et al., 2003). Originally this facet was seen

254

Driver Behaviour and Training – Volume III

in the context of hazard detection, but Groeger states that implied goal interruption reﬂects processes ‘that are also more generally involved in aspects of driving where other goals and expectations might be threatened’ (Groeger, 2000, 192). This process is described in more detail later as situation awareness. Environment

IMPLIED GOAL INTERRUPTION

APPRAISAL OF FUTURE INTERRUPTION

ACTION PLANNING

IMPLEMENTATION

Hypothetical Forward links Hypothetical Feedback links

Figure 21.1 Processes involved in responding to interruptions Source: Groeger, 2000.

The second process, appraisal of future interruption, appraises or evaluates detected changes. This process may be affected by personality characteristics (danger seeking, impulsiveness), beliefs about one’s self, level of conﬁdence (selfefﬁcacy), feelings of control (self-assessment of ability), the inherent controllability of the interruption, the seriousness of the consequences and what this means for the individual, and the assessment of the likelihood of the consequences if no action is taken. The outcome of this process is to act or not to act (Grayson et al., 2003). The third process, action planning, selects and constructs the most appropriate form of action under the actual circumstances. During driving, most actions are routinely performed under the control of a hierarchy of goals. There is a discontinuity in the normal hierarchy of control when an interruption is detected and appraised as sufﬁciently serious to require a response. Action planning depends on various types of reaction speed (simple reaction time, choice reaction time and un-alerted reaction time), mental processing (information-processing speed, inspection time and intelligence) and attention selection and inhibition (allocation). The fourth process of Groegers’ model of driving behaviour, implementation, is responsible for the implementation of the selected actions. Individual differences in motor performance and reaction times are expected to inﬂuence the successful implementation.

Development of Multimedia Tests for Responsive Driving

255

Situation awareness According to Groeger’s model the ﬁrst process, implied goal interruption, consists of several elements. Those elements are judgement of trafﬁc scenes, hazard perception, judgement of speed and trajectories, estimating time-to-arrival, and style of decision making. Clearly, these elements are related to perceiving relevant information, interpreting this information and predicting or projecting the implications of this information in the future. This suggests a close relation with the concept of situation awareness. The term ‘situation awareness’ comes from the aviation domain. Although the recognised need for situation awareness can be traced back to World War I, an increased interest in the construct began in the 1980s and accelerated through the 1990s (Endsley, 2000). The emphasis on situation awareness and its measurement originated in the ﬁghter aircraft domain, but the use and application of the term has rapidly spread to other domains like medicine, space operations and also to the domain of driving (Endsley, 2000). Endsley (1995a; 1995b) deﬁnes situation awareness as ‘the perception of elements in the environment within a volume of time and space, the comprehension of their meaning, and the projection of their status in the near future’. Consequently, Endsley developed a model of situation awareness in human decision making. According to this model, situation awareness encompasses three levels of information processing: •

•

•

Perception: perceiving and attending to important cues or elements in the environment. On this level scanning plays a crucial role. Scanning the environment should be systematic and active and leads to selectively attending to relevant cues. The quality of scanning increases when a driver regularly shifts focus. An analysis of mistakes made by airplane pilots show that three-quarters of all situation awareness mistakes are made during the ﬁrst stage of perceiving. Overlooking the relevant information is the most common mistake. Other mistakes include perceiving the wrong information, difﬁculty in detecting the information, the available information is insufﬁcient and memory mistakes (Jones and Endsley, 1996).• Comprehension: combining, interpreting, storing and retaining information. The second level is concerned with the correct interpretation and integration of the perceived information, and determining their relevance to the person’s goals. This is mostly done based on experience with previous task situations that are very similar to the one encountered. A situation is recognised, compared with the ones that are stored in the long-term memory and the best matching situation is activated (template-matching). A common mistake is that this match is based only on one or two cues or variables. This leads to misinterpretation of the situation. Of all situation awareness mistakes made by pilots, 20 per cent are made during this stage (Jones and Endsley, 1996). Projection: anticipating future events and their implications. The third level is being ahead of what might happen and being prepared to operate in advance of real time events. Drivers who are able to anticipate the course of events have

256

Driver Behaviour and Training – Volume III

more time and space to take action themselves or leave more time and space for others to operate in case of critical escalation of a situation. To anticipate knowledge and insight is crucial. It is impossible to predict something you don’t know. The things we have no knowledge of are the most threatening, because we don’t see them coming or can’t imagine a potential scenario as being possible. Of all situation awareness mistakes made by pilots, ﬁve per cent are made during this stage (Jones and Endsley, 1996). Novice drivers base their predictions too much on a deterministic vision of the course of events in trafﬁc situations and too little on a realistic vision (Harrison, in Vlakveld, 2005). Loss of situation awareness often results in poor performance and even causes devastating accidents (Jones and Endsley, 1996). Recent research sheds some light on the role experience has in the development of situation awareness (Whelan, Senserrick, Groeger, Triggs and Hosking, 2004). Novice drivers have less short-term remembrance of important details in the trafﬁc environment and less recollection of the behaviour of other road users. Also, novice drivers were more easily distracted when trying to recall the position of other road-users. Training and testing of responsive driving The aspects of driving Cito Drive focuses on tests for responsive driving such as a change detection and action selection. Change detection can be learned and tested. For example, a short training programme for increased knowledge, scanning and anticipation has shown to improve the quality of search strategies and patterns. Drivers were informed about their typical pattern of visual search and the need for scanning multiple locations in the visual ﬁeld for potential dangers. They found notable changes in search patterns for horizontal spread compared with a control group (Chapman, Underwood and Roberts, in Engström et al., 2003). Responding to changes, by selecting an appropriate action, is part of the action planning in Groeger’s (2000) model. Successful action planning can only take place if the detection of changes is successful. In addition, the comprehension, projection and evaluation of cues has to be successful in order to plan an appropriate action. Both change detection and response selection can be viewed as cognitive aspects of driving competence. Assessment of competence The concept of competence has gained much attention in the many ﬁelds of education: general, vocational, higher education and learning on the job. On the basis of a study of dozens of deﬁnitions of competence (for example, Bunk, 1994; Spencer and Spencer, 1993; Parry, 1996) Mulder (2001) comes to the following deﬁnition of competence: ‘competence is the ability of a person or organisation to achieve particular levels of performance’ (76). According to Mulder, the competencies of individuals consist of: (a) integrated action proﬁciencies, (b) which are made up of

Development of Multimedia Tests for Responsive Driving

257

clusters of knowledge structures, (c) cognitive, interactive, emotional, and where necessary psychomotor skills, (d) and attitudes and values which are necessary for, (e) the performance of tasks, (f) solving of problems, (g) and more generally the ability to function in a particular occupation, (h) a particular organisation, (i) a particular position, (j) or a particular role. In the cited literature on professional competence, it is emphasised that competence is not achieved per se, but must be kept up with changing requirements from society. The same holds for driving competence. The complex task of driving does not mean the same now as it meant 40 years ago, and will not bear the same meaning in the future. An eclectic model of driving competence It is widely accepted that theoretical models of task performance facilitate the construction of assessment instruments. These models can be taxonomic in nature or functional: the ﬁrst describing domains of tasks and subtasks of the target domain; the latter describing the psychological processes underlying task performance. Theoretical models enable assessors to set up arguments to arrive at sound decisions about candidates, starting from observations of performance (Kane, 1992). As Kane (1992) states on valid assessment: it is not so much the test score itself that is valid or reliable, but rather the interpretations of test scores that can be more or less valid. The psychological theory on driving behaviour is explained in the previous section. The construction of assessment instruments of cognitive aspects of driving competence will be based on the general model of competent performance developed by Roelofs and Sanders (2003; 2007). The model is now serving the construction of various instruments for professional competence, including driving performance (cf. Nägele, Vissers and Roelofs, 2006). The model for assessment of driver competence (Figure 21.2) consists of ﬁve parts, the ﬁrst being the driver’s task environment, consisting of strongly differing and speciﬁc task situations in which the other parts are embedded. The other parts are: Decision making, the process that involves action planning, selection and (mental) construction of the most appropriate form of action in the circumstances; the Actions that are actually carried out in speciﬁc situations; Consequences of the actions for other people, materials, or processes within a situation; and Base, which comprises the knowledge, skills, and attitudes of a candidate. The assessment of the change detection and action selection aspects of driving competence can be classiﬁed as assessments of the decision making process. Aspects of assessment of driver competence In driver competence assessment, the ﬁrst question to answer is for what purpose drivers are assessed. Depending on the answer of the purpose question, different directions can be chosen for the content and type of assessment instrument. A ﬁrst purpose is supporting driver learning. This can relate to candidates who learn to drive different types of vehicles (motorcycle, car, bus, trucks). The assessments will usually involve informal observations of actual driving behaviour resulting

Driver Behaviour and Training – Volume III

258

Drivers’ task environment

Base • • • •

Knowledge Skills Conceptions – Attitudes Personality characteristics

Actions Consequences

Decision making

Figure 21.2 An eclectic model for the assessment of driving competence Source: Based on Roelofs and Sanders, 2003; 2007. in immediate feedback and support for the pupils. As in school situations, many types of teacher-made assessments can be part of driver training. The results of these assessments can be used for adapting driving instruction to the pupils’ level of performance. A second purpose is to make high-stakes decisions about drivers, like licensing. A third purpose is scientiﬁc in nature, acquiring knowledge about the nature of the driving task and all kinds of variables inﬂuencing driving performance. A ﬁnal purpose is to monitor and evaluate trafﬁc policy. Not only the purpose of a test can vary, but also the methods for assessing professional competence may vary. They can be distinguished along different dimensions (Kane, 1992; Roelofs and Sanders, 2003): •

•

•

Authenticity of the assessment situation and task structure. The assessment can take place in the real task environment, in simulated or in symbolised situations. In simulated situations, aspects of the natural trafﬁc situation are manipulated or omitted, as in the case of driving simulators. When driving competence is measured in symbolised trafﬁc situations, a candidate is presented with a hypothetical situation, with a speciﬁc accompanying question or assignment. Type of evaluative data collection. The assessment data yielded in various situations can be statistics that are free of subjective judgment, (self) observations using diaries, logs or observation forms, or judgmental data. Actors involved in the data collection and judgment. In all types of data collection the drivers themselves are involved anyway. Other actors who score and judge the quality of the task performance may be educators, colleagues, external assessors or research scholars. In addition to human actors, ICT can also function as an assessor, although it uses inserted man-made performance standards.

Development of Multimedia Tests for Responsive Driving

259

Research questions The current paper discusses the development of computer-based tests (CBTs) to assess important aspects of cognitive competence that underlie driver performance involving responding to changes in a complex task environment. Two CBTs will be developed: one is designed to assess a driver’s perception and detection of relevant cues for changes (change detection) and the other is designed to assess a driver’s competence to select an adequate response to changes in a complex task environment (action selection). Both change detection and response selection are viewed as important cognitive aspects of driver competence and are related to two out of four facets of Groeger’s model of driving behaviour (‘implied goal interruption’ and ‘action planning’). The tests are developed to support training of novice and experienced drivers. The results of the assessments can be used for adapting driving instruction to the pupils’ level of performance or to advise experienced drivers about speciﬁc types of additional training. The cognitive aspects of driver competence are assessed in symbolised trafﬁc situations (a video). A driver is presented with digitally recorded situations accompanied by a question. The tasks are highly explicit: whether they detected changes in given situations and how the candidate would handle certain tasks in a given situation. They have to give a selective response. The assessment is free of subjective judgment; the evaluation is carried out automatically by means of performance standards. The argument to choose computer-based tests is that it is expected to be impossible to develop an on-road assessment instrument that covers enough critical situations for solid conclusions and at the same time avoids dangerous situations. Using CBTs, it is expected that this combination is a possible option. This leads to the following research questions: 1. Is it possible to construct computer-based tests for situation attentiveness (perception and detection of relevant cues) and action selection that cover a wide range of competence levels? Can the tests discriminate between pupils preparing for their driving licence and experienced drivers? 2. In which way are scores on the CBTs related to external criteria, for example to road assessments of defensive driving and driving simulator assessments of defensive driving? We expect that inexperienced drivers will perform more poorly on the tests of situation attentiveness and action selection than experienced drivers. In addition, we expect the correlations between the two CBTs will be substantial. More speciﬁcally, a hierarchical relationship is expected between the performance measures used. Good performance on the situation attentive test is necessary for good performance on the action selection test. Subsequently, good performance on the action selection test is necessary for good performance during on-road and in simulator performance.

260

Driver Behaviour and Training – Volume III

Construction of instruments The development of a test is usually a multi-phase procedure. First, a trial version is developed and a small pilot study is carried out to ensure that candidates understand the instructions and show the intended test behaviour. If the pilot results are far from desirable, a new trial version may be developed and tested before proceeding to the next construction phase. After the pilot study, a large-scale study is carried out. In this study, the potential test is administered to the target group. The ﬁnal test is speciﬁed after analysing the results of the large-scale study. Before publishing the ﬁnal test, a manual also has to be written and the results of the large-scale study have to be documented. The present test development project is still in the ﬁrst indicated construction phase. In this phase, the adjustment of the test to the test purpose, the intended assessment methods and the theoretical framework is central. Test material Both Cito Drive responsive driving tests will be computer based. That is, the items are presented on a monitor screen and the candidate has to select responses with a mouse click. In both tests, a stimulus consists of a short movie, lasting between 5 and 15 seconds. It is displayed on a monitor and shows what a driver sees when looking through the front windshield. The rear-view mirror view is displayed in a small area centred in the upper area of the windshield view. The response format of both tests differs, resulting in the assessment of two different aspects of driving competence. The CBTs start with an instruction and trial items, to enable the candidate to practice with the item format. After the instruction, a candidate has to respond to 40 items. Including instruction and trial items, each CBT takes 25 to 30 minutes. A response format for the change detection test which requires limited language capacities is to ask the candidate to indicate a spot where the most attention is needed at the moment that the movie freezes or stops. Another possibility is to ask the opinion on statements like ‘a car is driving behind me now’. When the movie freezes, the task in the action selection test is to indicate which driver behaviour would be most appropriate at this moment. The response options are restricted to ‘Brake’, ‘Accelerate’, ‘Move left’, ‘Move right’, ‘Slow down gently’ and ‘Continue as before’. Item selection Two main options for item selection in the ﬁnal tests are possible: a ﬁxed format or a computerised adaptive test format (see for example Sands, Waters and McBride, 1997). The ﬁxed format is most common: all candidates have to respond to the same item set, or several equivalent sets are available. In a computerised adaptive test, the presented items are selected during the test after a small number of ﬁxed items, depending on the responses of the candidate to the previous items. This allows for adaptation of the difﬁculty of the items to the level of the candidate.

Development of Multimedia Tests for Responsive Driving

261

A wide range of trafﬁc situations has to be covered in each CBT. The situations which are ﬁlmed for the items are classiﬁed in ﬁve different trafﬁc environments or road types (SWOV, 2006). 1. 2. 3. 4. 5.

urban access road in residential areas (30 km/h zones) urban distributor road (50 km/h) rural access road in residential areas (60 km/h zones) rural distributor road (80 km/h) through-road (100 / 120 km/h)

The main function of environments 1 and 3 is living. Relatively high numbers of pedestrians and bicycles can be expected. The function of environments 2 and 4 is mainly transport. Environment 5 comprises the motorways and regional throughroads. For each environment, multiple item-stems have to be ﬁlmed. In each ﬁnal CBT, all environments have to be presented several times to cover the entire range of possible trafﬁc situations. Finally, the item-stems or stimuli have to be ﬁlmed in the country where the ﬁnal test is published. In this project, this means that the stimuli will be ﬁlmed in the Netherlands. This requirement ensures that the candidate is familiar with the presented trafﬁc situations. Also, the camera car has to obey the local trafﬁc rules, in order to avoid confusing the candidates. Participants The CBTs are intended for two main populations of candidates: pupil drivers and experienced drivers. The pupil driver population consists of pupils who vary in the amount of training they have had. The least experienced pupils have recently started driving instruction, mastering barely the technical aspects of car driving. The most experienced pupils are measured on the day they pass the driving test. The experienced driver population consists of lease car drivers, among others. Lease car drivers usually have several years of driving experience and they have a large annual mileage. These experienced drivers may be invited to participate in a post-licence training by the lease company. Statistical analysis All items in the CBTs have one response which is considered ‘correct’ and all other responses are considered ‘incorrect’. The one-parameter logistic model (OPLM, Verhelst and Glas, 1995), which is one of the item response theory (IRT) models, can be used to estimate the competence levels of all participants and the item parameters. IRT models have several advantages over more traditional analysis methods, like classical test theory. IRT models are probabilistic models, separating the observed responses from the underlying latent trait or competence level. For each candidate, the competence level can be estimated and the difﬁculty level of each item can be

262

Driver Behaviour and Training – Volume III

estimated. Also, missing data do not complicate the estimation process and IRT models allow for easy construction of parallel tests. Finally, computerised adaptive testing is not possible without IRT models. Discussion Responsive driving is an element of driving performance which is especially relevant for driver training and examination. Young novice drivers are slower in perception of changes than more experienced drivers, and their response selection is often inadequate (Grayson et al., 2003). Several studies have shown that experienced drivers are also more competent in detecting hazards and that adequate perception of complex, hazardous situations can be trained (McKenna and Crick, 1991; 1994; 1997). However, hazards are uncommon and thus the detection of them might not be based on training or on driving experience (Groeger, 2000). Improvement of hazard perception may ‘be a result of learning to identify situations in combination with automation of other driving tasks, thus reducing the mental workload and leaving more mental capacity for the hazard detection task’ (Engström et al., 2003, 33). The focus in the development of driver assessment instruments in the Cito Drive project is not explicitly on hazards, but supports Groeger when he states that implied goal interruption reﬂects processes ‘that are also more generally involved in aspects of driving where other goals and expectations might be threatened’ (Groeger, 2000, 192). Therefore, the competence assessment which is developed in this project will be based on the cognitive abilities that underlie driver performance in accomplishing multiple driver goals. In particular, we focus on the competence of detecting and responding to changes that imply some discontinuity in currently active goals. References Bunk, G.P. (1994). ‘Competentie-ontwikkeling in de Duitse beroepsopleidingen.’ [Development of competence in German vocational education]. Beroepsopleiding, 1, 8–15. Deery, H.A. (1999). ‘Hazard and risk perception among young novice drivers.’ Journal of Safety Research, 30(4), 225–35. Dierick, S. and Dochy, F. (2001). ‘New lines in edumetrics: new forms of assessment lead to new assessment criteria.’ Studies in Educational Evaluation, 27(4), 307– 29 Drummond, A.E. (1990). ‘An overview of novice driver performance issues: a literature review.’ Monash University Accident Research Centre. Endsley, M.R. (1995a). ‘Measurement of situation awareness in dynamic systems.’ Human Factors, 37(1), 65–84. Endsley, M.R. (1995b). ‘Towards a theory of situation awareness in dynamic systems.’ Human Factors, 37(1), 32–64. Endsley, M.R. (2000). ‘Theoretical underpinnings of situation awareness: a critical review.’ In M.R. Endsley and D.J. Garland (eds). Situation Awareness Analysis

Development of Multimedia Tests for Responsive Driving

263

and Measurement. Mahwah, NJ: Lawrence Erlbaum Associates. Engström, I., Gregersen, N.P., Hernetkoski, K., Keskinen, E. and Nyberg, A. (2003). ‘Young novice driver education and training.’ Literature review, VTI-rapport 491A., Swedish National Road and Transport Research Institute, Linköping. Grayson, G.B., Groeger, J.A., Maycock, G., Hammond, S.M. and Field, D.T. (2003). Risk, Hazard Perception and Perceived Control. Report TRL560. Transport Research Laboratory TRL, Crowthorne. Gregersen, N.P. (1996). ‘Young drivers’ overestimation of their own skill – an experiment on the relation between training strategy and skill.’ Accident Analysis and Prevention 28, 243–50. Gregersen, N.P. and Bjurulf, P. (1996). ‘Young novice drivers: towards a model of their accident involvement.’ Accident Analysis and Prevention 28, 229–41. Groeger, J.A. (2000). Understanding Driving: Applying Cognitive Psychology to a Complex Everyday Task. Routledge: Psychology Press. Hatakka, M., Keskinen, E., Baugahn, C., Goldenbeld, Ch., Gregersen, N.P., Groot, H., Siegrist, S. Willmes-Lenz, G. and Winkelbauer, M. (2003). Basic Driver Training: New Models. Turku, University of Turku. Hatakka, M., Keskinen, E., Gregersen, N.P., Glad, A. and Hernetkoski, K. (2002). ‘From control of the vehicle to personal self-control: broadening the perspectives to driver education.’ Transportation Research Part F, 5, 201–15. Joly, P., Gilbert, M., Paquette, M., Perraton, F., Bergeron, J. (1997). ‘Performance in a driving simulator and intention to take risk on the road among learner and experienced young drivers.’ In Brookhuis, De Waard and Weikert (eds), Simulators and Trafﬁc Psychology. HFES Europe Chapter. Jones, D.G. and Endsley, M.R. (1996). ‘Sources of situation awareness errors in aviation.’ Aviation, Space and Environmental Medicine, 67(6), 507–12. Kane, M.T. (1992). ‘An argument-based approach to validity.’ Psychological Bulletin, 112(3), 527–35. Mayhew, D.R. and Simpson, H.M. (1996). Effectiveness and Role of Driver Education and Training in a Graduated Licensing System. Trafﬁc Injury Research Foundation, Ottawa. McGwin, G. and Brown, D.B. (1999). ‘Characteristics of trafﬁc crashes among young, middle-aged and older drivers.’ Accident Analysis and Prevention, 31, 181–98. McKenna, F.P. and Crick, J. (1991). ‘Experience and expertise in hazard perception.’ In G.B. Grayson and J.F. Lester (eds), Behavioural Research in Road Safety. Crowthorne, TRL Limited. McKenna, F.P. and Crick, J. (1994). Hazard Perception in Drivers: A Methodology for Testing and Training. Report CR313. Crowthorne, TRL Limited. McKenna, F.P. and Crick, J. (1997). Developments in Hazard Perception. TRL Report 297. Crowthorne, TRL Limited. Mesken, J. (2006). Determinants and Consequences of Drivers’ Emotions. Dissertation Rijksuniversiteit Groningen. Mulder, M. (2001). Competentieontwikkeling in organisaties. Perspectieven en praktijk. [Development of Competence in Organisations. Perspectives and

264

Driver Behaviour and Training – Volume III

Practice]. The Hague: Elsevier bedrijfsinformatie. Nägele, R., Vissers, J. and Roelofs, E. (2006). Herziening WRM. Een model voor competentiegericht examineren [Revision of the Law for Driving Education for Motor Vehicles: A Model for Competence Based Certiﬁcation Testing] Amersfoort/Arnhem: DHV/Cito. Parry, S.B. (1996). ‘The quest for competencies.’ Training, July, 48–56. Preusser, D.F., Ferguson, S.A. and Williams, A.F. (1998). ‘The effect of teenage passengers on the fatal crash risk of teenage drivers.’ Accident Analysis and Prevention, 30, 217–22. Roelofs, E.C. and Sanders, P. (2003). ‘Beoordeling van docentcompetenties.’ [Assessment of teacher competences]. In M. Mulder, R.Wesselink, H.Biemans, H.Nieuwenhuis and R. Poell (eds), Bevoegd, maar ook bekwaam? Groningen: Wolters-Noordhoff, 277–99. Roelofs, E.C. and Sanders, P. (2007). ‘Towards a framework for assessment of teacher competence.’ European Journal for Vocational Training (forthcoming). Sands, W.A., Waters, B.K. and McBride, J.R. (eds). (1997). Computerized Adaptive Testing. Washington, DC: American Psychological Association. Spencer, L.M. and Spencer, S.M. (1993). Competence at Work: Models for Superior Performance. New York: Wiley and Sons. Summala, H. (1987). ‘Young driver accidents: risk taking or failure of skills?’ Alcohol, Drugs and Driving, 3, nr. 3–4. SWOV Institute for Road Safety Research (2006). ‘Advancing sustainable safety: national road safety exploration for 2005–2020; the advanced vision in brief.’ http://www.swov.nl/rapport/DMDV/Advancing_Sustainable_Safety_brief.pdf; May 14, 2007. Twisk, D.A.M. (1998a). ‘Kansrijke maatregelen voor beginnende bestuurders. Eindrapport: uitgangspunten, effectiviteit en uitvoerbaarheid’ [Promising regulations for novice drivers. Final report: assumptions, effectivity and practical applicability]. Rapportnummer R-98-63. Leidschendam, SWOV. Twisk, D.A.M. (1998b). ‘Verkeersonveiligheid van jonge bestuurders in de periode 1985–1994’ [Trafﬁc unsafety of the novice driver in the period 1985–1994]. Rapportnummer R-98-18. Leidschendam, Stichting Wetenschappelijk Onderzoek Verkeersveiligheid SWOV. Verhelst, N.D. and Glas, C.A.W. (1995). ‘The one-parameter logistic model.’ In G.H. Fischer and I.W. Molenaar (eds), Rasch Models: Their Foundations, Recent Developments and Applications. New York: Springer. Vissers, J. (2004). ‘Modernisering rijexamens, Probleemanalyse beginnende bestuurders’ [Modernizing driving certiﬁcation tests, problem analysis of beginning drivers]. Report number TT04-021. Veenendaal: Trafﬁc Test. Vlakveld, W. (2005). ‘Jonge beginnende automobilisten, hun ongevalsrisico en maatregelen om dit terug te dringen. Een literatuurstudie’ [Young novice car drivers, their accident risk and regulations to reduce this. A literature study]. Leidschendam, Stichting Wetenschappelijk Onderzoek Verkeersveiligheid SWOV. Whelan, M., Senserrick, T., Groeger, J., Triggs, T. and Hosking, S. (2004). Learner Driver Experience Project. Report No. 221, Victoria, Monash University Accident Research Centre.

Chapter 22

The Effect of Simulation Training on Novice Driver Accident Rates R. Wade Allen, George D. Park and Marcia L. Cook Systems Technology, Inc. Introduction The premise of this research project was that simulator training can reduce novice driver accident rates when compared with the accident rates of traditionally trained drivers. The simulator training included orientation and knowledge presentation and multiple exposures to simulator drives containing a high rate of hazardous and time critical encounters with trafﬁc, pedestrians, signalised intersections, varying roadway conﬁgurations and trafﬁc control devices (signs and markings). Driving scenarios were 12–15 minutes long and subjects had to complete six scenarios before being considered for graduation. Graduation was judged on a number of performance criteria (accidents, speeding violations, turn indicator usage, and so on) and subjects could meet the graduation criteria after six trials. They were given up to three additional trials to pass the graduation criteria. About 79 per cent of the subjects were able to meet the graduation criteria by the ninth trial. The overall training objectives were: (1) transmit knowledge relevant to driving (rules of the road, trafﬁc control devices, hazard perception, and so on); (2) train situation awareness and hazard perception skills; (3) train decision making and appropriate control response skills under time pressure. Background Simulators with three levels of ﬁdelity were used in this research. As illustrated in Figure 22.1 the simulator conﬁgurations included: (1) a desktop, single monitor or narrow ﬁeld of view conﬁguration (NFOVD); (2) a desktop, three monitor or wide ﬁeld of view conﬁguration (WFOVD); (3) an instrumented cab with projected wide ﬁeld of view display (WFOVC). The ﬁrst phase of this project was reported on by Allen, Rosenthal et al. (2003), and the detailed training results were reported by Allen, Cook and Park (2005). Data on the accident rates of our simulator trained subjects were obtained from the California State Department of Motor Vehicles (DMV). The analysis of this data is reported in this paper and compared with general teen accident rates reported elsewhere (Janke, Masten et al., 2003; Mayhew, Simpson and Pak, 2003).

266

Driver Behaviour and Training – Volume III

NFOVD: Single monitor, desktop; 45° ﬁeld of view; 50% image size; sidewinder game controls; iconic side view mirrors; two school districts; students in driver education classes; one trained in computer lab; second trained in back of classroom. WFOVD: Three monitor, desktop; 135° ﬁeld of view; 50% image size; sidewinder game controls; real image side-view mirrors; students recruited at local DMV ofﬁce; trained in laboratory environment and supervised by researchers WFOVC: Three channel projected image; 135° ﬁeld of view; 100% image size; instrumented cab; real image side-view mirrors; students recruited at local DMV ofﬁce; trained in laboratory environment and supervised by researchers

Figure 22.1 Simulator conﬁgurations and deployment milieu

Previous studies have failed to show the efﬁcacy of traditional driver education and training, although it still has popular appeal (Williams and Ferguson, 2004; Hedlund and Compton, 2005). For this reason many states in the US have instituted graduated driver licensing (GDL), which restricts novice driver exposure and the conditions under which they are trained. Studies have shown that GDL seems to reduce novice driver accident rates (Hedlund and Compton, 2005) and it is being advocated for general application. However, GDL does not directly inﬂuence the acquisition of skills required for safe driving, including situation awareness, hazard perception, decision making and psychomotor control, all of which must be applied under time pressure in critical situations in order to avoid accidents. In military training, simulation is accepted as an essential tool for teaching situation awareness, hazard perception, psychomotor skills and coping with malfunctions (Fletcher, 1998). Given that there is a signiﬁcant concern for driver education (NTSB, 2003), it is relevant to consider the application of driving simulation to driver education. The US Center for Disease Control (CDC) has looked

The Effect of Simulation Training on Novice Driver Accident Rates

267

at the causes of injury and death for teenagers, and found that automobile crashes by far outweigh all other considerations as documented in NHTSA (2003). Based on this epidemiological ﬁnding, the CDC commissioned the research described herein to determine the efﬁcacy of simulator training on novice driver accident rates. Methods Training system An automated training system was developed that would log in subjects, establish a database, administer orientation material, administer driving simulator scenarios, record driving performance, and compare subject data to criteria that would determine acceptable training performance (Allen, Guibert et al., 2006). After logging in subjects were administered orientation material that presented information necessary for driving the simulation, including: trafﬁc control devices (signs, signals and markings), rules of the road, lane changing, turning and use of turn indicators, and hazard recognition, situation awareness and defensive driving. The orientation ended with information on the performance scoring system and use of the driving controls. Driving simulation The driving simulation has been described elsewhere (Allen, Rosenthal et al., 2002). The ﬁrst exposure was a familiarisation run which slowly introduced the student to steering and speed control, then intersections with trafﬁc control devices, and ﬁnally trafﬁc and pedestrian conﬂicts. After the familiarisation run the training system presented the student with standardised training scenarios. The students were presented with six 12–15 minute training scenarios with performance scores displayed at the end of each run. Performance was evaluated at the end of the sixth scenario, and if the students met the performance criteria (for example, no accidents, no tickets and acceptable average speed) then they were graduated. If not, the application presented up to three additional trials. On any additional trial the students were graduated if they met the performance criteria. If the students drove all of the additional trials and did not meet the performance criteria, they were admonished to drive carefully in the future and acknowledged for their participation. Driving scenarios Driving scenarios were created with a scenario deﬁnition language (SDL) that allowed the speciﬁcation and control of critical hazards (Park, Rosenthal and Aponso, 2004). The SDL allowed driving scenarios to be conveniently described in terms of roadway alignment, and include events for trafﬁc control devices (signals, signs and markings), roadside objects, trafﬁc and pedestrians. In addition the temporal properties of the trafﬁc, trafﬁc signals and pedestrians were triggered relative to the subject’s own vehicle in order to control the severity of hazard conﬂicts. The SDL also allowed

268

Driver Behaviour and Training – Volume III

situations to be counter balanced between scenarios so that subjects could not learn patterns of upcoming events over successive trials. The application was designed to select a different variation every time a given subject was administered an additional run. The characteristics of the driving scenarios, including critical events, have been described previously (Allen, Park and Cook, 2007). The SDL also allowed speciﬁcation of performance measures. Performance measures included elements such as lane and speed deviations, speed limit and trafﬁc signal violations, accidents, run completion time and median time to collision for all vehicle and pedestrian encounters. These performance measures were used as graduation criteria (that is, no accidents, no violations, and nominal values for lane and speed deviations and median time to collision). Scenario characteristics The driving scenarios were designed to be about 10–15 minutes in length depending on the driver’s speed. Participants drove along routes that covered distances of approximately 34 000 feet in length. At a speed of 45 mph the scenarios could be completed in about 8.5 minutes. There were sections where subjects could go faster, and sections where subjects had to slow or stop for intersections, trafﬁc and pedestrians which extended the driving time to roughly 12–15 minutes. The scenarios involve 155 approaching vehicles and 107 interacting vehicles going in the same direction as the subjects. The vehicles were relatively evenly distributed throughout the scenarios. There were also 67 pedestrians distributed throughout the scenarios. The trigger times for pedestrians moving in front of the subject’s vehicle were designed to present challenging decisions to subjects (for example, typical time to encounter of three to six seconds). The signal light trigger timings were designed to give relatively critical stop or go decisions (for example, typical time to intersection of three to six seconds). Curve severity was designed to be challenging as subjects went faster than the general speed limit of 45 mph. Subject population and training sites Subjects trained in two research laboratories and two school districts as summarised in Figure 22.1. The research laboratory subjects were recruited from local Department of Motor Vehicle ofﬁces, while the high school district subjects were all registered in high school driver education classes. The population had a slight bias towards females, and there was some variation the total number of subjects driving each conﬁguration as shown in Figure 22.2. Table 22.1 shows the participant distribution by simulator conﬁguration and gender. The recruitment of subjects was limited by time and logistics and in each case we attempted to maximise the number of subjects.

The Effect of Simulation Training on Novice Driver Accident Rates

269

Number of participants

250 200 150 100 50 0

14

15

16

17

18

Females

35

126

81

56

19

Males

20

111

67 Age (years)

25

14

Figure 22.2 Subject population by age and gender Table 22.1 Subject population by simulator conﬁguration and gender Simulator conﬁguration

Recruiting method

Number of subjects

Males

Females

Wide FOV vehicle cab

DMV ofﬁce

159

59

100

Wide FOV desktop

DMV ofﬁce

180

74

106

Single monitor desktop

High school drivers ed.

215

104

111

554

237

317

Totals

Accident results Accident data for our student population were obtained from the state of California Department of Motor Vehicles. Figure 22.3 shows time distribution for training date to licensure date. The median subjects were generally receiving their licences about eight months from their training completion date. The distribution tails off signiﬁcantly, however, so that some subjects were waiting two years or more before getting their licence. The wide ﬁeld of view (WFOVC, WFOVD) groups have the tightest distributions, with the single monitor high school group (NFOVD) having a wider time range to licensure. Figure 22.4 shows the distribution of accidents over time. For total accidents across all training groups in Figure 22.4(a) we see that a signiﬁcant number occur prior to licensure. After licensure the rate is high for the ﬁrst three months, then falls off but the rate of occurrence is clearly sporadic. This is consistent with novice driver accident rates reported elsewhere (Mayhew, Simpson and Pak, 2003). The accidents after licensure are broken down by simulator training group in Figure 22.4(b), where we again see a signiﬁcant rate in the ﬁrst few months of driving. Beyond six months

Driver Behaviour and Training – Volume III

270 50

WFOVD

WFOVC

Number participants

40 NFOVD

30

20

10

0 -8

-2

4

10

16

22

28

34

40

Simulator training date to licensure date

Figure 22.3 Time distributions for subject licensure date Number of accidents

30 25 20 15 10 5 0 -36

-30

-24

-18

-12

-6 3 9 15 Months from licensure

21

27

33

39

a) Total accident count for 3 month intervals

Number of accidents

8 Wide FOV instrumented cab

Wide FOV desktop

Single monitor desktop

6 4 2 0 1

6

11

16

21

26

31

36

Months from licensure

b) Accident count for each simulator configuration

Figure 22.4 Accidents relative to licensure accident occurrence is somewhat sporadic which will require special analysis for comparative purposes. Figure 22.5 compares the accident rate (number of accidents per number of drivers) of the simulator training groups against previously published general novice

The Effect of Simulation Training on Novice Driver Accident Rates

271

driver accident rates over the ﬁrst year of driving. The simulator trained subjects have lower accident rates (accidents per driver), however, there are problems with this comparison. The databases from the literature are fairly large and consistent over time. The accident database for our subjects has a dwindling number of subjects as illustrated in Figure 22.6, which shows the number of subjects in our subject population as a function of time beyond licensure. The subject population size falls off with time due to the length of time each subject waited to be licensed. Due to the decline in the subject population over time we computed accident rates on a monthly basis given the subject population at a given point in time. The 50 per cent point of each distribution as shown in Figure 22.6 was arbitrarily chosen as the limit of the length of time over which we will carry out this analysis. Figure 22.7 compares the cumulative accident rate of our simulator training groups with previously published data. The accident rates for the simulator trained subjects were calculated on a month by month basis by taking the accident that month shown in Figure 22.4(b) and dividing by the subject population size shown in Figure 22.6. The cumulative accident rates for the data from the literature (Janke, Masten et al., 2003; Mayhew, Simpson and Pak, 2003) were computed from their reported data. In Figure 22.7 we see that the wide FOV conﬁgurations show a lower accumulation rate of accidents than the previously published data, and the wide FOV and vehicle cab gave the lowest cumulative accident rates. Comparison of the cumulative accident rates in Figure 22.7 was carried out with linear regression analysis in which the intercept gives the initial accident rate and the slope gives the average accident rate beyond the initial period. The data are summarised in Table 22.2, which compares the intercepts, slopes and conﬁdence intervals for each of the simulation conﬁguration groups with general novice driver accident rates from the literature. Note that the accident rates from the two studies in the literature (Janke, Masten et al., 2003; Mayhew, Simpson and Pak, 2003) are quite comparable. The Janke data is from California, USA while the Mayhew data is

10

Accident rate (%)

8 6 4 2 0 WFOVD

WFOVC

NFOVD

Janke

Mayhew

Display configuration

Figure 22.5 Accident rates for each simulator conﬁguration compared with previously published North American accident rates

Driver Behaviour and Training – Volume III

272

Number of subjects

250 WFOVC

200

WFOVD

NFOVD

150 50% 100

50%

50 50 0 0

10

20

30

40

50

Months since licensure

Figure 22.6 Number of subjects as a function of time beyond licensure

Accident rate (% accidents/driver)

30%

Mayhew

25%

Janke WFOVC

20%

WFOVD NFOVD

15%

10%

5%

0% 0

5

10

15

20

25

30

35

40

Months from licensure

Figure 22.7 Cumulative accident rate plots of simulator training groups as compared with rates from the literature Source: Janke, Masten et al., 2003; Mayhew, Simpson and Pak, 2003.

from Nova Scotia, Canada. The Mayhew data were reported on a monthly basis over the ﬁrst two years of driving, while the Janke data were reported on a yearly basis over the ﬁrst four years of driving. The slopes for the data from the literature give an accident rate of 0.0070 accidents per driver per month or 0.084 accidents per driver per year which is consistent with Janke’s reported ﬁrst four year rates of 9.10, 8.88, 8.45 and 7.68 accidents per year per hundred drivers respectively (Table 3 in Janke, Masten et al., 2003).

The Effect of Simulation Training on Novice Driver Accident Rates

273

In Table 22.2 and Figure 22.8 the regression slope generally describes the accident rate beyond the initial few months. Here we see that the simulator conﬁguration groups have lower regression slope accident rates than the general novice driver accident rates reported in the literature. The subject group trained with the wide FOV and vehicle cab conﬁguration had an accident rate that was about one-third (34.3 per cent) of the general novice driver accident rates from the literature (Janke, Masten et al., 2003; Mayhew, Simpson and Pak 2003). The wide FOV desktop group had an accident rate that was 75 per cent of the accident rate from the literature. The single monitor group had an accident rate of 87.1 per cent of the rate reported in the literature. The conﬁdence intervals in Table 22.2 show that the accident rate differences between the simulator training data sets are generally reliable. In Table 22.2 and Figure 22.8, the intercept of the regression function general describes the initial accident rate over the ﬁrst few months. Here we see that the single monitor simulation group (NFOVD) had a signiﬁcantly higher initial accident rate than all other data sets, and this result is reliable based on non-overlapping conﬁdence intervals. This is also apparent in the cumulative accident rate data plotted in Figure 22.7. The narrow FOV group initial accident rate is over two times greater than the other data sets as indicated by the large intercept. This high initial accident rate is also evident in the accident counts shown in Figure 22.4. Discussion These results have implications for training simulator display conﬁgurations and more broadly for simulator ﬁdelity. At a minimum it would appear that full sized projected displays are signiﬁcantly superior in their training value to miniﬁed monitor presentations. The wide ﬁeld of view also seems to be important, as the groups trained on the single monitor display only had some slight improvement in accident rate. Previous analysis of the training data has shown some downsides to Table 22.2 Accident rate regression analysis Data Set Mayhew Janke

R2 0.993 0.999

WFOV cab

0.918

WFOV desktop

0.986

NFOV desktop

0.968

Coefﬁcients

P-value Lower 95% Upper 95%

Intercept

0.0164

0.000

0.0128

0.0199

Slope

0.0070

0.000

0.0067

0.0072

Intercept

0.0104

0.187

-0.0122

0.0329

Slope

0.0070

0.001

0.0063

0.0076

Intercept

0.0178

0.000

0.0130

0.0225

Slope

0.0024

0.000

0.0021

0.0026

Intercept

0.0080

0.001

0.0035

0.0124

Slope

0.0053

0.000

0.0051

0.0056

Intercept

0.0354

0.000

0.0283

0.0424

Slope

0.0061

0.000

0.0056

0.0066

Driver Behaviour and Training – Volume III

274

the single monitor training where drivers seem to drive faster than with the wide ﬁeld of view displays which could relate to peripheral cues (Park, Allen et al., 2005). 0.008 Slope

Accident rate

0.006

0.004

0.002

0.000 WFOVC

WFOVD

NFOVD

Janke

Mayhew

Janke

Mayhew

Data Set

a) Slope

0.040 Intercept

Accident rate

0.030

0.020

0.010

0.000 WFOVC

WFOVD

NFOVD

Data set

b) Intercept

Figure 22.8 Cumulative accident rate regression analysis trends The value of the instrumented cab and surround is unknown at this time, and is a signiﬁcant confounding variable. It would appear at a minimum that simulators for driver training should have wide aspect ratio displays scaled and positioned to give real world image sizing and including real image side view mirrors (not icons as presented by the single monitor desktop conﬁguration herein). There are other confounding variables with each of the simulator training groups involving their selection, supervision and administration of training as listed in Figure 22.1. The simulator training with the higher ﬁdelity conﬁgurations occurred in research laboratories and was administered by research staff, and subjects were recruited at

The Effect of Simulation Training on Novice Driver Accident Rates

275

local California Department of Motor Vehicle ofﬁces. The single monitor systems were deployed in high schools as part of their driver education classes. Students and teachers did report enthusiasm for the simulator training system. Conclusion The analysis presented herein shows that simulator training may lower teen driver accident rates. The analysis also shows that simulator ﬁdelity may have a signiﬁcant inﬂuence. The current results do not lead to a deﬁnite description of an appropriate simulator conﬁguration, but near full size imagery is deﬁnitely indicated. Full size wide aspect ratio images may provide an affordable answer to this requirement and can be provided by ﬂat panel and projected displays in the future. Acknowledgement This research was funding by the US Centers for Disease Control and Prevention under research grant number 5 R44 CE00111-03 entitled ‘A PC Based Low Cost Simulator for Driver’s Education’. References Allen, R.W., Cook, M.L. and Park, G.D. (2005). ‘Novice driver performance improvement with simulator training.’ Proceedings of the 2nd International Conference on Driver Behaviour and Training, Edinburgh, UK, 15–17 November. Allen, R.W., Guibert, M.R., et al. (2006). ‘A user conﬁgurable PC platform for driver assessment and training.’ Proceedings of the Driving Simulation Conference Asia/ Paciﬁc, Tsukuba, Japan, May/June 2006. Allen, R.W., Park, G.D. and Cook, M.L. (2007). ‘The effect of display conﬁguration on the training value of novice driver simulators.’ Proceedings of the Image 2007 Conference, Scottsdale, AZ, July. Allen, R.W., Rosenthal, T.J., et al. (2002). ‘A low cost platform for training and evaluating driver behaviour.’ Proceedings of the Driving Simulation Conference, DSC 2002, Paris, France, September. Allen, R.W., Rosenthal, T.J., et al. (2003). ‘Experience with a low cost, PC based system for young driver training.’ Proceedings of the 1st International Conference on Driver Behaviour and Training, Stratford-upon-Avon, UK, 10–12 November. Fletcher, J.D. (1998). ‘Measuring the cost, effectiveness, and value of simulation used for military training.’ Proceedings of the Simulation Industry Association of Australia, http://www.siaa.asn.au/get/2410061680.pdf. Hedlund, J. and Compton, R. (2005). ‘Graduated driver licensing research in 2004 and 2005.’ Journal of Safety Research, 36, 109–19. Ivancic, K. and Hesketh, B. (2000). ‘Learning from errors in a driving simulation:

276

Driver Behaviour and Training – Volume III

effects on driving skill and self-conﬁdence.’ Ergonomics, 43(12), 1966–84. Janke, M.K., Masten, S.V., et al. (2003). ‘Teen and senior drivers.’ California Dept. of Motor Vehicles Report CAL-DMV-RSS-03-194, July. Joung, W. and Hesketh, B. (2006). ‘Using “war stories” to train for adaptive performance: is it better to learn from error or success.’ Applied Psychology: An International Review, 55(2), 282–302. Mayhew, D.R., Simpson, H.M. and Pak, A. (2003). ‘Changes in collision rates among novice drivers during the ﬁrst months of driving.’ Accident Analysis and Prevention, (35), 683–91. National Transportation Safety Board (2003). Report of Proceedings, Public Forum on Driver Education and Training, October 28–29, 2003, NTSB/RP-05/01, PB2005-917003, Notation 633A. NHTSA (2003). ‘Trafﬁc safety facts 2002 young drivers.’ NHTSA National Center for Statistics and Analysis, Washington, D.C., DOT HS 809 619, 2003. Park, G.D., Cook, M.L., et al. (2006). ‘Automated assessment and training of novice drivers.’ Advances in Transportation Studies an International Journal, 2006 Special Issue, December, 87–96, Aracne, Rome, Italy. Park, G., Rosenthal, T.J. and Aponso, B.L. (2004). ‘Developing driving scenarios for research, training and clinical applications.’ Advances in Transportation Studies an International Journal, 2004 Special Issue, December, 19–28, Aracne, Rome, Italy. Williams, A.F. and Ferguson, S.A. (2004). ‘Driver education renaissance?’ Injury Prevention, (10), 4–7.

Chapter 23

Driving Experience and Simulation of Accident Scenarios Catherine Berthelon, Claudine Nachtergaële and Isabelle Aillerie Institut National de Recherche sur les Transports et leur Sécurité, France Introduction Novice drivers have a strong probability of being involved in accidents, but the interactions between driving experience and age are hard to analyse. In fact, beginner drivers are often young and predisposed to taking risks, which can be seen in the fact that, for the same amount of time taken to earn their driving licence, beginners aged 20 years and over have fewer accidents than beginners under the age of 20 (Deery, 1999; McCarrt, Shabanova and Leaf, 2003; Williams, 2003). The probability of being involved in an accident is also connected to the driver’s gender, with an over-representation of young men (Williams, 2003; Monarrez-Espino, Hasselberg and Laﬂamme, 2006). The effect of experience may be linked to the high-level cognitive activities required in dealing with difﬁcult situations (Ranney, 1994; Gregersen and Bjurulf, 1996; Fisher et al., 2002), but more elementary aspects of performance may also vary. Thus, a certain level of perceptive expertise is characterised by a better ability to foresee events from the context and to detect cues that are relevant to this prediction (Mourant and Rockwell, 1972; Crundall, Underwood and Chapman, 1999; Underwood et al., 2002; Underwood et al., 2003; Berthelon, Mestre and Taramino, 1995). Concerning driving skills themselves, while basic vehicle control and trafﬁc rules can be learnt quite quickly (Hall and West, 1996), experience calls on the acquisition of complex coordination of motor and perceptive activities (Mac Leod, 1987), which can be seen in the fact that experienced drivers show more effective integration and coordination than beginner drivers (Salthouse and Somberg, 1982; Mayhew, Simpson and Pak, 2003). Thus, research carried out in this ﬁeld suggests that the performances of beginner drivers are inferior to those of experienced drivers, which is corroborated by the fact that the strong probability of young people being involved in an accident drops as the time needed to obtain the driving licence increases (Turner, McClure and Pirozzo, 2004; Williams, 2003). To address the over-involvement of novice drivers in trafﬁc accidents, France adopted a reform of the driver training system in 1988 with the possibility of taking Early Driver Training (AAC – Apprentissage Anticipé de la Conduite) at the age of 16. AAC is a step in the system for gradual access to driving. This driver training includes initial training in driving schools after which the level is assessed to be equivalent to those of young people who follow the traditional training to obtain

278

Driver Behaviour and Training – Volume III

their licence. The learner acquires additional experience consisting of driving with an adult for at least 3 000 km, along with educational meetings organised with the driving school. In any case, the driving test is taken no earlier than the age of 18. Although the success rate among those who choose the AAC when taking their driver test appears to have decreased since 1997, it is still much higher than for those who follow the traditional driver training approach. Drivers who follow this type of training could be less likely to be involved in accidents than drivers who follow the traditional training, and for whom the probability of an accident is high in the ﬁrst months after receiving their licence (Mayhew et al., 2003; Sagberg and Bjørnskau, 2006). We shall focus part of the work presented here on the consequences that this training could have on driving behaviour itself. The objective is therefore to determine whether AAC favours the ability to perceive and foresee the behaviour of other users and enables the learner driver to develop behaviour that is closer to that of an experienced driver following the traditional driver training approach. Today simulators can be used to reproduce vehicle dynamics. Drivers have also been found to exhibit behaviours close to those observed in reality, due to the realism of the simulation. Simulators reproduce extreme conditions that would be difﬁcult, or even impossible, to put into practice under normal conditions of trafﬁc and safety. The methodology presented below aims to assess the relevance of introducing scenarios identiﬁed as generating accidents in the simulators and to assess whether the simulator discriminates between drivers according to experience and driver training method. Method Subjects The subjects are males who are not sensitive to travel sickness. The beginner drivers took traditional driver training (DCT; N = 12) or early driver training (DCA; N = 7). Their average age is, respectively, 19.6 years (δ = 1.4) and 18.4 years (δ = 0.5). They had passed their driver test within the last two months and do not have their own personal vehicle. The experienced drivers (EXP; N = 11) have an average age of 24.5 years (δ = 2.9) and have had their driving licence and a vehicle for more than three years. Experimental system and task The subjects sit at an interactive driving station comprising one quarter of a vehicle with a ﬁxed base (Figure 23.1). The image projection surface underlies an angular opening that spans 58° horizontally and 49° vertically. The images are generated at a frequency of approximately 30 Hz.

Driving Experience and Simulation of Accident Scenarios

279

Figure 23.1. The pedestrian shoots out from the right 2.4 seconds before the driver crosses his trajectory Each subject undergoes three rounds of the circuit, lasting approximately seven minutes for each round. Some autonomous vehicles with non-conﬂicting behaviours are driving on the circuit. The ﬁrst round is an opportunity to get used to the simulator commands. The scenarios are introduced randomly in the second and third rounds to avoid any order effect. The recommendation is to follow the road signs indicating the town centre and to drive at a speed of approximately 50 km/h. No indications are given suggesting that unexpected events will occur. Description of the scenarios and their implementation In road safety literature, the scenario concept concerns a group of accidents presenting similarities from the point of view of the chain of events and relations of causality in the different phases leading to the collision (Fleury and Brenac, 2001). The sequential analysis method used to group the accidents in the form of scenarios is based on a segmentation of their progression in several phases, thus providing a characterisation of the prototypical situations that generate accidents (Brenac and Fleury, 1999; Brenac, Nachtergaële and Reigner, 2003; Brenac et al., 1996), making it possible to undertake a selection for use in driving simulators. Once the selection is made, the scenarios are implemented with Detailed Accident Study data from INRETS/MA, which constitute reference situations. These highly precise data provide the description of the infrastructure and time positioning of the vehicles and/or pedestrians involved in a scenario (Ferrandez et al., 1995; Perrin et al., 2004). The scenarios implemented can also be introduced into the simulator by the Modelling, SImulation and driving Simulator team at INRETS (MSIS – Modélisations, SImulations et Simulateurs de conduite). The ‘hidden pedestrian crossing’ scenario corresponds to approximately 19 per cent in terms of the population involved in pedestrian bodily injury accidents in France (Brenac, Nachtergaële and Reigner, 2003). A vehicle is travelling in an urban area on a straight main through-road, generally not in an intersection. A pedestrian

280

Driver Behaviour and Training – Volume III

is getting ready to cross the road outside a pedestrian crossing. The pedestrian, initially hidden from the driver’s view, usually by a parked vehicle, starts to cross as the driver arrives. The driver does not detect the pedestrian or does so very late. The pedestrian does not see the vehicle. In our simulation, the pedestrian is hidden by a bus parked on the right-hand side of the carriageway; he suddenly crosses the carriageway in front of the bus as the driver is arriving from his left (Figure 23.1). The pedestrian appears in the driver’s ﬁeld of vision 2.4 seconds before he reaches him. This conﬁguration remains effective only if the driver does not take any manoeuvre to avoid him. The ‘sudden stop at a trafﬁc light’ scenario corresponds to 36 accident cases (Brenac et al., 1996; Blancher, Brenac and Fleury, 1998). The driver is driving in a queue on a straight two-way carriageway. Vehicles and heavy trucks are driving ahead of their car, blocking the upcoming T-junction and the trafﬁc light regulating the junction. This is the ﬁrst trafﬁc light on their itinerary. The light turns red, and the vehicles in front of the subject’s car brake at a distance of approximately 25m to stop at the light. The driver is taken by surprise and brakes too late. To implement this scenario, we have deﬁned a queue of three vehicles, the ﬁrst of which is a bus hiding the upcoming junction and trafﬁc light. The speed of the queue depends on that of the subject, so that the distance between the subject and the last vehicle in the queue is 25 m over a distance of approximately 500 m. The bus, followed by the second vehicle, turns left at the junction. The vehicle preceding the subject brakes for 2.9 seconds to stop at the trafﬁc light. Two other scenarios are directly inspired by the accident data available at INRETS/ MA. They correspond to situations that are less representative in accidentology but which are nonetheless typical of what all drivers encounter on the road. In the ‘vehicle overtaking and pulling back into the lane’ scenario, the subject is driving in the right-hand lane of a straight urban main road with three lanes, two lanes being allocated to their travel direction. A vehicle overtakes the subject; its speed is 10 km/h greater than theirs. When this vehicle is positioned 20 m in front of the subject, it starts to pull back in front of them while slowing down; three seconds later, it is completely back in the right-hand lane (therefore in front of the subject) and adopts a speed of 30 km/h. The ‘vehicle pulling out of parking’ scenario corresponds to the following situation: a vehicle parked on the right-hand side pulls out of its parking space as the subject has just turned left at an intersection and is speeding up. This manoeuvre is performed when the subject is 20 m away. Results The variables analysed concern the speed and lateral position of the drivers and their response time when confronted with the event. Inter-vehicle times (TIV) were recorded in all scenarios except for ‘hidden pedestrian crossing’. The variables are analysed using analysis of variance (ANOVA), given the large number the standard deviations cannot be illustrated. Only signiﬁcant results are presented here.

Driving Experience and Simulation of Accident Scenarios

281

‘Hidden pedestrian crossing’ scenario Response times following the appearance of the pedestrian are signiﬁcantly shorter for the EXPs than for the DCAs when releasing the accelerator pedal, but does not differ signiﬁcantly for stepping on the brake pedal (see Table 23.1). Table 23.1

Hidden pedestrian crossing

Stop at a trafﬁc light

Vehicle overtaking and pulling back in Vehicle pulling out of parking

Average response times as a function of driving experience and scenario; standard deviations in parentheses Accelerator (s) EXP DCT DCA N= N= N=5 11 12 0.79 0.64 0.7 (0.08) (0.1) (0.23) N= 11 1.09 (0.65) N=8 1.77 (0.36)

N= 12 1.06 (0.6) N= 10 1.58 (0.79)

N= 10 0.61 (0.21)

N= 12 1.32 (0.78)

N=3 1.03 (0.48) N=5 1.26 (0.41)

N=7 0.79 (0.33)

T EXP < DCA T= –2.77 p<.01 NS

EXP > DCA T= 2.38 p < 0.04 EXP < DCT; T= –2.77 p < 0.01

EXP N= 11 0.98 (0.14)

Brake (s) DCT N= 12 1.11 (0.38)

T

N= 10 2.08 (1.28) N=6 2.83 (0.79)

N= 11 1.85 (0.81) N=9 2.51 (0.59)

N=7 2.09 (1.29)

NS

N=4 2.06 (0.54)

NS

N= 10 0.93 (0.24)

N=8 1.29 (0.40)

N=7 1.14 (0.34)

DCA N=7 0.96 (0.19)

NS

EXP < DCT T= –2.35 p < 0.03

Time inﬂuences speed (F[97,2619] = 165.65; p < 0.0001) and lateral positions (F[97,2522] = 5.75; p < 0.0001). Speeds start to drop 0.7 seconds after the pedestrian becomes visible. The decrease continues for 3.5 seconds, then the speed stabilises for one second and the subjects accelerate. Swerving toward the right of the carriageway becomes signiﬁcant approximately three seconds after the pedestrian becomes visible. The swerving increases signiﬁcantly for two seconds, then the lateral positions remain stable (see Figure 23.2). The interaction between the group and time (F[194,2522] = 1.72; p < 0.0001) shows that the swerve toward the right is due to the EXPs who start swerving two seconds after the pedestrian becomes visible. The swerving increases for 1.5 seconds and its maximum amplitude is 43 cm. The lateral position of DCTs and DCAs does not change signiﬁcantly during progression (see Figure 23.3).

Driver Behaviour and Training – Volume III

282

*

** ***

16

lateral position

14 190

12 10

180

8 170

6 4

160

speed

Speed (m/s)

Lateral position (cm)

200

2 150

0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7

Time(s)

Figure 23.2 Average speeds and lateral positions as a function of time ‘Sudden stop at a trafﬁc light’ scenario The subjects’ response times do not vary signiﬁcantly as a function of driving experience (Table 23.1). Speeds evolve over time (F[139,3753] = 55.74; p < 0.0001). They decrease signiﬁcantly 0.5 seconds after the vehicle being followed brakes; the decrease continues to the end of the recording (see Figure 23.4). The lateral positions and the TIVs do not evolve signiﬁcantly as a function of time. EXP

Lateral position (cm)

*

** ***

210

DCT

200

DCA

190 180 170 160 150 140 0 0.5 1 1.5 2 2.5 3

3.5 4 4.5 5 5.5 6 6.5 7

Time(s)

Figure 23.3 Average lateral positions as a function of time and driving experience

Driving Experience and Simulation of Accident Scenarios

lateral position

10

175 8

172

6

169 166

4

163

2

160

speed

Speed (m/s)

Lateral position (cm)

12 178

283

0 0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

Time(s)

Figure 23.4 Average speeds and lateral positions as a function of time ‘Vehicle overtaking and pulling back into the lane’ scenario The analyses for this scenario were performed on 24 subjects. Response times do not vary signiﬁcantly as a function of driving experience (see Table 23.1). Speeds and lateral positions evolve as a function of time (F[139,2919] = 62.20; p < 0.0001 and F[139,2919] = 1.98; p < 0.0001, respectively). Speeds decrease signiﬁcantly one second after the start of the manoeuvre of the adverse vehicle; the decrease lasts 4.5 seconds, then speeds stabilise. The subjects swerve towards the left 1.5 seconds after the vehicle is totally back in the lane (see Figure 23.5).

Lateral position (cm)

*

**

100 50

16

5 subjects

14

average

12

19 subjects

10

0

8 6

-50

4 -100

Speed (m/s)

150

speed

2

-150

0 0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

5.5

6

6.5

7

Time(s)

Figure 23.5 Average speeds and lateral positions as a function of time We can see that ﬁve subjects (three EXPs and two DCTs) swerved into the left lane when the other vehicle pulled into their lane in front of them (Figure 23.5). The TIV recording stopped prematurely for these ﬁve subjects. The TIV analysis was therefore performed on 19 subjects.

Driver Behaviour and Training – Volume III

284

4 3.5

TIV (s)

3 2.5

**

*

2 1.5 1 0.5 0 0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

Time(s)

Figure 23.6 Average TIVs as a function of time

12

3.5

* **

3.0 2.5

8

2.0 6 1.5 4

TIV (s)

Speed (m/s)

10

1.0

2

0.5

0

0.0 0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

speed TIV

5

Time(s)

Lateral position (cm)

Figure 23.7 Average speeds and TIVs as a function of time 60 70

left

80 90 100

EXP

110

DCT DCA

120 130

right

140 0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

Time(s)

Figure 23.8 Average lateral position as a function of time and driving experience

Driving Experience and Simulation of Accident Scenarios

285

TIVs evolve over time (F[99.1485] = 20.48; p < 0.0001). They decrease signiﬁcantly 0.5 seconds after the vehicle starts to merge back into the right-hand lane, then stabilise until the vehicle is totally back in the lane (duration: 1.5 seconds), then increase for the last three seconds (see Figure 23.6). ‘Vehicle pulling out of parking’ scenario The DCTs take more time than the EXPs to release the accelerator. Response times for pressing the brake pedal do not vary signiﬁcantly as a function of the group but certain subjects did not brake, notably three DCTs who hit the vehicle merging into their lane (see Table 23.1). Speeds and TIVs evolve signiﬁcantly as the subjects progress (F[139.3753] = 38.26; p < 0.0001 and F[139.3197] = 21.55; p < 0.0001, respectively). A signiﬁcant drop in speed occurs during the ﬁrst second and lasts for 2.5 seconds; the speed stabilises for 0.5 seconds, then increases again. TIVs decrease 0.5 seconds after the vehicle starts to pull out of the parking space and for one second, then they stabilise for 0.5 seconds, increasing signiﬁcantly for the remaining three seconds (see Figure 23.7). Lateral positions do not evolve over time but we can observe an interaction between the group and time on this variable (F[278.3753] = 3.91; p < 0.0001). DCTs are positioned signiﬁcantly more to the right of their lane than EXPs for the ﬁrst two seconds (23 cm difference). They swerve to the left of the lane 0.5 seconds after the other vehicle starts pulling out of the parking space; this swerve lasts one second, then their position stabilises. EXPs swerve to the right of the lane four seconds after the vehicle starts to pull out of the parking space. The lateral position of DCAs is stable (see Figure 23.8). Singular behaviours These behaviours can be attributed to ﬁve EXPs who often avoid conﬂict by swerving laterally and ﬁve DCTs who often cause collisions. We can observe an absence of singular behaviours amongst the DCAs (Table 23.2). Moreover, ﬁve of the six subjects who collide (four DCTs and one EXP) drive at high speeds and consequently have short TIVs, which reduces the possibility of avoiding collision simply by braking. The ‘vehicle pulling out of parking’ scenario leads to collision for three DCTs and the ‘vehicle overtaking and pulling back into the lane’ scenario is the only one for which no collision is observed, but it leads to more swerving into the left-hand lane (see Table 23.2). One DCT collides in two scenarios; he drives at high speed, leading to a TIV that is too short to avoid collision. One EXP systematically swerves laterally, with or without braking, in three scenarios out of four. He is the only driver to try and avoid the pedestrian to the left and by doing so collided with the pedestrian. We have therefore hypothesised that he was sensation seeking, ampliﬁed by driving in simulation. In a natural road environment, his behaviour would be assimilated with risk-taking understood in terms of excessive conﬁdence in his own abilities or in terms of increased exposure

Driver Behaviour and Training – Volume III

286

Table 23.2 Singular behaviours

Hidden pedestrian crossing EXP EXP

Sudden stop at a trafﬁc light

Vehicle overtaking and pulling back in

Vehicle pulling out of parking

C C

AR

Total

1C AL

EXP

1C + 2A AL

1A

EXP

AL

1A

EXP

AL

1A

Total EXP

1AL

2C + 5A

DCT

C

1C

DCT

C

1C

DCT

C

1C

1C

1C + 1AR

3AL

DCT

AL

1A

DCT

AL

1A

Total DCT

2AL

3C

3C + 2A

5A

3C + 1A

5C + 7A

Total

1C

1C + 1A

C = Collision; A = Avoidance, AL = Avoidance to the left; AR = Avoidance to the right due to the travel conditions (Summala, 1987). These data represent one of the limitations of simulator-based research. Discussion After selecting prototypical accident-generating situations or accident scenarios (Fleury and Brenac, 2001) using data from real accident cases, we recreated them for the driving simulator.

Driving Experience and Simulation of Accident Scenarios

287

Experienced and beginner drivers were asked to drive the simulator and their behaviour was analysed. Insofar as the origin of accidents has many factors, of course, a small percentage of the drivers actually collided in the situations presented. However, two scenarios clearly show the effect of experience on driving; the subjects are confronted with an unforeseeable event. Faced with the ‘hidden pedestrian crossing’ scenario, all subjects brake in the emergency situation in equivalent timeframes, but only experienced drivers combine braking with swerving toward the right before reaching the pedestrian. This probably reﬂects greater forecasting abilities and greater skill than beginner drivers (Berthelon, Mestre and Taramino, 1995; Underwood, Chapman, Bowden and Crundall, 2002). Faced with a ‘vehicle pulling out of parking’, DCTs take more time than EXPs to take their foot off the accelerator, this long response time, combined with high speed, led to collisions for three of them. Moreover, during the two ﬁrst seconds of this scenario, DCTs drive more to the right of their trafﬁc lane than EXPs; they then swerve to the left of their lane to avoid conﬂict with the other vehicle. Swerving to the right observed with the EXP group comes later and corresponds to calculated avoidance more than to an emergency manoeuvre. DCAs were not differentiated from the other two groups. As for the ‘sudden stop at a trafﬁc light’ scenario, none of the variables analysed can be used to clearly differentiate driver behaviour as a function of their driving experience. This scenario causes an accident with an experienced driver whose high speed and the related low TIV did not enable him to react quickly enough to avoid collision. In the ‘vehicle overtaking and pulling back into the lane’ scenario, most drivers slow down sharply when the other vehicle merges in front of them, then they follow the vehicle until it turns right at the following intersection. Five drivers (three EXPs and two DCTs), however, adopt a strategy of overtaking the other vehicle. Singular behaviours attributable to EXPs mainly correspond to localised avoidances in all scenarios, but half of these avoidances were performed by a single driver with playful behaviour. For DCTs, singular behaviours notably correspond to collisions with the ‘vehicle pulling out of parking’. The absence of singular behaviours in the group of DCA drivers leaves us to suppose that they may adopt a more ‘safety-oriented’ driving style enabling them to manage situations better. It may be that the additional practice provided by this type of training enables them to acquire a certain number of skills. The small number of people in this group, however, allows us only to formulate this hypothesis and more research will be needed to conﬁrm it. Lastly, the simulator used in this research does not have a gearbox, which may have masked inter-group differences. Our beginners in fact probably had more cognitive resources to allocate to the events occurring around them than if they had had to concentrate on changing gears, which we know can be hard to manage when people ﬁrst start driving (Groeger and Clegg, 1997). The absence of a rear-view mirror should not have had any major effects on the behaviours observed. Consulting the information it provides was not an essential factor in the longitudinal and lateral checks made in the vehicle (except for changing lanes), which we know requires greater concentration among beginner drivers than experienced drivers (Lansdown, 2002).

288

Driver Behaviour and Training – Volume III

The results obtained in this study thus appear to be promising. Other scenarios need to be included in the driving simulators and, with a view to improving driver training, could be used to confront young drivers with ranges of difﬁcult situations not commonly encountered in natural driving. These scenarios could then be used in the context of driver training insofar as they could increase awareness of the unusual situations which could potentially occur. Along these lines, Fisher et al. (2002) showed that inexperienced drivers trained to react to scenarios on a computer have less risky behaviour on the simulator than inexperienced drivers who have not had such training. Their behaviour, moreover, is closer to that of experienced drivers who have not had this training. Likewise, the work by Chapman, Underwood and Roberts (2002) suggests a transfer of skills learnt by video to driving behaviour.

References Berthelon, C., Mestre, D. and Taramino, R. (1995). ‘Anticipation visuelle de collisions en situations simulées: effets de l’expérience de la conduite automobile.’ Le Travail Humain, 58, 311–38. Blancher, G., Brenac, T. and Fleury, D. (1998). Scénarios types d’accidents liés à une interaction entre véhicules hors agglomération. INRETS/DRAST Convention Report No. 95MT0051/1. November 1998. Brenac, T., Delcamp, J., Pelat, S. and Teisseire, G. (1996), Scénarios types d’accidents de la circulation dans le département des Bouches-du-Rhône. INRETS Report, MA 9611-2, November 1996. Brenac, T., Nachtergaële, C. and Reigner, H. (2003), Scénarios types d’accidents impliquant des piétons et éléments pour leur prévention. INRETS Report No. 256, ISBN 2-85782-599-4, December 2003. Chapman, P., Underwood, G. and Roberts, K. (2002) ‘Visual search patterns in trained and untrained novice drivers.’ Transportation Research Part F 5, 157–67. Deery, H.A. (1999). ‘Hazard and risk perception among young novice drivers.’ Journal of Safety Research, 30, 225–36. Ferrandez, F., Brenac, T., Girard, Y., Lechner, D., Jourdan, J.L., Michel, J.E. and Nachtergaële, C. (1995). ‘L’étude détaillée d’accidents orientée vers la sécurité primaire : méthodologie de recueil et de pré-analyse.’ Presses de l’école nationale des Ponts et Chaussée, Paris. Fisher, D.A., Laurie, N.E., Glaser, R., Connerney, K., Pollatsek, A. and Duffy, S.A. (2002). ‘Use of a ﬁxed-base driving simulator to evaluate the effects of experience and PC-based risk awareness training on drivers’ decisions.’ Human Factors, 44, 287–302. Fleury, D. and Brenac, T. (2001). ‘Accident prototypical scenarios: a tool for road safety research and diagnostic studies.’ Accident Analysis and Prevention, 33, 267–76. Gregersen, N.P. and Bjurulf, P. (1996). ‘Young novice drivers: towards a model of their accident involvement.’ Accident Analysis and Prevention, 28, 229–41. Groeger, J.A. and Clegg, B.A. (1997). ‘Automaticity and driving: time to change

Driving Experience and Simulation of Accident Scenarios

289

gear?’ In R. Rothengatter and E. Carbonell Vaya (eds), Trafﬁc and Transport Psychology: Theory and Application , Oxford: Pergamon, 137–46. Hall, J. and West, R. (1996). ‘The role of formal instruction and informal practice in learning to drive.’ Ergonomics, 39, 693–706. Kemeny, A. and Panerai, F. (2003). ‘Evaluating perception in driving simulation experiments.’ Trends in Cognitive Sciences, 7, 31–7. Lansdown, T.C. (2002). ‘Individual differences during driver secondary task performance: verbal protocol and visual allocation ﬁndings.’ Accident Analysis and Prevention, 34, 655–62. Mac Leod, C. (1987). ‘Visual reaction time and high-speed ball games.’ Perception, 16, 49–59. Mayhew, D.R., Simpson, H.M. and Pak, A. (2003). ‘Changes in collision rates among novice drivers during the ﬁrst months of driving.’ Accident Analysis and Prevention, 35, 683–91. Monarrez-Espino, J., Hasselberg, M. and Laﬂamme, L. (2006). ‘First year as a licensed driver: gender differences in crash experience.’ Safety Science, 44, 75– 85. Mourant, R.R. and Rockwell, T.H. (1972). ‘Strategies of visual search by novice and experienced drivers.’ Human Factors, 14, 325–35. Perrin, C., Van Elslande, P., Fleury, D., Girard, Y., Canu, B., Magnin, J., Parraud, C., Delage, B., Pascal, N. and Preget, A. (2004). Les études détaillées d’accidents: Renouvellement de la base. INRETS/DRAST Final Convention Report No. 03MT-32. November 2004. ISRN: INRETS/RE-04-914-FR. Ranney, T.A. (1994). ‘Models of driving behaviour: a review of their evaluation.’ Accident Analysis and Prevention, 26, 733–50. Salthouse, T.A. and Somberg, B.T. (1982). ‘Skilled performance effects of adult age and experience on elementary processes.’ Journal of Experimental Psychology: General, 111, 176–207. Sagberg, F. and Bjørnskau, T. (2006). ‘Hazard perception and driving experience among novice drivers.’ Accident Analysis and Prevention, 38, 407–14. Summala, H. (1987). ‘Young drivers’ accidents: risk taking or failure of skills.’ Alcohol, Drugs and Driving, 3, 79–91. Turner, C., McClure, R. and Pirozzo, S. (2004). ‘Injury and risk-taking behavior – a systematic review.’ Accident Analysis and Prevention, 36, 93–101. Underwood, G., Chapman, P., Bowden, K. and Crundall, D. (2002). ‘Visual search while driving: skill and awareness during inspection of the scene.’ Transportation Research F, 5, 87–97. Underwood, G., Chapman, P., Berger, Z. and Crundall, D. (2003). ‘Driving experience, attentional focusing and the recall of recently inspected events.’ Transportation Research F, 6, 289–304. Williams, A.F. (2003). ‘Teenage drivers: patterns of risk.’ Journal of Safety Research, 34, 5–15.

This page intentionally left blank

Chapter 24

Investigating the Contexts in which In-Vehicle Navigation System Users Have Received and Followed Inaccurate Route Guidance Instructions Nick Forbes and Gary Burnett University of Nottingham, UK Introduction Automation has evolved within automobiles over the past few decades. Initially, subtle modiﬁcations supported the driver in braking (for example, ABS) and steering (for example, power steering) control tasks. More recently, a range of automated systems has been introduced to support (and even replace) higher-level tactical (for example, lane keeping, collision avoidance) and strategic (for example, navigation) tasks. Few authors have speciﬁcally considered in-vehicle navigation systems as automation, and most vehicle-automation research has focused on driver support systems that automate control tasks (for example, adaptive cruise control). Navigation system research has primarily focused on issues such as distraction, usability and interface design (see Burnett, 2000 for a review). However, in most automation taxonomies (for example, Parasuraman et al., 2000; Sheridan, 2002), navigation systems would be classiﬁed as a form of automated system replacing certain strategic tasks (for example, optimal route selection) whilst supporting tactical (for example, turn by turn route following) aspects of normal driving behaviour. Although current forecasts suggest that the global GPS marketplace (of which vehicle navigation is the largest sector) is yet to perform as well as expected (GPS market update, 2006), it is clear that millions of drivers have already adopted this technology (DfT, 2005; ITS, 2006). Surveys can provide a useful tool for understanding some of the issues faced by an established user population, especially from those who use these systems regularly and have integrated them into their daily lives. A literature review revealed two non-commercial (DfT, 2005; Svahn, 2004) and two commercial (Privilege, 2006, J.D. Power and Associates, 2003; 2004; 2005; 2006) surveys of navigation system users. The Department for Transport (2005) survey examined key user demographics, and the main focus of Svahn (2004) was navigation system usage in familiar and unfamiliar areas. The Privilege (2006) survey investigated distraction and system interaction while driving, and the J.D. Power and associates (2003–2006) surveys examined usage and user satisfaction. These surveys have revealed some interesting demographic trends. The DfT (2005) survey considered a wide selection of UK drivers’ and passengers’ attitudes

292

Driver Behaviour and Training – Volume III

towards transport. It revealed that an equal proportion of males and females reported using a navigation system (7 per cent) and although they were used by drivers of all ages, the highest using age bands were 26–44 years (9 per cent) and 45–54 years (9 per cent). Although the ﬁndings failed to attain statistical signiﬁcance, data concerning job type and annual income also suggested a tendency for navigation system users to occupy higher socio-economic positions. Svahn (2004) surveyed German drivers who used the Volvo RTI (road and trafﬁc information system) only. This is an integrated (that is, built into the car) navigation system. Just over three-quarters of participants were male (77.5 per cent), and their mean age was 45 years (45–49 years was the highest using age band). Most participants drove 30–35 000 kilometres per year (mean = 34 000 kilometres). Nearly three-quarters of participants (72.4 per cent) had a university degree or higher, suggesting that most navigation system users in this survey also occupied higher socio economic positions. Nearly two-thirds of navigation system users considered themselves expert computer users or as possessing considerable skills. Skilled computer users tended to use their systems more frequently, utilise greater functionality and worry less about safety considerations than those less skilled at computing. The Privilege (2006) survey aimed to compare the distraction potential of navigation systems and traditional maps while the vehicle is in motion. A higher proportion of navigation system users (19 per cent) reported losing concentration than those who used maps (17 per cent). One in ten drivers admitted to using system controls while driving and over half of these participants thought that in doing so, their eyes were taken off the road. Navigation system users took their eyes off the road for an average of ten seconds while travelling at an average speed of 60 mph. The survey also revealed that nearly one in eight drivers did not check a route they were unfamiliar with in advance and simply relied on their navigation systems to help them to reach their destinations. The J.D. Power and Associates surveys are carried out each year and are of particular interest to the automobile industry. In the survey for 2003, nearly half of all the participants who purchased a vehicle with a factory installed navigation system reported high levels of satisfaction, and many indicated that the presence of a navigation system could positively affect their vehicle purchase decisions. The J.D. Power and Associates (2004) survey identiﬁed ease of use (35 per cent) and routing capabilities (20 per cent) as the most popular components of user satisfaction. Just over three-quarters of participants in Svahn (2004) considered route guidance from their navigation system to be ‘of great value’ and 65 per cent thought there was ‘signiﬁcant’ or ‘reasonable’ correspondence between system routing advice and their own individual preference when travelling in familiar areas. Although these ﬁndings indicate general user-satisfaction, clearly some navigation system users are less satisﬁed with their systems’ routing capabilities. In the J.D. Power and Associates (2006) usage and satisfaction survey, route or map related issues accounted for over 50 per cent of all problems cited by navigation system owners. The authors suggested dissatisfaction will diminish as the time between map production and delivery to customers decreases. However, in a continually evolving urban and rural transport system, it is unlikely that navigation system users will ever

In-Vehicle Navigation System Users

293

experience completely reliable route guidance. In some cases, drivers may simply arrive at incorrect destinations, but recently, anecdotal evidence has emerged in the news which suggests that drivers have received and consequently followed, a range of inaccurate and sometimes dangerous route guidance instructions. For example: Scenario 1 An 80-year-old German motorist has obediently followed his navigation system all the way into a huge pile of sand … ‘The driver was following the orders from his navigation system and even though there was a sufﬁcient number of warnings and barricades, he continued his journey into the construction site,’ a police spokeswoman has said. (Source: http://www.news.com.au/story/0,23599,20555319-13762,00.html, accessed 24 May, 2007)

Scenario 2 [A] student’s car was wrecked by a train after she followed her sat nav system onto a railway track … she was trying to cross the line in the dark when she heard a train horn, realised she was on the track, and the train smashed into her car … ‘I put my complete trust in the sat nav and it led me right into the path of a speeding train’ she said … ‘the crossing wasn’t shown on the sat nav … I was using the sat nav completely dependent on it’. (Source: http://news.bbc.co.uk/1/hi/wales/south_west/6646331.stm, accessed 24 May, 2007)

Scenario 3 [S]atellite navigation systems have caused so many problems in one corner of the country that road signs have been put up to tell drivers they are heading for trouble. The bright yellow signs have gone up in the village of Exton, near Winchester in Hampshire, after lorries repeatedly got stuck in a narrow lane hardly wide enough for one car. (Source: http://www.metro.co.uk/news/article.html?in_article_id=37839andin_page_id=34, accessed 24 May, 2007)

The research reported in this paper aimed to investigate this phenomenon in light of these anecdotal accounts. Previous surveys have neither directly addressed perceived system reliability nor adequately highlighted instances in which inaccurate route guidance instructions were received or followed. Furthermore, surveys such as Svahn (2004) and J.D. Power and Associates (2003–2006) have only investigated drivers who used factory installed (integrated) navigation systems. A survey was designed which aimed to overcome the limitations of previous work by using a large, more diverse sample of navigation system users. A sub-section of a wider online navigation system user survey (Forbes and Burnett, 2006) was designed to examine perceived reliability and to highlight inaccurate route guidance instructions that participants have both received and followed. Participants were asked to report features they particularly liked and disliked about their navigation systems to determine the components of user satisfaction and dissatisfaction. Routing/map issues were expected to feature prominently in both lists. The survey also aimed to determine the frequency with which navigation system users updated

294

Driver Behaviour and Training – Volume III

their system maps and some of their thoughts about this. Although maps may never be completely reliable, there may be an association between the frequencies with which users update their maps and the extent to which they receive inaccurate route guidance instructions. It was envisaged that this study would form the foundation of future research to both explain the phenomenon, and suggest potential remediation. Method Design An online survey was developed with extensive piloting ensuring that all questions and options had good clarity with a clear structure. A subset of 27 variables was considered relevant to the present paper. Age, number of years driving, mileage, months using navigation system and months using current map were ratio level variables. Computing skill, map update frequency and map update cost were ordinal level variables. Gender, employment status, navigation system type, navigation system experience, map update, reasons for map update/non-update, future map update, map update importance, route guidance reliability/efﬁciency, and dangerous/ illegal route guidance instructions received and followed were nominal variables. Nationality, system features particularly liked or disliked and the ‘other, please specify’ responses for variables in italics required text entry. Participants Eight hundred and seventy-two participants (844 male, 28 female, mean age = 46 years, range = 17–79 years) provided data for the survey. All participants were self-selecting, having responded to an online advertisement placed on driving and navigation system related Internet forums, bulletin boards and mailing lists. Apparatus Participants used their own computers to complete the questionnaire from remote locations. The questionnaire was designed using Microsoft Excel and html authoring software. Procedure Participants responded to online advertisements and were directed to follow a link to the questionnaire, provided that they had held a full driving licence for at least six months and used an in-vehicle navigation system. They were asked to answer questions truthfully and were reminded that their responses were completely anonymous.

In-Vehicle Navigation System Users

295

Results Participant demographics Figure 24.1 shows all age bands were well represented, but female participants were severely under-represented in the sample. Most participants were from the UK (59 per cent), USA/Canada (26 per cent) or the rest of Europe (10 per cent). The majority (96 per cent) were employed or retired, and most considered themselves as expert computer users (35 per cent) or possessing considerable skills (43 per cent).

120

Number of participants

100

80 Female

60

Male

40

20

0 <=29

30-34

35-39

40-44

45-49

50-54

55-59

60+

Age band

Figure 24.1 Graph showing participants’ age distribution and gender (N = 712) Participants had held their driving licences for an average of 26 years, and had driven an average of about 18 000 miles (approximately 29 000 kilometres) in the previous year. Following Rothengatter et al. (1993), most participants were experienced (35 per cent) or very experienced (55 per cent) drivers. Participants had been using their current system for an average of 12 months (median value reported) and 42 per cent had used other systems previously. Most participants used separate (nomadic) (52 per cent), PDA-based (30 per cent) or integrated (14 per cent) navigation systems. Inaccurate route guidance instructions Only 15 per cent of participants thought that routing instructions generated by their navigation systems were always completely reliable. Eighty-two per cent reported that they had received route guidance instructions that were inefﬁcient or wrong and 37 per cent reported that they had received inaccurate instructions that were dangerous or illegal. Nearly a quarter of these participants (23 per cent) admitted to having followed dangerous or illegal instructions on at least one occasion. Those

Driver Behaviour and Training – Volume III

296

who had followed them were signiﬁcantly older (t = 2.35, df = 101.8, p < 0.02) and held their driving licences for signiﬁcantly longer (t = 2.784, df = 83.5, p < 0.007) than those who had not. Seventy-three per cent of them were over the age of 40. Following dangerous or illegal route guidance instructions was not signiﬁcantly associated with navigation system type, navigation system experience, driving experience, computing skill or any other demographic variables. Dangerous or illegal route guidance instructions that participants have both received and followed are illustrated generally in Figure 24.2 and more speciﬁcally in Table 24.1. Bold text in Table 24.1 represents inaccurate instructions participants have followed. Speciﬁc

Do you think that the routing instructions generated by your navigation system are always completely reliable? (N=872)

85% have received unreliable route guidance instructions (n=744)

94% of these participants have received inefficient or wrong route guidance instructions (n=699)

43% have received dangerous or illegal route guidance instructions (n=321)

23% have followed these on at least one occasion (n=74)

81% of these participants have followed instructions guiding them into no-entry or one way streets (n=60)

15% have never received unreliable route guidance instructions (n=128)

77% of these participants have followed instructions guiding them onto private roads or roads for restricted vehicles only (n=57)

77% have never followed them (n=247)

12% of these participants have followed instructions guiding them onto roads with vehicle restrictions (n=9)

34% of these participants have followed instructions guiding them to perform prohibited manoeuvres (n=25)

Figure 24.2 Tree diagram highlighting contexts in which drivers have followed inaccurate route guidance instructions

In-Vehicle Navigation System Users

297

examples noted by participants of inaccurate instructions they have both received and followed (also shown in bold) are highlighted in Table 24.1. Maps Most participants who have received inaccurate route guidance instructions had never updated the map on their navigation system (58 per cent), although a signiﬁcant proportion had (42 per cent). Figures 24.3 and 24.4 highlight some of the reasons why participants had or had not updated their maps. Those who had purchased a map update were signiﬁcantly older (mean = 47.7 years) than those who had not (mean = 42.5 years) (t = 2.878, df = 713, p < 0.004). Participants had their current map installed for an average of ten months, although there was a wide range (0–84 months). A ﬁfth of participants who had not updated their maps would not consider updating in the future or were unsure. Most of those who updated did so once a year or more than once a year (61 per cent), although more than a third updated once every two years or more (39 per cent). Just over a third of participants (36 per cent) reported that they particularly liked their systems’ (re)routing and mapping capabilities. However, a similar proportion (34 per cent) indicated that they disliked the capabilities of their systems in this regard. 52 per cent of participants who particularly disliked them had never updated the map on their navigation system, although 84 per cent indicated they would update some time in the future.

Percentage of participants

45 40 35 30 25 20 15 10 5 Other (inc. forget, loss of software features, area not covered)

Not available

Don’t know how to/don’t know how often updates released

Don’t need to (yet)

Don’t use system often enough

Didn’t know they needed updating

System too new

Too expensive/awaiting offer

0

Reasons for not updating map

Figure 24.3 Reasons why participants have not updated the map on their navigation system (N = 498)

Driver Behaviour and Training – Volume III

298

Table 24.1

Speciﬁc contexts and participants’ examples of occasions where they have received and followed inaccurate route guidance instructions

Routing problem No entry/oneway street *58% ** 22%

Speciﬁc routing problems Routed into streets signposted as no entry or no permitted turn (42%)

Private road/ road for restricted vehicles 51% ** 19%

Routed into one way streets speciﬁcally (39%) Routed into pedestrianised zones and other city areas cars are not permitted (20%)

Routed onto a private road or road for emergency vehicles only (8%)

Routed on bus / tram / train lines (17%) Routed on cycle tracks / farm tracks / fjords / rivers / woodland (32%)

Speciﬁc examples based on participants’ comments Indicated turns where these are not permitted (the map is out of date!) Told me in my own neighbourhood to drive into streets that do not exist Suggested taking a slip road on a motorway that has been blocked off for nearly a year Told me to drive where there are no roads It once told me to turn right while I was on a bridge It sent me down a street which had been boarded off by the council. Suggested I turn where there is no road to turn into Indicated a turn at a non-existent street drive to a dead-end Told me to turn into a newly designated car-free bus station Directed me into a cul de sac where the only forward exit was for pedestrians Suggested I drive on a non-road (it was an alley) The system frequently directs me along roads prohibited for use except for access Instructed me to drive on an access restricted business roadway network which only employees are permitted on Suggested I use police/highways agency slip roads to access the M11!! Indicated a route onto a private road I was told to enter/leave motorway via works access Once used a motorway junction that was for service vehicles only Told me to drive onto railroad tracks Told me to go onto train tracks Suggested I drive down unmettalled roads Told to drive down farm tracks I was routed through farmyards and across ﬁeld tracks Directed me across a farmer’s ﬁeld that had an exit on a different road that I would travel down. Tried to send me down farm tracks I was routed down a road that had been reclaimed and allowed to grow wild which was also blocked by a barrier I was routed down public footpaths across a ﬁeld – no vehicle access possible The system frequently directs me along footpaths Bicycle paths It frequently routes along farm tracks

In-Vehicle Navigation System Users

Table 24.1

299

continued

Vehicle restrictions 3% ** 1%

Width restrictions (7%)

Weight restrictions (1%) Height restrictions (1%) Other vehicle unsuitability issues (3%)

Other prohibited manoeuvres 44% ** 17%

Instructed to perform prohibited manoeuvres such as U-turn, lane issues, unauthorised reverse (34%)

Other 11% ** 4%

System efﬁciency issues

I was routed down extremely narrow roads It frequently routes along unusable non-vehicular narrow tracks/lanes Routes through width and weight restriction Routes through weight limits Directed me under low bridges Suggested a drive down a steep narrow road with 180 degree bends not suitable for large vehicles Routed me down streets that were not acceptable for the vehicle we were in Directed me onto HOV section of beltway in Washington DC. Illegal but not dangerous We took a road in the motor home that we should not have taken In Australia you can’t usually do a U-turn at trafﬁc lights I was told to make a 180 degree turn Indicated a U-turn on a dual carriageway Tried to suggest a drive across a dual carriage way where the crossing is only for cars turning from the opposite side I was told to reverse direction at a motorway junction that did not allow reverse directions (in France) The system occasionally does not give sufﬁcient warning to move into freeway exit lane before manoeuvre I was directed to the wrong address in opposite direction Sometimes it will specify a roundabout when in fact it’s a junction Route causes a drive through town centres rather than take the by-pass or ring road Does not inform me where a route is interrupted by natural or man-made obstacles or detours or where street name changes en route

* Percentage of participants who have received dangerous/illegal instructions **Percentage of all participants Bold text represents dangerous/illegal route guidance instructions that participants have followed. Bold percentage represents percentage of those who have followed dangerous/illegal route guidance instructions

Driver Behaviour and Training – Volume III

300

30 25 20 15 10

Other (inc. gift, extra features in new maps))

Like to keep up to date

Awareness of roadworks

Advice from other sources

Manufacturer advice

Holiday or trip/change of address

0

Free/warranty/part of software or hardware upgrade

5 Previously given inaccurate instructions

Percentage of participants

35

Reasons for updating map

Figure 24.4. Reasons why participants have updated the map on their navigation system (N = 374) Discussion User demographics and validity Most participants (82 per cent) used separate or PDA based navigation systems and only a minority used integrated systems (14 per cent). Since previous user surveys (Svahn, 2004; J.D. Power 2003–2006) have exclusively sampled integrated navigation system users, the wider sampling frame makes the present study a valuable extension to previous work. In line with Svahn (2004) and DfT (2005), most navigation system users were aged between 26 and 54 years. However, most participants were employed or selfemployed and only a quarter held managerial positions. This ﬁnding differs from previous surveys and may be due to the wider sampling frame and more recent date of the present study. Although female drivers were represented in almost every age band, they were particularly under-represented in the entire sample. This restricts external validity, and was probably an artefact of the Internet sampling methodology. Internet user surveys (for example, GVU, 2000) have revealed a trend towards gender equality. However, in a review, Krantz and Dalal (2000) noted a wide range in the dispersion of females across studies, from 8 per cent to 71 per cent. Although some authors have contested the validity of Internet-mediated research (for example, Coomber, 1997; Stanton, 1998), Hewson (2002) concluded that at the very least, the ﬁndings in favour show ‘the issue is not as problematic as has been previously suggested’ (30). She suggests comparability between samples will very much depend on the sampling procedures employed. Female drivers may therefore have been under-represented in the present sample because they were under-represented in the sampling frame (that

In-Vehicle Navigation System Users

301

is, driving and navigation system enthusiast web sites), rather than the Internet in general. A wider sampling frame might ensure a more representative sample. Inaccurate route guidance instructions and maps Clearly most participants questioned the fallibility of route guidance instructions they have received from their navigation systems, with 85 per cent having received unreliable guidance. Moreover, over 40 per cent of these respondents report that they have received inaccurate instructions that were dangerous or illegal, in a range of different contexts. Participants have been instructed to perform prohibited driving manoeuvres, or to drive into areas that are prohibited for legal, safety or other reasons. Relatively few participants have received instructions guiding them into areas for which their vehicle dimensions are unsuitable, although this may be an artefact of the sampling frame. Inaccurate guidance will arise due to a variety of reasons; for instance, poor mapping information and erroneous routing algorithms. Clearly, the currency of the underlying map data is of particular importance. In this survey, it was revealed that inaccurate route guidance instructions were by no means exclusively received by users who had not regularly updated their maps – the survey revealed that 42 per cent of those who had updated maps had still received poor guidance. This is not particularly surprising, as internationally even the most accurate maps will become outdated very quickly. Many participants received free map upgrades. Although it is unlikely that upgrades will ever be universally free, the general consensus of respondents in this survey was that they’re presently overly expensive. The results suggest manufacturers should increase the appeal of map upgrades to younger drivers, although greater foreign mobility (for example, family holidays, business trips) may explain why participants who had updated were signiﬁcantly older than those who had not. Why might some drivers follow inaccurate guidance instructions? In order to explain why some drivers follow inaccurate guidance instructions, it is important to consider the extent to which drivers actually process road signs when presented with consistent and contradictory navigation system instructions. If road signs are not sufﬁciently processed, this would imply an attention-based explanation. However, if they are processed and drivers still follow inaccurate system instructions, the explanation may not be purely cognitive. It would indicate inappropriate reliance on the navigation system. With respect to the attention argument, it is important to differentiate between different modes of human attention. Selective attention concerns the ability to focus on relevant information and ignore irrelevant information. Divided attention concerns the ability to maintain focus on more than one task simultaneously in multitask situations (for example, driving while using a navigation system). Sustained attention concerns the ability to maintain focus over a period of time. Following inaccurate instructions could respectively represent a failure to focus on relevant information (that is, road signs), or a failure in driving task performance (that is,

302

Driver Behaviour and Training – Volume III

inappropriately responding to road signs) or a failure to respond appropriately to system errors. Failures in each of these forms of attention could plausibly explain why some drivers follow inaccurate guidance instructions. To illustrate, consider sustained attention speciﬁcally. Sustained attention is typically measured using vigilance tasks. These require participants to respond to infrequent ‘targets’, presented among frequent ‘nontargets’. Several studies (for example, Warm and Dember, 1998; Parasuraman and Molloy, 1996) have indicated a temporal decline in vigilance performance, known as the vigilance decrement, which occurs fairly rapidly after presentation of initial stimuli (Craig et al., 1987), and can be measured using short-duration tests. The sustained attention to response task (SART: see Robertson et al., 1997) is a particularly short vigilance task with proven validity. In most vigilance tasks, participants are required to respond to a target, but in the SART they are required to withhold a response to a target (the digit 3) and respond to non-targets (digits 1–9, excluding 3). Typically task performance declines fairly rapidly. Matthews et al. (1998) note that automated tasks are similar to vigilance tasks because they require sustained monitoring and infrequent response. For navigation system users, they will typically receive and follow accurate route guidance instructions. In the rare cases in which they receive inaccurate instructions, they should withhold the usual response. Following inaccurate instructions could therefore represent a vigilance failure. Reliance is strongly associated with trust in automation (Parsuraman and Riley, 1997; Dzindolet et al., 2003). The concept of trust in automation has developed from social psychological research concerning trust in interpersonal interactions (for example, Ross and LaCroix, 1996). Although a comprehensive review of the trust literature is outside the scope of this paper (see Lee and See, 2004, for a review), it is sufﬁcient to highlight that trust is also an important component of humanautomation interaction (Lee and See, 2004). Trust is the attitude – it is a purely psychological state that can be measured subjectively (Wickens and Xu, 2002; Jian et al., 2000). Reliance is the behaviour, and it can be measured using objective task performance measures. Reliance may be appropriate (for example, when operators trust automation that is either reliable or more reliable than manual operation) or inappropriate (for example, when operators trust automation that is either inaccurate or less reliable than manual operation) (see Dzindolet et al., 2003). Inappropriate reliance may lead to automation-induced complacency (Parasuraman et al., 1993). Complacency has been deﬁned as ‘a Psychological state characterised by a low index of suspicion’ (Weiner, 1981, 117). Sheridan and Parasuraman (2006, 99) illustrated a ‘classic case of automation complacency’ by citing a naval accident in which a cruise ship ran aground because the GPS system had malfunctioned by switching to dead reckoning mode (that is, it didn’t account for tides, weather, and so on). An accident report showed the crew were over-reliant on the (malfunctioning) automated position display, and failed to utilise other navigation aids or environmental information for navigation. Complacency has been associated with excessive trust (Parasuraman and Riley, 1997), low self-conﬁdence (Lee and See, 2004; Lee and Moray, 1994), sub-optimal system monitoring for malfunctions (Parasuraman et al., 1993) and a poor mental model of the automated system (Lee and Moray, 1992). It may explain scenarios

In-Vehicle Navigation System Users

303

1–3 in the introduction. In scenario 1, the elderly driver may have over-trusted his navigation system because he continued following instructions despite contradictory environmental information (that is, ‘warnings and barricades’). In scenario 2, the student speciﬁcally stated that she trusted her navigation system too much. In scenario 3, the council response suggests drivers may have become stuck in the narrow lane because they over-trusted their navigation systems. In this case over-trust may have been caused by an inaccurate mental model, which prevented some drivers from realising their vehicle characteristics would not be taken into consideration in route planning algorithms. Attention, trust, complacency and age Age was the only variable signiﬁcantly associated with following inaccurate instructions. Increasing age is typically associated with declines in physical and mental ability (see Matthews et al., 2000). It has also been linked to failures in a range of driving performance abilities, including perception, memory and attention. Speciﬁcally it has been associated with impaired performance on tasks of selective (Rogers, 2000), divided (Korteling, 1991) and sustained (Mouloua and Parasuraman, 1995) attention. When performance is not impaired, research suggests that older adults must work harder to maintain similar performance to younger counter-parts (Bunce and Sisa, 2002). In their model of trust and reliance in automated technology for older adults, Ho et al. (2005) suggest that complacency may be the result of age-related cognitive deﬁcits in attention-allocation, working memory, mental workload, decision making and interpreting stochastic information. These cognitive changes may reduce selfconﬁdence in manual performance. Additionally they suggest that older adults may be less familiar with computer technology and less aware of potential unreliability. A range of driving research using advanced traveller information systems (Fox and Davies, 1998) and gauge warning monitoring tasks (Sanchez et al., 2004) has indicated that older drivers trust automated vehicle systems more than their younger counterparts. Using a ﬂight simulation task, Vincenzi and Mouloua (1999) showed that older adults were less likely to notice automation errors and correct for them when they occurred. Ho (2005) also found that older adults placed greater trust in an automated medical management system, and made more errors because they relied on the system too much. Future research Research should aim to determine whether situations in which drivers have followed inaccurate route guidance instructions from an in-vehicle navigation system may be explained by attention, trust and/or complacency. Given the safety considerations, it would be advisable for initial investigations to use driving simulators until psychological characteristics are more fully explained. This would entail fewer ethical considerations, and the degree of experimental control they offer should make it possible to design scenarios that closely approximate some of the contexts illustrated in the present paper.

304

Driver Behaviour and Training – Volume III

A simple investigation could identify and administer to participants, valid and reliable measures of attention, trust and complacency. After completing these measures, participants could be presented with a range of urban scenarios in a driving simulator. Over the course of a few trials, a pseudo-navigation system could provide frequent accurate turn-by-turn guidance, and infrequent inaccurate guidance. Comparing the scores on attention, trust and complacency measures of drivers who do and do not follow inaccurate guidance instructions in different contexts, or correlating scores with task performance measures, may highlight the most plausible explanations for this phenomenon, and may enable future work to investigate potential remediation. Attention-based remediation may include techniques to increase arousal or ensure more efﬁcient monitoring of salient environmental information such as road signs. Trust and complacency-based remediation may include techniques to increase selfconﬁdence or to remind drivers that that system information can be unreliable (for example, De Vries, 2004) Measuring attention, trust and complacency Although auditory ﬁltering tasks (Craik, 1977) have traditionally measured selective attention, visual search tasks (Plude and Doussard-Roosevelt, 1989) would be most appropriate, as driving requires a greater visual component. Divided attention could be measured by examining dual task performance on a tracking task and a signal detection task. This is a traditional combination, and would be relevant as driving has central tracking (Matthews et al., 2000) and perceptual (Hills, 1980) components. Driving requires attention to both cognitive and sensory stimuli, and vigilance tests exist using both types. The SART is a cognitive test and may be appropriate due to its apparent similarity with the phenomenon under investigation. The cognitive failures questionnaire (CFQ) examines the frequency with which participants experience lapses in attention, perception, memory and motor function (Broadbent, 1982). The mindful attention awareness scale (MAAS) concerns actions that are performed ‘automatically’ or ‘without being aware’ (Brown and Ryan. 2003). The attention-related cognitive errors scale (ARCES) assesses everyday performance failures caused by brief malfunctions of sustained attention or absent-mindedness (Cheyne et al., 2006). Each of these self-report scales consists of everyday instances of attention failure. Following inaccurate guidance instructions may be a symptom of a wider attention problem. Drivers’ scores on these scales may reveal trends in particular types of attention failure. Trust in automation can only be measured subjectively as it is a purely psychological state (Wickens and Xu, 2002). Most attempts to measure trust (for example, Lee and Moray, 1994; Muir and Moray, 1996) have been based on theoretical rather than empirical notions (Bisantz and Seong, 2001). However, Jian et al. (2000) and Madsen and Gregor (2000) developed empirically based scales using a bottom up approach in which participants rated words as relevant to interpersonal trust, trust in automation or both. Madsen and Gregor’s (2000) scale would only be appropriate for long-term research as it assumes several months system experience. The Jian et al. (2000) general purpose scale has been used to measure trust in a range of automated devices including automated decision aids and adverse condition

In-Vehicle Navigation System Users

305

warning systems (Bisantz and Seong, 2001; Gupta et al., 2002), and may also be appropriate for navigation systems. Complacency is typically inferred from task performance (for example, Parasuraman et al., 1993; Singh et al., 1997). However, Moray and Inagaki (2001) refuted some previous evidence, and made a compelling case that complacency should only be inferred when an optimal sampling strategy has already been deﬁned. Nevertheless, Singh et al. (1993) created a subjective scale to measure complacency potential (an individual’s tendency to over-rely on automation). It is a general-purpose scale in which participants must rate their attitude towards a range of everyday automated systems (for example, ATMs, aircraft automation). This scale may be appropriate as following inaccurate guidance instructions may be a symptom of a wider potential for complacency in other contexts. Validity of future research Driving simulators are appealing as a research tool as they provide a safe, convenient and comprehensive environment for research (Kaptein et al. 1996). In relation to the present work, they also would allow researchers to study conditions which may otherwise be rare. However, data collected may be confounded by learning effects (for both the simulator and the navigation system) and experimenter effects (NHTSA, 2000). Many participants (particularly older ones) also suffer simulator sickness (Goodman et al.,1997). The main drawback however, concerns validity. Several authors have questioned the extent to which simulated driving performance corresponds with real driving performance (for example, Reed and Green, 1999; Carsten et al., 1997). Goodman et al., (1997) suggests that driving behaviour, particularly allocation of attention to in-vehicle tasks, may signiﬁcantly differ from real world performance because no serious consequences result from driving errors in a simulator. Simulator participants may not observe legal or socially accepted codes of behaviour (for example, turning into a one-way street) as stringently as they would in the real world. They’re also unlikely to experience the same motivations, time constraints and other pressures as real drivers who may want to reach a destination for many different reasons. Consequently, driving simulator research may only be appropriate for initial investigations. In contrast to simulator studies, it may be possible to observe real driving behaviour using longitudinal on-road research. However, these studies are often expensive and time consuming, particularly as the receiving and following of inaccurate route guidance instructions are relatively rare events. Further survey research may be useful, and could be combined with online measures of attention, trust and complacency. Alternative qualitative research techniques may also yield rich data. Drivers could for example be asked to complete a diary of their navigation system use. Such an approach could provide interesting accounts of situations in which inaccurate guidance was followed, and drivers’ thoughts and cognitions about contexts they have experienced. Nevertheless, an adequate sample size and time span will be required.

306

Driver Behaviour and Training – Volume III

Conclusions This chapter describes a survey of 872 navigation system users which revealed that many drivers follow inaccurate route guidance instructions from their navigation systems in a variety of different contexts. Importantly, the survey indicated that the likelihood of drivers following poor guidance increases with age. Attention, trust and/or complacency could explain why drivers behave in this way. It will be important to investigate the role of these factors in an empirical setting, particularly so that potential remediations can be tested. However, future research must be carefully designed as simulator studies may lack validity in this speciﬁc area, and on-road studies are likely to be time consuming and expensive. Qualitative research techniques will be important for understanding driver behaviour in this context. References Bisantz, A.M. and Seong, Y. (2001). ‘Assessment of operator trust in and utilization of automated decision aids under different framing conditions.’ International Journal of Industrial Ergonomics, 28(2), 85–97. Broadbent, D.E., Cooper, P.F., FitzGerald, P. and Parkes, K.R. (1982). ‘The cognitive failures questionnaire (CFQ) and its correlates.’ British Journal of Clinical Psychology, 21, 1–16. Brown, K.W. and Ryan, R.M. (2003). ‘The beneﬁts of being present: mindfulness and its role in psychological well-being.’ Journal of Personality and Social Psychology, 84, 822–48. Bunce, D. and Sisa, L. (2002). ‘Age differences in perceived workload across a short vigil.’ Ergonomics, 45, 949–60. Burnett, G.E. (2000). ‘Usable vehicle navigation systems: are we there yet?’ In Vehicle Electronics Systems 2000 – European Conference and Exhibition (Stratfordupon-Avon, UK) 29–30 June, 2000 (ERA Technology Limited, Leatherhead), 3.1.1–3.1.11 Carsten, O.M.J., Groeger, J.A., Blana, E. and Jamson, A.H. (1997). Driver Performance in the EPSRC Driving Simulator [LADS]: A Validation Study. Final report for EPSRC project GR/K56162, December 1997. Cheyne, J.A., Carriere, J.S.A. and Smilek, D. (2006). ‘Absent-mindedness: lapses of conscious awareness and everyday cognitive failures.’ Consciousness and Cognition 15, 578–92. Coomber, R. (1997). ‘Using the internet for survey research.’ Sociological Research Online, vol. 12. Craig, A., Davies, D.R. and Matthews, G. (1987). ‘Diurnal variation, task characteristics and vigilance performance.’ Human Factors, 29, 675–84. Craik, F.I.M. (1977). ‘Age differences in human memory.’ In J.E. Birren and K.W. Schaie (eds), Handbook of the Psychology of Aging. New York: Van Nostrand Reinhold. Department for Transport (2005). ‘Results of the ONS omnibus survey, May 2005.’ http://www.dft.gov.uk/stellent/groups/dft_transstats/documents/page/dft_

In-Vehicle Navigation System Users

307

transstats_611003.hcsp. DeVries, P. (2004). Trust in Systems: Effects of Direct and Indirect Information. Eindhoven: Technische Universiteit Eindhoven, 2004. Proefschrift. Dzindolet, M.T., Peterson, S.A., Pomranky, R.A., Pierce, L.G. and Beck, H.P. (2003). ‘The role of trust in automation reliance.’ International Journal of Computer Studies, 58, 697–718. Forbes, N. and Burnett, G.E. (2006). ‘Online in-vehicle navigation system user survey.’ Unpublished study. School of Computer Science and IT, University of Nottingham, UK. Fox, J.E. and Davis, D.A.B. (1998). ‘Effects of age and congestion information accuracy of advanced traveler information systems on user trust and compliance.’ Transportation Research Record, 1621: 43–9. Goodman, M., Bents, F., Tijerina, L., Wierville, W., Lerner, N. and Benel, D. (1997). ‘An investigation of the safety implications of wireless communications in vehicles.’ NHTSA. http://www.nhtsa.dot.gov/people/injury/research/wireless/. GPS market update (2006), ID: CICQ129054. RNCOS. http://www. researchandmarkets.com/reportinfo.asp?report_id=328836. Graphics, Visualization and Usability Center (2000). ‘GVU’s user survey graphs.’ http://www-static.cc.gatech.edu/gvu/user_surveys/. Gupta, N., Bisantz, A.M. and Singh, T. (2002). ‘The effects of adverse condition warning system characteristics on driver performance: an investigation of alarm signal type and threshold level.’ Behaviour and Information Technology, 21(4), 235–48. Hewson, C. (2003). ‘Conducting research on the internet.’ Psychologist, 16, 290–3. Hills, B.L. (1980). ‘Vision, visibility and perception in driving.’ Perception, 9, 183– 216. Ho, G. (2005). ‘Age differences in trust and reliance of a medication management system.’ PhD thesis, University of Calgary, Calgary, AB, Canada, 2005. Ho, G., Kiff, L.M., Plocher, T. and Haigh, K.Z. (2005). ‘A model of trust and reliance of automation technology for older users.’ Proceedings of the AAAI Fall Symposium ‘Caring Machines: AI in Eldercare,’ 3–5 Nov 2005, Washington, DC, USA. ITS (2006). Japanese Ministry of Land, Infrastructure and Transport. http:// c11z7m9s.securesites.net/ITS/topindex/topindex_navi.html. J.D. Power and Associates (2003–2006). Navigation Usage and Satisfaction Study. http://www.jdpower.com/corporate/news/releases/. Jian, J.J., Bisantz, A.M., and Drury, C.G. (2000). ‘Foundations for an empirically determined scale of trust in automated systems.’ International Journal of Cognitive Ergonomics, 4(1), 53–71. Kaptein, N.A., Theeuwes, J. and van der Horst, A.R.A. (1996). ‘Driving simulator validity: some considerations.’ Transportation Research Record 1550, 30–36. Korteling, J.-E. (1991). ‘Effects of skill integration and perceptual competition on age-related differences in dual-task performance.’ Human Factors, 33, 35–44. Krantz, J.H. and Dalal, R.S. (2000). ‘Validity of web based psychological research.’ In M.H. Birmbaum (ed.), Psychological Experiments on the Internet San Diego,

308

Driver Behaviour and Training – Volume III

CA: Academic Press, 35–60. Lee, J.D. and Moray, N. (1992). ‘Trust, control strategies and allocation of function in human-machine systems.’ Ergonomics, 35, 1243–70. Lee, J. and Moray, N. (1994). ‘Trust, self-conﬁdence, and operators’ adaptation to automation.’ International Journal of Human-Computer Studies, 40, 153–84. Lee, J.D., Stone, S., Gore, B.F., Colton, C., McCauley, J., Kinghorn, R., Campbell, J.R., Finch, M. and J.G. (1997). ‘Advanced traveller information systems and commercial vehicle operations components of the intelligent transportation systems: design alternatives for in-vehicle information displays.’ Technical Report FHWA-RD-96-147, Federal Highway Administration, McLean, VA, 1997. Lee, J.D. and See, K. (2004). ‘Trust in automation: designing for appropriate reliance.’ Human Factors, 46(1), 50–80 Madsen, M. and Gregor, S. (2000). ‘Measuring human-computer trust.’ In Proceedings of 11th Australasian Conference on Information Systems, Brisbane. Matthews, G., Campbell, S.E., Desmond, P.A., Huggins, J., Falconer, S. and Joyner, L.A. (1998). ‘Assessment of task induced state change: stress, fatigue, and workload components.’ In M.W. Scerbo and M. Mouloua (eds), Automation, Technology and Human Performance, Mahwah, NJ: Erlbaum, 199–203. Matthews, G., Davies, D.R., Westerman, S.J. and Stammers, R.B. (2000). Human Performance: Cognition, Stress and Individual Differences. London: Psychology Press. Maule, A. and Sanford, A. (1980). ‘Adult age differences in multi-source selection behaviour with partially predictable signals.’ British Journal of Psychology, 71, 69–81. Moray, N. and Inagaki, T. (2001). ‘Attention and complacency.’ Theoretical Issues in Ergonomics Science, 1, 354–65. Mouloua, M., and Parasuraman, R. (1995). ‘Aging and cognitive vigilance: effects of spatial uncertainty and event rate.’ Experimental Aging Research, 21, 17–32. Muir, B.M. and Moray, N. (1996). ‘Trust in automation part II: experimental studies of trust and human intervention in a process control simulation.’ Ergonomics, 39, 429–60. NHTSA (2000). NHTSA Driver Distraction Internet Forum: In-Vehicle Technologies: Experience and Research (Other) ‘Question Those Statistics.’ NHTSA (2002). Automotive Collision Avoidance System Field Operational Test Phase I Interim Report. NHTSA Technical Report DOT HS 809 453. Parasuraman, R., Molloy, R., and Singh, I.L. (1993). ‘Performance consequences of automation-induced “complacency”.’ International Journal of Aviation Psychology, 3, 1–23. Parasuraman, R. and Molloy, R. (1996). ‘Monitoring an automated system for a single failure: vigilance and task complexity effects.’ Human Factors, 38, 311– 22. Parasuraman, R., and Riley, V. (1997). ‘Humans and automation: use, misuse, disuse, abuse.’ Human Factors, 39, 230–53. Parasuraman, R., Sheridan, T.B. and Wickens, C.D. (2000). ‘A model for types and levels of human interaction with automation.’ IEEE transactions on Systems, Man

In-Vehicle Navigation System Users

309

and Cybernetics. Part A: Systems and Humans, 30, 286–97. Plude, D.J. and Doussard-Roosevelt, J.A. (1989). ‘Aging, selective attention, and feature integration.’ Psychology and Aging, 4, 98–105 Privilege Insurance (2006). ‘Unsafe use of navigation equipment.’ http://www. privilege.com/aboutus/Unsafeusenavigation.htm. Reed, M.P. and Green, P.A. (1999). ‘Comparison of driving performance on-road and in a low-cost simulator using a concurrent telephone dialling task.’ Ergonomics, 42(8), 1015–37. Robertson, I.H., Manly, T., Andrade, J., Baddeley, B.T. and Yiend, J. (1997). “‘Oops!” Performance correlates of everyday attentional failures in traumatic brain injured and normal subjects.’ Neuropsychologia, 35, 747–58. Rogers, W.A. (2000). ‘Aging and attention.’ In Cognitive Aging: A Primer. London: Psychology Press, 57–73. Ross, W. and LaCroix, J. (1996). ‘Multiple meanings of trust in negotiation theory and research: a literature review and integrative model.’ The International Journal of Conﬂict Management, 7, 314–60. Rothengatter, T., Alm, H., Kuiken, M.J., Michon, J.A. and Verwey, W.B. (1993). ‘The driver.’ In J.A. Michon (ed.), Generic Intelligent Driver Support. London: Taylor and Francis, 33–52. Sanchez, J., Fisk, A. and Rogers, W. (2004). ‘Reliability and age-related effects on trust and reliance and reliance of a decision support aid.’ In Proceedings of the 48th Human Factors and Ergonomics Society, 586–9. Sheridan, T. (2002). ‘Humans and automation: system design and research issues.’ Wiley Interscience. Sheridan, T. and Parasuraman, R. (2006). ‘Human-automation interaction.’ Reviews of Human Factors and Ergonomics, 1, 89–129. Singh, I.L., Molloy, R. and Parasuraman, R. (1993). ‘Automation-induced “complacency”: development of the Complacency Potential Rating Scale.’ The International Journal of Aviation Psychology, 3, 111–22. Singh, I.L., Molloy, R. and Parasuraman, R. (1997). ‘Automation-induced monitoring inefﬁciency: role of display location.’ International Journal of Human-Computer Studies, 46, 17–30. Stanton, J.M. (1998). ‘An empirical assessment of data collection using the internet.’ Personnel Psychology, 51(3), 709–725. Svahn, F. (2004). ‘In-car navigation usage: an end-user survey on existing systems.’ In Proceedings of IRIS27, Falkenberg, Sweden. August 2004. Vincenzi, D. and Mouloua, M. (1999). ‘Monitoring automation failures: effects of age in performance and subjective workload.’ In Automation Technology and Human Performance: Current Research and Trends, Lawrence Erlbaum Associates, 253–7. Warm, K.S. and Dember, W.N. (1998). ‘Tests of vigilance taxonomy.’ In R.R. Hoffman, M.F. Sherrick and J.S. Warm (eds), Viewing Psychology as a Whole: The Integrative Science of William N. Dember. Washington, D.C.: American Psychological Association. Wickens, C.D. and Xu, X. (2002). ‘Automation, trust, reliability and attention.’ HMI

310

Driver Behaviour and Training – Volume III

02 03, AHFD-02-14/MAAD-02-2, AHDF Technical Report. Weiner, E.L. (1981). ‘Complacency: is the term useful for air safety?’ In Proceedings of the 26th Corporate Aviation Safety Seminar, Denver Flight Safety Foundation, 116–25.

Chapter 25

Comparison of Novice Drivers in Austria and the Czech Republic With and Without the Use of Intelligent Speed Adaptation Christine Turetschek1 and Ralf Risser2 1 Factum Chaloupka 2 Risser OHG, Austria Introduction Speed adaptation via in-car devices has been studied for over 15 years. Estimates of the safety effects of a fully implemented automatic speed management system in Sweden and the UK vary from a 10 per cent reduction in injury accidents with an advisory system, to a reduction in the range of 20–40 per cent with a system that enforces current speed limits and automatically limits the speed in critical conditions (for example, slippery road, poor visibility; see Várhelyi, 1996; Carsten and Comte, 1997; Carsten and Fowkes, 2000). Earlier ﬁeld trials (Persson et al., 1993; Almqvist and Nygård, 1997) and simulator experiments (Comte, 1996; Comte, 1998a; 1998b) have demonstrated positive effects of Intelligent Speed Adaptation (ISA), with a focus on the adaptive accelerator pedal (AAP Várhelyi et al., 2002). In studies about ISA, it is often suggested that speed reduction affects communication between road users. Car drivers, for instance, yield more often for pedestrians, while pedestrians are less often forced to stop when vehicle speeds are lower (Varhelyi, 1996). There is also empirical evidence for modiﬁed visual behaviour under different speed conditions (Berger, 1996). When driving at higher speeds the focus of human vision tends to be more distant, and thereby peripheric vision in the vicinity deteriorates. At a lower speed the focus is more on the nearby environment and peripheric vision is more efﬁcient and wider in angle, which facilitates the registration of and the communication with, for example, pedestrians. The question of communication is important against the fact that around 75 per cent of all accidents take place between two or more road users, as can be seen in most of the European accident statistics. It can be assumed that those 75 per cent represent a breakdown of communication. This study began by observing behaviour (using the ‘Wiener Fahrprobe’ as an observation instrument) of drivers aged 18 to 25 years while driving a driving school car along a standardised route in Vienna and Brno. Additionally, all participants completed a questionnaire about their view on trafﬁc safety, speed in general and ISA. After about one month, the participants drove an ISA equipped car along the

312

Driver Behaviour and Training – Volume III

same standardised test route. Again their driving behaviour was observed and the young drivers completed a standardised questionnaire based on their answers from the ﬁrst round of questionnaires. As a next step, psychological group training was developed. The training was based on a discussion with the goal of shaping participants’ opinion about trafﬁc safety in general, but also about speeding in particular. Two sessions were planned, each of them lasting two hours. The focus of the ﬁrst part of the training was on motives and emotions related to driving. In this discussion, novice drivers became aware, among other things, of the discrepancy between personal motives (convenience, fun, and so on) and motives of society (trafﬁc safety). ISA would represent society’s interests in keeping speed limits. The second part of the training focused on speeds: on reasons for and problems of speeding. The group discussions were carried out in addition to the use of ISA for half of the participants. The other half only had to carry out a test drive with ISA, without any group discussion. Method Participants Novice drivers in Austria and Czech Republic aged between 18 and 25 years were asked to participate in this study. The mean age of the Czech Republic participants was 19 years and in Austria 22. In Austria, participants were recruited via the Internet. In the Czech Republic, a driving instructor of a local driving school asked some of his former students to participate. All in all, 75 novice drivers took part in the study, 48 in Austria and 27 in the Czech Republic: 28 were women and 46 men. The distribution of sex in each countries was signiﬁcantly different [Chi-Quadrat (df = 1, p = 0.009)]. In Austria about the half of the participants were women, while in the Czech Republic fewer females took part. Six of the participants did not have a valid driving licence at the time of completing the questionnaire. Sixty-eight participants had a driving licence allowing them to drive cars and some also held a driving licence for other vehicles, like motorbikes or lorries. The driving experience of both groups was quite different. In Austria participants had acquired their licence on average 58 months ago, while in the Czech Republic this was 11 months ago. Seven participants in Austria and six in the Czech Republic dropped out of the study for different reasons. Procedure Behaviour observation The behaviour observation was carried out using the standardised behaviourobservation method Wiener Fahrprobe (Vienna Driving Test; see Risser and Lehner, 1998). This method was developed to ﬁnd out whether a person, according to observers’ assessment, is able to drive a car or not. The development of the Wiener Fahrprobe was based on other behaviour observation methods such as Quenault’s

Comparison of Novice Drivers in Austria and the Czech Rebublic

313

driving test or the Kölner Fahrverhaltens test. The Wiener Fahrprobe has been used in different studies in order to assess behaviour of road users for different age groups, use of new intelligent speed technology, driving along dangerous routes, and so on. The procedure has been adapted slightly in order to achieve respective goals, depending on the content of the study. The observation usually took place on a standardised route with a length between 25–50 km which included densely inhabited areas, but also rural roads (approximately 1/4 of the route) and motorways (approximately 1/8 of the route). The chosen routes in Vienna and Brno are only 17 km long, but otherwise fulﬁl all conditions deﬁned in the literature on the Vienna Driving Test. To make the data analysis easier, the route is usually divided into sections, so that one always knows from the checklists on which part of the route certain behaviours occurred. The Wiener Fahrprobe should be carried out between the peak hours to avoid trafﬁc jams, which helps to keep the time schedule, but also facilitates the occurrence of different types of driving behaviour. It takes between 40 and 60 minutes to drive along the test route depending on driver characteristics and trafﬁc volume. The original version of the Wiener Fahrprobe suggests that two observers are seated in the back of the car. One of them registers behaviour variables in a standardised way (the Coding Observer); the other one concentrates more on communication aspects which are non-standardised (the Free Observer). The standardised variables are measured with the help of a standardised observation checklist by the Coding Observer. One checklist per route is completed. The standardised observation variables record exclusively erroneous types of behaviour. The list of standardised variables consists of those types of behaviour that can be speciﬁed and anticipated. For example, distance keeping behaviour: on each section of any route it will be possible to say whether a participant drove too close to the car in front or not. Other standardised variables are the use of the indicator, the accuracy of lane use, speed, driving on bends, and so on. Every event is registered on the checklist and if no deﬁned event takes place there is also no recording of the absent behaviour. Non-standardised variables like errors, explicit interaction/communication processes and trafﬁc conﬂicts are registered by the Free Observer. One checklist per event is completed and the number of the route section where the event took place is added. Errors are events such as neglecting to obey a stop sign that represent a serious violation of the trafﬁc law and/or risk. Another example of an error is if the driver almost knocks over a pedestrian who crosses a pedestrian crossing. For interaction/communication, events are not easy to predict. At the same time it is critically important to register communication processes since they inﬂuence driver behaviour to a very high degree. The trafﬁc conﬂicts that take place along the route are registered as well. A trafﬁc conﬂict is an event in which the time to collision is shorter than one second and where an accident is only avoided due to the evasive action of at least one of the involved road users. The total set of variables is meant to be a reﬂection of the observed subjects’ driving behaviour, or driving style. Changes in driving behaviour that are expected from any intervention, or brought about by any other changes in the preconditions for driving should be detected in the scores for these variables.

314

Driver Behaviour and Training – Volume III

In practice, the actual recording of behaviour and interaction starts after 15 minutes of the observation trip, as research on driver observation suggests that the presence of the observers is forgotten after some 10 to 20 minutes. The hypothesis is that it is not possible to hide types of behaviour that have been internalised (that is, habits or automatised behaviour) for a long period. For ﬁnancial reasons only one observer recording the data for standardised and non-standardised variables was used, but with a reduced number of variables to record for each trip. Small adaptations to these variables were also made. Questionnaires To understand how novice drivers estimate their own driving style, their behaviour, and so on in comparison to their on road observed driving behaviour, four questionnaires were used. Firstly, the Manchester Driving Behaviour Questionnaire (DBQ) contains questions about driving style, risk taking and lapses, errors and violations. Secondly, a questionnaire measuring self-evaluation of driver type was used developed by psychometrics for the Axa Gruppe (see http://www.autofahrertypen.de/). Participants had to answer questions about emotional reactions, their attitudes towards the vehicle, driving-related motives, and so on. Thirdly, participants took part in a group discussion and an evaluation sheet was developed for the purpose of recording responses to questions. It can be seen as a barometer of participants’ opinion about the discussion and the lessons they believe they have learned. Finally, the ISA questionnaire was developed to measure age, gender, driving experience and the participant’s own driving style, and what parents and friends think about their driving style. But predominantly, questions about trafﬁc safety, speed and the ISA system were asked. The work is based on the recent literature (as Engström et al., 2003; Lamszus, 2002; Mienert, 2002; 2003; Gregerson, 1996; Mourant et al., 1972; Prochaska and DiClemente, 1983; Veliver et al., 1998) on the topic and in-depth interviews with experts and a test–retest within the same sample. This assessment suggested that the right questions had been asked, but that the answers gave the impression that more and different answer categories would be possible. Therefore, in the ﬁrst round of interviews mostly open questions were asked. The answers were categorised, and standardised questions were added to the more open version of the ISA questionnaire so that the instrument consisted of both open questions as well as standardised ones using a tick box response format. For the open questions, the participants were asked to write their opinion in their own words as their answers to several questions, including ﬁnishing the phrase: ‘Trafﬁc safety means to me …’. Some other questions could be answered by ticking in one of several given boxes. Other questions were answered with the help of scales. Participants had, for instance, to decide to what extent they agreed with a given statement, on a ﬁve-step scale. For example, they would respond to ‘I generally feel safe in trafﬁc’ within a range from ‘strongly agree’ to ‘strongly disagree’. The ISA questionnaire was completed with each step of the study. All participants completed the instrument three times. Those who took part in the psychological

Comparison of Novice Drivers in Austria and the Czech Rebublic

315

group discussion completed the ISA questionnaire four times. The results reported here derive from the analyses of data from the ISA questionnaire. Psychological group training Half of the participants were asked to participate in a group discussion where issues like trafﬁc safety, speeding, different roles in trafﬁc and so on were discussed. The training was split into two parts and took place on two different days with a two week period in between; each part lasted two hours. As well as the group discussion method, small group team work and a role-play were also conducted. The main focus of the ﬁrst day was on motives, emotions and expectations related to driving. It was discussed how a car can be thought of, for instance, as a transport mode, as a hobby and so on, and which motives may be important while driving. Participants should think about possible consequences of certain motives; for example, a stressed driver may put pressure on other road users (for instance by driving too close), thus making them feel anxious. Another point of interest was how important safety was considered to be and what motives could affect safe behaviour. The last part of this session was a discussion about advantages and disadvantages of the use of an ISA-system and what needs, interests and feelings may be affected by the system. For example, the system could make driving more comfortable but frustrate the need for freedom of choice. The second part of the discussion was more focused on trafﬁc safety, speeding and the consequences of speeding, but also on communication with other road users. Participants had to describe, in their opinion, an unsafe driver, and how such a person could be characterised. Additionally, some information about speeds and consequences of speeds was given. Participants were informed that inappropriate speed is the main reason for accidents, and that pedestrians will die if they are hit by a car driving 50 km/h or more. They were also shown some photographs of cars that had hit a tree at different speeds. Participants were asked if they thought that speed limits were necessary, if there were speciﬁc conditions where limits are more or less important and so on. A major component of this section also was a role-play where participants had to think about the behaviour of different road users (car driver, cyclist and pedestrian) in crossing-situations where they would meet. For the present chapter only preliminary descriptive analyses for the ISA questionnaire is presented. For the future, more in-depth parametric analyses will reveal relationships between the battery of questionnaire measures and on-road behaviour measures. Results First of all it must be clariﬁed that most of the ﬁndings are probably not contributed to by cross-cultural differences, but from the different composition of the samples with respect to age, driving experience and gender. Concerning the self-evaluation part of the questionnaire, it was found that there was no difference between the Austrian and the Czech Republic sample regarding the

Driver Behaviour and Training – Volume III

316

questions about whether they believe they are safe drivers or not. All of them were quite sure that they drove safely. But there was a signiﬁcant difference concerning the answer to the question about how safe novice drivers felt in trafﬁc between the countries. In the Czech Republic, participants felt less safe than in Austria; the participants in the Czech Republic have less driving experience than those in Austria (see Figure 25.1).

5

4

2.81

3

I am a safe driver I generally feel safe in traffic

1.93

1.98

2

1.98

1 Czech Republic

Austria

Figure 25.1 Self-assessment and general feeling of safety on a ﬁve point scale (1 = ‘very safe’, 5 = ‘not safe at all’)

Behave according to law Ability to react fast

50%

14% 10% 7%

46%

Safety

16%

Czech Republic Austria

Good overview 0%

16% 23% 19%

Tolerance Driving anticipatory 0% 0%

38% 10%

20%

30%

40%

50%

60%

Figure 25.2 Frequency of responses to the question: ‘What is a good driver in your opinion?’

Comparison of Novice Drivers in Austria and the Czech Rebublic

317

It was very interesting how the question ‘how a good driver could be deﬁned’ was answered between the countries (see Figure 25.2). In the Czech Republic, it seems quite important that a driver behaves according to the law, drives safely (whatever ‘safe’ means in this context) and is tolerant. In Austria, novice drivers mostly think that a good driver drives in an anticipatory way, is tolerant, has a good overview and drives safely. The differences between the two countries may inﬂuence the willingness to use ISA. Concerning trafﬁc safety, in the Czech Republic novice drivers think that aggression, law breaking and speeding endanger others. Again there is a difference from Austria, where most people think that a driver who drives too fast endangers others. This is followed by driving recklessly, driving under the inﬂuence of alcohol, driving aggressively and not driving according to the law (see Figure 25.3).

Drives not according to law

32% 17% 50%

Drives aggressively Drives under the influence of alcohol

19% Czech Republic

0% 26%

Austria

23% 29%

Drives recklessly

32% 36%

Drives too fast 0%

20%

40%

60%

80%

100%

Figure 25.3 Responses to the question: ‘What types of drivers endanger other road users? Concerning speed, participants think that speed limits are necessary, although there is a trend for Austrian drivers to think that speed limits are more important (see Figure 25.4). On average, novice drivers think that speed limits in their country are appropriate, and they say that they mostly drive according to the law. Concerning this question, a signiﬁcant difference between the Czech and the Austrian sample can be found [(p < 0.034). In Austria, novice drivers are less willing to drive according to the law. Again, this may be due to the difference in age and driving experience between the samples. In both countries, participants say that they often overtake other vehicles, but signiﬁcantly more often in Austria (p < 0.013). In both countries most of the participants are not aware of the ISA system (see Figure 25.5). After a short explanation on how ISA works, participants provide possible advantages and disadvantages of ISA, although they had no experience of it so far. As shown in Figure 25.6, the expected advantages are quite common. In both countries novice drivers generally think that the system makes the driver aware of

Driver Behaviour and Training – Volume III

318

Speed limits are necessary

2,04 1,57

Speed limits in Austria/Czech Republic are ok

2,19 2,53

I drive according to speed limits

Czech Republic Austria

1,93 2,47

I often overtake other vehicles

3,56 2,96 1

2

3

4

5

Figure 25.4 Responses concerning speed limits measured on a ﬁve-point-scale 100% 90% 78%

80% 70%

64%

60% No

50%

Yes

36%

40% 30%

22%

20% 10% 0% Austria

Czech Republic

Figure 25.5 Awareness of ISA on a ﬁve-point scale speed limits. This is followed by the thought that ISA would allow the possibility to control one’s behaviour (in Austria) and that ISA can increase trafﬁc safety. Drivers also think that ISA has the potential to reduce average speed, it can be considered an assistance for the driver, as they do not have to look at the speedometer all the time and ISA can give feedback about the allowed speed at any speciﬁc point of the road. The disadvantage of ISA assumed by novice drivers shown in Figure 25.7 was that it could be a handicap if someone wanted to overtake. They explained that the system would take too much control so that the driver could not behave as they liked. In Austria, people also commented that the system could be ignored and therefore might be useless. They also assumed that there could be some delegation of responsibility and they worried about possible errors the system might produce (for

Comparison of Novice Drivers in Austria and the Czech Rebublic Increase in traffic safety

25%

5%

Reduction of speed

319

12%

19%

0%

Relieve the driver

12%

Czech Republic

Feedback about speed limits

19% 20%

Possibility to control one´s behaviour

19%

Austria

34%

Make aware of speed limits

31%

0%

20%

39%

40%

60%

80%

100%

Figure 25.6 Most frequent responses regarding advantages of ISA example, setting an erroneous limit and so on). However, some of the participants believe that ISA does not have any disadvantages at all. In sum, drivers in both countries were quite neutral concerning the question of whether they would use such a system or not, as shown in Figure 25.8. Discussion

14% 10%

Overtaking Too much control of the system Can be ignored

10% 10% 0%

10%

Czech Republic Austria

14% 12%

None Delegation of responsibility

10% 12% 5%

Errors 0%

14% 20%

40%

60%

80%

100%

Figure 25.7 Most frequent responses concerning assumed disadvantages of ISA Differences between samples of novice drivers in the Czech Republic and in Austria were found with respect to several aspects. But since the samples differ signiﬁcantly in age, gender distribution and driving experience, results have to be interpreted very

320

Driver Behaviour and Training – Volume III Mean

5

4 3.00

2.81

Austria

Czech Republic

3

2

1

Figure 25.8 Willingness to use ISA (1 = ‘certainly not’, 5 = ‘yes, certainly’) carefully. Further steps will be to apply more elaborate statistical methods toﬁnd out, for instance, whether differences between the samples are due to attitudinal differences or from demographic differences such as age and gender, and how the results from questionnaires correspond to observed on-road behaviour. Not least, the extent to which driving an ISA-equipped car and participating in a psychological group discussion training inﬂuence both the opinions and the behaviour of the subjects who take part in the study. References Almqvist, S., and Nygård, M. (1997). ‘Dynamic speed adaptation: ﬁeld trials with automatic speed adaptation in an urban area.’ Bulletin 154, Department of Trafﬁc Planning and Engineering, Lund University, Sweden. Berger, W.J. (1996). ‘Informationsaufnahme im Straßenverkehr.’ PhD Thesis, Universität für Bodenkultur, Austria. Carsten, O. and Comte, S. (1997). ‘UK work on automatic speed control.’ Proceedings of the ICTCT 97 conference. 5–7 November 1997, Lund, Sweden. Carsten, O. and Fowkes, M. (2000). ‘External vehicle speed control.’ Executive summary of project results. University of Leeds, UK. Comte, S.L. (1996). Response to Automatic Speed Control in Urban Areas: A Simulator Study. Institute for Transport Studies, University of Leeds. ITS Working Paper, no. 477. Comte, S.L. (1998a). Simulator Study on the Effects of ATT and non-ATT Systems and Treatments on Driver Speed Behaviour. Working Paper R 3.1.2 in the MASTER project. VTT, Espoo, Finland. Comte, S.L. (1998b). Evaluation of In-Car Speed Limiters: Simulator Study. Working

Comparison of Novice Drivers in Austria and the Czech Rebublic

321

Paper R 3.2.1 in the MASTER project. VTT, Espoo, Finland. Engström, I., Gregerson, N.P., Hernetkoski, K., Keskinen, E. and Nyberg, A. (2003). ‘Young novice drivers, driver education and training.’ Literature review, VTI rapport, 491A. Gregerson, N.P. (1996). ‘Young drivers’ overestimation of their own skill: an experiment on the relation between training strategy and skill.’ Accident Analysis and Prevention, 28(2), 243–50. Lamszus, H. (2002). Die Problematik junger Fahranfänger und Möglichkeiten zur Verringerung ihres hohen Unfallrisikos; Zeitschrift für Verkehrssicherheit. Heft 3, Seite 121; TÜV-Verlag GmbH, Köln. Mienert, M. (2002). Merkmale potenzieller Risikofahrer vor dem Führerscheinerwerb; Zeitschrift für Verkehrssicherheit. Heft 4, Seite 145; TÜV-Verlag GmbH, Köln. Mienert, M. (2003). Entwicklungsaufgaben Automobilität – Teil1; Psychische Funktionen des Pkw-Führerscheins für Jugendliche im Übergang ins Erwachsenenalter. Heft 1, Seite 26, TÜV-Verlag GmbH, Köln. Mourant, R.R. and Rockwell, T.H. (1972). ‘Strategies of visual search by novice and experienced drivers.’ Human Factors, 14, 325–35. Persson, H., Towliat, M., Almqvist, S., Risser, R. and Magdeburg, M. (1993). Hastighetsbegränsare i bil. Fältstudie av hastigheter, beteenden, konﬂikter och förarkommentarer vid körning i tätort. Lund University, Sweden. Prochaska, J.O. and DiClemente, C.C. (1983). ‘Stages and processes of self-change of smoking: toward an integrative model of change.’ Journal of Consulting and Clinical Psychology, 51, 390–5. Risser, R. and Lehner, U. (1996). The Wiener Fahrprobe. Internal Paper; Leeds. Várhelyi, A. (1996). ‘Dynamic speed adaptation based on information technology: a theoretical background.’ PhD Thesis, Lund University, Sweden. Várhelyi, A., Hydén, C., Hjälmdahl, M., Almqvist, S., Risser, R. and Draskóczy, M. (2002). ‘Effekterna av aktiv gaspedal i tätort.’ Sammanfattande rapport. LundaISA. [‘The effects of active accelerator pedal in built-up areas.’ Summary report.] Bulletin 210. Lund University, Sweden. Várhelyi, A., Hydén, C., Hjälmdahl, M., Draskóczy, M. and Risser, R. (2002). The Effects of Active Accelerator Pedal on Driver Behaviour and Trafﬁc Safety After Long Time Use in Urban Areas. Department of Technology and Society, Lund University, Sweden. Veliver, W.F., Prochaska, J.O., Fava, J.L., Norman, G.J. and Redding, C.A. (1998). ‘Smoking cessation and stress management: applications of the transtheoretical model of behaviour change.’ Homeostasis, 38, 216–33. http://www.uri.edu/ research/cprc/TTM/detailedoverview.htm. Internet: http://www.autofahrertypen.de/

This page intentionally left blank

PART 5 Human Factors and the Road Environment

This page intentionally left blank

Chapter 26

What Factors are Involved in Crashes, How Do We Measure Them and What Shall We Do About Them? Frank McKenna Reading University, UK Introduction The aim of the present paper is, as noted in the title, to address three separate questions. The ﬁrst aim is to provide an indication of which factors are involved in crashes. The second aim is to indicate how the factor might be measured and whether there are signiﬁcant individual differences. In the process I will focus on a project that is designed to provide computer assessment of the relevant risk factors at the individual level. The ﬁnal aim is to consider some of the countermeasures that might address the particular risk factors identiﬁed. Speed The relationship between speed and crash involvement can be documented in a variety of ways. For example, it is possible to look at the association between average speed and crash involvement, or one can examine the change in crash involvement when an intervention changes speed. Whichever way one addresses the issue, the conclusion remains that there is a straightforward relationship such that as the speed goes up so also does the crash involvement (Aarts and van Schagen, 2006; Finch et al., 1994; Richter et al., 2006). The next question is measurement. Here we are fortunate in that speed is transparent and can readily be measured. Indeed, in the UK free ﬂowing speeds have been shown to be decreasing. For example, in 1998, 70 per cent of car drivers were observed to break the 30 mph speed limit, whereas in 2005 this ﬁgure has come down to 50 per cent (Department for Transport, 2006). These ﬁgures tell us about speeding in general but not about the speed choices of one particular individual versus another. Is it the case that different individuals in the same circumstances choose very different speeds? Clearly it would be possible to put measurement devices into the vehicles and track the speed choices of the individual. We could force the drivers to drive the same route and hope that they do not come across different trafﬁc conditions. The alternative, which we will consider here, is the viability of presenting digital moving images of real driving scenes and asking drivers to make their speed choices using a computer. The advantage here is that trafﬁc conditions can be controlled. Another potential attraction is the simplicity with which measurement might proceed. The question of course is whether a

326

Driver Behaviour and Training – Volume III

measurement taken in this way is related to crash involvement. To assess this, a sample of 6870 was considered. They went through a range of measures including speed. It was found that drivers who were in the top ten per cent were 2.7 (Conﬁdence Interval CI 1.9–3.9) times more likely to have reported a crash in the preceding three years. Cronbach’s alpha for the measure was 0.85. It is clear then that there is a relationship between speed and crash involvement and that speed can readily be measured even at the individual level. The ﬁnal question concerns countermeasures. Here there is a wide range of measures such as speed humps and cameras (Hirst et al., 2005). Indeed technology is now available to control the speed of the vehicle. The prominence of speed is due to a combination of a number of features. One of course is the clear causal role in both the occurrence and severity of crashes. Another important feature is that it can readily be measured. And of course there is a wide range of countermeasures. It is this combination of features that make speeding so critically important. Close following It has been noted that in the United States about 30 per cent of all car crashes are rear-end collisions (Ben-Yaacov et al., 2002). In an observational study Evans and Wasielewski (1983) found a relationship between close following and crash involvement. This conﬁrms the common perception that the distance that you follow the vehicle in front will change your risk and raises the possibility that following behaviour may be trait-like. It has been reported that close following is the most frequent driving error on the road (Harvey et al., 1975). Different countries have different recommendations about the time gap that should be left between you and the vehicle in front. Many countries recommend a gap of two seconds. However, these recommendations are rarely observed. For example, Taieb and Shinar (2001) found that the gap that drivers judged as comfortable for their normal driving was just less than one second. In terms of measurement, the time gap between vehicles can be measured from the roadside or from within the vehicle. Similar to speed, there is an issue as to how to provide a simple measure of an individual’s close following. Again, as with speed, an attempt was made to examine a digital video measure of close following. It was found that drivers who were in the top ten per cent in terms of close following were 1.6 times more likely to have reported a crash in the preceding three years (CI 1.2–2.2). Cronbach’s alpha for the measure was 0.95. There is a range of countermeasures for close following. For example, Sivak and Flannagan (1993) explored fast-rise brake lamps that facilitate fast reaction times by providing drivers with a brake light that lights up faster. In-vehicle warning devices have been shown to increase following times (Ben-Yaacov et al., 2002). Another measure is high mounted brake lights which are designed to hasten the response of following drivers (McKnight and Shinar, 1992). A fourth measure is roadside vehicle activated signs that indicate to following vehicles that they are driving too close (Helliar-Symons, 1983). Another feedback method is chevrons on the road, which provide drivers with a simple method of determining a recommended following distance (Helliar-Symon et al., 1995).

What Factors are Involved in Crashes

327

Alcohol The association between drink driving and accident involvement has been known for some time and will not be considered in any detail here. Many societies have made considerable progress in reducing drink driving. A signiﬁcant advance in enforcement came with the introduction of per se laws, which allowed a prosecution on the basis of the presence of alcohol. (It was no longer necessary to prove that the individual was impaired.) This meant that measurement could proceed through simple devices such as the breathalyser. A current challenge for most societies is to determine the maximum level of alcohol that is permitted for drivers. Mann (2002) has noted that the level adopted will depend on the criterion chosen. For example, if skill impairment is chosen then the research evidence would support a low level of alcohol such as 0.02 blood alcohol concentration (BAC). If the criterion chosen is an elevated collision risk then it has been noted that BACs between 0.05 and 0.07 is associated with an increased crash risk (Zador et al., 2000). Legal and enforcement countermeasures have been dominant. A range of measures have been shown to be effective. These include providing legal blood alcohol concentration levels, having lower permitted BAC levels for younger drivers, raising the minimum age for legal drinking and random breath testing (Shults et al., 2001). Violations It has been proposed, with factor analytic support, that human error can be divided into three factors: errors, lapses and violations (Reason et al., 1990). Errors have been deﬁned as deviations from planned actions, lapses as failures of attention and violations as deliberate infringements of safe practices. It has been argued that it is only violations that are related to accident involvement (for example, Lawton et al., 1997). It is fairly clear that speeding and drink driving can readily be subsumed under the heading of violations. It is less certain whether close following should be subsumed under this category. Close following could be considered to reﬂect thrill seeking and hence be a violation (Taieb and Shinar, 2001). Alternatively, it might reﬂect perceptual difﬁculties in determining headway when the vehicle in front appears relatively stationary in one’s visual ﬁeld (Reinhardt-Rutland, 1985). Other factors, such as running trafﬁc lights on red and failure to use seat belts, can readily be considered as violations. There is little doubt that violations are associated with a higher crash rate (see Little (1966) for a discussion of early work). More recently Gerbers and Peck (2003) have found that those drivers who have committed a violation have a subsequent accident rate that is more than three times higher than those with no violations. In terms of developing an individual difference measure, clearly we could consider the violation history of the individual. As just noted, that clearly works. However, the majority of drivers will have no violations. This, of course, may simply mean that they have not yet been caught. Recent work has focussed on self-reports. Parker et al. (1995), for example, have demonstrated the relationship between a self-report violation measure and crash involvement. In my own version of a violation measure, I ﬁnd that those who are in the top ten per

328

Driver Behaviour and Training – Volume III

cent are 2.8 times more likely to have been in a crash than those in the bottom ten per cent (CI 2.1–3.7). Cronbach’s alpha for the measure was 0.82. Countermeasures for particular violations such as speeding have already been discussed. More general countermeasures include a punishment system that provides ﬁnes and a points system that can lead to disqualiﬁcation. Clearly deterrence can work, but one has to be clear about the conditions under which deterrence generally has an effect. Deterrence theory proposes that a change in behaviour will follow from increases in punishment severity, celerity (imminence) and certainty. However, as Nagin and Pogarsky (2001) note, there is sound empirical support only for certainty. For example, the role of severity of punishment while maintaining a high proﬁle in criminology has not been supported with solid evidence. One prominent review concluded that there is no association between the severity of punishment and the level of crime in society (Doob and Webster, 2003). These results have important implications for trafﬁc violations, which historically have been characterised by levels of detection certainty that are probably too low to deter. Speed cameras get round this problem by providing an area where certainty of detection can be directly controlled. However, for speed violations outside the camera area and other violations such as use of a mobile phone while driving, it is likely that certainty of detection is too low to provide signiﬁcant deterrence. Fatigue In order to perform most tasks some degree of attention is required. At one end of the attention continuum individuals are unresponsive to environmental stimuli and may be asleep. McKenna (forthcoming) reported that in a survey of over 10 000 drivers, 16 per cent indicated that they had fallen asleep at the wheel in the last two years. Estimates of sleep-related crashes vary between 9 and 16 per cent for all accidents and between 15 and 20 per cent for motorway accidents (Horne and Reyner, 1995; Maycock, 1996). Horne and Reyner (1999) have argued that sleep-related accidents are more liable to be serious, in large part because there is little evidence of avoidance behaviour in the form of steering or braking so recipients receive the full impact of the force. Measurement of individual differences in susceptibility to sleep-related crashes presents some daunting challenges. Sleepiness is clearly a transient state. If sleeprelated crashes were due to random transient sleep deprivation states, then it is difﬁcult to see how they could be predictable. However, it is possible that these states are not random and may be predictable. For example, we might anticipate that those with a sleep disorder may be more vulnerable. It has been found that those who suffer sleep apnoea (a breathing disruption while sleeping) are more than six times more likely to have had a trafﬁc accident (Teran-Santos et al., 1999). Although we might anticipate sleep disorders may be rare in the population, there is epidemiological work indicating that sleep apnoea may occur often unrecognised in one in 20 adults and in a milder normally unrecognised form in one in ﬁve adults (Young et al., 2002). A different approach to assessing susceptibility to fatigue related crashes may be to consider a factor that might be expected to disrupt normal sleep patterns, namely

What Factors are Involved in Crashes

329

shift work. In an analysis of 6870 drivers attending speed awareness workshops after being caught speeding, I ﬁnd that those who report regularly doing a night shift were 1.4 times more likely to have been in a crash (CI 1.1–1.7) and 2.1 times more likely to have fallen asleep at the wheel (CI 1.7–2.6). Following up the prospect that it might be possible to devise a measure of susceptibility to sleep-related crashes, I developed several measures. The rationale for the measures was as follows. First, a measure of sleepy driving was developed on the basis that those who drive more frequently when they are tired or who drive at times of the day associated with more sleepy driving might be expected to be exposed to more risk. A second measure was that of daytime sleepiness, on the basis that those who generally feel sleepy may be at more risk. Two other measures were developed to try to understand the factors that might underlie daytime sleepiness and sleepy driving. One measure was designed to assess sleep quality on the basis that if individuals suffer poor sleep, then it would be expected that this would have a knock on effect on daytime sleepiness and potentially falling asleep at the wheel. The other measure that was developed was sleep hygiene, on the basis that those individuals who have poor sleep habits may well suffer excessive daytime sleepiness and may be more prone to fall asleep at the wheel. Let us ﬁrst consider the measure of sleepy driving. One obvious criterion against which to assess the validity of the measure is whether it is associated with having fallen asleep at the wheel. Those who are in the top ten per cent on this measure were 26 times more likely to have fallen asleep at the wheel than those in the bottom ten per cent (CI 13.1–50.8). Those who are in the top ten per cent on sleepy driving were 3.5 times more likely to have been in a crash than those in the bottom ten per cent (CI 2.4–5.0). Reliability was assessed using Cronbach’s alpha and found to be 0.83. Those who suffer excessive daytime sleepiness (in the top ten per cent) were 12.5 times more likely to have fallen asleep at the wheel than those in the bottom ten per cent (CI 7.6–20.5). Those who suffer excessive daytime sleepiness (in the top ten per cent) were also 2.4 times more likely to have had a crash than those in the bottom ten per cent (CI 1.7–3.4). Cronbach’s alpha for the measure was 0.87. Interestingly, a measure that has nothing to do with driving is predicting crash risk. Another measure that clearly has nothing to do with driving is sleep quality. What we ﬁnd is that those who are in the group with the poorest ten per cent of sleep quality were 4.5 times more likely to have fallen asleep at the wheel than those in the best ten per cent (CI 3.0–6.6). Perhaps even more surprising is that those in the poorest ten per cent, in terms of sleep quality, were twice as likely to have been in a crash than those in the best ten per cent (CI 1.5–2.7). Cronbach’s alpha for the measure was 0.85. Our ﬁnal measure was sleep hygiene, which considers those sleep habits that might promote or detract from sleep quantity and quality. Here we ﬁnd that those who are in the poorest ten per cent in terms of sleep hygiene were 4.5 times more likely to have fallen asleep at the wheel than those in the best ten per cent. Those who are in the poorest ten per cent in terms of hygiene were also 2.8 times more likely to have been in a crash than those in the top ten per cent (CI 2.0–3.8). Cronbach’s alpha for the measure was 0.64. (It is likely that the absence of a high alpha is due to fact that a construct such as sleep hygiene is likely to be multidimensional, consisting of

330

Driver Behaviour and Training – Volume III

factors as diverse as exercising before going to bed and whether one drinks alcohol shortly before going to bed.) Overall, it is clear that sleep is related to driving risk and that individual difference measures can be constructed that have a signiﬁcant relationship with driving risk. Our ﬁnal question concerns countermeasures. Perhaps the most obvious countermeasure emerges as a function of the sleep hygiene result. The clear strategy would be to create good sleep habits that avoid exposure to excessive sleepiness. In other words, maximising the opportunity for adequate sleep is the obvious ﬁrst step. Of course, for some individuals sleep disruption and exposure to risk is not under their control. However, a number of other countermeasures are also available. For example, the beneﬁcial effects of naps have been identiﬁed (for example, Macchi et al., 2000). There is a range of other measures that have received empirical support. For example, caffeine has long been known to have an effect on alertness and has a range of cognitive effects (Lorist and Tops, 2003). Rest breaks have been rather less well researched, but there is evidence, for example, that accident risk in the last half-hour of two hours of continuous work is twice that in the ﬁrst half-hour (Tucker, Folkard and Macdonald, 2003). There is a last ditch countermeasure which attempts to alert drivers as they drift off to sleep. Rumble strips on the shoulder of highways have been shown to reduce run-off-the-road accidents (Grifﬁth, 1998). Hazard perception The topic of hazard perception has been covered in a number of papers (McKenna and Horswill, 1999; Horswill and McKenna, 2004) and will be considered only brieﬂy here. The idea is that the ability to read the road and anticipate what other drivers might do is a skill that develops with experience and training. A variety of measurement approaches have been explored. Pelz and Krupat (1974) and Watts and Quimby (1979) employed a technique in which drivers moved a lever to indicate how safe or unsafe they considered ﬁlmed hazards. McKenna and Crick (1994) removed the simulator that Watts and Quimby had employed and measured the time it took drivers to respond to particular hazards. The obvious countermeasures are training and experience. In small-scale studies it has been shown that it is possible to train people in hazard perception (McKenna and Crick, 1994; 1997). More recently, McKenna et al. (2006) examined the effect of hazard training on risk taking. They noted that there are arguments in the literature that skill-based training could lead to an increase, a decrease, or no change in risk taking. They found that training in hazard perception led to a decease in risk taking. Experience as a countermeasure is, of course, problematic because it involves exposure to risk. One solution as noted by Gregersen et al. (2000) is to increase supervised experience. This did result in a reduction in accidents though the extent to which hazard perception is directly involved is unknown. What underlies these varies risk factors? For hazard perception and sleepy driving we have speculated on some of the causes that might underlie these factors. For sleepy driving we have considered factors

What Factors are Involved in Crashes

331

such as sleep quality, sleep hygiene and shift work, all of which can be expected to have an impact on this risk factor. For hazard perception, we considered experience and training. What, then, about factors such as speed, close following, alcohol and violations in general? First, we might consider whether these measures reﬂect a single underlying trait. In principle they might all be considered as violations. (The alcohol measure is the number of units consumed prior to driving.) Table 26.1 presents the correlations. Table 26.1 The intercorrelation of speed, close following, alcohol and violations Speed Speed Close following Alcohol Violations

Close following 0.141

Alcohol 0.204 0.047

Violations 0.358 0.058 0.368

Given the fact that the sample is over 6000, all of the correlations are signiﬁcantly different from zero. What is of more interest is the magnitude of the correlations. It is fairly clear that the different factors are relatively independent of each other. The general violation factor does share some variance with speed and alcohol. Close following shares relatively little variance with the other factors raising the possibility that close following may have a perceptual component in addition to a small violation component. Given the relative independence of the different factors, there is a rationale for considering each of the factors separately. Sensation seeking has been proposed as an underlying inﬂuence for at least speed, alcohol and violations (Jonah, 1997). While no measure of sensation seeking was included in the database under current consideration, there was a domain speciﬁc measure of thrill seeking within driving. Following a similar line of enquiry, there was a module measuring the use of the vehicle as an emotional outlet. Table 26.2

The correlations between the two potential underlying factors (thrill and emotional outlet) and the risk factors

Speed Thrill Emotional outlet

0.252 0.264

Close following 0.059 0.160

Alcohol

Violations

Fatigue

0.182 0.174

0.354 0.405

0.240 0.251

Hazard Perception –0.065 0.012

332

Driver Behaviour and Training – Volume III

The correlations are presented in Table 26.2. Again, since the sample is over 6000, all correlations are signiﬁcant (except the relationship between emotional outlet and hazard perception) so the focus is on the magnitude of the relationship. It is worth adding that the correlation between thrill and emotional outlet was r = 0.425, p < 0.001 so while there is overlap between the measures they are by no means identical. Let us consider each of the risk factors. For speed there is a relationship with measures of emotion (thrill and emotional outlet), though it is not substantial. This reinforces the ﬁnding of McKenna (2005) that the majority of speeding offenders report that enjoying speed had little inﬂuence on their speeding offence. For close following there is a signiﬁcant but insubstantial relationship with measures of emotion, again suggesting that close following is different from the other factors. For alcohol, there is a relationship with measures of emotion, but again the relationship is not large. The relationship between the violation factor and measures of emotion was more substantial. For hazard perception the relationship with emotion was minimal; but perhaps the most intriguing result was the relationship between measures of emotion and fatigue. This signiﬁcant, though not large, relationship raises the prospect that sleepy driving is not just a biological phenomenon. Clearly, if we disrupt the sleep of anyone and then require them to drive we will have sleepy drivers. However, the relationship with measures of emotions raises the possibility that individuals vary in the extent to which they permit themselves to be exposed to sleepy driving. This possibility was explored in the following way. First, if sleepy driving were a purely biological phenomenon, then we would expect no relationship with violations. By contrast, if sleepy driving were in part a function of willingness to expose oneself to risk by breaching normal safety conditions, then one might anticipate a relationship. The correlation between sleepy driving and violations was r = 0.428, p < 0.001. Another way of examining this relationship is to consider whether the relationship between sleepy driving and violations is different when there is a biological impediment to sleep. Consider the relationship when we compare those who regularly do night shift versus those who do not. We might predict that for those who regularly do night shift there will be more of a biological component and hence the relationship with violations should be diminished. Consistent with this prediction, the correlation between violations and sleepy driving was r = 0.248, p < 0.001 for the 666 individuals doing nightshift but for those not doing nightshift (6061) the correlation was r = 0.458. It is clear that when drivers are more free from the biological impediment of sleep deprivation, there is a more powerful relationship between violations and sleepy driving. Conclusion The human factors associated with crash involvement are relatively clear. It is argued that we can not only identify these factors, but that we can provide individual difference measures for these factors. It is also argued that we have a wide range of countermeasures. As our understanding of the causal factors underlying these factors increases, so also will our ability to reﬁne the available countermeasures.

What Factors are Involved in Crashes

333

References Aarts, L., van Schagen, I. (2006). ‘Driving speed and the risk of road crashes: a review.’ Accident Analysis Prevention, 38, 215–24. Ben-Yaacov, A., Malz, M. and Shinar, D. (2002). ‘Effects of an in-vehicle collision avoidance warning system on short-and long-term driving performance.’ Human Factors, 44, 335–42. Department for Transport (2006). Transport Statistics Bulletin: Vehicle Speeds in Great Britain, (2005). Department for Transport: London, UK. Doob, A.N. and Webster, C.M. (2003). ‘Sentence severity and crime: accepting the null hypothesis.’ In M. Tonry (ed.), Crime and Justice: A Review of Research, vol. 25, Chicago: University of Chicago Press, 143–95. Evans, L. and Wasielewski, P. (1983). ‘Risky driving related to driver and vehicle characteristics.’ Accident Analysis and Prevention, 15, 121–36. Finch, D.J., Kompfner, P., Lockwood, C.R. and Maycock, G. (1994). ‘Speed, speed limits and crashes.’ Project Record S211G/RB/Project Report PR 58, Transport Research Laboratory TRL, Crowthorne, Berkshire. Gerbers, M.A. and Peck, R.C. (2003). ‘Using trafﬁc conviction correlates to identify high accident-risk drivers.’ Accident Analysis and Prevention, 35, 903–12. Gregersen, N.P., Berg., H.Y., Engstrom, I., Nolen, S., Nyberg, A. and Rimmo, P.A. (2000). ‘Sixteen years limit for learner drivers in Sweden: an evaluation of safety beneﬁts.’ Accident Analysis and Prevention, 32, 25–35. Grifﬁth, M.S. (1999). ‘Safety evaluation of continuous shoulder strips installed on Freeways.’ Transport Research Board, No. 990162. Harvey, C.F., Jenkins, D. and Sumner, R. (1975). ‘Driver error.’ Transport and Road Research Laboratory. Supplementary Report 149UC. Helliar-Symons, R.D. (1983). ‘Automatic close-following warning sign at Ascot.’ Laboratory Report, LR 1095. Crowthorne: TRL Limited. Helliar-Symons, R., Webster, P. and Skinner, A. (1995). ‘The M1 chevron trial.’ Trafﬁc Engineering and Control, 36, 563–7. Hirst, W.M., Mountain, L.J. and Maher, M.J. (2005). ‘Are speed enforcement camera more effective than other speed management measures? An evaluation of the relationship between speed and accident reductions.’ Accident Analysis and Prevention, 37, 731–41. Horne, J. and Reyner, L., (1995). ‘Sleep-related vehicle accidents.’ British Medical Journal, 310, 565–7. Horne, J. and Reyner, L. (1999). ‘Vehicle accident related to sleep: a review.’ Occupational and Environmental Medicine, 56, 289–94. Horswill, M.S. and McKenna, F.P. (2004). ‘Drivers’ hazard perception ability: situation awareness on the road.’ In S. Banbury and S. Tremblay (eds), A Cognitive Approach to Situation Awareness, Aldershot: Ashgate, 155–75. Jonah, B.A. (1997). ‘Sensation seeking and risky driving: a review and synthesis of the literature.’ Accident Analysis and Prevention, 29, 651–65 Lawton, R., Parker, D., Stradling, S.G. and Manstead, A.S.R. (1997). ‘Predicting road trafﬁc accidents: the role of social deviance and violations.’ British Journal

334

Driver Behaviour and Training – Volume III

of Psychology, 88, 249–63. Little, A.D. (1966). The State of the Art of Trafﬁc Safety: A Critical Review and Analysis of the Technical Information on Factors Affecting Trafﬁc Safety. Cambridge, Massachusetts: Arthur D. Little Inc. Lorist, M.M. and Tops, M. (2003). ‘Caffeine, fatigue, and cognition.’ Brain and Cognition, 53, 82–94. Macchi, M.M., Boulos, Z., Ranney, T., Simmons, L., and Cambell, S.S. (2000). ‘Effects of an afternoon nap on night-time alertness and performance in long-haul drivers.’ Accident Analysis and Prevention, 34, 825–34. Mann, R.E. (2002). ‘Choosing a rational threshold for the deﬁnition of drunk driving: what research recommends.’ Addiction, 97, 1237–8. Maycock, G. (1996). ‘Sleepiness and driving: the experience of UK car drivers.’ Journal of Sleep Research, 5, 229–37. McKenna, F.P. (2005). ‘Why do drivers break the speed limit?’ In Behavioural Research in Road Safety, 15. Department for Transport. http://www.dft.gov.uk/ pgr/roadsafety/research/behavioural/ﬁfteenthseminar/ﬁfteenthseminarpdf. McKenna, F.P. (forthcoming). ‘Human factors in road trafﬁc accidents.’ In Ayers et al. (eds), Cambridge Handbook of Psychology, Health and Medicine (2nd ed.), Cambridge, UK: Cambridge. McKenna, F.P. and Crick, J.L. (1994). ‘Hazard perception in drivers: a methodology for testing and training.’ Transport Research Laboratory, Contractor Report 313. Crowthorne: TRL Limited. McKenna, F.P. and Crick, J.L. (1997). ‘Developments in hazard perception.’ Transport Research Laboratory Report 297. Crowthorne: TRL Limited. McKenna, F.P. and Horswill, M. (1999). ‘Hazard perception and its relevance for driver licensing.’ IATTS Research, 23, 36–41. McKenna, F.P., Horswill, M. and Alexander, J.L. (2006). ’Does anticipation training affect drivers’ risk taking?’ Journal of Experimental Psychology: Applied, 12, 1–10. McKnight, A.J. and Shinar, D. (1992). ‘Brake reaction time to high-mounted stop lamps on vans and trucks.’ Human Factors, 34, 205–13. Nagin, D.S. and Pogarsky, G. (2001). ‘Integrating celerity, impulsivity and extralegal sanction threats into a model of general deterrence: theory and evidence.’ Criminology, 39, 404–30. Peltz, D.C. and Krupat, E. (1974). ‘Caution proﬁle and driving record of undergraduate males.’ Accident Analysis and Prevention, 6, 45–58. Parker, D., Reason, J.T., Manstead, A.S.R. and Stradling, S.G. (1995). ‘Driving errors, driving violations and accident involvement.’ Ergonomics, 38, 1036–48. Reason, J.T., Manstead, A.S.R., Stradling, S.G., Baxter, J.S. and Cambell, K. (1990). ‘Errors and violations on the road: a real distinction.’ Ergonomics, 33, 1315–32. Reinhardt-Rutland, A.H. (1985). ‘Nonveridical factors of visual perception and close following on the road.’ Perceptual and Motor Skills, 61, 255–8. Richter, E.D., Berman, T., Friedman, L., Ben-David, G. (2006). ‘Speed, road injury, and public health.’ Annual Review of Public Health, 27, 125–52. Shults, R.A., Elder, R.W., Sleet, D.A., Nichols, J.L., Aloa, M.O., Carande-Kulis, V.G., Zaza, S., Sosin, D.M. and Thompson, R.S. (2001). ‘Reviews of evidence

What Factors are Involved in Crashes

335

regarding interventions to reduce alcohol-impaired driving.’ American Journal of Preventive Medicine, 21, 66–88. Sivak, M. and Flannagan, M. (1993). ‘Fast-rise brake lamps as a collision-prevention device.’ Ergonomics, 36, 391–5. Taieb, M. and Shinar, D. (2001). ‘Minimum and comfortable driving headways: reality versus perception.’ Human Factors, 43, 159–72. Teran-Santos J., Jimenez-Gomez, A. and Cordero-Guevaro, J. (1999). ‘The association between sleep apnea and the risk of trafﬁc accidents.’ New England Journal of Medicine, 340, 847–51. Tucker, P., Folkard, S. and MacDonald, I. (2003). ‘Rest breaks and accident risk.’ The Lancet, 361, 680. Watts, G.R. and Quimby, A.R. (1979). ‘Design and validation of a driving simulator for use in perceptual studies.’ TRRL Report LR907. Transport and Road Research Laboratory, Crowthorne. Young, T., Peppard, P.E. and Gottlieb, D.J. (2002). ‘Epidemiology of obstructive sleep apnoea.’ American Journal of Respiratory and Critical Care Medicine, 165, 1217–39. Zador, P.L., Krawchuk, S.A. and Voas, R.B. (2000). ‘Alcohol-related relative risk of driver fatalities and driver involvement in crashes in relation to driver age and gender: an update using 1996 data.’ Journal of Studies on Alcohol, 61, 387–95.

This page intentionally left blank

Chapter 27

Driver Training and Assessment: Implications of the Task-Difﬁculty Homeostasis Model Ray Fuller Trinity College Dublin, Ireland Introduction The aim of this paper is to explore some of the implications of the Task-Difﬁculty Homeostasis model for the development of driver training and assessment. The present formulation of the model is based on earlier representations (see Fuller, 2000 and Fuller and Santos, 2002) described as the Task-Capability Interface model. However the model has undergone some recent development and this has called for a new emphasis on the model’s properties, hence the revised nomenclature. I will begin with a brief exposition of the driver’s decision-making processes that the model attempts to capture. Task-difﬁculty homeostasis The model starts with the premise that drivers drive in such a way as to maintain perceived task difﬁculty within a preferred range. Perceived task difﬁculty (box 3 in Figure 27.1) arises out of the difference between the perceived demands of the task and the driver’s perceived available capability for it (boxes 1 and 2). It is inversely related to the size of the difference between the two: the smaller the separation, the greater is task difﬁculty. Speed directly affects the demands of the driving task. For any given road and trafﬁc scenario, the faster one travels the less the available time to take information in, process it, make decisions, execute those decisions and make any necessary error corrections. Thus drivers may adjust their speed to maintain perceived difﬁculty within their preferred range. Increases in task difﬁculty may simultaneously be experienced as increases in feelings of risk (Fuller, 2005a). This is hardly surprising, given the likely punishing consequences of loss of control of the task. However this enables us to refer to the upper limit of task difﬁculty which a driver is prepared to accept, as the driver’s risk threshold (see box 5). Thus, drivers generally choose a speed (boxes 6 and 7) such that the difﬁculty of the task falls within the range they are prepared to accept and which in particular does not exceed their risk threshold. This process is known as Task-Difﬁculty Homeostasis. It involves an ongoing comparison (box 4) between perceived task

Driver Behaviour and Training – Volume III

338

difﬁculty (box 3) and the range of acceptable difﬁculty (box 5) and the use of variation in speed as the principal method of homeostatic control. disposition to adopt particular range of task difficulty and risk threshold

disposition to comply with speed limit

immediate influences on risk threshold effort motivation

immediate effort

goals of journey

immediate driving goals

range of acceptable task difficulty and risk th reshold 5

immediate influences on compliance comparator 4

decision and response 6 perceived task difficulty 3

perceived capability 2

human factor variables

education training experience

physiological competence

effects on vehicle speed 7

perceived t ask demand 1

road environment behaviour of other road users

route time-of-day

vehicle characteristics

Figure 27.1 Representation of the process of Task-Difﬁculty Homeostasis What are the key determinants of each of the main elements involved in this process? From more distal to more proximal determinants, perceived task demand arises out of a complex combination of the handling characteristics of the driver’s vehicle, the nature of the route chosen, time of day on that route, characteristics of the road environment (particularly visibility and surface adhesion) and the presence and behaviour of other road users (see sequence of inputs to box 1). Perceived capability arises fundamentally out of the driver’s physiological competence (for example, visual acuity, information processing and reaction speeds, strength, ﬂexibility) and the accumulation of declarative and procedural knowledge resulting from education, training and experience (see sequence of inputs to box 2). However this level of competence may be undermined to varying extents by a host of human factor variables, such as fatigue and drowsiness, mood state, stress and

Driver Training Assessment

339

attentional capture. The resulting effect may thus be an available level of capability which is something less than the driver’s unimpaired competence. The level of perceived capability is one key determinant of the range of task difﬁculty and risk threshold that a driver is prepared to accept (see link from box 2 to box 5). But there are several others, such as the goals of the driving task (which can be general, for example a rush to work versus leisurely exploration of countryside, or more immediate, such as a desire to overtake a slow truck that has been holding the driver up). The amount of effort the driver is prepared to assign to the driving task may also affect the range of acceptable task difﬁculty (see pathways to box 5). Evidence strongly suggests, however, that there are individual differences in the disposition to adopt a particular range of task difﬁculty and risk threshold, with some drivers opting for a low level of difﬁculty and others opting for a very much higher level (see Fuller et al., 2006). Overlaid on these dispositional inﬂuences are more immediate inﬂuences such as an acute emotional state of frustration and/or anger (see upper pathway to box 5). Finally the model needs to take account of the fact that this entire process may result in a speed choice which is higher than the posted limit for the road segment the driver is travelling. Thus we need to include the driver’s disposition to comply with the limit in such circumstances and include also any immediate inﬂuences which might affect compliance, such as the presence of a police vehicle or the sighting of a safety camera up ahead (see pathway to box 6). The calibration problem As can be seen (in the pathway to box 2) the model takes explicit account of the contribution of education and training to the determination of driver capability. However it must be noted that the driver’s perception of task difﬁculty will be partly based on his or her perceived level of capability, and partly based on his or her perceived level of task demand. If the driver overestimates his or her capability, this will translate into an underestimate of task difﬁculty and this may in turn lead to the adoption of too high a speed. A similar result arises if the driver underestimates task demand. Again, task difﬁculty would be underestimated and compensatory adjustments in speed may be too high. This is sometimes referred to as the ‘calibration’ problem and is typically associated with inexperienced drivers and a greater likelihood of loss of control and collision (Matthews and Moran, 1986; Brown and Groeger, 1988; Gregersen and Bjurulf, 1996; Deery, 1999). Because of the calibration problem arising out of poor estimation of objective task demand, some jurisdictions have introduced some form of hazard perception test into formal driver assessment for licensing. Apart from providing a measure of the driver’s competence in this area, such formal assessments may encourage more emphasis in driver training on this relatively neglected aspect of the driving task (Groeger and Clegg, 1994). However such tests, albeit well-intentioned, are designed in such a way that the testee is a passive observer of unfolding hazards to which she or he has to respond in some way. This process obviously disregards the fact that a signiﬁcant, and probably

340

Driver Behaviour and Training – Volume III

the larger proportion of hazards normally arises out of the interaction between what the driver does and some other event(s) on, or feature of, the roadway. To capture this important feature therefore, it would seem worth reiterating the potential beneﬁts of the development of simulation training and assessment for drivers, and for that matter, motorcyclists (see Fuller, 2005b). Simulator driving experience provides, of course, for the dynamic creation of hazards which arise out of the behaviour of the trainee. A simple example would be the creation of a hazardous situation arising out of the driver attempting to take a corner at too high a speed. But simulator training has the potential to offer so much more than dynamic learning experiences: • • • • • •

accelerated exposure to varieties of road experience provision of enhanced and multiple perspective feedback unlimited repetition of elements automated and objective assessment demonstration of model driving safe environment for practice

Simulators are already widely accepted for training in basic driving control skills and advanced moving-base simulation technology is now available for much more comprehensive training of professional bus and truck drivers over a wide range of road and trafﬁc scenarios. Thus road transport is already following the evident success of simulation training for aircrew. To extend the application of this technology into standard training practice, it needs, ﬁrst of all, research to demonstrate its effectiveness. After that, market pressures and economies of scale may achieve its introduction, even without any political push for it. But beyond this, the extension of the application of simulator driving to driver assessment offers really important further beneﬁts, not the least of which are its objectivity, repeatability, comprehensiveness and the opportunity for systematic evaluation of key competencies in developing drivers. Although there have been moves to address the issue of improving the perception of task demand by the trainee driver, a similar response to the calibration problem on the perceived capability side of the task difﬁculty equation (box 2) does not seem to have taken place to the same extent. Certainly in these islands it would be difﬁcult to ﬁnd systematic training and assessment in knowledge of human factor variables and of how to manage their inﬂuence on driving safety. This issue will be reviewed below, but before that it is worth noting that the model describes a critical problem for the novice driver as they develop ever increasing competence: the problem of shifting from reactive to anticipatory responding. From reactive to anticipatory responding In the early stages of learning, task difﬁculty will be great, in part because the driver is unfamiliar with vehicle controls – their use demands a signiﬁcant proportion of conscious attention because such use is not yet automated. Thus novice drivers typically drive more slowly over any road segment than more experienced drivers.

Driver Training Assessment

341

What the driver learns in these early stages is that he or she usually has time to react to unexpected obstructions on the roadway ahead (such as a leading vehicle suddenly slowing or a pedestrian moving out from a parked vehicle) or unexpected changes in task demand (such as a bend having a much smaller radius than expected). As the driver develops more competence with automatisation of vehicle control responses, and as perceived task difﬁculty therefore simultaneously decreases (with the perceived gap between capability and demand ever widening), the driver will progressively drive the same segment faster. However this increase in speed has very important implications for the time available to respond to any unexpected hazard. Thus if speed is increased from 60 km/h to 100 km/h on a dry road, the increase in stopping distance (reaction plus braking distance) is over 53 metres (Road Safety Authority, 2007). This places a premium on the driver shifting from a reactive to an anticipatory driving style. What might have been avoided in the past by simply reacting to it when it occurred will no longer work. The driver now needs to look well ahead and prepare for possible increases in task difﬁculty, rather than wait for them to happen before responding. The disturbing possibility is that it may be just as the driver is entering this phase of development that they undertake and pass the driving test, creating the conditions of being left alone to discover through perhaps painful experience the need to ‘expect the unexpected’, as an old public service message used to proclaim. Is it no wonder, then, that drivers are most vulnerable to collision and single vehicle accidents in the months following passing the driving test? If there was ever a fundamental theoretical argument for the introduction of graduated training and assessment, this must be it. Human factor variables Reverting to human factor variables which may undermine driver competence, the ﬁrst issue for driver education and training is to raise driver awareness of these variables and of their impact on capability. At this point in time, both legislation and widespread publicity have highlighted the inﬂuence of alcohol, although it is remarkable that in Ireland drivers’ knowledge of what the legal permitted limit is still has some way to go: most drivers overestimate the limit (Hibernian Motoring Report, 2007). Nevertheless there is strong support for the introduction of a zero blood alcohol limit and only 17 per cent of drivers believe the limit should stay as it is (80 mg/100 ml) (Gormley and Fuller, 2005). Human factors, however, encompasses a potentially large agenda, including for example the effects of: • • • • •

prescription and illicit drugs fatigue, drowsiness and circadian variation in alertness mood and emotion illusions (such as speed adaptation and overestimation of others’ speeding and aggressive driving) fallacies (such as the Speed/Time Fallacy; see Stradling et al., 2007).

342

Driver Behaviour and Training – Volume III

But of course awareness of these factors and their effects alone is usually not enough. Apart from being able to detect the presence of particular factors, it is usually recommended to teach trainees what to do about them. A little of this teaching has already ﬁltered through to public information campaigns, such as what to do when feeling drowsy to prevent falling asleep at the wheel. However, the same cannot yet be said for dealing, for example, with the management of emotion and stress (McKenna, 2005), the misperception of others’ behaviour (Åberg et al.,1997; Gerrard et al., 1996) or the inﬂuence of fallacious thinking (Stradling et al., 2007). With regard to the effects of emotional states, such as anger, on capability, there is of course the well known Yerkes-Dodson Law (Yerkes and Dodson, 1908) which relates level or arousal to level of performance, typically in the form of an inverted U curve. However, much more recently it has been suggested that elements of a road and trafﬁc scenario may elicit emotional responses which can direct attention and mediate rapid driver actions; a process that does not necessarily involve some intervening conscious cognitive component (Damasio, 1994; 2003; Slovic et al., 2002). One implication of this is that current emotional states that are independent of the road environment may swamp out this feedback, or alternatively that feedback may be misattributed to a source of non-driving related emotion (Fuller, 2007). In effect, the driver loses the sense of risk normally triggered by the unfolding events in the road and trafﬁc scenario ahead. Thus emotions aroused by critical events may lose their potential to direct driver behaviour. Examples of this process are perhaps reﬂected in these comments, made by participants in a recent focus group study, which was part of a larger investigation of the conditions for inappropriate high speed (Fuller et al., 2007): You might have had an argument that can make you speed … well it makes me do anyway … cause you’re in a bit of a mood (Driver on Speed Awareness course). I’ve done it when I’ve had an argument with my girlfriend. You know it’s like, your like aah just hit the boot and later on you’re like ‘why did I do that?’ you know, because really it was stupid to do it but it’s just anger because you’ve been arguing in the car or whatever (Professional driver).

Decreases in the driver’s capability have implications for the risk threshold a driver is prepared to accept: he or she needs to opt for a lower threshold, which will require lower speeds to keep task difﬁculty within tolerable bounds. Incidentally, in relation to this, some drivers allege that when drowsy they deliberately increase speed in an attempt to drive up their arousal and thereby make their capability more compatible with task demands. The safety consequences of this strategy may well be illusory if all that changes is the driver’s speed, while maintaining the previous separation between task demand and capability (that is, safety margin) (see Figure 27.2). The possibility that this is the actual outcome is an empirical issue and warrants further research. However, in principle, the prediction from the model would be that the driver’s statistical risk of loss of control might not change, but because of the increase in speed, the severity of the consequences of loss of control would increase.

Driver Training Assessment

343

20 18 16 14 12 Task demand Capability

10 8 6 4 2 0 1

3

5

7

9 11 13 15 17 19 21 time

Figure 27.2 Possible effects of increasing task demand (by increasing speed) to reverse a progressive decline in arousal (capability) Range of acceptable task difﬁculty and risk threshold If we turn from the determinants of perceived capability to the determinants of the range of acceptable task difﬁculty and risk threshold in the model (box 5), then it is clear that training might usefully encompass both distal and immediate effects of driving goals and of more direct inﬂuences on risk threshold. Research shows conclusively that saving time is a major motive for faster driving (see Adams-Guppy and Guppy, 1995; Campbell and Stradling, 2003; McKenna, 2005; Fylan et al., 2006; Musselwhite, 2006). In terms of the model, saving time is thus a motive for raising risk threshold. And put yet another way, saving time is a motive for a reduced safety margin. One training solution to part of this problem is, of course, instruction in effective time management. Sometimes however, unexpected delays on a journey create a more immediate pressure to drive faster in order to make up for lost time. Understanding the nature of the speed/time fallacy, as indicated in the discussion of human factor variables, could clearly help reduce the effects of this supposed time-earning motivation. To be more explicit, Stradling et al. (2007) discovered that drivers imagining travelling at or above 60 mph (approximately 100 km/h), and increasing or decreasing their speed by 10 mph for ten miles, signiﬁcantly and grossly overestimated the time gained in the ﬁrst condition by over ﬁve minutes, and signiﬁcantly overestimated the time lost in the second by the same margin. Emotions can affect capability, as discussed. They can also have a fairly direct effect on risk thresholds. In our recent focus group study (Fuller et al., 2007),

344

Driver Behaviour and Training – Volume III

professional drivers, drivers on speed awareness courses and bikers all pointed out that one’s speed and reaction to events and the behaviour of others may crucially depend on the prevailing mood you happen to be in: [The speed I drive at] depends on my mood depending on, um, whether I’m yeah relaxed or if I’ve had kind of a bad day or anything, stuff like that. That actually deﬁnitely inﬂuences it (Professional driver). It’s a lethal weapon we’ve got in our hands. It depends what mood you’re in and how you feel behind that lethal weapon that’ll kill anybody, and kill you if you want to really try (Driver on Speed Awareness course). But then again to be honest, it depends how I felt, if I was in the mood, I would maybe take it up a bit faster, then again depending how I feel I might go a bit slower, you know, just how I felt on the day (Biker). It’s an automatic reaction. Nine times out of ten you control it but the tenth time you don’t and you end up chasing them down the road and ﬂashing your lights and he’s giving you two ﬁngers and that or even she for that matter and I think most of us after about a mile you think ‘this is bloody stupid, it’s dangerous’ and you slow up (Professional driver).

On the basis of this qualitative evidence and the quantitative evidence relating emotional state to inappropriate high speed choice and other violations (for example, Wells-Parker et al., 2002; Mesken et al., 2002; Oltedal and Rundmo, 2006), a strong case can be made for the introduction of emotion management elements into standard driver training. More direct inﬂuences on risk threshold include factors such as social inﬂuences. Stradling et al. (2003) found that 36 per cent of 17–20 year old males said they would drive faster with people their own age in the car. Parker et al. (1992) found that attitude and subjective norm (that is, perceived social pressure) together accounted for nearly 33 per cent of the variance in intention to speed. Younger drivers perceived more approval from salient others for their speeding (as well as for their close-following and risky overtaking). These drivers were also more motivated to comply with the perceived wishes of their referents. Trying to deal with the unsafe inﬂuence of these social forces is likely to prove a serious challenge to driver training. Perhaps even more intractable will be the problem of dealing with drivers who have a dispositional preference for adopting a high risk threshold. Individual differences and the limits of driver training Our recent review of the literature on driver speed choice over the past decade concluded that there is a group of drivers who characteristically drive with a relatively high risk threshold and a relatively weak disposition to comply with trafﬁc regulations such as speed limits (see pathway leading to box 6 in Figure 27.1): Their driving style typically involves not just higher speeds but more extreme speed limit violations and other risky behaviours such as close following, dangerous overtaking and aggressive driving. They tend to hold positive attitudes to such behaviours, are prone to

Driver Training Assessment

345

sensation seeking, vulnerable to normative inﬂuences from their peers and may be part of a driving subculture which uses driving for recreational purposes other than getting from A to B. They are most likely to be young, male and immature. Not surprisingly they are more likely to be convicted for violations and are more likely to be involved in collisions (Fuller et al., 2007).

The behaviour, values, vulnerabilities and attitudes of these drivers may be characterised as a distinct lack of maturity in their engagement as drivers with the road and trafﬁc system. A typical proﬁle is suggestive of impulsive behaviour with lack of consideration of its consequences, and incomplete development of selfknowledge, self-control, social responsibility and independence of judgement. This group perhaps deﬁnes the limits of the possibilities offered by driver training, as currently conceived. There seems to be no way in which a driving instructor might somehow accelerate the process of maturation. Three implications follow. One is that in the short term, some kind of restraint on the behaviour of these high risk drivers needs to be considered and applications of Intelligent Speed Adaptation offer a real technical possibility, operating in the manner of reins on a toddler pedestrian: restraining the road user for his or her own safety. A second implication is that the existence of this group provides more evidence for the introduction of a graduated training and licensing approach. At the very least such an approach might delay the granting of a full licence and thereby allow for greater progress to mature adulthood. And third, it becomes even more apparent that attention should be paid to the development of attitudes and values in relation to the role of driver well before the adolescent is eligible to apply for a license. This concept has already been taken on board in many jurisdictions. It is noteworthy, however, that in Ireland, driver education and training have thus far been excluded as required formal elements of the secondary school syllabus (in part because of competition with more academic demands on the timetable). The Road Safety Strategy 2004–2006 (2004) states that the National Council for Curriculum and Assessment recommended that road safety be addressed within the context of the Social Personal Health and Education programme and that driver education (and speciﬁcally learning to drive for pupils aged 17) does not become part of the school curriculum. Careful evaluation of the effects of programmes in other jurisdictions, where driver education in schools is more proactively pursued, will be important in determining whether or not such educational programmes actually save lives and reduce the incidence of permanent disabilities in the late adolescent and young adult populations. Summary In summary, this paper has outlined the Task-Difﬁculty Homeostasis model in order to explore some of its implications for driver training and assessment. The so-called calibration problem is easily identiﬁed as arising out of either or both of an underestimation of task demand or an overestimation of capability. In dealing with the former, it is argued that there is a need to assess the driver’s disposition to create hazards, rather than rely exclusively on the ability to identify them as a

346

Driver Behaviour and Training – Volume III

passive observer. In dealing with the overestimation of capability, there is a clear need to expand education and training to include the development of awareness and management of human factor variables. The model also suggests that as drivers gain in capability and drive faster over any particular road segment, they will need to learn that a reactive response to hazards will not enable them to keep task difﬁculty within manageable bounds: a transition to an anticipatory style of driving will be required. This conclusion provides clear theoretical support for the introduction of graduated training and assessment. In considering the determination of a driver’s risk threshold, the model also identiﬁes a need to provide training in how to manage factors which motivate increases in this threshold, factors such as saving time, mood state, emotional reactions and social pressures. However, individual differences in risk threshold, particularly where the disposition to a high threshold is associated with a lack of maturity, provide a virtually impossible challenge for driving instructors. This observation provides further support for the introduction of graduated training and licensing. But more than this, it emphasises the potential importance of the contribution to driver safety of early educational interventions. References Åberg, L., Larsen, L., Glad, A. and Beilinsson, L. (1997). ‘Observed vehicle speed and drivers’ perceived speed of others.’ Applied Psychology: An International Review, 46, 287–302. Brown, I.D. and Groeger, J.A. (1988). ‘Risk perception and decision taking during the transition between novice and experienced driver status.’ Ergonomics, 31, 585–97. Damasio, A.R. (1994). Descartes’ Error: Emotion, Reason and the Human Brain. New York: Putnam. Damasio, A.R. (2003). Looking for Spinoza: Joy, Sorrow and the Feeling Brain. London: Heinemann. Deery, H.A. (1999). ‘Hazard and risk perception among young novice drivers.’ Journal of Safety Research, 30, 225–36. Fuller, R. (2000). ‘The Task-Capability Interface Model of the driving process.’ Recherche Transports Sécurité, 66, Jan–Mars 2000, 47–59. Fuller, R. (2005a). ‘Towards a general theory of driver behaviour.’ Accident Analysis and Prevention 37, 461–72. Fuller, R. (2005b). ‘The future of driver training and assessment.’ In L.W. O’Sullivan and S.W.L. Chan (eds), Irish Ergonomics Review 2005: Proceedings of the Irish Ergonomics Society Annual Conference 2005, Dublin: IES, 1–15. Fuller, R. (2007). ‘Motivational determinants of control in the driving task.’ In C. Cacciabue and C. Re, (eds), ‘Critical issues’ in Advanced Automotive Systems and Human-Centred Design. Guildford: Springer, 165–88. Fuller, R., Bates, H., Gormley, M., Hannigan, B., Stradling, S., Broughton, P., Kinnear, N. and O’Dolan, C. (2006). ‘Inappropriate high speed: who does it and why?’ In Behavioural Research in Road Safety 2006. Sixteenth seminar. London:

Driver Training Assessment

347

Department for Transport, 70–84. Fuller, R., Bates, H., Gormley, M., Hannigan, B., Stradling, S., Broughton, P., Kinnear, N. and O’Dolan, C. (2007). ‘The conditions for inappropriate high speed: a review of the research literature from 1995 to 2006.’Report under Contract Number PPRO 4/001/015 Improved Driver Information on Speed /Accident Risk (T201G), London: Department for Transport, (forthcoming). Fuller, R., Hannigan, B., Bates, H., Gormley, M., Stradling, S., Broughton, P.S., Kinnear, N. and O’Dolan, C. (2007). ‘Understanding inappropriate high speed: qualitative results from the HUSSAR Project.’ In Behavioural Research in Road Safety, Seventeenth Seminar, London: Department for Transport (forthcoming) . Fuller, R. and Santos, J.A. (2002). Human Factors for Highway Engineers. Amsterdam: Pergamon. Gerrard, M., Gibbons, F.X., Benthin, A.C. and Hessling, R.M. (1996). ‘A longitudinal study of the reciprocal nature of risk behaviors and cognitions in adolescents: what you do shapes what you think, and vice versa.’ Health Psychology, 15, 344–54. Gormley, M. and Fuller, R. (2005). SARTRE 3 Ireland. Dublin: NRA. Gregersen, N.P. and Bjurulf, P. (1996). ‘Young novice drivers: towards a model of their accident involvement.’ Accident Analysis and Prevention, 28, 229–41. Groeger, J.A. and Clegg, B.A. (1994). ‘Why isn’t driver training contributing more to road safety?’ In G.B. Grayson (ed.), Behavioural Research in Road Safety lV. Crowthorne: Transport Research Laboratory. Hibernian Motoring Report (2007). Moving in the Right Direction? Motorists Views on Motoring in Ireland. Dublin: Hibernian. Matthews, M.L. and Moran, A.R. (1986). ‘Age differences in male drivers’ perception of accident risk: the role of perceived driving ability.’ Accident Analysis and Prevention, 18, 299–313. McKenna, F.P. (2005). ‘What shall we do about speeding – education?’ In G. Underwood (ed.), Trafﬁc and Transport Psychology. Oxford: Elsevier, 521–8. Mesken, J., Lajunen, T. and Summala, H. (2002). ‘Interpersonal violations, speeding violations and their relation to accident involvement in Finland.’ Ergonomics, 45, 469–83. Oltedal, S. and Rundmo, T. (2006). ‘The effects of personality and gender on risky driving behaviour and accident involvement.’ Safety Science, 44, 621–8. Parker, D.M.A., Stradling, S.G., Reason, J.T. and Baxter, J.S. (1992). ‘Intention to commit driving violations – an application of the Theory of Planned Behaviour.’ Journal of Applied Psychology, 77, 94–101. Road Safety Authority (2007). Rules of the Road. Government Ofﬁces, Ballina, Co. Mayo: Road Safety Authority. Road Safety Strategy 2004–2006 (2004). Dublin: Department of Transport. Slovic, P., Finucane, M.L., Peters, E. and MacGregor, D.G. (2002). ‘Risk as analysis and risk as feelings: some thoughts about affect, reason, risk and rationality.’ Paper presented at the Annual Meeting of the Society for Risk Analysis, New Orleans, Louisiana, December 10, 2002. Stradling, S.G., Campbell., M., Allan, I.A., Gorrell, R.S.J., Hill, J.P., Winter, M.G. and Hope, S. (2003). The Speeding Driver: Who, How and Why. Research

348

Driver Behaviour and Training – Volume III

Findings No. 170/2003, Edinburgh: Scottish Executive Social Research. Stradling, S., Fuller, R., Gormley, M., Broughton, P., Kinnear, N., O’Dolan, C. and Hannigan, B. (2007). ‘Understanding inappropriate high speed: a quantitative analysis.’ Report under Contract Number PPRO 4/001/015, Improved Driver Information on Speed/Accident Risk (T201G), London: Department for Transport. Wells-Parker, E., Ceminsky, J., Hallberg, V., Snow, R.W., Dunaway, G., Guiling, S., Williams, M. and Anderson, B. (2002). ‘An exploratory study of the relationship between road rage and crash experience in a representative sample of US drivers.’ Accident Analysis and Prevention, 34, 271–8. Yerkes, R.M. and Dodson, J.D. (1908). ‘The relation of strength of stimulus to rapidity of habit formation.’ Journal of Comparative Neurological Psychology, 18, 459–82.

Chapter 28

Do We Really Drive by the Seat of Our Pants? Neale Kinnear,¹ Steve Stradling¹ and Cynthia McVey² ¹ Napier University, UK ² Glasgow Caledonian University, UK Introduction In a paper entitled ‘Driving by the seat of your pants: a new agenda for research’, Fuller (2005b) reported ﬁnding that drivers rate their feelings of risk in almost exactly the same way in which they would determine the difﬁculty of the driving task. This ﬁnding suggested that feelings may be a key component of the information processing feedback loop which inﬂuences a driver’s behavioural decision making and speed choice. In discussion of these ﬁndings, it was noted that neurological research was already complimenting this notion (Damasio, 2004). The current paper aimed to replicate Fuller’s (2005b) study and explore whether there were any differences due to a driver’s experience level. Task Capability Interface (TCI) The Task Capability Interface model is a complicated sounding title for a straightforward representation of the driving task (Fuller and Santos, 2002). The model starts with the self-evident fact that loss of control will occur when the demand of the driving task becomes greater than the capability of the driver. The capability of the driver is constrained by their personal characteristics, which creates a capability range. This range will have its foundations in a driver’s experience and training, but may also be mediated, at any time, by factors such as fatigue and stress. Meanwhile, task demand is inﬂuenced by many on-the-road factors that can make it somewhat unpredictable. However, one of the most important inﬂuences over task demand is managed by the driver, and that is speed. Driving is a self-paced activity; hence speed has a crucial role to play in the maintenance of the gap between task demand and capability. A change in speed will have a direct inﬂuence on the demand of the driving task. This control over speed allows a driver to maintain a preferred level of task demand and therefore within a preferred range of task difﬁculty. The inﬂuences and interactions of task demand, capability and preferred task demand are shown in Figure 28.1.

Driver Behaviour and Training – Volume III

350

Additional tasks

Environment Road Position & Trajectory

Vehicle

SPEED Other road users

Human Factors

Preferred Range of Task Difficulty

Task Demand

Loss of Control

Motivation, journey purpose & Compliance Weather

Task Demand Preference

Physical Environment

Capability Range

Predicted road & traffic conditions

Training & Education

Social Environment Passengers

Vehicle Experience

Human Factors

Distraction

Drugs & Alcohol

Motivation Fatigue

Mood & Emotions Stress

Figure 28.1 Authors’ illustration of the Task Capability Interface (TCI) Model (Fuller & Santos 2002) with inﬂuences Task difﬁculty Task difﬁculty is the real-time gap between level of task demand and level of capability. Task difﬁculty is inversely proportional to this gap as when the gap decreases, task difﬁculty increases. Therefore, as task demand approaches capability, the driver will experience that the task of driving is becoming progressively more difﬁcult and that he/she is more at risk. Hence, this gap could otherwise be termed as the driver’s safety margin. It is proposed that we drive within a preferred range of task difﬁculty that we are prepared to engage with (Fuller, 2005a; 2005b). This process has been conceptualised in the form of the Task-Difﬁculty Homeostasis (Fuller, 2000).

Do We Really Drive by the Seat of our Pants?

351

Task-Difﬁculty Homeostasis (Fuller, 2000) Based on the capability of the driver and the motivation of a particular journey, a driver engages with a range of difﬁculty within which they are comfortable (Fuller, 2005a). The driver will therefore drive in such a way as to maintain the level of task demand within that range. Manipulation of speed is seen as the primary mechanism for achieving this, although variations in effort or undertaking or dumping other secondary tasks may also be used (that is, making or ending a mobile phone call). A representation of this process can be seen in Figure 28.2.

effort motivation

goals of journey

range of acceptable task difficulty

comparator

decision and response capability task difficulty

effects on vehicle speed

task demand

Figure 28.2 Illustration of Task-Difﬁculty Homeostasis Source: Fuller, 2000.

There is signiﬁcant empirical support for this process, as detailed in Fuller et al. (2006). Studies suggest that changes in a driver’s environment cause a driver to compensate through increased or decreased speed (Evans and Charlton, 2006; van Driel, Davidse and Maarseveen, 2004; Smiley, 2003). In Evans and Charlton’s (2006) study of experienced New Zealand drivers, the road width was subtly manipulated on a simulator. All drivers drove on four sections of road that were technically identical other than minor differences in road width. The results indicated signiﬁcant effects of road width on participants’ speeds, such that narrow roads were associated with lower speeds, while wider roads were associated with higher speeds. Curiously, when interviewed about what inﬂuenced their driving on the four sections of road, not one participant stated that the width of the road had any bearing on their driving. Similarly, other changes to the driver’s environment, like internal car noise and road lighting, also appear to alter a driver’s perception and speed choice (Horswill and McKenna, 1999; Evans, 1970a; 1970b; Assum et al., 1999).

Driver Behaviour and Training – Volume III

352

An important part of the representation of Task-Difﬁculty Homeostasis is what Fuller terms the ‘comparator’. This area of the model signiﬁes its depth into understanding driver behaviour. The TCI and Task-Difﬁculty Homeostasis provide conceptualisation of the driving task but do not explain how or why a driver may reach their decision or behavioural response. Essentially, understanding this is therefore crucial as to whether the TCI and Task-Difﬁculty Homeostasis are valid representations of the driving task. Without understanding what exactly is fed back to the driver that requires ‘comparing’, one can only speculate as to whether this is a working model or not. To investigate this, one must question how a driver actually senses and evaluates how difﬁcult and/or safe the driving task is at any particular moment. And how does this inﬂuence the ensuing decision and behavioural response? In testing the TCI model, Fuller (2005b) provided an intriguing angle for answering such questions. Thirty participants were presented with footage of three short driving sequences from a residential road, a country road and a dual carriageway in Dublin, Ireland. The driving sequences were each digitally altered to be shown at different speeds, increasing at 5 mph increments. Participants viewed the sequences at the different speeds (without knowing the actual speeds) and rated the sequence for task difﬁculty, feelings of risk and statistical risk estimates of the probability of collision. Fuller hypothesised that task difﬁculty ratings would relate to increases in speed; that probability of collision ratings would not relate to speed at low speeds but would after some threshold was reached; and that feelings of risk (risk perception) would track probability of collision ratings. However, the results did not support all of the hypotheses, as illustrated in Figure 28.3.

high

Driver ratings

Task difficulty

Rating

Preferred speed

Probability of collision

low

Risk perception

low

Speed

high

Figure 28.3 Illustrated summary of results from Fuller Source: Summary of Fuller, 2005b.

Do We Really Drive by the Seat of our Pants?

353

Task difﬁculty ratings were very highly correlated with speed, and there was no relationship between probability of collision ratings and speed at low speeds. Only after some threshold was reached would ratings of statistical risk begin to rise and relate to further increases in speed, just as Fuller had hypothesised. However, feelings of risk (risk perception) ratings were correlated with speed in the same way as task difﬁculty. In fact, ratings between task difﬁculty and feelings of risk were correlated to the order of 0.97 (Fuller, 2005b). This deﬁed the hypothesis and prompted further investigation. The results suggest that feelings can represent the difﬁculty of the task, hence feelings may provide feedback to the driver, and likewise the ‘comparator’. In discussion, Fuller cited neurological evidence that not only are feelings crucial in providing feedback to humans, but that they also play a crucial role in our decision making processes (Damasio, 2004). However, before the results can be reliably discussed as providing such an important insight into driver behaviour, they must ﬁrstly be validated. A critique of the study may therefore focus on the reliability and validity of the results; the order effect of the stimuli; and the use of the probability of collision to measure statistical risk estimate. Reliability has been somewhat demonstrated through replication (Lynn, 2006) although validity could be added through the use of new stimuli. Meanwhile, the order effect of the stimuli has been found to be minimal (Lynn, 2006). Nevertheless, whether using the probability of collision is beneﬁcial over the probability of loss of control is debatable. As it is postulated within the TCI that a loss of control will occur when task demand exceeds capability, it could be inferred that the probability of loss of control may be a better measure. Based on the critique of the original paper, the current study aimed to advance the understanding of the initial results. Newly produced video clips were created to add validity. Further, the current study also utilised a question measuring the probability of loss of control from previous large scale motorcycle research to be used as the measure of statistical risk (Sexton et al., 2006). The probability of risk response was also made more sensitive by using a scale of 0–100 rather than the 0–30 used in the original paper. In addition, the present study also asked participants to state their maximum speed for driving each road type, rather than their ‘preferred speed’ as used in the original experiment. It is postulated that this demonstrates a driver’s maximum acceptance of task demand and therefore demonstrates a greater understanding of their minimum accepted safety margin. To further advance the understanding of any potential ﬁndings, different levels of driver experience (learner, inexperienced and experienced) were tested on four road types: residential, straight country, bendy country and dual carriageway. In summary, it was hypothesised that: • • •

task difﬁculty and feelings of risk will be associated with speed and each other; the probability of loss of control will rise in relation to speed only after some threshold is reached; there will be differences in response to task difﬁculty, feelings of risk, probability of loss of control and maximum stated speed by experience level.

Driver Behaviour and Training – Volume III

354

Method Design To test for drivers’ appraisal of task difﬁculty and risk in relation to speed choice, participants were shown movie clips of a car driving at different speeds on the same section of road, from the driver’s perspective. A repeated measures design was used to investigate task difﬁculty, feeling of risk and probability of loss of control across nine different speeds. This design was repeated for four types of road: residential, straight country, bendy country and dual carriageway. For each road type, the same clip was digitally altered to represent nine different speeds. The speed range was dependent on the type of road; as shown in Table 28.1. The speeds were spaced at 5 mph increments. Table 28.1

Speed range across the different road types (clips set at 5 mph increments)

No. of clips

Speed range (mph)

Residential

9

20–60

Straight country

9

30–70

Bendy country

9

30–70

Dual carriageway

9

60–100

As previous research has demonstrated that the order effect of video clips is minimal (Lynn, 2006), and to give the experiment ecological validity, it was decided to remain with participants viewing slow-to-fast for each road type. However, to control for any order effect arising due to the order of the road type or the order of the questions, these factors were taken into account. The road type order was set out as residential, followed by straight country, followed by bendy country, followed by dual carriageway. Half of the participants viewed the road types in this order and half in reverse order. Similarly, the questions were originally set out asking for ratings of task difﬁculty, followed by feeling of risk, followed by probability of loss of control. Half of the participants answered in this order and half answered with the order reversed. There were therefore four conditions under which a participant could carry out the experiment: 1. 2. 3. 4.

normal order questions – normal order road type normal order questions – reverse order road type reverse order questions – normal order road type reverse order questions – reverse order road type

Do We Really Drive by the Seat of our Pants?

355

No signiﬁcant differences were found between ratings of these four experimental groups. Participants One hundred and ﬁfty-two participants took part in the study. Opportunistic quota sampling was used to ensure adequate numbers of male and female, learner, inexperienced and experienced drivers were included for analysis (see Table 28.2). Experienced drivers were deﬁned as having held a UK driving licence for three or more years; inexperienced drivers were deﬁned as holding a UK driving licence for less than three years; with learner drivers currently seeking to learn to drive. Participants were recruited from within the Glasgow area. Table 28.2 Breakdown of sample by experience and gender

Learner

Inexperienced

Experienced

Total

Male

17

23

28

68

Female

23

29

32

84

Total

40

52

60

152

The mean age for learner drivers was 21.3 years (sd = 3.6, range 17.9–33.4). The mean age for inexperienced drivers was 20.8 years (sd = 5.4, range 17.9–52.5). And the mean age for experienced drivers was 29.75 years (sd = 9.2, range 20.3–62). For participants with a driving licence, the mean duration that the licence had been held was 17.31 months (sd = 9.3, range 1–35) for inexperienced drivers and 118.3 months (sd = 84.6, range 36–480) for experienced drivers. Learner drivers stated they had been learning to drive for an average of 16.9 months (sd = 15.7, range 1–60). Learner drivers reported that they had driven an average of 112.6 miles in the last 12 months (sd = 100, range 0–350); with inexperienced drivers reportedly having driven an average of 3786.8 miles in the last 12 months (sd = 5287, range 2–23 000); and experienced drivers reporting they had driven an average of 6366.8 miles in the last twelve months (std dev = 6047, range 0–30 000). Materials The video clips were constructed by the researchers using roads around the south side of Glasgow, UK, and once digitally altered for speed, were put together in sequences using Microsoft PowerPoint. Each original clip was around 25 seconds in length before being digitally altered to the different speeds.

Driver Behaviour and Training – Volume III

356

The answer booklet had a new page for each clip that was being rated, so that participants could not see their previous rating. The same three questions were asked on each page of the answer booklet: 1. How difﬁcult would you ﬁnd it to drive this section of road at this speed? Extremely difﬁcult

Extremely easy 1

2

3

4

5

6

7

2. How risky would it feel to drive this section of road at this speed? Not at all risky 1

2

Extremely risky 3

4

5

6

7

3. Imagine if 100 drivers like you, of the same age and experience, were to drive this section of road at this speed and in these conditions. How many do you think would lose control of the vehicle? Answer (0 – 100): _____________________ Participants also completed a further questionnaire entitled ‘You and Your Driving’, which gained general demographic information about the participant and their driving attitudes and behaviours. Procedure Participants who signed the disclaimer were then asked to read the instructions for the experiment. The presentation started when participants pressed the space bar. A clip would start and then stop automatically, prompting the participant to answer the corresponding page in the answer booklet. The participant repeated this procedure for the nine speeds on each of the four road types. At the end of the presentation, the participant was prompted on-screen to complete the ‘You and Your Driving’ questionnaire. Results 1. Task difﬁculty and feelings of risk will be associated with speed and each other Graphical representation of the mean scores for task difﬁculty, feelings of risk and probability of loss of control can be seen in Figure 28.4. Ratings of task difﬁculty and feelings of risk appear to increase simultaneously with increases in speed on all

Do We Really Drive by the Seat of our Pants?

357

road types. However, at higher speeds feelings of risk appears to be greater than that of task difﬁculty. Probability of loss of control ratings also demonstrates a gradual increase with speed on all road types but not to the magnitude of task difﬁculty or feelings of risk. Road type also demonstrates differences in the magnitude of ratings although the overall interaction remains similar. Straight country road 7

6

6

5

5

Rating

Rating

Residential 7

4 3

4 3

2

2

1

1 20

25

30

35

40

45

50

55

60

30

35

40

45

Speed (mph) Task difficulty

Feelings of risk

Task difficulty

Prob of loss of control

Bendy country road

55

60

65

70

Feelings of risk

Prob of loss of control

Dual carriageway

7

7

6

6

5

5

Rating

Rating

50 Speed (m ph)

4

4

3

3

2

2

1

1 30

35

40

45

50

55

60

65

70

60

65

70

Speed (m ph) Task difficulty

Feelings of risk

75

80

85

90

95

100

Speed (m ph) Prob of loss of control

Task difficulty

Feelings of risk

Prob of loss of control

Figure 28.4 Means plot of task difﬁculty, feelings of risk and probability of collision across speed for the four road types Spearman’s Rho correlation demonstrated that increases in speed were strongly associated to increases in task difﬁculty and feelings of risk ratings on all road types (residential: rho = 0.95, p < 0.001; straight country: rho = 0.61, p = 0.001; bendy country: rho = 0.71,1 p = 0.031; dual carriageway: rho = 0.89, p = 0.001). Correlation coefﬁcients between task difﬁculty and feelings of risk ratings for each speed and across all road types can be seen in Table 28.3. All relationships were signiﬁcant at the p < 0.001 level. The data suggest that the relationship between task difﬁculty and feelings of risk becomes stronger as speed increases and also as the speed potential of the road type increases. 2. The probability of loss of control will rise in relation to speed only after some threshold is reached Graphical representation of the overall mean scores for the probability of loss of control can be seen in Figure 28.4 above. The trend lines demonstrate that the probability of loss of control estimates remain low despite the increases in speed but that it does rise gradually on all road types. However it is not apparent from the mean scores whether or not there is a clear ‘threshold’.

Driver Behaviour and Training – Volume III

358

Table 28.3

Road type / speed (mph) 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Co-efﬁcient average

Table 28.4

Correlation coefﬁcients of task difﬁculty and feelings of risk ratings by speed and road type (all coefﬁcients signiﬁcant at the p < 0.001 level)

Residential

Bendy country

0.62 0.66 0.67 0.71 0.70 0.77 0.74 0.79 0.78

0.71

Straight country

0.65 0.72 0.73 0.70 0.83 0.78 0.71 0.82 0.82

0.63 0.75 0.79 0.83 0.78 0.77 0.79 0.84 0.83

0.75

0.78

Dual carriageway

Co-efﬁcient average 0.62 0.66 0.65 0.73 0.74 0.77 0.78 0.78 0.74 0.80 0.81 0.77 0.76 0.84 0.83 0.83 0.86

0.69 0.73 0.77 0.77 0.76 0.84 0.83 0.83 0.86 0.79

Mean speeds in mph for ratings threshold and maximum speed by road type (correlation coefﬁcient between the two variables also shown: p = ns for all road types)

Ratings threshold (mph) N Standard deviation Maximum speed (mph) N Standard deviation Correlation coefﬁcient (p = ns)

Residential

Bendy country

Straight country

Dual carriageway

43

51

55

83

82 9.63 37 151 8.56

104 10.77 40 151 10.05

66 9.47 52 151 12.44

58 10.59 69 151 12.26

0.01

0.10

0.06

–0.05

Do We Really Drive by the Seat of our Pants?

359

The current study has two ways in which to determine drivers’ thresholds: from the ratings data or from the participants stated maximum speed. A participant’s ratings threshold was taken as the ﬁrst increase in response from their baseline measure given on clip 1 of each road type. However, there were many participants whose ratings did not increase with speed and remained constant. These participants were excluded as no threshold could be established. The mean thresholds for the remaining participants can be seen in Table 28.4, along with participants’ mean stated maximum speeds. There was no relationship between the ratings threshold and maximum stated speed on any road type. Given the large number of participants from which a threshold could not be established and that there appears to be no obvious change in the data around either the ratings threshold or the mean maximum speed, the current results do not support earlier work that noted a clear diversion from a zero rating at some threshold (Fuller, 2005b). 3. There will be differences in response to task difﬁculty, feelings of risk, probability of loss of control and maximum stated speed by experience level

Task difﬁculty Graphical comparison of task difﬁculty mean scores by experience level can be seen in Figure 28.5. The mean scores suggest little difference between the three experience groups although there is a minor trend across all road types suggesting experienced driver’s mean ratings are slightly lower than that of learner and inexperienced drivers. Straight country road 7.00

6.00

6.00

5.00

5.00

Rating

Rating

Residential road 7.00

4.00 3.00 2.00

4.00 3.00 2.00

1.00

1.00

20

25

30

35

40

45

50

55

60

30

35

40

Speed (mph) Learner

Inexperienced

50

55

60

65

70

Speed (mph)

Experienced

Learner

Bendy country road

Inexperienced

Experienced

Dual carriageway

7.00

7.00

6.00

6.00

5.00

5.00

Rating

Rating

45

4.00 3.00 2.00

4.00 3.00 2.00

1.00

1.00 30

35

40

45

50

55

60

65

70

60

65

70

Speed (mph) Learner

Inexperienced

75

80

85

90

95

100

Speed (mph) Experienced

Learner

Inexperienced

Experienced

Figure 28.5 Means plot of task difﬁculty ratings across speed for the four road types

Driver Behaviour and Training – Volume III

360

Repeated measures multivariate analysis of variance (MANOVA) was performed to test for differences in task difﬁculty ratings at all speed levels by experience. The assumption of ‘sphericity’ was examined, but this assumption was not met. Therefore, in reporting results, the Greenhouse-Geisser statistic is used. There was no signiﬁcant difference between the experience groups ratings of task difﬁculty at any speed and on any road type (residential: F(5, 387) = 0.41, p = ns; straight country: F(5, 717) = 1.01, p = ns; bendy country: F(5, 358) = 0.43, p = ns; dual carriageway: F(4, 303) = 1.11, p = ns). Feelings of risk Graphical comparison of feeling of risk mean scores by experience level can be seen in Figure 28.6. The mean scores suggest little difference between the three experience groups although again, there is a minor trend across all road types suggesting experienced driver’s mean ratings are slightly lower than that of learner and inexperienced drivers. Straight country road

7.00

7.00

6.00

6.00

5.00

5.00

Rating

Rating

Residential road

4.00 3.00 2.00

4.00 3.00 2.00

1.00

1.00 20

25

30

35

40

45

50

55

60

30

35

40

Speed (mph) Learner

Inexperienced

50

55

60

65

70

Speed (mph) Experienced

Learner

Bendy country road

Inexperienced

Experienced

Dual carriageway

7.00

7.00

6.00

6.00

5.00

5.00

Rating

Rating

45

4.00 3.00 2.00

4.00 3.00 2.00

1.00

1.00 30

35

40

45

50

55

60

65

70

60

65

70

Speed (mph) Learner

Inexperienced

75

80

85

90

95

100

Speed (mph) Experienced

Learner

Inexperienced

Experienced

Figure 28.6 Means plot of feelings of risk ratings across speed for the four road types Repeated measures multivariate analysis of variance (MANOVA) was performed to test for differences in feeling of risk ratings at all speed levels by experience. The assumption of ‘sphericity’ was examined, but this assumption was not met. Therefore in reporting results, the Greenhouse-Geisser statistic is used. There was no signiﬁcant difference between the experience groups’ ratings of feeling of risk at any speed and across any road type (residential: F(6, 478) = 1.6, p = ns; straight

Do We Really Drive by the Seat of our Pants?

361

country: F(5.8, 427) = 0.53, p = ns; bendy country: F(6, 431) = 0.66, p = ns; dual carriageway: F(4, 303) = 1.11, p = ns). Probability of loss of control Graphical comparison of probability of loss of control mean scores by experience level can be seen in Figure 28.7. The mean scores suggest that with increases in speed the gap between the experienced group and the inexperienced and learner groups increases. Whilst the experienced group’s ratings rise with speed, they never match the magnitude of the learner or inexperienced groups. Conversely, the learner and inexperienced groups ratings demonstrate greater increases with speed and somewhat track each other. It is also of note that the inexperienced group also tends to rate the probability of the loss of control greater than that of the learner group, except on the bendy country road. Residential road

Straight country road 3.00

Rating

Rating

3.00

2.00

2.00

1.00

1.00 20

25

30

35

40

45

50

55

30

60

35

40

Learner

Inexperienced

45

50

55

60

65

70

Speed (mph)

Speed (mph)

Learner

Experienced

Bendy country road

Inexperienced

Experienced

Dual carriageway 3.00

Rating

Rating

3.00

2.00

2.00

1.00

1.00 30

35

40

45

50

55

60

65

70

60

65

70

Speed (mph) Learner

Inexperienced

75

80

85

90

95

100

Speed (mph) Experienced

Learner

Inexperienced

Experienced

Figure 28.7 Magniﬁed means plot of probability of loss of control ratings across speed for the four road types (ratings re-coded into 1–7 rating scheme) Repeated measures multivariate analysis of variance (MANOVA) was performed to test for differences in probability of loss of control at all speed levels by experience. The assumption of ‘sphericity’ was examined, but this assumption was not met. Therefore, in reporting results, the Greenhouse-Geisser statistic is used. There was a signiﬁcant difference between the experienced groups ratings of the probability of loss of control across all road types (residential: F(3, 207) = 3.88, p = 0.012; straight country: F(3, 200) = 5.24, p = 0.003; bendy country: F(3, 204) = 3.76, p = 0.014; dual carriageway: F(4, 266) = 2.638, p = 0.040).

Driver Behaviour and Training – Volume III

362

Maximum speed There was also a difference in the maximum stated speed by experience level on all road types, as can be seen in Figure 28.8. The difference between groups on the residential and bendy country roads was not signiﬁcant. However, there were signiﬁcant differences between the experienced groups on the straight country road (F(2, 148) = 3.43, p = 0.035) and dual carriageway (F(2, 148) = 9.24, p < 0.001). Post hoc Tukey analysis demonstrated signiﬁcant difference between the experienced group and the learner group on the straight country road (p = 0.032) and between the experienced group and both the learner (p = 0.001) and inexperienced (p = 0.002) groups on the dual carriageway.

Maximum speed by experience Speed (mph) 0

10

20

30

40

50

60

70

80

Residential road Bendy country road Straight country road Dual carriageway

Learner

Inexperienced

Experienced

Figure 28.8 Maximum speed comparison by experience level on each road type Discussion The present study aimed to replicate and extend the understanding of results reported by Fuller (2005b). The key ﬁnding reported by Fuller was that participants rated their feelings of risk as they would the difﬁculty of the task. Furthermore, both were related to increases in speed. The current results would support the original ﬁndings and further validate this interaction. It was also found that this relationship was consistent across different road types and driver experience levels, thus suggesting that it is a naturally occurring phenomenon not mediated by driving experience. It is of interest that although the strength of the relationship between task difﬁculty and feelings of risk is strong at low speeds, it appears to become even stronger as speed increases. It is, however, inﬂuenced by the environment and the type of road being driven on, suggesting a relationship with speed sensation rather than with absolute speed itself. In terms of the Task-Capability Interface model, the increases in speed would push task demand closer towards capability and reduce the

Do We Really Drive by the Seat of our Pants?

363

driver’s safety margin. It would therefore be logical that as a driver’s safety margin is reduced, the relationship between the sensation of risk and the demand of the task is required to be more exact. This could be an area for further investigation. The current study did not ﬁnd support for the notion that statistical risk estimates will only increase after a threshold is reached. This could be for a number of reasons, although notably the use of a differently worded question may be measuring a different response. For example, the original study asked for the probability of collision, whereas the present study asked for the probability of loss of control. This again is a further area for investigation given that drivers appear to rate these as different concepts. Whilst no threshold was found within the data and there were gradual increases with speed, the probability of loss of control was not related to task difﬁculty in the way that feelings of risk were. The key point to note is that if drivers were to use probability of loss of control estimates to deduce the risk of the driving task then they would not have accurately determined the demand characteristics of the task. It is therefore intriguing that driver experience demonstrated a signiﬁcant difference in response to the probability of loss of control. Of further note is that inexperienced drivers rated this factor comparatively to that of learner drivers. If we theoretically relate this to the ‘comparator’ and the ensuing decision making process that deﬁnes the behavioural outcome, then it could be postulated that experienced drivers rely less on a cost–beneﬁt analysis of the situation than less experienced drivers. Cost–beneﬁt analysis is obviously more taxing and takes longer to process (Gilovich, Grifﬁn and Kahneman, 2002), therefore, it would be advantageous for the driving task to become more of an automated procedure as a person gains the experience of doing so. This in itself would rely on the premise that automated and faster decision making must be learnt through experience of the task. Existing driver behaviour research suggests initial solo driver experience is crucial to the reduction in crash risk, yet is unable to report exactly why (Twisk, 2006; Maycock, Lockwood and Lester, 1991; Forsyth, Maycock and Sexton, 1995). Meanwhile, education and training of inexperienced drivers has failed to be effective as a substitute for this crucial stage (Christie, 2001). Within the early period of solo driving, it would appear that through experience of the task, drivers are continuing to learn a very important component of driving that reduces their crash risk and increases their ability to drive safely. Hence, could it be that through this experience, drivers learn to rely less on cost–beneﬁt analysis and instead on their feelings or risk? Neurological theory would appear to support this premise (Damasio, 2004; Le Doux, 2004). In conclusion, the current study has demonstrated support for the original ﬁndings of Fuller (2005b) and extended our understanding of the results. The interaction between task difﬁculty, feelings of risk and speed has been upheld. Meanwhile the measurement of statistical risk estimate remains an area for further research, although it does not relate to task difﬁculty in the way that feelings of risk do. The original results have been extended in ﬁnding that the experience level of the driver is important in the response to the probability of loss of control. When these results are applied to existing driver behaviour research one may conclude that we do appear to drive by the seat of our pants, but only once we have gained the experience to do so.

364

Driver Behaviour and Training – Volume III

References Assum, T., Bjørnskau, T., Fosser, S. and Sagberg, F. (1999). ‘Risk compensation – the case of road lighting.’ Accident Analysis and Prevention, 31, 545–53. Christie, R. (2001). ‘The effectiveness of driver training as a road safety measure: a review of the literature.’ (No. 01/03): Royal Automobile Club of Victoria (RACV) Ltd. Damasio, A.R. (2004). Looking for Spinoza: Joy, Sorrow and the Feeling Brain. London: Heinemann. Evans, B.L. and Charlton, S.G. (2006). ‘Explicit and implicit processes in behavioural adaptation to road width.’ Accident Analysis and Prevention, 38, 610–17. Evans, L. (1970a). ‘Automobile-speed estimation using a movie-ﬁlm simulation.’ Ergonomics, 13, 231–7. Evans, L. (1970b). ‘Speed estimation from a moving automobile.’ Ergonomics, 13, 219–30. Forsyth, E., Maycock, G. and Sexton, B. (1995). Cohort Study of Learner and Novice Drivers: Part 3, Accidents, Offences and Driving Experience in the First Three Years of Driving. Project Report PR111. Crowthorne: TRL Limited. Fuller, R. (2000). ‘The Task-Capability Interface model of the driving process.’ Recherche Transports Sécurité, 66, 47–59. Fuller, R. (2005a). ‘Towards a general theory of driver behaviour.’ Accident Analysis and Prevention, 37, 461–72. Fuller, R. (2005b). ‘Driving by the seat of your pants: a new agenda for research.’ In Behavioural Research in Road Safety 2005. London: Department for Transport. Fuller, R., Bates, H., Gormley, M., Hannigan, B., Stradling, S., Broughton, P., Kinnear, N. and O’dolan, C. (2006). ‘Inappropriate high speed: who does it and why?’ In Behavioural Research in Road Safety 2006, Sixteenth Seminar, London: Department for Transport. Fuller, R. and Santos, J.A. (2002). ‘Psychology and the highway engineer.’ In R. Fuller and J.A. Santos (eds), Human Factors for Highway Engineers. Oxford: Pergamon, 1–10. Gilovich, T., Grifﬁn, D. and Kahneman, D. (2002). ‘Heuristics and biases: the psychology of intuitive judgement.’ New York: Cambridge University Press. Horswill, M.S. and McKenna, F.P. (1999). ‘The development, validation, and application of a video-based technique for measuring an everyday risk-taking behaviour: drivers’ speed choice.’ Journal of Applied Psychology, 84, 977–85. LeDoux, J. (2004). The Emotional Brain. London: Phoenix (Orion Books). Lynn, P. (2006). ‘The relationship between driver experience and driver decision making in an Irish population.’ Unpublished Dissertation Thesis, Trinity College Dublin. Maycock, G., Lockwood, C.R. and Lester, J.F. (1991). The Accident Liability of Car Drivers (No. 315). Crowthorne: Transport Research Laboratory. Sexton, B., Hamilton, K., Baughan, C., Stradling, S. and Broughton, P. (2006). Risk and Motorcyclists in Scotland. Scottish Executive Social Research. Smiley, A. (2003). ‘Driver adaptation to the road.’ In Seminaire International En Sécurité Routère, Québec City: Association Québéciose du Transport et des

Do We Really Drive by the Seat of our Pants?

365

Routes. Twisk, D. (2006). ‘Findings of the OECD/ECMT working group on young drivers and effective countermeasures.’ In Behavioural Research in Road Safety 2006, Sixteenth Seminar, London: Department for Transport. van Driel, C.J.G., Davidse, R.J. and van Maarseveen, M.F.A.M. (2004). ‘The effects of edgeline on speed and lateral position: a meta-analysis.’ Accident Analysis and Prevention, 36, 671–82.

This page intentionally left blank

Chapter 29

The Impact of Subjective Factors on Driver Vigilance: A Driving Simulator Study Jérémy Vrignon,1,2 Andry Rakotonirainy,2 Dominique Gruyer1 and Guillaume Saint Pierre1 1 LIVIC, Versailles, France 2 CARRS-Q, Brisbane, Australia Introduction Lack of vigilance decreases drivers’ performances thus reducing safe driving tolerances, as well as the ability to react to unexpected events and increasing the likelihood of a crash (Wierwille, 1994; Wierwille et al., 1996), and in Australia between 1995 and 2005, vigilance decline can be estimated as a contributing factor in seven per cent of all reported crashes and 15 per cent of fatal crashes (Queensland Government). Thus, assessing and preventing vigilance decline can reduce the number of road accident and fatalities and therefore has been a major issue in the past few years (Bekiaris, Amditis and Wevers, 2001). Existing technology-based solutions used to assess vigilance show deﬁciencies when deployed in real driving situations. They are often based on a single device (PERCLOS, Lane Position, Time to Line crossing, and so on) and they offer a limited representation of the driving context which has been pointed as a critical aspect of the reliability and user acceptability of such systems. Deeper analysis of the problem suggests that combination of several devices should provide a more reliable estimation of the driver’s state, through a more complete representation of the driving situation. To date, little research has examined this approach, but recent ones tend to support it (Zhu, Ji and Lan, 2004). Three main kinds of factors that impact on vigilance have been identiﬁed. They are task factors, environmental factors and subjective factors (Johns and Counselling, 2004; Wellbrink, Zyda and Hiles, 2004). In this context, our approach consists in combining information related to factors and consequences of vigilance coming from the driver, the vehicle and the environment in order to build a reliable and robust system able to assess and prevent driver vigilance decline (Gruyer, Rakotonirainy, and Vrignon, 2005a, 2005b) This chapter investigates the impact of subjective factors on drivers’ vigilance, using the driver’s psychomotor performances assessed using an unobtrusive Psychomotor Vigilance Task (PVT). We ﬁrst described our experiment, with the design of the scenario used on a driving simulator and the design of a non-obtrusive Psychomotor Vigilance Test. This innovative design enables the assessment of the

368

Driver Behaviour and Training – Volume III

performance of the driver while driving. Then, the impact of subjective factors reported by participants is analysed with ANOVA. Finally, the outcomes of this study are discussed to estimate the potential of each of the subjective factors while used in combination with existing vigilance detection systems. Methodology Participants A panel of 12 drivers (six females and six males) took part in the study. Participants were recruited through media release and were required to hold a current full driver’s licence for at least one year, and drive a car on a weekly basis. They gave their written informed consent and received 20 Australian dollars as an incentive. Ethics approval for the study was granted by the Human Research Ethics Committee of the Queensland University of Technology. Driving simulator Experiments were conducted using the SiVIC™ driving simulator (Gruyer, Royere, Lac, Michel and Blosseville, 2006) supplied for the proposed research by the LIVIC laboratory (Laboratory on Vehicle-Infrastructure-Driver Interaction located in Versailles, France). The rendering capabilities of this simulator enabled the display of a realistic driving environment of highways and country roads. The SiVIC™ allowed us to programme and monitor all the elements of the environment (point of view, mirror, speedometer, cars, pedestrians, trees, buildings, road signs, lights, smog, and so on). Participants sat in front of the 2m x 2m screen where the driving environment was displayed with a projector. The ﬁeld of view was 120 degrees, which provided a realistic visual motion and a good perception of the driving environment (Allen, Cook and Park, 2005). A six-speaker audio system was used to reproduce the sound of the vehicle, the driving environment, including Doppler effect and three-dimensional position of sources, and with a position which ensured audio immersion of the driver in the driving environment. Then a view from the inside of the vehicle, with a speedometer and a rear view mirror was displayed. Participants were driving an average automatic sedan car using a steering wheel which provided force feedback, and a two pedals set (brake and accelerator). Psychomotor vigilance test We used the psychomotor vigilance test (PVT) to assess performance of the participants throughout the experiment. According to the aim of the project, the PVT was speciﬁcally designed in order to minimise its intrusiveness on driving behaviour. For this design, we took into account the multiple resources model proposed by Wickens (2002), as well as experimental results which supports part of this model (Sigman and Dehaene, 2005). The resulting PVT exploits the parallelism of the

The Impact of Subjective Factors on Driver Vigilance

369

perception stage and response stage of decision making and information processing of stimuli in order to interfere as little as possible with the driving task. The resulting PVT was a choice reaction time task. Stimuli provided were audial, and participants were asked to respond verbally to the stimuli. Two kinds of stimuli, called signal and noise, were displayed to the participants. The stimuli were simple tones of 15ms, with frequencies of 440 Hz and 300 Hz for respectively the signal and the noise. The participants were asked to answer yes or no to respectively the signal, and the noise. Stimuli were displayed through a pair of dedicated stereo speakers placed on both sides of the steering wheel. Frequency, regularity and probability of the stimuli were the same for each participant, in order to avoid impact on performance of these known confounding factors. Stimuli were displayed every eight seconds, with a standard deviation of 1.3 seconds to avoid anticipation from the participants. The signal-to-noise ratio was 0.69. These parameters were chosen during a pilot study in order to get the best compromise between impact on driving behaviour, temporal deﬁnition and subjective bias. Procedure A test period of 15 minutes took place for each participant before participating in the experiment. The aim of this test period was for the participant to feel comfortable with the simulator and the PVT in order to limit the impact on normal driving behaviour. Firstly, each participant took part in a practice session for 15 minutes on the simulator, to familiarise themselves with the driving interface. This training session used the same scenario as the experiment. Secondly, each participant completed a training phase for the PVT without driving. This phase starts with the adjustment of the volume of the stimuli to enable a good perception. Performance feedbacks were continuously provided to the participant during practice. Performance was expected to improve during PVT practice. Hence, after approximately ﬁve minutes of training, performance converged to a maximum and stopped increasing (Benedetto, Blasiis and Benedetto, 2004; Lamond and Dawson, 1999), ending the training phase. Thirdly, the participants took part in a driving session with PVT, to practise dual task events for approximately ﬁve minutes. This last training session reproduced the condition of the experiment. After the test period, the participant started the experiment as described below performing on the PVT without feedback. The data collected were used to compute performance metrics to investigate the impact of subjective factors on vigilance. Driving scenario The driving scenario was identical for each participant, designed to induce decreasing vigilance. The driving environment was monotonous, with little and repetitive stimulus, reproducing the driving condition on rural roads in Australia. Such monotonous environments have high crash exposure, and vigilance decrement

Driver Behaviour and Training – Volume III

370

are among the factors contributing to these crashes (Meuter, Rakotonirainy, Johns, Tran and Wagner, 2005; Thiffault and Bergeron, 2003). The road environment consisted of a two-way rural road with a speed limit of 100 km/h, one lane in each way, and width of the road, lane marking and road shoulder width (Figure 29.1). Road trajectory was designed using a real road, with a light curve to control impact for the speed.

Australian rural road

Driving Simulator Scenario

Figure 29.1 Road condition reproduced in the driving simulator Experiments took place during the afternoon, between 1pm and 4pm, at a critical phase of the circadian cycle in terms of vigilance decrement (Lenné, Triggs and Redman, 1997; Loh, Lamond, Dorrian, Roach and Dawson, 2004). Finally, road signs were added to take into account the frequency of road signs in a rural environment and to make sure drivers knew the speed limit; a car coming in the opposite direction was added on average every two kilometres; and participants were asked to drive in the centre of the left lane. Subjective factors Subjective factors were collected for each participant before the experiment took place using standard questionnaires: •

• •

personality proﬁle was assessed using an abbreviated form of the EPQR (Eysenck, Eysenck and Barrett, 1985) proposed by Francis, Brown and Philipchalk (1992), enabling the assessment of four personality dimensions: Extraversion, Neuroticism, Lie Scale, and Psychoticism; cognitive failure score was collected using the Cognitive Failure Questionnaire (Broadbent, Cooper, FitzGerald and Parkes, 1982); general information about participants related to age, gender and body mass index (BMI).

The Impact of Subjective Factors on Driver Vigilance

371

Participant vigilance We based our approach on the Human Information Processing Model (Broadbent, 1958a; Mackworth, 1950; Sigman and Dehaene, 2005; Wickens, 2002). We assumed that vigilance inﬂuences the allocation of attention needed to perform any task, and that several factors (environmental, subjective or the task) can inﬂuence vigilance levels, impacting on the ability to perform tasks. During the experiment, participants performed the unobtrusive psychomotor vigilance task (PVT) detailed in above, in the section ‘Psychomotor vigilance test’. According to the model of human processing (Broadbent, 1958b), and previous research (Mackworth, 1950; Meuter et al., 2005; Wellbrink et al., 2004), task performance was expected to be inﬂuenced by vigilance. Various performance metrics have been proposed in previous research to study performance on similar tasks. Mackworth (1950) studied participants’ vigilance using RT and frequency of misses. Wellbrink et al. (2004) assessed vigilance in terms of mean RT and standard deviation of percentage of false alarms and misses. Loh, Lamond, Dorrian, Roach and Dawson (2004) studied the validity of a short psychomotor test. They used the mean RT, the fastest ten per cent of RT, the percentage of lapses, and the slowest ten per cent of RT. Williamson et al. (2000) combined the RT and the accuracy (error and misses percentage) in order to make an analogy between effect of alcohol and effect of sleep deprivation on several psychomotor tasks. Meuter et al. (2005) analysed mean RT to investigate the effect of monotony on vigilance, and found variation of up to 50 per cent of RT depending on the monotony of the performed task. Sigman and Dehaene (2005) analysed the mean and the inter-quartile range of RT to investigate effect of task manipulation on performance. For the present research, we have computed the following performance metrics for each participant: most probable RT, mean RT, standard deviation of RT, lapses percentage, misses percentage and errors percentage computed for each. These performance metrics were then used in analysis to evaluate the impact of subjective factors on vigilance (Table 29.1). Table 29.1 Expected impact of low vigilance on performance metrics Metric Mode of RT Mean of RT Standard deviation of RT Errors % Misses % Lapses %

Impact Low Low High High High High

Driver Behaviour and Training – Volume III

372

Statistical analysis In order to investigate the effect of subjective factors on the performance metrics, the effect of subjective factors on the six performance metrics were analysed across sub-groups using separate analyses of variances (ANOVA). For ANOVAs, participants were divided into two groups for each subjective factor. For age, participants were divided in two groups based on the median age of the entire pool (Age = 45). For the body mass index, participants were divided in two groups based on the median of the entire group (BMI = 27). It happened that this separation is equivalent to that of the population for normal weight (BMI < 25) and overweight participant (BMI > 25). For each of the personality dimension extracted from the EPQR-test, as well as for the cognitive failure scores, participants were time separated in two groups based on the median value of the entire group. A transformation was made to participants’ responses to ensure normality of distribution according to assumptions of the ANOVA method. Inverse of square roots was used for mode and mean of RT, and square root was used for percentage of error, misses and lapses. Results Variance of reaction time Separate ANOVAs indicated that the RT distribution was signiﬁcantly affected by age group [F(1,12) = 5.81, p < 0.05], gender [F(1,12) = 5.35, p < 0.05], extraversion score [F(1,12) = 4.4, p < 0.1], and Body Mass Index (BMI) [F(1,12) = 3.22.4, p < 0.1] (Table 29.2). Results show that younger participants, females, participants with low extraversion score and participants with BMI lower than 25, have a shorter RT than, respectively, older participants, males, participants with high extraversion score, and participants with BMI higher than 25 (Figure 29.2). No other signiﬁcant effects were found for the other subjective factors. Table 29.2 Impact of extraversion on performance metrics

Metric Mode RTa Mean RTa Variance Lapses %b

Age F (1,12) p 5.811 0.033* 9.105 0.011* 5.887 0.032* 5.217 0.041*

Genre F (1,12) p 5.355 0.039* 4.417 0.057 5.205 0.042 4.937 0.046*

a

ANOVA conducted on transformed data ( 1

b

ANOVA conducted on transformed data ( x ). p < 0.05

*

x ).

Extraversion F (1,12) p 4.406 0.057 5.808 0.033* 5.402 0.038 7.539 0.017*

BMI F (1,12) p 3.224 0.098 3.953 0.070 4.978 0.045* 9.869 <0.001

The Impact of Subjective Factors on Driver Vigilance

373

Figure 29.2 Effects of subjective factors on performance metrics Standard deviation of reaction time Standard deviation of RT distribution was signiﬁcantly affected by age group [F(1,12) = 9.11, p < 0.05], extraversion score [F(1,12) = 5.81, p < 0.05], gender [F(1,12) = 4.41, p < 0.1] and BMI [F(1,12) = 4.98, p < 0.05] (Table 29.2), with younger participants and males showing less variation in their RT than respectively older participants and females, and high extraversion score and high BMI participants showing less variations of RT than respectively low extraversion score and low BMI participants (Table 29.2). Mean reaction time Signiﬁcant effects were found for mean RT for age group [F(1,12) = 5.89, p < 0.05], gender [F(1,12) = 5.20, p < 0.05], extraversion score [F(1,12) = 5.40, p < 0.05], and BMI [F(1,12) = 3.95, p < 0.1] (Table 29.2). No other signiﬁcant effects or relations were found for the other subjective factors.

374

Driver Behaviour and Training – Volume III

Percentage of lapses Percentage of lapses was signiﬁcantly affected by age group [F> (1,12)=5.22, p<.05], extraversion [F(1,12)=7.54, p<.05], gender [F > (1,12)=4.94, p<.05], and BMI [F(1,12)=9.87, p<.001]. Results show that younger participants, females, participants with low extraversion score and participants with BMI lower than 25, have a shorter RT than, respectively, older participants, males, participants with high extraversion score, and participants with BMI higher than 25 (Figure 2). No other signiﬁcant effects were found for the other subjective factors.

Figure 29.3 Interaction of age group and gender on sigma

Percentage of error and misses For percentage of error and percentage of misses, no signiﬁcant effects were found for any of the subjective factors. Discussion Our aim was to investigate the impact of subjective factors on driver vigilance. We proposed an innovative experiment as design to measure the psychomotor performance of the driver while driving. We collected data on a driving simulator and asked the participants to perform a non intrusive psychomotor vigilance test while driving. This test enabled us to identify and assess performance metrics. Subjective factors were collected for each participant using standard questionnaires. ANOVA identiﬁed signiﬁcant effects of some subjective factors on performance metrics. Age and gender impacting signiﬁcantly on mean RT, most probable RT, standard deviation of RT, and percentage of lapses. Analysis indicates that younger participants perform better than older ones, and that males perform better than

The Impact of Subjective Factors on Driver Vigilance

375

females. An interaction between age and gender has also been found, with female performance on these metrics being more affected by age than male performance (Figure 29.3). However, no effects of these two subjective factors have been found on error percentage and lapses percentage. Cognitive failure, which was expected to impact on performance metrics, has not been found to have a signiﬁcant effect on any of the performance metrics collected during the experiment. Each participant answered the EPQR-A test (Francis et al., 1992), a shorter variant of the Eysenck personality questionnaire to assess four personality dimensions: Neuroticism, Lie Scale, Psychoticism, and Extraversion. From these four subjective factors, only Extraversion affected performance. Signiﬁcant effects were found for mean RT, most probable RT, standard deviation of RT, and percentage of lapses. Analysis showed that participant with a high extraversion score performed better than participants with a low extraversion score. Finally, body mass index (BMI), which is computed using the reported mass and size of each participant, has been found to have a signiﬁcant effect on several performance metrics (on mean RT, most probable RT, standard deviation of RT, and percentage of lapses). Results of analysis suggest that participants reporting a BMI higher than 25 (overweight persons) perform better on these metrics than participants reporting a BMI lower than 25 (normal weight persons). These results support and expand the notion that subjective factors affect vigilance performance. This information can be used to estimate, for a given driver, a relative probability to experience vigilance decrement. Then, while combined with other indicators of vigilance (Gruyer et al., 2005b), reliability of the ﬁnal estimation of vigilance is inﬂuenced, leading to an increase in accuracy and robustness of the overall system. This is the approach of our main project and this present study is a ﬁrst stage. Given the present ﬁndings future work on driver vigilance could investigate interaction of subjective factors with environment and task factors. This could then lead to either a conﬁrmation of the effects of these subjective factors on vigilance, or an extension of the theory and a fuller appreciation of the interaction of all these factors on vigilance. References Allen, R.W., Cook, M.L. and Park, G.D. (2005). ‘Novice driver performance improvement with simulator training.’ In L. Dorn (ed.), Driver Behaviour Training, Vol. II, Aldershot: Ashgate, 81–91. Bekiaris, E., Amditis, A. and Wevers, K. (2001). Advanced Driver Monitoring: The AWAKE Project. Paper presented at the 8th World Congress on ITS, Sydney, Australia. Benedetto, A., Blasiis, M.R.D. and Benedetto, C. (2004, 5th to 9th September). Evaluation of Reaction Time in Virtual Reality Environment for Road Safety Increasing. Paper presented at the 3rd International Conference on Trafﬁc and

376

Driver Behaviour and Training – Volume III

Transportation Psychology (ICTTP3), Nottingham, UK. Broadbent, D.E. (1958a). ‘The nature of vigilance.’ In Perception and Communication. London: Pergamon Press, 108–39. Broadbent, D.E. (1958b). Perception and communication. London: Pergamon Press. Broadbent, D.E., Cooper, P.F., FitzGerald, P. and Parkes, K.R. (1982). ‘The Cognitive Failures Questionnaire (CFQ) and its correlates.’ British Journal of Clinical Psychology, 21, 1–16. Eysenck, S.B.G., Eysenck, H.J. and Barrett, P. (1985). ‘A revised version of the psychoticism scale.’ Personality and Individual Differences, 6, 21–29. Francis, L.J., Brown, L.B. and Philipchalk, R. (1992). ‘The development of an abbreviated form of the revised eysenck personality questionnaire (EPQR-A); its use among students in England, Canada, the U.S.A. and Australia.’ Person. Individ. Diff., 13(4), 443–49. Gruyer, D., Rakotonirainy, A. and Vrignon, J. (2005a, February). Advancement in Advanced Driving Assistance Systems Tools: Integrating Vehicle Dynamics, Environmental Perception and Driver’s Behaviour to Assess Vigilance. Paper presented at the Intelligent Vehicles and Road Infrastructure Conference (IVIRI’05), Melbourne, Australia. Gruyer, D., Rakotonirainy, A. and Vrignon, J. (2005b). The Use of Belief Theory to Assess Driver’s Vigilance. Paper presented at the Australasian Road Safety Research, Policing and Education Conference, Wellington, New Zealand. Gruyer, D., Royere, C., Lac, N.D., Michel, G. and Blosseville, J.-M. (2006, October). SiVIC and RTMaps, Interconnected Platforms for the Conception and the Evaluation of Driving Assistance Systems. Paper presented at the ITSC’06, London, UK. Johns, B. and Counselling, Q. U. o. T. S. o. P. a. (2004). Monotony and Personality Differences: An Examination of Performance in Singular and Dual Vigilance Tasks. Lamond, N. and Dawson, D. (1999). ‘Quantifying the performance impairment associated with fatigue.’ Journal of Sleep Research, 8, 255–62. Lenné, M.G., Triggs, T.J. and Redman, J.R. (1997). ‘Time of day variations in driving performance.’ Accident Analysis and Prevention, 29(4), 431–7. Loh, S., Lamond, N., Dorrian, J., Roach, G. and Dawson, D. (2004). ‘The validity of psychomotor vigilance task of less than 10-minute duration.’ Behaviour Research Methods, 36(2), 339–46. Mackworth, N.H. (1950). Researches on the Measurement of Human Performance. London: Medical Research Council. Meuter, R.F., Rakotonirainy, A., Johns, B., Tran, P. and Wagner, P. (2005, 11–15 September). Dual Vigilance Task: Tracking Changes in Vigilance as a Function of Changes in Monotonous Contexts. Paper presented at the Proceedings International Conference on Fatigue Management in Transportation Operations, Seattle, USA. Queensland Government, Q.T. Webcrash 2.3. Retrieved April 2005, from https:// www.webcrash.transport.qld.gov.au/webcrash2/ Sigman, M. and Dehaene, S. (2005). ‘Parsing a cognitive task: a characterization of

The Impact of Subjective Factors on Driver Vigilance

377

the mind’s bottleneck.’ PLoS BIOLOGY, 3(2), 334–49. Thiffault, P. and Bergeron, J. (2003). ‘Fatigue and individual differences in monotonous simulated driving.’ Personality and Individual Differences, 34, 159– 76. Wellbrink, J., Zyda, M. and Hiles, J. (2004). ‘Modelling vigilance performance as a complex adaptive system.’ Journal of Defense Modeling and Simulation (1). Wickens, C.D. (2002). ‘Multiple resources and performances prediction.’ Theoretical Issues in Ergonomics Science, 3(2), 129–77. Wierwille, W.W. (1994). Overview of Research on Driver Drowsiness Deﬁnition and Driver Drowsiness Detection. Paper presented at the 11th International Conference on Enhanced Safety of Vehicles, Munich, Germany. Wierwille, W.W., Tijerina, L., Kiger, S., Rockwell, T., Lauber, E. and Bittner, A.J. (1996). Heavy Vehicle Driver Workload Assessment (Research Paper): US Department of Transportation, NHTSA. DOT HS 808 467 (4). Williamson, A., Feyer, A.-M., Friswell, R. and Finlay-Brown, S. (2000). Development of Measures of Fatigue: Using an Alcohol Comparison to Validate the Effects of Fatigue on Performance. Zhu, Z., Ji, Q. and Lan, P. (2004). ‘Real time non-intrusive monitoring and prediction of driver fatigue.’ IEEE Transactions on Vehicular Technology.

This page intentionally left blank

Chapter 30

The Use of Local Case Review Panels to Determine Contributory Factors Crash Data Peter Hillard, David Logan and Brian Fildes Monash University, Australia Introduction Since 2004, the Monash University Accident Research Centre, MUARC, has been conducting an in-depth crash investigation programme for the Victorian Roads Corporation designed to improve understanding of the factors that contribute to the occurrence and severity of serious injury crashes. Drivers and motorcycle riders who have sustained serious injuries in a crash are recruited in hospital and comprehensive data are collected from them and their medical records. The vehicle(s) involved are located and inspected to determine the performance of the vehicle structure and safety systems, and to identify how the occupants sustained their injuries. The crash site is visited to record both the layout of the road, and associated infrastructure, and any evidence of the crash itself. The previous crash history of the site is also researched. Police, ambulance and other emergency service reports on the crash are accessed. Where the crash involves more than one vehicle the other driver(s) involved are also interviewed, if they are willing to participate. The data collected from all these sources are compiled and then evaluated on a ‘no-blame’ basis by case review panels convened in the locality where the crash occurred. Panels are primarily composed of local stakeholders such as the state and council trafﬁc engineers responsible for the road, representatives from the local police, emergency services, receiving hospital and local community organisations. Where possible, the actual emergency service personnel who attended the scene serve on the panel. Where particular specialist knowledge seems likely to aid consideration of a particular crash, appropriate experts are invited to sit on the panel. Panels are facilitated by senior MUARC staff. The objectives of the panel are, ﬁrstly, to determine the factors that contributed to the occurrence of the crash and inﬂuenced the severity of the injuries sustained, and secondly, to determine actions which will prevent similar crashes occurring in the future. There are many in-depth crash investigation programmes operating around the world. Many of these collect and analyse contributory factors data. The uniqueness of the current study is the extent to which local stakeholders, with their superior knowledge of the local road network and conditions, are involved in determining the

380

Driver Behaviour and Training – Volume III

contributory factors data. This paper summarises the outcomes of the ﬁrst 79 crash investigations conducted and reviewed under the programme. Methods Crash investigations The protocol for the data collection element of the programme is similar to that of other retrospective real-world crash investigation studies, and in particular the Australian National Crash In-depth Study that has been run by MUARC since 2000 (Fildes et al., 2003). However, more emphasis is placed on collecting contributory factors data than is the case in many crash investigation studies, which often focus primarily on vehicle crashworthiness and assessment of the performance of safety systems. Potential study participants are identiﬁed through regular screening of the admissions databases in three hospitals in Melbourne and 14 hospitals in the regional centres of rural Victoria. The inclusion and exclusion criteria are designed both to target the most commonly occurring types of crashes in Victoria and to avoid the potential legal and ethical pitfalls inherent in a study of this nature. Participants are drivers/riders of cars, motorcycles and light commercial vehicles (the most common vehicles on the state’s roads) who have sustained serious injury as a result of a crash on a public road. Serious injury crashes are deﬁned as those that lead to the hospitalisation of the driver for a minimum of one night. As one focus of the project is contributory factors, the driver interview forms an important part of the dataset and so participants who are not capable of being interviewed are excluded. Drivers/ riders without a valid licence, those driving/riding stolen vehicles, and those with a history of psychiatric illness are also excluded from the study. Crashes that result in a fatality in any of the vehicles involved within 30 days, those that were under continued police investigation, and those involving a bicycle, pedestrian, bus, taxi, train, tram or ridden horse, are also excluded from the study. Having provided informed consent, participants are interviewed by a MUARC research nurse. The interview is structured and usually takes about 40 minutes depending on the individual and the nature of their vehicle and involvement (that is, different interviews are used for motorcyclists, case vehicle drivers, and noncase vehicle drivers). The interview questions cover, inter alia, the location of the crash; prevailing weather conditions; the participant’s recollection of events preduring, and post-crash; details of the vehicle driven; details of the participant’s driving experience; details of the trip; their perceptions of the causes of the crash; and their physical and psychological well-being immediately prior to the crash. The participants’ medical records are also accessed and the injuries they sustained are coded using the Abbreviated Injury Scale (AAAM, 1998). In addition, blood alcohol concentration and details of pre-existing medical conditions and medications are recorded. Using information provided by the participant, their vehicle is then located and inspected by a MUARC engineer, having ﬁrst done a degree of background

The Use of Local Case Review Panels

381

research on the model speciﬁcation, Used Car Safety Rating (Newstead, Cameron and Watson, 2005), New Car Assessment Programme rating (NCAP), and so on. The inspection involves measurement and photography of the interior and exterior damage to the vehicle, recording of the passive safety systems present on the vehicle and investigation of their performance during the crash event, and assessment of the pre-crash roadworthiness of the vehicle. Crash severity and direction of principal force/impact angle are then computed from the crush proﬁle and other vehicle data using CRASH3 software. In about three-quarters of the crashes involving more than one vehicle it has been possible to locate the other vehicle(s) involved and similar inspections were carried out on them. Having inspected the vehicle(s) involved in the crash, the MUARC engineer responsible for the case visits the site of the crash within two weeks of its occurrence. The objectives of the site inspection are to collect data on the road infrastructure (for example, surface material and condition, layout, trafﬁc controls, and so on), the roadside setting (for example, adjacent land use, topography, presence of trees and poles, sight distances, and so on) and to locate evidence of the crash (for example, damage to infrastructure and trees, skid marks, and so on). A detailed photographic record of the site is also made, along with drive-through videos from the perspective of each of the vehicles involved. Once the driver, vehicle and site data have been collected, simulations are conducted using the PC Crash and MADYMO modelling packages. The models are used both to estimate details such as the approach speeds of the vehicle(s) and also to produce animated computer graphics depicting the kinematics of the vehicle(s) and occupant(s), which are shown to the local panel reviewing the crash. The previous crash history of the crash site is also researched using Victorian Roads Corporation’s crash database, Crashstats, which contains records of all police reported crashes in Victoria since 1987. At this stage in the investigation the police crash report form is also accessed. Finally, the ﬁndings of the investigation are compiled into a presentation of about 30 minutes’ duration. A standardised format is used for the presentations and they are concluded with a draft list of the contributing factors leading to the occurrence of the crash prepared by the crash investigation team. Local case review panels To date, 28 local case review panels have been convened to review a total of 79 crashes. Collectively, the panels have involved the participation of 435 individuals from 23 different organisations and a range of local government authorities. The majority of these panellists were trafﬁc engineers from local government and Victorian Roads Corporation regional ofﬁces (271) or local members of Victoria Police and the emergency services (Rural and Metropolitan Ambulance Services, State Emergency Service, Country Fire Authority, hospital emergency departments) (109). Most panels also had a representative from the local Road Safety Community Group. The remaining panellists came from a range of organisations and were invited to attend for the specialist knowledge they could provide to the consideration of a particular crash. For example, when a motorcycle crash was reviewed a representative of the Victorian Motorcycle Advisory Council was invited to attend, and when a crash

382

Driver Behaviour and Training – Volume III

involved roadside trees a representative from the local Department of Sustainability and Environment ofﬁce was invited to attend. Cases are presented to the panels in a de-identiﬁed format, that is, details such as the name(s) of the individual(s) and registration number(s) of the vehicle(s) involved, and the exact date of the crash are withheld. To further protect study participants, all panellists are required to complete a deed of conﬁdentiality in favour of MUARC at the commencement of the panel. Typically, one-and-a-half to two hours is devoted to reviewing each crash. After being presented with the information collected on the crash, the panels consider the draft list of the contributing factors that has been prepared by the crash investigation team. Due to their superior knowledge of the local road network and conditions, the panels are often able to rank these factors in relation to their relative importance and in some cases determine additional ones which would not be apparent to an outsider. The panels then use a consensus approach to prepare a list of action items. The panel moderators encourage panellists to think outside the conﬁnes of their particular professional viewpoints (that is, the established policies, standards, and guidelines, and so on, that they were used to working to) and adopt a broader, safe systems approach. The action list is then distilled down to those items which are practicable and designated ‘high priority’ by the group. Panellists then take responsibility for carrying these actions out and reporting back progress at the next meeting of the panel. Case study The accident occurred early on a Monday morning on a minor road in the vicinity of a Victorian rural city. The single vehicle involved was a large passenger car, with four occupants, which left the road at a bend in the road and struck two trees before rolling and coming to rest on its roof. Two of the occupants were admitted to the hospital in the nearby city and the driver, who had been trapped in the vehicle, was taken by air ambulance to one of the main trauma hospitals in the capital and remained as an in-patient for 12 nights. The driver, who had been driving with his window down, had serious injuries to the right arm and the right side of the head. The driver worked in the hospitality industry and had had a busy weekend. Following a 16-hour workday and only four hours’ sleep on Saturday, the driver had worked all day on Sunday and then had what he described as a ‘big night’ with friends. Having only had four hours’ sleep again on Sunday, the driver was persuaded by friends to get up early on Monday and drive them into the city. The crash happened 25 minutes into the one-hour journey. The driver was experienced, covered around 125 000 kilometres a year, and used the road on which the crash occurred about once a month. The driver admitted falling asleep but also complained about the lack of warning signage at the bend where the crash occurred. The driver’s medical records indicated a blood-alcohol concentration of 0.13 g/100ml (the legal limit for driving in Victoria is 0.05 g/100ml). The vehicle was only a month old and had a four-star NCAP rating. It had a number of active safety systems, including ABS and a system that on detecting an emergency stop applies maximum brake force. Passive safety devices included dual

The Use of Local Case Review Panels

383

front airbags and side head-thorax airbags. None of these deployed during the crash but the driver’s injuries were primarily caused by contact with the ground when the vehicle rolled and so would have been unlikely to be mitigated by deployment. The crash team concluded that the vehicle had performed very well given the severe nature of the crash. The road at the crash site was eight metres wide, had recently been re-surfaced, and was in good condition though no line markings were present. The radius of the bend was about 40 metres and the road had a 100 kilometres per hour speed limit. The two trees struck were about eight metres from the edge of the road. Due to the subsequent damage caused by the rollover it was not possible to calculate the speed at which the trees had been struck, but it was estimated to be between 60 and 80 kilometres per hour. Research on the previous crash history of the site revealed one previous crash of a very similar nature. When interviewed, the driver had cited fatigue and insufﬁcient warning of the bend as the main contributing factors to the occurrence of the crash. In reviewing the case the local review panel determined that the driver’s alcohol impairment, and an associated loss of concentration, played the main role in the vehicle leaving the road. The panel also suspected that the vehicle had been travelling above the 100 kilometres per hour speed limit, though it had not been possible to establish this during the site inspection due to the absence of braking marks (a characteristic of anti-lock brake-equipped cars even under heavy braking). Nevertheless, the panel also concluded that the absence of line markings had also contributed to the occurrence of the crash as, if they had been present, would have made it easier for the driver to perceive the bend. The main action arising from the review process was against the organisation responsible for the road, which undertook to carry out infra-structural improvements at the site. These included installation of curve alignment markers and a 40 kilometre per hour advisory sign, and re-marking of the centre and edge line markings. The improvements had been carried out by the time the next panel was convened in the region two months later; see Figures 30.1 and 30.2. The second action related to the setting up of a regular monthly meeting between representatives of the Victorian Roads Corporation regional ofﬁce and local government representatives for the purposes of information exchange on road safety issues. The other actions arising were recommendations for a review of the default speed limits for minor rural roads and the greater application of alcohol interlocks. Outcomes Aggregate contributory factors data Collectively, the regional panels reviewed 79 crashes and identiﬁed a total of 295 factors that had either led to their occurrence or inﬂuenced their severity. Table 30.1 shows an analysis of the frequency with which individual factors occurred, broken down into human, vehicle and environmental groupings. Hence, the most common human factor identiﬁed was fatigue impairment (27 crashes), the most common

384

Driver Behaviour and Training – Volume III

Figure 30.1 Case study crash site before treatment identiﬁed by the local case review panel

Figure 30.2 Case study crash site after treatment identiﬁed by the local case review panel vehicle factor identiﬁed was poor vehicle crashworthiness (12 crashes), and the most common environmental factor identiﬁed was proximity of a tree or pole to the road (22 crashes).

The Use of Local Case Review Panels

385

Fifty-four per cent of the factors identiﬁed related to the actions, behaviour, or the mental or physical state of one or more of the drivers involved. Eleven per cent related to one or more of the vehicles involved. The remaining 35 per cent of the Table 30.1 Frequency of occurrence of contributory factors across the 79 crashes reviewed Factor

Frequency

Human Impairment (Fatigue) Distraction/inattention Speeding Inexperience Impairment (prescription drug) Impairment (drugs and alcohol) Impairment (medical) Driver error (including deliberate violation of road rules) Driving at an inappropriate speed for the prevailing conditions

158 27 26 24 22 13 12 12 12 10

Vehicle Poor vehicle crashworthiness Clothing provided inadequate protection (motorcyclists) Poor roadworthiness (including tyre condition) Low motorcycle conspicuity

32 12 8 6 6

Road, roadside setting and environment Proximity of pole/tree to road Road layout (topography, alignment, lane width, etc) Speed limit too high (for quality of road or nature of roadside setting) Poor line markings and/or edge delineation Insufﬁcient level of control at intersection Weather conditions Inadequate lighting Unsealed shoulders Road surface condition

105 22 22 21 12 7 6 6 5 4

factors related to the road, roadside setting and environment. These ﬁgures may appear to differ from a number of previous studies of crash causation, such as the Indiana Tri-level and TRRL studies (Treat et al., 1980; Sabey and Taylor, 1980), which usually indicate a much higher percentage of factors related to the humans involved. The main reason for this is that the panels were not solely identifying factors that had led to the occurrence of the crash but also those that had inﬂuenced its severity. However, other issues, such as the exclusion of pedestrian crashes, could also play a part.

386

Driver Behaviour and Training – Volume III

Panel action items To date, the panels have established 225 ‘high priority’ action items. The most frequent type of action has related to infra-structural improvements at the crash site under consideration. In many cases the organisation responsible for the road has undertaken to carry out improvements directly. In other cases a commitment has been given to carry out a detailed site inspection to assess the need for, and most appropriate type of, improvements, or to prepare a funding proposal for the works required. The next most frequent type of action has been to conduct research on a particular issue. In 11 cases this research has been site speciﬁc and related to detailed and/or comparative analyses of the crash history of lengths of road or intersections. The other research actions have been more universal; examples include two literature reviews, one on systems of curve warning signage and the other on the effects of prescription drugs on driving. The majority of the remaining actions have related to undertakings to report back to the next panel on an issue, to develop campaigns or promotional material, or to conduct reviews of guidelines, policies, or future priorities at an organisational level. The panels have also made 17 recommendations for regulatory changes to the state government. With respect to the topic area of the actions, 65 of the actions were site speciﬁc in that they related to infrastructure, speed limit, and so on, at the site at which the particular crash under consideration had occurred. A further 33 also related to infra-structure but in a more general way; for example a number of the panels identiﬁed the need for the development of effective short-length barrier systems for use in urban environments. A total of 42 actions related to various types of driver impairment (medical, drug and alcohol, and distraction, and so on) and 16 related to various aspects of driver training. Thirty-one actions concerned vehicle design standards, safety systems, and the potential beneﬁts of Intelligent Transport Systems. Twelve of the actions speciﬁcally concerned motorcyclists and these mainly related to protective clothing and conspicuity. The remaining topic areas with more than a single action concerned: accident reporting and analysis (seven actions), used car safety (ﬁve actions), and ﬂeet purchasing (four actions). Discussion The programme discussed in this paper was conceived to fulﬁl a number of objectives. Firstly, the project provided a means of collecting more detailed information on critical crash types. Secondly, it was intended to be an educational tool for regional practitioners that made up the local review panels. Thirdly, it was hoped it would provide a mechanism for increasing both the amount and effectiveness of the cooperation among the diverse range of key stakeholders. Finally, it was expected that it would facilitate the implementation of new innovative countermeasures at the crash sites investigated. It is still early days to assess the outcomes of the project as many of the issues identiﬁed are wide-ranging, and many of the projects initiated could take years to complete. Hence, many of the beneﬁts arising from the project will only become apparent in the long term. However, it is clear that the four project

The Use of Local Case Review Panels

387

objectives have already been substantially met and any long-term beneﬁts arising should be a bonus. The outcomes in relation to the four project objectives will be brieﬂy discussed below. Information on critical crash types The case review process has identiﬁed a number of the key factors leading to the occurrence of vehicle crashes in Victoria today and a range of issues which need to be tackled in order to reduce the incidence of serious injury crashes. Speciﬁcally, the need for further research and development is indicated in the following areas: • • • • •

the extent of the involvement of fatigue in urban crash causation and development of an objective measure of fatigue; development of effective short-length barrier treatments for urban environments; development of low-cost mass action treatments for sharp bends in minor rural roads and for intersections between them and more major routes; the effects of prescription drugs, particularly anti-depressants, on driving; methods to improve motorcycle conspicuity

Educational objectives In relation to the educational outcomes of the project, participation clearly gave panellists a more holistic understanding of road trauma and its consequences. A programme evaluation exercise was conducted on a sample of 86 of the panellists, that is, just under 20 per cent of the total to date of 435. Ninety-seven per cent of respondents rated the process of in-depth crash investigation and review as excellent or good overall, 97 per cent said they felt that their organisations’ involvement had been beneﬁcial, and 90 per cent said that their involvement in the project had increased their understanding of the factors that contributed to serious crash causation. Many also commented that they regarded their participation in the programme as a form of professional development. Development of strategic road safety partnerships The third objective of the programme relates to the development of new and more effective strategic road safety partnerships between stakeholder organisations. Fortyone per cent of those surveyed in the programme evaluation exercise indicated that involvement in the programme had assisted in the development of new partnerships and a further 13 per cent indicated that it had consolidated existing partnerships. It seems likely that in the future further partnerships will arise between stakeholders and that development of these will be facilitated by acquaintances made at local panel meetings.

388

Driver Behaviour and Training – Volume III

Infrastructural improvements at crash sites The ﬁnal objective of the programme relates to the implementation of infra-structural improvements. Fifty-four of the 79 crash sites reviewed by the panels to date have received attention as a result of the ECI process. The panels felt that improvements at the remaining 25 crash sites were not warranted because they would be unlikely to prevent similar crashes occurring again. Conclusions The programme originally ran from November 2004 until June 2006. The process of crash investigation and review was considered to be highly worthwhile by all of the stakeholders involved. The importance of on-going programmes for the collection of in-depth crash data to improve understanding of crash causation was also universally acknowledged. At the time of writing, two extensions to the programme have recently started. One of these is a two-year programme speciﬁcally focusing on motorcycle crashes; the other is a ﬁve-year programme looking at all other classes of light vehicle. Acknowledgements The authors would like to thank staff in the seventeen participating hospitals for assistance with the recruitment of participants. Support for the programme is received from the Victorian Roads Corporation under Professional Services Agreement No. 6018. Ethics approval was obtained from the Monash University Standing Committee for Ethical Research Involving Humans (Approval No. 2004/903A) and each of the participating hospitals. The views expressed in this paper are those of the authors, and not necessarily those of Monash University or the Victorian Roads Corporation. References AAAM (1998). The Abbreviated Injury Scale (AIS), Des Plaines, Association for the Advancement of Automotive Medicine. Fildes, B., Logan, D., Fitzharris, M., Scully, J. and Burton, D. (2003). ANCIS – The First Three Years, Melbourne, Monash University Accident Research Centre. Newstead, S., Cameron, M. and Watson, L. (2005). Vehicle Safety Ratings Estimated from Police Reported Crash Data: 2005 Update. Australian and New Zealand Crashes during 1987–2003, Melbourne, Monash University Accident Research Centre. Sabey, B.E. and Taylor, H. (1980). ‘The known risks we run: the highway.’ In Schwing, R.C. and Albers, W.A. Jr., (eds), Societal Risk Assessment: How Safe Is Safe Enough? New York, Plenum Press. Treat, J.R., Jones, R.K. and Joscelyn, K.B. (1980). ‘Analysis of unsafe driving actions – data requirements and methodological issues.’ In Evans, L. (ed.), Accident Causation, SAE/SP-80/461, Warrendale, Society of Automotive Engineers.

Chapter 31

The Effectiveness of New Seat Belt Legislation in Northern Ireland A.R. Woodside, J.R. Seymour and C. Gallagher University of Ulster, Northern Ireland Introduction Child passenger safety while travelling within the car has become a major issue as road safety organisations seek to continue the downward trend of fatalities on our roads. New legislation has been passed in the United Kingdom and subsequently been implemented in Northern Ireland (February 2007), as a measure to help assist the safety of children whilst in the car. The aim of this law is to try and limit the number of young children who suffer neck injuries in the event of a collision. The law applies to all children under the age of 13 unless they are taller than 135cm (4ft 5in). The aim of this project is to examine parental attitudes and compliance with the newly introduced seat belt and booster seat legislation in Northern Ireland. Seat belt legislation The seat belt legislation was implemented in February 2007 in the hope of creating a safer environment for young children whilst in a motor vehicle, and overall in the UK it has been estimated that the legislation ‘could save up to 2,000 children per year from death or injury in road collisions.’ (BBC News, 2006). One of the controversies with the new legislation is that certain types of vehicle are exempt from complying with the law in certain circumstances, such as taxis, emergency vehicles and for unexpected journeys and when faced with a lack of space. Taxis Children aged less than three years old may travel unrestrained in the rear of a taxi and children aged three or above are required to use an adult seat belt, in the rear of the vehicle only. In addition to this it is the taxi driver who his held legally responsible to ensure that all children aged less than 14 use available seat belts and or child restraints, except where there is a ﬁxed partition.

390

Driver Behaviour and Training – Volume III

Distance A child of three years or more may use an adult seat belt, if there is an unexpected journey which is regarded as necessary. This does not include regular school runs or planned journeys. Occupancy In smaller cars, if ‘two occupied child seats or boosters prevent the ﬁtting of a third, and the front seat is not available, a third child aged three years and over may then use just an adult belt in the rear’ (DfT, 2006). Emergency vehicles Child passengers travelling in emergency vehicles are exempt from the legislation whatever the child’s age or size. Exemptions The exemptions in the legislation have been a cause of controversy, with taxis at the forefront. From the earlier unpublished study, many people (50 per cent) felt it is unfair that taxis should be excluded from the legislation. Objective of the legislation The primary objective of the law is to limit the amount of neck injuries suffered by children in the event of a collision. According to the Minister of State for Transport, Dr Stephen Ladyman, ‘sitting higher in the vehicle – rather than using seatbelts designed for adults, should mean children suffer fewer injuries to their spine and internal organs if involved in a crash, seated passengers aged 14 and above must use seatbelts where they are ﬁtted’ (Nottingham Evening Post, 2006). Despite the publicity surrounding the new legislation there is still some confusion. According the Essex Chronicle (2006) more than two-thirds of drivers did not know the fundamental rules surrounding the law; that a child has to be over 135cm tall to legally wear a standard seat belt. Over 75 per cent of the general population in the United Kingdom think that the adult driver is responsible for making sure that all passengers under the age of 18 travelling in their car are appropriately restrained, when the legislation clearly states that any person over the age 14 is responsible for their own actions within the motor vehicles concerning seatbelt wearing. Coupled with this ‘More than half (57 per cent) of drivers believe that no child can travel in a vehicle that is not ﬁtted with seat belts’ (Essex Chronicle, 2006). According to the Minister for the Environment (NI), David Cairns, ‘one in four backseat passengers in Ulster aged between 14 and 29 do not wear seatbelts and 15 per cent of children aged between ﬁve and 15 are not properly belted in’ (Belfast Telegraph, 2006).

The Effectiveness of New Seat Belt Legislation in Northern Ireland

391

It is evident that there is a relatively low level of compliance with the previous seat belt laws. This study seeks to establish if such a resistance and non-compliance has persisted. Why is the legislation needed? In a post-conﬂict situation, Northern Ireland is a country with a rapidly growing economy, with results on many levels. Statistically, unemployment throughout the region is on the decline and as a direct reﬂection of this there is an increase in the number of cars purchased and our car dependence has increased. The Regional Development Strategy (2006) states that since 1960 the number of vehicles in the region has increased 400 per cent, and by 2025 the number of cars on our roads today may double again. The implementation of the new seat belt legislation is another step in trying to cut down the number of casualties that may be a result of this increasing car use (DOE, 2001). The Northern Ireland Transport Statistics (2005–2006) showed that in 2004 there were 331 injury collisions per 100 000 population in Northern Ireland, 369 injury collisions per 100 000 population in England, 274 injury collisions per 100,000 population in Scotland, and 325 injury collisions per 100 000 population in Wales (DRD, 2006). In addition to this in 2004, 8.6 people were killed in Northern Ireland per 100 000 population due to a road trafﬁc collision. England had 5.4 deaths, Scotland had 6.1 deaths and Wales had 6.8 deaths. Northern Ireland had the highest number of deaths, again highlighting the need for this new measure to help reduce the number of casualties. According to the PSNI (2007), the number of RTCs involving children (under 16) has decreased 45 per cent and fatalities have halved in the last nine years. Also, two thirds (67.2 per cent) of all child casualties were passengers and half of the child fatalities were vehicle passengers. The Northern Ireland Seat Belt Survey (2005) showed that seat belt usage varies between drivers, front seat passengers and rear seat passengers. Drivers have the highest wearing rate (93 per cent), followed closely by front seat passengers (92 per cent), however, there is a considerable decrease regarding wearing rate of back seat passengers (81 percent) (DoE, 2005). The driver of the vehicle also has an impact on rear seat-belt usage: ‘The seat belt wearing rate for both front and back seat passengers was higher when the driver of the vehicle wore a seat belt, compared to when they did not… 83 per cent of back seat passengers wore a seat belt when the driver of the vehicle was wearing one, compared to only 53 per cent when the driver was not.’ Regarding children, the ‘back seat wearing rates for older children (5–9 years and 10–13 years) have risen steadily over the period. The rates for both 5–9-yearolds and 10–13-year-olds have risen to 82 per cent, the highest rate recorded for both age groups since the series began in 1994. However, around one in ﬁve children (18 per cent) aged between 5 and 13 years still travel unrestrained in the back seat.’ A reason for this may be due to an assumption that travelling in the back seat of the car is somewhat safer and that they are less aware of the risks of travelling unrestrained, despite the ongoing efforts of the DoE’s Road Safety Group and their television advertisement campaign.

Driver Behaviour and Training – Volume III

392

Method A number of schools in the Belfast Metropolitan Area and County Armagh were contacted to ascertain their willingness to participate in the study and two were chosen for an in-depth analysis of parental views and attitudes upon the topic. The questionnaires were distributed in late March after the introduction of the legislation. A questionnaire was developed for both rural and urban areas. The questionnaire consisted of a number of questions, all designed in either ‘tick box’ form or of numerical classiﬁcation. The main thrust of the questionnaire was to discover how parents viewed the legislation, and to discover whether there are any links between parents’ opinions living in rural areas compared to parents living in urban areas. 700 questionnaires were distributed and only 174 questionnaires were returned. Results The majority of parents who responded to the questionnaire (70 per cent) have two children, and a high proportion (39 per cent) have only one child. Forty-one per cent of parents own a car, 53 per cent own two and six per cent own at least three cars. Seventy60

50

51

Percentage

40

30

23

20

13

10

11

0 always

sometimes often

N/A

take it for granted

Do you ensure your child is restrained

Figure 31.1 Do you ensure your child is restrained? seven per cent of parents surveyed who own a car have already purchased a booster seat and only 64 per cent always or often check that their children are restrained. Figure 31.1 shows that 51 per cent of parents ‘always’ ensure that their child is appropriately fastened, 13 per cent ‘often’, two per cent ‘sometimes’, 11 per cent

The Effectiveness of New Seat Belt Legislation in Northern Ireland

393

‘take it for granted’ that they are always appropriately fastened. In additional to this 23 per cent of the respondents are deemed ‘not applicable’ due to the fact they had not already purchased a booster seat. These ﬁgures illustrate that 24 per cent of parents who have purchased a booster seat still do not take full precautions, ensuring that their booster seat and children are ‘always’ adequately belted in. More worrying is the admission by 45 per cent of parents that they ‘often’ or ‘sometimes’ travelled without belting in their children since the legislation came into force (see Figure 31.2). Thirty-three per cent of parents said they never begin a journey without ﬁrstly ensuring their child/children are suitably fastened in their booster seat. However 34 per cent and 11 per cent of the respondents said they ‘sometimes’ and ‘often’ go on a journey without ﬁrst of all placing their child/ children in a booster seat. This is only allowed on unexpected short journeys, however doing this regularly is not complying with the legislation, therefore parents could be penalised. Eighty-two per cent of one-car households have purchased a booster seat but only 70 per cent of two-car households have purchased booster seats (Figure 31.3).

40

34

Percentage

30

33

23 20

10

11

0 often

sometimes

never

n/a

Have you placed your child without restraining them

Figure 31.2 Have you placed your child without restraining them? Since the introduction of the legislation, eight per cent of the respondents have been involved in some type of a collision and in half the cases the child(ren) were also involved in the collision. More worryingly, 94 per cent of children were not using a booster seat when involved in the collision. When asked if their personal attitudes, regarding seat belts, had changed since the legislation had been introduced, one-third of parents (34 per cent) claimed that it had, however, over half (51 per

Driver Behaviour and Training – Volume III

394

cent) said that it had not. Despite this ﬁnding, nearly three out of four parents thought that the legislation would have a positive impact on child road safety. 100

18

30

100

90 80

82

Percentage

70

70

60 50 40 30

Purchased seat

20

no

10

yes

0 1

2

3+

Car ownership

Figure 31.3 Have you purchased a booster seat?

When a cross-tabulation was made of the drivers who had been involved in a collision and whether their personal attitudes had changed towards the new seat belt legislation it can be seen that 71 per cent of collision involved drivers are in favour but the majority of non-collision drivers have not changed their attitudes regarding the new legislation on seat belts (see Table 31.1).

Table 31.1 Attitudes of drivers towards seat belt legislation Collision Yes

Yes 71%

No

30%

Attitude No/Same Not sure 29% – 55%

Total 100.0%

15%

Discussion A focal area which needs to be considered from the data collected during the survey was the number of cars owned in each household. The results received from each of the questionnaires coupled with the literature reviewed identify a key trend. Car

The Effectiveness of New Seat Belt Legislation in Northern Ireland

395

80

72

Percentage

60

40

25

20

0 yes

no

not sure

Do you feel the legisation will have an impact on child safety?

Figure 31.4 Do you feel the legislation will impact child safety? ownership and dependence is continuously increasing. It is noticeable that most respondents owned two cars. This follows a similar pattern to that highlighted in the review of literature, this being the rise in car ownership in the last ﬁve decades. As stated by the Regional Development Strategy (2006), since 1960 the number of vehicles throughout the region has increased up to 400 per cent, and by 2025 the number of cars on our roads today may again double. These results suggest that there could be a rise in road trafﬁc collisions a result of the increasing levels of mobility on Northern Ireland’s roads. Consequently all measures required to further enhance the safety of motorists should be taken. The danger would be that parents may not equip both (or all) cars with adequate child restraint and may take a risk in transporting their children for short journeys. The statistics show that there is a signiﬁcant number of collisions recorded annually involving both adults and children on a national and multi-national level. From an earlier unpublished study over 30 per cent of all the respondents had been involved in a previous collision(s), and of these 40 per cent of them involved both parent and child. Prior to the implementation of the new legislation, all children who had been involved in a road trafﬁc collision should have been wearing a standard three-strap belt as a form of restraint against injury. It has been understood from the literature that these adult seat belts can in fact induce back and neck injuries in small children in the event of a collision, thus highlighting the required need for the new legislation. In addition to this, eight per cent of the respondents had been involved in a collision within the space of a month of the new legislation, highlighting the need for in-vehicle restraints.

396

Driver Behaviour and Training – Volume III

The study showed that even a month after the introduction of the legislation 23 per cent of parents had not yet bought a child seat belt/booster seat. In addition to this, 49 per cent of the respondents do not ‘always’ check to see if their child(ren) are appropriately fastened and 33 per cent of the respondents, who had purchased a booster seat, ‘never’ begin a journey without ﬁrstly restraining their child in the seat. Conclusions This initial study has shown that there should be concerns over parental compliance with, and attitudes to, the recently introduced seat belt legislation into Northern Ireland. The majority of households possess at least one car, and in many cases two, and they are using these vehicles more and more. Many parents are at risk of collision, and one in 12 drivers in the present study was involved in a collision since the legislation was introduced. These drivers are risking the safety of their passengers and in particular their children, by not observing the regulations, or ensuring that their passengers are secured. Only 33 per cent of respondents who had purchased a booster seat ‘never’ began a journey without ﬁrstly strapping their child in the seat. It is important that the DoE Road Safety Branch and PSNI are seen to enforce the legislation so that this and future generations of children are not disadvantaged before their independent lives begin. Complete compliance is crucial for the regulation to be fully effective, and therefore greater emphasis needs to be placed on both education initiatives and enforcement activities. There is also need for further and detailed research into why parents fail to comply with the seat belt legislation, and how they could be made more aware of the consequences of their actions or inactions. References DfT (2006). Law, www.thinkroadsafety.gov.uk/campaigns/childcarseats/pdf/lawleaﬂet.pdf. DoE (2001). Regional Development Strategy (2025), http://www.planningni.gov.uk/ AreaPlans_Policy/Strategies/Strategy.htm. DoE (2005). Seat Belt Survey, http://www.roadsafetyni.gov.uk/index/publicity/ seatbeltsportal/seatbeltsresearch/seatbeltssurvey.htm. DRD (2006) Northern Ireland Transport Statistics Annual 2005–2006, http://www. drdni.gov.uk/statistic-details.htm?publication_id=170. Essex Chronicle (2006). ‘Seat confusion’, November 2nd, p. 22. McAleese, D. (2006). ‘Over 20,000 ﬁxed penalties issued for failures to belt up’, Belfast Telegraph, October 3rd, p. 2. Nottingham Evening Post (2006). ‘Warning about car seats’, October 31st, p. 10. PSNI (2007). Injury Road Trafﬁc Collisions and Casualties, 1st April 2006 – 31st March 2007, Statistical Report No. 5, http://www.psni.police.uk/5._injury_road_ trafﬁc_collisions_and_casualties.pdf.

PART 6 Rider Behaviour

This page intentionally left blank

Chapter 32

An Evaluation of the Portuguese Moped Rider Training Programme Patrícia António1,2,3 and M. Matos3 Institute of Drug and Addiction (IDT DRN CAT Bragança), Portugal 2 Portuguese Road Prevention, Psychology Department, Portugal 3 Faculty of Psychology and Educational Sciences, University of Lisbon, Portugal 1

Introduction Young people within the age range 14–25 are at most risk of involvement in a road trafﬁc accident than any other age group, and their injuries tend to be severe. Data shows that a signiﬁcant proportion of victims sustain multiple injuries and require hospitalisation (SWOV, 2002; Arnett, 2002; Ulleberg and Rundmo, 2002; McKnight and McKnight, 2003; Williams, 2003; Goldenbeld, Twisk and Craen, 2004). Accidents involving mopeds represent a signiﬁcant source of injury-related mortality and morbidity, especially in populations where the moped is a popular means of transport amongst young people, such as Portugal and other Southern European countries (for example, Italy, Spain, Greece and France). According to the European Commission (DG TREN Road Safety Unit, 2004), the rate of moped road trafﬁc crashes among adolescents (14–25 years old) on all moped road accidents is 67 per cent and the rate on motorcycles road trafﬁc crashes on all motorcycles road accidents is 38 per cent. This is particularly prevalent amongst young men (WHO, 2004a; 2004b). Unfortunately these represent the leading causes of death among adolescents, followed by suicide and other self-destructive behaviour. Portugal has one of the highest motorised two-wheeler mortality rates, followed by Italy and the UK (Reto and Sá, 2003; DGV, 2005). However, in recent years this fact seems to have been overlooked. Despite this, it is not known how large the problem of moped injuries represents in terms of overall trafﬁc safety (Matos, 1991; Kopjar, 1999; SWOV, 2001a; 2001b). Adolescence is the period between childhood and adulthood, during which the individual acquires increased physical maturity, education and training enabling young people to fulﬁl a useful role in adult society. However, this physical progress does not always correspond to psychological development. The pubertal phase introduces discontinuity into the psyche of the adolescent. This discontinuity threatens the young person’s narcissistic foundations or object relations, and there is a risk of major breakdown (for example, disorientation, depression, committing self-destructive acts or other-oriented violence).

400

Driver Behaviour and Training – Volume III

Driving involves mastering not only the technical skills of vehicle handling, learning rules, both formal and informal acquired through experience, but also by balancing capability and task demands to avoid loss of vehicle control. Insufﬁcient driving skills, lack of experience, underestimation of risks, overestimation of self ability, high exposure to difﬁcult situations and willingness to take risks are thought to be the main causes of young drivers’ accidents (Gregersen, 1996; 2005; Deery, 1999; Ferguson, 2003; Rundmo and Iversen, 2004; Clarke, Ward and Truman, 2005). From a psychological point of view, one of the main reasons for young people being over-represented in road trafﬁc crashes seems to be the tendency to involve themselves in ‘risky’ behaviour, which is age related and has ‘developmental sources’ (SWOV, 2002; Arnett, 2002). Adolescents and young riders are usually high in sensation seeking. Riding a moped or a motorcycle acts like a passport, a rite of passage to adulthood: a mark of adult status, autonomy, independence, mobility, a way of being apart from the family unit. From a psychodynamic point of view, the adolescents may be more vulnerable to risk taking due to a sense of helplessness and the need to ﬁll this void with arousal states – to avoid feelings of depression and anxiety. Risk-taking is often a way of testing out personal determination; the discovery provides a sense of being alive. This serves to maintain self-cohesiveness, develop identity and maintain one’s very existence, while at the same time operating to destroy what is reprehensible about the self or a threat to a fragile sense of coherence. Abandonment and family indifference, but also overprotection, especially on the part of the mother, can be root causes of such acts. Discredited paternal authority is also a common theme. Sometimes the causes may be found in violence, parental disharmony, hostility shown by a step-father or mother in a reconstructed family (Laufer, 2000; Le Breton, 2004; Marcelli and Braconnier, 2005). So the term ‘risk-taking’, as applied to adolescents, is used to designate a series of different behaviours (for example, drug-taking, eating disorders, reckless driving) whose common feature lies in exposing oneself to the signiﬁcant probability of being injured or dying. Such risk-taking attitudes stem from an intention. Some are adopted as a lifestyle (where risky driving is a signiﬁcant component of a deviant lifestyle); others a reaction to speciﬁc circumstances. Others can involve unconscious motivations: a tension-releasing act (acting-out), which replaces verbalising their feelings. In summary, according to some researchers reckless behaviour becomes virtually a normative characteristic of adolescent development. On the other hand, reckless behaviour may be in some cases a reﬂection of psychopathology (for example, depression, anxiety, borderline disorders). Therefore the main point of interest for professionals working with adolescents is the determination and separation of factors that are features of a developmental stage, and those factors that can be a reﬂection of psychopathology. Considering these assumptions, since 1999 the Portuguese Road Prevention (PRP) has provided a moped rider training programme (FJC programme) for adolescents throughout the country. It is an educational road safety programme providing an opportunity for 14–15-year-olds to obtain a special driver’s licence for mopeds with engine volumes of 50 cm3 and under (António, Matos and Horta, 2005). In Portugal,

An Evaluation of the Portugese Moped Rider Training Programme

401

there are thousands of teenagers who are riding mopeds without a licence or any kind of formal training. This represents in itself an example of reckless behaviour and police enforcement is not effective. The FJC Programme tries to change this reality and reduce adolescent driving risk by easing beginner riders into the trafﬁc environment; providing road safety, driver education and psychological intervention if a psychopathology is diagnosed. The general goals are to create an attitude of concern for personal security, to promote the internalisation of responsibility in trafﬁc scenarios, to inform about legal, physical and psychological aspects of the driving task, to promote a self-awareness of the adverse consequences of their behaviour and to promote a self-awareness of the social dimension of driving. A full description of the FJC programme is reported by António and colleagues (2004). The aim of the present research is to evaluate the effectiveness of the FJC programme. More speciﬁcally, we have studied the effect of participation in the FJC programme on riding behaviour among adolescents by comparing our ﬁrst year applicants (from February 1999 to February 2000) with a control group. Furthermore, we have studied the impact of the FJC programme on riding behaviour by means of the presence or absence of signs of psychopathology revealed on the ﬁrst psychological assessment undertaken in 1999/2000. It is expected that this programme will have a positive effect on the risk level of moped riders, including reducing accident involvement. Method Participants A total of 190 ﬁrst year participants (131 males, 59 females) aged 19 to 21 years (M = 20.08; SD = 0.66) and 84 young riders not submitted to the FJC programme (68 males, 12 females) aged 17 to 24 years (M = 19.9; SD = 1.73) took part in the study. When considering the ﬁrst year participants the inclusion criterion was determined by the available data from the psychological assessment undertaken in 1999/2000. Control group participants were required to have riding experience without qualiﬁcation or any kind of formal training during 1999/2000. Questionnaires Data sets were collected four years after the FJC programme was initiated and obtained via mailed questionnaires. Subjects’ behaviour was assessed by self-report using a Road Behaviour Questionnaire developed speciﬁcally for this study, and the J. Stork’s Suicidal Risk Scale (1977) adapted for the Portuguese population. These two questionnaires together with a covering letter explaining the research and a freepost return envelope were sent to applicants. Questionnaires were treated conﬁdentially. Three advanced motorcycle training courses provided by PRP (€150 each) and 20 vouchers for gasoline (€20 each) were allotted among all those ﬁrst year applicants who returned the questionnaires.

402

Driver Behaviour and Training – Volume III

Road Behaviour Questionnaire (RBQ) The RBQ is a two section questionnaire. The ﬁrst section incorporates 25 questions related to socio-demographic measures including age, gender, educational attainment, residence area, family structure, living situation, driver qualiﬁcations, usual riding position (operator or passenger), urban or rural driving experience. The second section (21 questions) sought information on exposure measures (the estimated annual mileage in driving a moped, motorcycle and car) and driving behaviour in the past four years. More speciﬁcally, variables related to riding behaviour in the past four years were subdivided into: a) riding behaviour in public area – by means of presence or absence of trafﬁc offences; b) use of rider equipment and safety accessories – by means of use frequency and kind of equipment and accessories; and c) crash experience – by means of presence or absence of trafﬁc accidents. Participants were asked to categorise crashes by means of seriousness, hospitalisation and whether they were responsible for the crash (moped, motorcycle and car). A full description of the RBQ construction is reported in António et al. (2005). J. Stork’s Suicidal Risk Scale (STORK) The STORK scale measures depression in both dimensions: ‘acting out’ and ‘acting in’. This scale enables us to assess and study likely suicidal tendency. In the Portuguese version the STORK has been reduced from 175 to 76 true–false items. There are ﬁve categories of suicidal risk: one normal and four pathological (Matos, 1991). Higher scores represent suicidal risk and consequently depressive symptoms and a self-destructive tendency. Statistical Analysis Statistical analysis was conducted using SPSS software package. Signiﬁcant differences between FJC ﬁrst participants and those from the control group were tested. Comparisons were made using parametric and non parametric techniques, namely independent t–tests and chi-square tests. In a supplementary analysis, a logistic regression model using the Enter method (computing odds ratio and 95 per cent CI) was conducted using moped/motorcycle crash involvement as the dependent variable in order to obtain the important predictors of being involved in a moped crash amongst the FJC programme participants. This multivariate modelling technique has become the accepted method for regression analysis when the dependent variable is dichotomous (Shope et al., 1996; Blows et al., 2005). Results Comparative Study Considering the effect of participation in the FJC programme on riding behaviour among adolescents by comparing our ﬁrst year applicants with a control group,

An Evaluation of the Portugese Moped Rider Training Programme

403

Table 32.1 t-test results for public riding behaviour and trafﬁc offences

Variables (Min = 1; Max = 5)

Experimental group

Control group t

p

0.81

1.77

0.080

4.20

0.94

0.17

0.868

1.01

2.23

1.18

–3.22

0.002**

3.05

0.94

3.26

1.20

–1.36

0.177

respect transit lines

2.91

1.24

3.08

1.38

–0.97

0.335

weaving in and out of trafﬁc

1.80

0.89

2.33

1.27

–3.40

0.001**

respect distance to other vehicles

3.68

0.99

3.15

1.08

3.93

0.000**

riding parallel with another vehicle

1.71

0.78

2.07

0.95

–3.27

0.001**

manoeuvres

4.18

0.92

3.46

1.33

4.45

0.000**

racing

1.75

1.00

2.29

1.22

–3.54

0.001**

respect road signals

4.43

0.70

3.99

1.08

3.40

0.001**

checking rear view mirror

4.66

0.67

4.20

0.90

4.17

0.000**

driving without a legal permit

1.51

0.94

2.63

1.45

–6.40

0.000**

Driving in public area:

M

respect priorities

4.53

give way to pedestrians

SD

M

SD

0.58

4.36

4.22

0.71

risky driving/style

1.75

speeding

* p ≤ 0.05; ** p ≤ 0.01 statistically signiﬁcant differences between the two groups were found regarding behaviour in public numbers of trafﬁc offences, use of rider equipment and safety accessories, physical and psychological aspects of riding (internal risk factors) and crash involvement. Table 32.1 shows the t-test results obtained for behaviour in public and trafﬁc offences. The two groups differed negatively and signiﬁcantly for risky driving/style (t = –3.22, p = 0.002), weaving in and out of trafﬁc (t = –3.40, p = 0.001), riding parallel with another vehicle (t = –3.27, p = 0.001), racing (t = –3.54, p = 0.001) and riding without a legal permit (t = –6.40, p = 0.000) and differ positively and

404

Driver Behaviour and Training – Volume III

signiﬁcantly for distance to other vehicles (t = 3.93, p = 0.000), manoeuvres (t = 4.45, p = 0.000), respect of road signals (t = 3.40, p = 0.001) and checking trafﬁc in the rear view mirror (t = 4.17, p = 0.000). FJC participants showed statistically signiﬁcant lower mean on those variables related to trafﬁc offences and a statistically signiﬁcant higher mean on rider safety variables. These comparisons were in the expected direction. Considering use for rider equipment and safety accessories, statistically signiﬁcant differences between the two groups were found for all variables. Table 32.2 shows the results obtained for the t-test. These differences indicated that FJC participants revealed higher means regarding the use of rider equipment and safety accessories compared with the control group. Table 32.2 t-test results for rider equipment and safety accessories

Variables

(Min = 1; Max = 5)

Experimental Group M

SD

Control Group M

SD

t

p

helmet

4.83

0.50

4.26

1.25

3.91

0.000**

helmet closed

4.60

0.94

4.06

1.34

3.23

0.002**

safe jacket

3.30

1.18

2.81

1.23

3.01

0.003**

riding boots

2.65

1.30

2.17

1.24

2.79

0.006**

lights switched on

4.42

1.04

3.65

1.51

4.16

0.000**

* p ≤ 0.05; ** p ≤ 0.01

Regarding physical and psychological features associated with the driving task, the results obtained for independent t-tests shows statistically signiﬁcant differences between the two groups for riding while tired or exhausted (t = 2.83, p = 0.005), riding under the inﬂuence of alcohol (t = –3.17, p = 0.002) and riding while euphoric/ happy (t = –2.69, p = 0.007). These results suggested a FJC participant’s tendency for overestimating their driving skills while tired or exhausted (Table 32.3). The descriptive statistics for crash experiences (moped/motorcycle and car) are presented in Table 32.4. Of the 190 FJC participants, 51.9 per cent had at least one moped/motorcycle crash between 2000 and 2003 against 30.9 per cent of the control group. Chi-square tests shows statistically signiﬁcant differences between groups regarding moped/ motorcycle crash experience (χ2 = 10.070, p = 0.002). Gender differences were not signiﬁcant (χ2 = 1.274, p = 0.259 experimental group; χ2 = 0.041, p = 0.839 control group). Considering the total number of moped/motorcycle crashes, independent-

An Evaluation of the Portugese Moped Rider Training Programme

405

Table 32.3 t-test results on physical and psychological features while riding (internal risk factors)

Variables (Min = 1; Max = 5) Do you ride while:

Experimental Group M SD

Control Group M SD

t

p

tired or exhausted

2.42

0.80

2.11

0.88

2.83

0.005**

sleepy

1.85

0.72

1.72

0.82

1.18

0.242

under inﬂuence of medication

1.41

0.78

1.39

0.65

0.24

0.807

under inﬂuence of alcohol

1.30

0.58

1.66

0.94

–3.17

0.002**

under inﬂuence of drugs

1.12

0.44

1.24

0.62

–1.56

0.122

sad or mad

2.89

0.87

2.91

1.01

–0.17

0.867

euphoric or happy

3.32

0.89

3.64

0.88

–2.69

0.007**

* p ≤ 0.05; ** p ≤ 0.01

Table 32.4

Descriptive statistics for crash experience (moped/motorcycle and car) between 2000 and 2003

Varibles N Moped/motorcycle crashes Yes No

Car crashes Yes No

Experimental Group (100)

98 (51.9) 91 (48.1)

Control Group N (100)

χ2

25 (30.9) 56 (69.1)

10.07 0.002**

65 18 (39.1) (36.5) 28 (60.9) 113 (63.5) * p ≤ 0.05; ** p ≤ 0.01

0.107

p

0.744

sample t-tests results shows no statistically signiﬁcant differences between the two groups (t = –1.844, p = 0.077). The mean score in the FJC sample was 2.7 (SD = 1.87) and in the control sample was 5.84 (SD = 8.45). Regarding car crash experience, crash rates were similar for both groups: 36.5 per cent FJC participants

406

Driver Behaviour and Training – Volume III

and 39.1 per cent control participants had at least one car crash between 2000 and 2003. Chi-square tests shows no statistically signiﬁcant differences between groups (χ2 = 0.107, p = 0.744). The two groups were quite homogeneous regarding car crash experience. Considering the total number of car crashes, there were statistically signiﬁcant differences between the two groups (t = –2.332, p = 0.032). The control sample had higher mean (M = 2.94, DP = 2.16) than the FJC sample (M = 1.68, DP = 0.99). FJC participants’ driving behaviour; presence or absence of psychopathology signs In order to study FJC participants’ driving behaviour by means of the presence or absence of psychopathology signs, two main variables were considered: suicidal risk and moped/motorcycle crash experience. Suicidal risk can be divided into ﬁve categories: (1) [0–63] normal state; (2) [64–79] doubt state; (3) [80–97] small risk; (4) [98–107] important suicidal risk; (5) [> 107] critical suicidal risk. Of the 190 FJC participants, almost all revealed scores within the normal state, as in the psychological assessment undertaken in 1999/2000, and again in 2004. Furthermore, the mean scores in 1999/2000 and 2004 are relatively lower when compared with other studies (Matos, 1991; Almeida, 2004). Regarding moped/motorcycle crashes, of the 190 FJC participants, 91 subjects (48.1 per cent) had no crashes and 98 subjects (51.9 per cent) revealed crashes between 2000 and 2003. Of this last group, 27 individuals had only one and 67 individuals had several accidents (≥ 2) in the same period. According to t-tests analyses there was no statistically signiﬁcant difference between 1999/2000 (t = –0.276, p = 0.783) and in 2004 (t = –0.092, p = 0.927). Surprisingly, the crash involved FJC participants had lower suicidal risk scores than those who had not been involved in a crash in 1999/2000 and 2004. These ﬁndings were not in the expected direction considering previous studies. The results from the logistic regression model using moped/motorcycle crash experience as the dependent variable are presented in Table 32.5. Main effects are found for family background, moped/motorcycle riding experience and car crash experience. Sons of single-mother families (OR 5.53, 95 per cent CI, 1.39–21.93), sons of reconstructed families (OR 5.07, 95 per cent CI, 0.98–26.24) and sons of parent-mother (traditional) families (OR 2.55, 95 per cent CI, 1.02–6.429) show a higher increase in the probability of being involved in moped/motorcycle crash as compared with those from other family backgrounds. The odds ratio for sons of single-father families does not reach statistical signiﬁcance. The sons of singlemother families and reconstructed families are at greater risk of moped/motorcycle injuries. FJC participants who drove more than 10 000 kilometres are at greater risk of having moped/motorcycle crash (OR 4.35, 95 per cent CI, 1.67–11.31) compared with FJC participants who drove between 5000 and 10 000 kilometres (OR 2.35, 95 per cent CI, 0.87–6.39) and compared with those who drove at least 5000 kilometres (reference group). The probability of having a moped/motorcycle crash increases with mileage. FJC crash involved young riders increases the odds (OR 2.46, 95 per cent CI, 1.17–5–17) of having moped/motorcycle crash.

An Evaluation of the Portugese Moped Rider Training Programme

Table 32.5

407

Odds ratios (95 per cent CIs) for moped/motorcycle crash experience as dependent variable

Variables

OR (95% IC) FJC participants (n = 190)

Sex Male Female Educational attainment Upper-secondary education Higher education Basic education Occupation Student Employed Unemployed Family background Other Single-mother Traditional Reconstructed Single-father Mileage in 50cc (km) Under 5000 5000–10 000 More than 10 000 No mileage Trafﬁc offences 50cc/>50cc No Yes Car crash involvement No Yes Age began riding 50cc

1.00 0.95

(0.41; 2.17)

1.00 0.94 0.37

(0.41; 2.14) (0.09; 1.40)

1.00 0.55 0.00

(0.18; 1.68) (0.000)

1.00 5.53 (1.39; 21.93)* 2.55 (1.02; 6.42)* 5.07 (0.98; 26.24)* 4.9E+09 (0.00; .) 1.00 2.35 (0.87; 6.39) 4.35 (1.67; 11.31)** 1.1E+10 (0.00; .) 1.00 0.51

(0.21; 1.26)

1.00 2.46

(1.17; 5.17)*

0.99

Suicide risk 1999/2000

1.00

Observed cases Missing

166 24 * (p ≤ 0.05); ** p ≤ 0.01

(0.85; 1.17) (0.98; 1.02) (87.4%) (12.6%)

408

Driver Behaviour and Training – Volume III

For the FJC participants (G2 sample, 36.2 per cent) with more than two moped/ motorcycle accidents (M = 3.39, SD = 1.82, min 2 max 10), descriptive results show that 80.3 per cent had received higher education (university attendance), 34.3 per cent had upper-secondary education and only 7.5 per cent had basic education. Regarding family background, young riders from single-mother families were signiﬁcantly over represented (G2 16.4 per cent) compared with the FJC sample showing 50 per cent of all cases. For road behaviour, 97 per cent began riding a moped prior to 15 years old and in 92.5 per cent of cases the moped was a gift from parents and/or relatives. Until 2004, 58.9 per cent rode more than 10 000 kilometres and revealed a much higher proportion of trafﬁc offences than the FJC sample, and 55.2 per cent had been hospitalised. With regards to car driving experience, 50.8 per cent rode more than 25 000 kilometres. As before, the majority of G2 riders revealed suicidal risk within the normal state. Using the cut-off values method (Hill and Hill, 2000), 43.3 per cent showed risk values below inferior cut-off value (inferior value 24; superior value 34). Discussion This research aimed to evaluate the effectiveness of the FJC programme. Considering the results from the comparative study, there are positive effects of the FJC moped rider training programme. The results indicate that FJC participants showed higher rule following riding/driving behaviour, self-awareness of physical and psychological features associated with the riding task (internal risk factors) and were less likely to have trafﬁc offences when compared with young riders in the control group. Similar results were found by Goldenbeld et al. (2004). However, this study shows that the differences between trainees and the control group reduces with time and experience. With regards to the use of rider equipment and safety accessories, FJC participants revealed higher levels of responsibility towards vehicle knowledge, rider equipment and safety accessories compared with the control group. These ﬁndings are in line with the main goals of the FJC programme. With respect to moped/motorcycle crash experience, contrary to expectations, more than half of FJC participants (51.9 per cent) revealed they had been involved in a crash experience during the four years after the programme. As for the control group during the same period, that proportion was only 30.9 per cent. Regarding car crash involvement, rates were similar for both groups. One possible explanation could be the differential exposure in moped/motorcycle mileage. In car driving experiences, the two groups showed similar mileage. Accidents are, to some extent, random events, and several researchers have pointed out that the intensity and amount of exposure are major determinants of involvement in multiple crashes (Murray, 1998; Laapotti et al., 2001; Hasselberg and Laﬂamme, 2005). Within the FJC sample, the moped/motorcycle crash rate averages are a matter of concern considering the FCJ programme goals. Regarding the impact of the FJC programme on the presence or absence of psychopathology signs, the results showed no relation between moped/motorcycle crash involvement and suicidal risk (depressive symptoms). Surprisingly, and

An Evaluation of the Portugese Moped Rider Training Programme

409

contrary to two previous studies (Matos, 1991; Almeida, 2004), the FJC participants showed lower suicidal risk levels compared with those who had not been involved in a crash. A signiﬁcant proportion of those who had more than two moped/motorcycle accidents revealed suicidal risk values below 24 points. On the other hand, a minority of the FJC participants with pathological suicidal risk values did not have a higher crash rate. In our study, we found that the J. Stork’s Suicidal Risk Scale was not a predictive measure of future accident involvement in 1999/2000 and was not able to differentiate FJC participants on moped/motorcycle crashes in 2004. The FJC participants’ psychopathology was not revealed using this scale. One possible explanation could be a change in the actual psychopathology conﬁguration of young riders. The STORK scale is an old scale to measure depression. The primary and very legitimate motivation for FJC students is to obtain their licence. As a consequence, when reading STORK items and identifying possible signs of vulnerability, they may ‘fake good’ or deny their suffering in line with the notion of defence mechanisms that serve to deny inner reality and hide a self-destructive tendency. FJC participants with more than two moped/motorcycle crashes (G2 sample) showed several examples of risk-taking behaviours in the Road Behaviour Questionnaire (for example, age, early experience of the road environment, trafﬁc offences, car crash involvement and hospitalisations). This provides some evidence of the emergence of a borderline disorder being associated with moped/motorcycle crash involvement. Low results on the STORK, high rates of moped/motorcycle accidents and a single-mother family background are good indicators. Furthermore, our results for the FJC participants echo a surprising ﬁnding: for the majority of the participants, the moped was a gift from adults (parents or other relatives). At the age of 18 they were driving a car and a moped and showed high levels of educational attainment (university and upper-secondary education), which are indicators of a high standard socio-economic status. On the contrary, other studies show that young drivers with basic levels of education are considerably more involved in trafﬁc accidents (for example, Matos, 1991; Murray, 1998; Vaez and Laﬂamme, 2005; Hasselberg and Laﬂamme, 2005). Nowadays, we live in a consumer society. It is not just young people who are demanding material goods, but adults too. However, today’s adolescent needs one thing that adults seem to have the least surplus of – time. It takes time to listen and relate to an adolescent. Some recent studies also indicate that not enough time spent with their parents is one of teens’ top problems and also that adolescents whose parents are concerned about the lives of their children have signiﬁcantly lower rates of problem/risk behaviours (Shope et al., 1996; Murray, 1998; APA, 2002). Together our ﬁndings suggest that abandonment, family indifference and discredited paternal authority are potential sources of risk for being involved in a moped/motorcycle crash. Our results show that theoretical and practical knowledge about riding a moped is important and useful. Psychomotor and physiological aspects are essential to safe driving and must be a part of the training process. However, according to our study, it does not necessarily prevent crash involvement. It is also necessary to address the psychological and emotional attitudes of young people towards road safety and their level of responsibility. Over the past few years, research has lead to the recognition that young drivers’ goals and attitudes increases or decreases the risks of driving, and

410

Driver Behaviour and Training – Volume III

there is a close connection between unconscious motives, attitudes and personality development (Mayhew and Simpson, 2002; Broughton, 2005; Gregersen, 2005). Accordingly, it can be presumed that some of these signiﬁcant differences found may be associated with the adolescent’s inner world. Someone driving under the inﬂuence of alcohol or other drugs is presumably aware, to a certain degree, of the associated risks. However, given the nature of the different mental states such as depression anxiety or euphoria, the individual may not be fully aware of the implications and be at risk of being involved in a crash. Concluding remarks In summary, road prevention should help adolescents to develop their ‘thinking process’, to increase awareness of their at-risk behaviour, to temper risk taking tendencies and vulnerability, rather than only providing vehicle handling skills. Riding for greater excitement, speed and the youthful sense of invincibility needs to be tempered by adults. According to a psychodynamic approach, these behaviours can be understood as suicidal tendencies where young drivers often deny risky consequences for narcissistic reasons. The implications for moped riding training and effective road safety prevention, are, we believe to further our understanding of adolescent development, risk taking, and psychopathology to improve young riders’ road behaviour and, therefore, reduce death and injuries. Our investigation was the ﬁrst evaluation of the FJC programme and some limitations have been noted. Considering future psychological interventions new psychopathology measures should be investigated in order to identify which young drivers are at high risk of accident involvement. References Almeida, A. (2004). ‘Avaliação do risco suicidário em adolescentes candidatos à licença especial de condução de ciclomotores: estudo das diferenças masculino/ feminino’ (Suicidal risk assessment in adolescents candidates to obtain a special drivers licence for mopeds. Gender diferences). Dissertação de Mestrado. Lisboa: Faculdade de Psicologia e de Ciências da Educação (328 p.). American Psychological Association (APA, 2002). ‘Developing adolescents: a reference for professionals.’ Washington, DC: American Psychological Association. Retrieved June 17, 2004, from www.apa.org/pi/pii/develop.pdf António, P., Matos, M., and Horta, M. (2004). ‘Making roads safer: a model for promoting road safety among adolescents.’ Retrieved February 22, 2005, from http://www.psychology.nottingham.ac.uk/IAAPdiv13/ [Index of/IAAOdiv13/ ICTTP2004papers2/children and Adolescents]. António, P., Matos, M., and Horta, M. (2005). ‘Driving at ﬁfteen: assessment of moped rider training among teens.’ In L. Dorn (ed.), Driver Behaviour and Training, vol. II (chapter 21, 252–60). Aldershot, UK: Ashgate. Arnett, J. (2002). ‘Developmental sources of crash risk in young drivers.’ Injury Prevention, 8 (Suppl II), ii17–ii23. Retrieved August 10, 2004, from http://

An Evaluation of the Portugese Moped Rider Training Programme

411

ip.bmjjournals.com/cgi/reprint/8/suppl_2/ii17. Blows, S., Ameratunga, S., Ivers, R., Lo, S. and Norton, R. (2005). ‘Risky driving habits and motor vehicle driver injury.’ Accident Analysis and Prevention, 37, 619–24. Broughton, P. (2005). ‘Designing powered two wheeler training to match rider goals.’ In L. Dorn (ed.), Driver Behaviour and Training, vol. II (chapter 19, 233– 42). Aldershot, UK: Ashgate. Clarke, D., Ward, P. and Truman, W. (2005). ‘Voluntary risk taking and skill deﬁcits in young driver accidents in the UK.’ Accident Analysis and Prevention, 37, 523– 29. Deery, H. (1999). ‘Hazard and risk perception among young drivers.’ Journal of Safety Research, 30(4), 225–36. DG TREN Road Safety Unit (2004). European Comission. DG Energy and Transport. TREN Informations. Retrieved August 23, 2004, from http://europa.eu.int/comm/ transport/care/statistics/most_recent/detailed_breakdown/index_en.htm.s DGV, 2005. Sinistralidade rodoviária 2004. Elementos estatísticos (2004 road trafﬁc accidents. Statistical elements) Lisboa: Observatório de Segurança Rodoviária, Ministério da Administração Interna. Ferguson, S. (2003). ‘Other high-risk factors for young drivers – how graduated licensing does, doesn’t, or could address them.’ Journal of Safety Research, 34, 71–7. Goldenbeld, C., Twisk, D. and Craen, S. (2004). ‘Short and long term effects of moped rider training: a ﬁeld experiment.’ Transportation Research Part F, 7, 1–16. Retrieved May 10, 2004, from http://www.sciencedirect.com/science?_ ob=ArticleURLand_udi=B6VN8-49V3JWJ-2...pdf. Gregersen, N. (1996). ‘Young drivers’ overestimation of their own skill – an experiment on the relation between training strategy and skill.’ Accident Analysis and Prevention, 28(2), 243–50. Gregersen, N. (2005). ‘Driver education – a difﬁcult but possible safety measure.’ In L. Dorn (ed.), Driver Behaviour and Training, vol. II (chapter 12, 144–53). Aldershot, UK: Ashgate. Hasselberg, M. and Laﬂamme, L. (2005). ‘The social patterning of injury repetitions among young car drivers in Sweden.’ Accident Analysis and Prevention, 37, 163–68. Hill, M. and Hill, A. (2000). ‘Investigação por questionário’ (Research by questionnaire). Lisboa: Edições Sílabo. Kopjar, B. (1999). ‘Moped injuries among adolescents: a signiﬁcant forgotton problem?’ Accident Analysis and Prevention, 31, 473–78. Laapotti, S., Keskien, E., Hatakka, M. and Katila, A. (2001). ‘Novice drivers’ accidents and violations – a failure on higher or lower hierarchical levels of driving behaviour.’ Accident Analysis and Prevention, 33, 759–69. Laufer, M. (2000). ‘O adolescente suicida’ (Suicidal adolescent). Lisboa: Climpesi Editores. Le Breton, D. (2004). ‘The anthropology of adolescent risk-taking behaviours.’ Body and Society, 10(1), 1–15. Retrieved October 22, 2004, from www.sagepublications. com. Marcelli, D. and Braconnier, A. (2005). ‘Adolescência e psicopatologia’ (Adolescence

412

Driver Behaviour and Training – Volume III

and psychopathology). Lisboa: Climepsi Editores. Matos, M. (1991). ‘Factores de risco psicológico em jovens condutores de motorizada e sua inﬂuência relativa na ocorrência de acidentes’ (Psychological risk factors among motorcycle riders and their inﬂuence on accident involvement). Dissertação de Doutoramento. Lisboa: Faculdade de Psicologia e de Ciências da Educação (581 p.). Mayhew, D. and Simpson, H. (2002). ‘The safety value of driver education and training.’ Injury Prevention, 8 (Suppl II), ii3–ii8. Retrieved August 10, 2004, from http://ip.bmjjournals.com/cgi/reprint/8/suppl_2/ii3. McKnight, A.J. and McKnight, A.S. (2003). ‘Young novice drivers: careless or clueless.’ Accident Analysis and Prevention, 35, 921–25. Murray, A. (1998). ‘The home and school background of young drivers involved in trafﬁc accidents.’ Accident Analysis and Prevention, 30(2), 169–82. Reto, L. and Sá, J. (2003). ‘Porque nos matamos na estrada e como o evitar. Um estudo sobre o comportamento dos condutores’ (Why do we kill ourselves on roads and how to avoid it. A study about drivers behaviour). Lisboa: Editorial Notícias. Rundmo, T. and Iversen, H. (2004). ‘Risk perception and driving behaviour among adolescents in two Norwegian counties before and after a trafﬁc safety campaign.’ Safety Science, 42, 1–21. Retrieved July 27, 2004, from http://www. sciencedirect.com/science?_ob=MImgand_imagekey=B6VF9-4846N6X-2Vand_cdi=6005and_orig=browseand_coverDate=01%2F31%2F2004and_sk= 999579998andview=candwchp=dGLbVtzzSkWAand_acct=C000050221and_ version=1and_userid=10andmd5=0e40407cd55626b246e1a1cc147005acandie= f.pdf. Shope, J., Waller, F. and Lang, S. (1996). ‘Alcohol-related predictors of adolescent driving: gender differences in crashes and offenses.’ Accident Analysis and Prevention, 28(6), 755–64. Stork, J. (1977). ‘Échelle d’évaluation de risque suicidaire.’ Psychiatrie de l’Enfant, XX(2), 493–517. SWOV (2001a). ‘Promotion of mobility and safety of vulnerable road users. Final report of the European research project PROMISING.’ Wittink, R. (ed.). Leidschendam: SWOV (97 pp.). Retrieved August 10, 2004, from http://www. swov.nl/rapport/d-2001-03.pdf. SWOV (2001b). ‘Integration of needs of moped and motorcycle riders into safety measures. Review and statistical analysis in the framework of the European research project PROMISING, Workpackage 3.’ Noordzij, P., Forke, E., Brendicke, R. and Chinn, B. (eds.). Leidschendam: SWOV (212 pp.). Retrieved August 10, 2004, from http://www.swov.nl/rapport/d-2001-05.pdf. SWOV (2002). ‘“Hardcore” problem groups among adolescents. Their magnitude and nature and the implications for road safety policies.’ Wurst, T. (ed.). Leidschendam: SWOV (33 pp.). Retrieved August 10, 2004, from http://www. swov.nl/rapport/r-2002-25.pdf. Ulleberg, P. and Rundmo, T. (2002). ‘Risk-taking attitudes among young drivers: the psychometric qualities and dimensionality of an instrument to measure young

An Evaluation of the Portugese Moped Rider Training Programme

413

drivers’ risk-taking attitudes.’ Scandinavian Journal of Psychology, 43, 227–37. Vaez, M. and Laﬂamme, L. (2005). ‘Impaired driving and motor vehicle crashes among Swedish youth: an investigation into drivers’ sociodemographic characteristics.’ Accident Analysis and Prevention, 37, 605–11. Williams, A. (2003). ‘Teenage drivers: patterns of risk.’ Journal of Safety Research, 34, 5–15. WHO (2004a). ‘World report on road trafﬁc injury prevention.’ Geneva: World Health Organization. Retrieved July 29, 2004, from http://www.institutoarade. org/arquivo/oms_full_report.pdf. WHO (2004b). ‘Preventing road trafﬁc injury: a public health perspective for Europe.’ Copenhagen: WHO Regional Ofﬁce for Europe. Retrieved July 29, 2004, from http://www.institutoarade.org/arquivo/oms_europa_report.pdf.

This page intentionally left blank

Chapter 33

Flow, Task Capability and Powered TwoWheeler (PTW) Rider Training Paul Broughton Napier University, UK Introduction According to the Department for Transport, motorcyclists are at a much greater risk of death or serious injury than other road users. In 2004, the relative risk of a motorcycle rider being killed or seriously injured per kilometre travelled was almost 46 times higher than for car drivers (DfT, 2006). The Government, using a 1994 to 1998 baseline, aims to reduce KSI accidents by 40 per cent by 2010 (DfT, 2000). The ﬁrst (DfT, 2004) and second review (DfT, 2007), reported that good progress was being made, but not the case of powered two-wheelers (PTWs) where there was an increase in KSI accidents. The statistical evidence indicates that there is a problem with PTW safety; therefore interventions that reduce the KSI rate would be prudent. This paper reviews the author’s research into the goals of PTW riders and discusses some of the implications for rider training. Rider Motivation The PTW safety statistics shows that the government’s 2010 target is not going to be met unless effective interventions are put into place; to aid in these interventions being effective they must address rider motivations. Therefore the question of why PTWs are chosen as a form of transport must be ﬁrst understood. Enjoyment Types What is the motivation for PTW riders? PTW use has been associated with many reasons, for example convenience (ACEM, 2000) and sensation seeking (Sexton, Hamilton, Baughan, Stradling and Broughton, 2006). Broughton and Stradling (2005) found that riding was primarily expressive with enjoyment being a major motivation for participation in the activity. This enjoyment can be found as ‘rush based enjoyment’ or ‘challenge based enjoyment’ (Broughton, in press). ‘Rush based enjoyment’ is adrenaline related and generally experienced by riding fast in a straight line in areas with good visibility. This type of enjoyment is often found when the riding task is not too difﬁcult. ‘Challenge based enjoyment’ is less related to speed and more with bends, with enjoyment being found by the rider matching their skill set to the challenge of the ride (Broughton, 2006a). This type of

416

Driver Behaviour and Training – Volume III

enjoyment is more likely to be found when the riding task is more difﬁcult; however neither enjoyment types are prominent at high levels of task difﬁculty, that is, when the challenge is too great then riding ceases to be pleasurable The type of machine that a rider uses is related to the types of enjoyment sought; it may be expected that those seeking ‘rush based enjoyment’ would prefer high performance machines. However as bike performance increases, one might expect that ‘rush based enjoyment’ decreases. This dichotomy may be due to experienced riders being more likely to ride the higher performance machines and as these riders generally have a better skill set they are less likely to gain enjoyment just from ‘rush’ and more likely to gain enjoyment from ‘challenge’. Younger riders’ tendency towards ‘rush based enjoyment’ further supports this. The two enjoyment types can be compared with the difference between bungee jumping and rock climbing (Broughton, 2006b). Both of these activities can be enjoyable, yet enjoyment is found in completely different ways. Bungee jumping does not require much skill in throwing oneself off a high place with a piece of elastic saving one from certain death, yet it is very enjoyable for those who are seeking an adrenaline buzz. Rock climbing, on the other hand, is a sport where a climber pits one’s skills against the challenge presented by the rock face with enjoyment predominantly found in the skill/challenge match. Enjoyment can be found in either, or both, types of enjoyment. The ‘challenge based enjoyment’ is akin to what is described by Csikszentmihalyi (1990) as the ‘ﬂow state’. Enjoyment and ﬂow Csikszentmihalyi’s theory of ﬂow (Csikszentmihalyi, 1990) suggests that when a person has a ‘high skill level’ and is faced with a ‘high challenge’ then this person can enter into a state called ‘ﬂow’. Csikszentmihalyi describes this state as ‘The Holistic Sensation that people feel when they act with total involvement’ (Csikszentmihalyi, 2000, 36), as being ‘an almost automatic, effortless, yet highly focused state of consciousness’ with ‘no sense of time or worry of failure’ (Csikszentmihalyi and Csikszentmihalyi, 1988). Riders often cite ‘freedom’ as a reason for riding; this description of ﬂow reﬂects this feeling of freedom (Broughton, 2005). While in the state of ﬂow, concentration is so intense that there is no attention left over to think about anything irrelevant or to worry about problems. Flow is an almost automatic, effortless, yet highly focused state of consciousness. The theory of ﬂow describes four states (Table 33.1): apathy, boredom, anxiety and ﬂow. The ﬂow state is entered into when one’s skills are matched by the challenge faced. When both the skill level and the challenge is low then an apathetic state is entered into, however if the level of skill is higher than the level of challenge then boredom is the result; conversely when the skill level does not meet the challenge then anxiety is experienced. For the ﬂow state to be entered into not only must an individual’s skills be matched to the challenge, but these challenges and the skills needed to confront them must exceed the normal levels of daily occurrence (Csikszentmihalyi and Csikszentmihalyi, 1988). As well as the skill/challenge match, clear goals and instant feedback are also required conditions to enable the ﬂow state (Csikszentmihalyi, 2000).

Flow, Task Capability and Powered Two-Wheeler

417

Table 33.1 The four states of ﬂow

Challenge Low High

Low Apathy Anxiety

Skill High Boredom Flow

A modiﬁed model of ﬂow is shown in Figure 33.1, with this linear model showing the states a person goes through as task difﬁculty changes. When task difﬁculty is low, boredom results, as task difﬁculty increases then the state of ‘challenge based enjoyment’ is passed through and on to anxiety. The change of states is not instantaneously going directly from boredom to enjoyment or enjoyment to anxiety; rather, the boundaries are fuzzy. At the peak of enjoyment, just before the anxiety state begins, is the ﬂow state. Once the ﬂow state has been exited then enjoyment drops off rapidly while anxiety, experienced as risk, rapidly increases.

Flow

Boredom

Enjoyment

Anxiety

Task Difficulty Figure 33.1 Linear Model of Task Difﬁculty and Flow

There is obviously a neurocognitive process that takes place when a person enters into a state of ﬂow, Dietrich (2004) comments that ‘A necessary prerequisite to the experience of ﬂow is a state of transient hypofrontality that enables the temporary suppression of the analytical and meta-conscious capacities of the explicit system.’ (Dietrich, 2004, 746). Therefore for ﬂow to exist the brain must be running on the automatic implicit system (Broughton, 2006a). While in the ﬂow state, the person will be in a fully automatic mode where there is no processing power left over to carry out other activities, such as day dreaming or analysing the task that is being undertaken. This is in agreement with the description of the ﬂow state (Csikszentmihalyi and Csikszentmihalyi, 1988) as being ‘an almost automatic, effortless, yet highly focused state of consciousness’ where the ‘task is performed, without strain or effort, to the best of the person’s ability’ and that there is also ‘no sense of time or worry of failure’. Flow is tied in with the automatic, implicit brain functions. Therefore a rider who is seeking a ‘challenge based enjoyment’ experience will be attempting to ride in a ﬂow, or near ﬂow, state. This means that the riding process will be being

Driver Behaviour and Training – Volume III

418

carried out automatically, without any interference from the conscious elements of the brain. The way enjoyment is found while riding is a complex combination of enjoyment methods with the age and experience of riders tending to affect how enjoyment is achieved. More experienced riders tend to ﬁnd enjoyment in situations where task difﬁculty is higher compared with less experienced riders. Enjoyment types are therefore also associated with task difﬁculty. Task difﬁculty Fuller (2005) in his appraisal of car driver behaviour suggests that drivers subconsciously attempt to keep task difﬁculty constant. In Fuller’s model task difﬁculty is the ‘dynamic interface between the demands of the driving task and the capability of the driver.’ As seen from Figure 33.2, while task demand is lower than their capability then the driver is in control. However when task demand exceeds capability then loss of control results, culminating in either ‘a lucky escape’ or a collision. Although riding is more complex than driving in terms of skills required (Mannering and Grodsky, 1995), similar issues of capability and task demand apply. For riders seeking ‘Challenge Based Enjoyment’ they would be riding close to where capability equals demand (C≈D).

LUCKY ESCAPE

compensatory action by others

CAPABILITY (C)

COLLISION

LOSS OF CONTROL

CD

TASK DEMANDS (D)

CONTROL

Figure 33.2 Outcomes of the dynamic interface between task demand and capability Source: Fuller, 2005, 464.

Flow, Task Capability and Powered Two-Wheeler

419

As task difﬁculty increases, or capability decreases, it would be expected that there would be a degradation of performance rather than a sudden loss of control (Wickens and Hollands, 2000), and lower priority tasks, such as checking mirrors, may be neglected. As task difﬁculty further exceeds capability then more important tasks may not be carried out, such as proper forward observation. Speed is a major way of controlling task difﬁculty, with an increase in speed making the riding task more difﬁcult, therefore it is not surprising that riders tend to ride slower when task difﬁculty is high. Where task demand exceeds capability then an out of control situation exists, therefore it may be possible to explain PTW crashes in terms of task difﬁculty. Task difﬁculty and PTW crashes The goal of most riders is to enjoy the riding experience with the enjoyment being found in a combination of two ways: ‘rush based enjoyment’ and ‘challenge based enjoyment’. Riders who are seeking one of these enjoyment types will be riding in a corresponding manner, therefore the types of PTW crashes that occur are likely to reﬂect the type of enjoyment being sought. Two common types of PTW crashes are loss of control on bends and crashes while overtaking. Loss of control by the rider on bends is a major cause of KSI crashes. Clarke, Ward, Bartle and Truman (2004) reported that loss of control accidents on bends accounted for around 12 per cent of all accidents, 7 per cent on left-hand bends and 5 per cent on right-hand bends. In a similar study that looked at Scottish PTW accidents between 1992 and 2002 Sexton, Fletcher and Hamilton (2004), reported that 9 per cent were ‘going ahead on at right hand bend’ and 11 per cent ‘going ahead on at left hand bend’. Bends are a major factor for ‘challenge based enjoyment’, and as this type of enjoyment is ﬂow based, riders will be attempting to match their skill level with the demands presented by the environment (C ≈ D). If the rider makes even a small mistake in assessing either the skill level required or the demand level of the task, then loss of control will result. A similar loss of control may occur where an event increases task demand, such as a reduction in traction between tyres and the road surface, or where a rider’s capability reduces, for example, through being distracted. Another common PTW manoeuvre being carried out during an accident is overtaking. Sexton, Fletcher and Hamilton (2004) reported 9 per cent of PTWs were carrying out this manoeuvre just prior to the accident, with Clarke, Ward, Bartle and Truman (2004) reporting this ﬁgure as 14 per cent in their research. Despite ‘rush based enjoyment’ not being skill based, task demand must still exceed rider capability for loss of control to result. Speed is an enhancer of task difﬁculty, and also a major element of this type of enjoyment. As PTWs can generally accelerate signiﬁcantly quicker than cars (for example, a Ford Focus ST car can accelerate from 0 to 60 mph in 6.8 seconds, while a BMW F800s PTW can do 0–60mph in 3.5 seconds), task difﬁculty can rapidly increase to a point where task demand exceeds rider capability. The resulting loss of control can occur before the rider is aware of what is happening or has time to reduce task demand. Less experienced riders are more likely to be seeking ‘rush based enjoyment’, and these riders may be less able

420

Driver Behaviour and Training – Volume III

to quickly interpret the signs that they are heading towards a situation where demand exceeds capability (C < D) and therefore are unlikely to take corrective action before they are out of control. These two examples show that a sudden change in capability or demand can place a rider into an out of control situation, which may result in a crash. While this is not desirable in a driving situation, it is even more of an issue for PTW riders as the PTW affords little protection (ACEM, 2004; DfT, 2005), making any such lapses more likely to result in serious injury or death. Task demand, task capability and training As an out of control situation is the result of a rider’s capability being outstripped by task demand, then can this be addressed by increasing a rider’s capability with training? Training is an area that is being used to try and reduce the number of rider casualties, but there is a need to underpin skills training by getting ‘inside the mind of a motorcyclist’ (Kipling, 2005). Skills training alone can increases the risk of the rider being involved in an accident due to an over estimation of skills (Rutter and Quine, 1996). The frequency of training should also be considered, as motorcycle training may only have short-term effects. Goldenbeld, Twisk and de Craen (2004) found that the effects of PTW training were not detectable after a period of 11 months. For those who like to feel ‘rush based enjoyment’, training could provide a rider with the aptitude to prevent a misjudgement of speed that could cause task demand to exceed capability (C < D). Younger and less experienced riders are more likely to gain enjoyment from ‘rush based enjoyment’, as well as being over-represented in the KSI accident ﬁgures, therefore speed and task demand awareness training for younger riders may be an effective way to target accident reduction strategies. Speed is a severity enhancer of accidents (Aarts and Van Shagen, 2006) as the resultant energy (Ek) of an impact is related to the mass of the object (M) and velocity (V) squared. Ek = ½(MV2). ‘rush based enjoyment’ is positively correlated to speed, whereas ‘challenge based enjoyment’ is not related to speed but to bends. As it is the less experienced riders who tend to look for ‘rush based enjoyment’, then training aimed at raising the skills of these riders may alter the type of enjoyment they seek to ‘challenge based’. This change may slow riders down, and may be an effective intervention. However for the more experienced riders who enjoy a ‘challenge based enjoyment’ experience, an increase in capability may be counter productive. Increasing the skills, or capability, of ‘challenge based enjoyment’ type riders could mean that riders seek a higher task difﬁculty to obtain the desired state of ﬂow and hence may not ride in a safer manner. As skills training may increase rider risk, these interventions should seek ways to inoculate against this phenomenon (Mannering and Grodsky, 1995; Ormston, Dudleston, Pearson and Stradling, 2003; RoSPA, 2006). This need for ‘inoculation’ suggests that some form of behavioural training could be beneﬁcial, addressing rider attitudes in an attempt to make PTW users ride in a safer manner (Cunliffe and Stradling, 2006). Training needs to aim

Flow, Task Capability and Powered Two-Wheeler

421

at getting riders to ride with a greater margin of safety, that is ensuring that C > D, not C ≈ D, therefore reducing the likelihood of a rider ﬁnding himself in a C < D situation when task demand changes. When a rider who is in a ﬂow state is operating on automatic, the skills being used must also be automatic and have been learnt by repeated practice. If any of these skills are inappropriate they should be replaced by correct skills, and the new skills will have to be well practiced too. Training that intends to alter the skill set of a rider, rather than just build upon existing skills, needs to take into account the automaticity that accompanies ﬂow riding (Broughton, 2006b). Ideally for interventions to be most productive they must be targeted at the correct audience, for example skills training for those who are more inclined to ‘rush based enjoyment’, and behavioural training for riders inclined to ‘challenge based enjoyment’. With enjoyment proﬁles for riders being complex, and liable to be different for individual riders, proﬁling prior to any non mass media interventions could be beneﬁcial. Proﬁling can allow interventions to be targeted to the needs and goals of the individual rider. The enjoyment proﬁle differences between riders also need to be considered when mass media interventions are being designed as one size does not ﬁt all! Conclusion There is evidence that riders may be seeking a ﬂow type experience by matching their riding skills to the challenge presented by the riding environment. This is equivalent to task demand being closely matched to capability (Fuller, 2005), which can be expressed as C ≈ D. When a rider is in this state then there is a very small safety margin. If task demand rises, or capability drops, (C < D) then a rider would be out of control which may result in a collision. Skills training may increase rider capability; therefore a higher task demand would be required to achieve a C ≈ D state. As speed is a major enhancer of task demand then it would be logical that skills training may entice riders who are seeking a ﬂow experience to ride faster. Because of this, any skills based intervention should incorporate methods to help prevent this from happening (Mannering and Grodsky, 1995; Ormston, Dudleston, Pearson and Stradling, 2003). As enjoyment is a major riding goal, then any safety intervention must respect this. If an intervention does not acknowledge, or attempts to remove, this goal, then it is likely that riders will reject the intervention and it will be ineffective (Broughton, 2006a). Conversely, as the majority of riders do not ride because of the risk, most would accept interventions that reduced the risk provided there was not an erosion of enjoyment. Interventions should also be targeted at speciﬁc rider enjoyment types, therefore rider proﬁling techniques need to be developed. By targeting interventions, and preserving rider goals, interventions can be developed that will aid in reducing PTW KSI and help to meet Government road safety targets.

422

Driver Behaviour and Training – Volume III

References Aarts, L. and Van Shagen, I. (2006). ‘Driving speed and the risk of road crashes: a review.’ Accident Analysis and Prevention, 38, 215–24. ACEM (2000). Smart Wheels for City Streets. Brussels: ACEM. ACEM (2004). Maids: In-depth Investigations of Accidents Involving Powered Two Wheelers. Brussels: ACEM. Broughton, P.S. (2005). ‘Designing PTW training to match rider goals.’ In Dorn, L. (ed.), Driver Behaviour and Training. Aldershot: Ashgate Publishing. Broughton, P.S. (2006a). ‘The Implication of the Flow State for PTW Training.’ In Behavioural Research in Road Safety 2006, Sixteenth Seminar. London: DfT. Broughton, P.S. (2006b, 27th Feb–1st March 2006). ‘Inﬂuencing Power Two Wheeler Behaviour.’ Paper presented at the 71st RoSPA Road Safety Congress: The Road to Safer Behaviour, Blackpool Hilton Hotel, Blackpool. Broughton, P.S. (in press). Risk and Enjoyment in Powered Two Wheeler Use. Edinburgh: Transport Research Institute, Napier University. Broughton, P.S. and Stradling, S.G. (2005). ‘Why ride powered two wheelers?’ In Behavioural Research in Road Safety 2005, Fifteenth Seminar. London: DfT. Clarke, D.D., Ward, P., Bartle, C. and Truman, W. (2004). In-depth Study of Motorcycle Accidents (No. 54). London: DfT. Csikszentmihalyi, M. and Csikszentmihalyi, I.S. (1988). Optimal Experience: Psychological Studies of Flow in Consciousness. Cambridge: Cambridge University Press. Csikszentmihalyi, M. (1990). Flow: the Psychology of Optimal Experience. New York: Harper and Row. Csikszentmihalyi, M. (2000). Beyond Boredom and Anxiety: Experiencing Flow in Work and Play. San Francisco: Jossey-Bass. Cunliffe, N. and Stradling, SG. (2006, 27th Feb–1st March 2006). Riders Improving and Developing Their Experience: RIDE Programme. Paper presented at the 71st RoSPA Road Safety Congress: The Road to Safer Behaviour, Blackpool Hilton Hotel, Blackpool. DfT (2000). Tomorrow’s Roads: Safer for Everyone. London: DfT. DfT (2004). Tomorrow’s Roads: Safer for Everyone – The First Three Year Review. London: DfT. DfT (2005). The Government’s Motorcycling Strategy. London: DfT. DfT (2006). Compendium of Motorcycling Statistics 2006. London: DfT. DfT (2007). Second Review of the Government’s Road Safety Strategy. London: DfT. Dietrich, A. (2004). ‘Neurocognitive mechanisms underlying the experience of ﬂow.’ Consciousness and Cognition, 13(4), 746–61. Fuller, R. (2005). ‘Towards a general theory of driver behaviour.’ Accident Analysis and Prevention, 37(3), 461–72. Goldenbeld, C., Twisk, D. and de Craen, S. (2004). ‘Short and long term effects of moped rider training: a ﬁeld experiment.’ Transportation Research Part F: Trafﬁc Psychology and Behaviour, 7, 1–16. Kipling, J. (2005). Inside the Mind of a Motorcyclist to Reduce Accidents. Retrieved

Flow, Task Capability and Powered Two-Wheeler

423

14 April, 2006, from http://www.devon.gov.uk/index/your_council/inside/ directorates_and_departments/chief_executive_s/communication/news_service/ press-releases/press_psychology.htm. Mannering, F.L. and Grodsky, L.L. (1995). ‘Statistical analysis of motorcyclists’ perceived accident risk.’ Accident Analysis and Prevention, 27(1), 21–31. Ormston, R., Dudleston, A., Pearson, S. and Stradling, S.G. (2003). Evaluation of Bikesafe Scotland (No. 169/2003): Scottish Executive. RoSPA. (2006). Motorcycling Safety Policy Paper. The Royal Society for the Prevention of Accidents. Rutter, D.R. and Quine, L. (1996). ‘Age and experience in motorcycling safety.’ Accident Analysis and Prevention, 28(1), 15–21. Sexton, B., Fletcher, J. and Hamilton, K. (2004). Motorcycle Accidents and Casualties in Scotland 1992–2002. Edinburgh: Scottish Executive. Sexton, B., Hamilton, K., Baughan, C., Stradling, S.G. and Broughton, P.S. (2006). Risk and Motorcyclists in Scotland. Edinburgh: Scottish Executive. Wickens, C.D. and Hollands, J.G. (2000). Engineering Psychology and Human Performance (Third ed.). New Jersey: Prentice Hall.

This page intentionally left blank

Chapter 34

Understanding Inappropriate High Speed by Motorcyclists: A Qualitative Analysis Barbara Hannigan,1 RayFuller,1 H. Bates,1 Martin Gormley,1 Steve Stradling,2 Paul Broughton,2 Neale Kinnear2 and C. O’Dolan2 1 Trinity College Dublin, Ireland 2 Napier University, UK Introduction Higher speeds are associated with increases in the probability of crashing and the severity of the outcome (Aarts and van Schagen, 2006; Federal Highway Administration, 1998). In the context of informing developments in road safety, these relationships make the understanding of the conditions for inappropriate high speed a particularly compelling issue. Proportionately, more motorcyclists exceed the speed limit than car drivers (Fuller et al., 2007). However, excessive speed does not appear to be a major cause of motorcycle accidents. What seems to be much more relevant is inappropriate high speed for conditions leading to loss of control, such as on a bend or in the event of encountering something hazardous and unexpected like an oil spill or debris on the road (Clarke et al., 2004). Just as with car drivers, lack of experience can reduce capability, and motivation for speed can increase task demand, increasing the difﬁculty of the task and narrowing the rider’s safety margin (Elliott et al., 2003; Baughan et al., 2004). There is also some evidence of an adolescent riding subculture that reinforces high risk behaviour (Bina et al., 2006). In order to further our understanding of inappropriate high speed (IHS), a threepronged strategy was adopted involving a review of the available English language literature published in the 12 years 1995 to 2006, a focus groups study and a national (Great Britain) survey of drivers. Preliminary results of the literature survey may be found in Fuller et al. (2006). The results of the national survey are reported in part in Stradling et al. (2007). This report is speciﬁcally concerned with the results from one group of the focus groups study: the riders of powered two wheelers (PTWs), who will also be referred to hereafter as ‘bikers’ or ‘riders’, as these are the terms the participants use. Before describing that study, a summary of theoretical developments in the Task-Difﬁculty Homeostasis model that were informed by the literature review (Fuller et al., 2007) will be presented. This model provides a theoretical framework for understanding speed choice and will be evaluated as a basis for the thematic organisation of the results from the focus group.

Driver Behaviour and Training – Volume III

426

The Task-Difﬁculty Homeostasis model In the context of speed choice, task difﬁculty homeostasis basically refers to the process where bikers adjust their speed so that the perceived difﬁculty of the bikeriding task falls within a range of difﬁculty that the biker is prepared to accept. Perceived task difﬁculty arises out of the interacting interface between perceived task demand and perceived capability (see Figure 34.1 below), hence the original name of Task-Capability Interface model (Fuller and Santos, 2002). Speed directly affects task demand (Fuller et al., 2007). Research also shows that in the bike-riding task, difﬁculty and feelings of risk co-vary, enabling us to refer to the upper level of difﬁculty a biker is prepared to accept as the risk threshold (see Figure 34.1).

range of acceptable task difficulty and risk threshold

comparator

decision and response perceived task difficulty/risk perceived capability

effects on vehicle speed

perceived task demand

Figure 34.1 Representation of the basic process of Task-Difﬁculty Homeostasis From distal to proximal determining factors, perceived capability arises out of the biker’s basic physiological competence, from which starting point the combined effects of education, training and experience produce the individual’s competence as a rider. However this competence may be undermined to varying extents by human factor variables such as drowsiness, stress and emotional state, resulting in the rider’s actual available capability and the biker’s perception of this (see pathway to ‘perceived capability’ in Figure 34.2). Perceived task demand arises out of the handling and other operating characteristics of the bike, driving route, time of day, the physical road environment (in particular visibility level, road alignment and road surface characteristics) and the presence

Understanding Inappropriate High Speed by Motorcyclists disposition to adopt particular range of task difficulty and risk threshold

disposition to comply with speed limit

immediate influences on risk threshold effort motivation

immediate effort

goals of journey

immediate driving goals

range of acceptable task difficulty and risk threshold

immediate influences on compliance comparator

decision and response perceived task difficulty perceived capability

human factor variables

education training experience

physiological competence

427

effects on vehicle speed

perceived task demand

road environment behaviour of other road users

route time-of-day

vehicle characteristics

Figure 34.2 Representation of the process of Task-Difﬁculty Homeostasis, distinguishing between proximal (clear boxes) and distal (grey boxes) determinants and inﬂuences on compliance and behaviour of other road users. It also critically involves the speed at which the biker is travelling (see pathways to ‘perceived task demand’ in Figure 34.2). The biker’s range of acceptable task difﬁculty (and its upper level deﬁned as the risk threshold) is partly determined by her or his perception of current capability. It is also determined in part by the distal and more immediate goals of the journey, the degree of effort the biker is prepared to invest in the task and the biker’s disposition to adopt a particular range of task difﬁculty and its related risk threshold. Beyond this dispositional characteristic, there may also be more immediate inﬂuences on risk threshold, such as an acute feeling of anger (see pathways to ‘range of acceptable task difﬁculty and risk threshold’ in Figure 34.2). Finally, the result of the inﬂuence of all of these components may determine a choice of speed that actually exceeds the posted speed limit. Thus we need to include some representation of the biker’s disposition to comply with the speed limit and also more immediate inﬂuences on compliance (see pathways to ‘decision and response’ in Figure 34.2).

428

Driver Behaviour and Training – Volume III

Speciﬁc aims of the study Questions arise, however, as to the extent to which this theoretical formulation coincides with bikers’ own perceptions and experiences regarding speed choice and how this compares with that of other road users. In order to capture a range of roaduser perspectives, four separate focus groups were established. One consisted of professional drivers, two consisted of participants who had been convicted for a speeding offence and who had opted for a speed awareness course as an alternative to receiving penalty points and one consisted of riders of powered two-wheeled vehicles. This report focuses mainly on the perspectives of the bikers. The use of focus groups appears to provide an appropriate method to explore this issue, providing at the same time an evaluation of both the model’s face validity and its content validity. Thus the principal aim of the study was to elicit rider experiences and perceptions regarding speed choice and speeding behaviour. A second aim was to determine the extent to which the themes identiﬁed by participants could be explained using the concepts of the Task-Difﬁculty Homeostasis model. A third aim was to examine the implications of the results for the content of speed-related media safety campaigns. Method Qualitative research methodology Qualitative research comprises a diversity of approaches to enquiry in the health and social sciences that address the meaning of verbal text. It is more subjective than quantitative research; more exploratory than conﬁrmatory; more descriptive than quantitative; more interpretative than positivist (see Denzin and Lincoln, 1994). It is predicated on the assumption that subjective experience is at least as important as ‘objective’ reality. Interpretative research nonetheless requires a trail of evidence throughout the research process to demonstrate credibility, transparency and trustworthiness (Koch, 1994; Aroni et al., 1999). The guiding theoretical framework underpinning the methodology was a humanistic-phenomenological one. The process of analysis described shows how thematic ﬁndings were derived from the participants’ original transcripts. Thematic analysis Thematic analysis is a search for themes that emerge as being important to the description of the phenomenon under study (Daly, Kelleher, and Gliksman, 1997). The process involves the identiﬁcation of themes through careful reading and rereading of the data (Rice and Ezzy, 1999). It is a form of pattern recognition within the data, where emerging themes become the categories for analysis.

Understanding Inappropriate High Speed by Motorcyclists

429

Participants The participants whose voices are represented in this report were nine motorcycle riders, ranging in age from 18 to 50 years, who participated in a focus group, held on February, 22nd 2006, in Edinburgh. They were from a convenience sample selected to provide homogeneity (all adult bikers) as recommended by Morgan (1997), and partly to ensure a diversity in range of ages and type of bike-riding experience. The powered two-wheelers focus group was included in the wider context of the study on IHS in particular for their potentially unique perspective on inappropriate high speed. No attempt was made to obtain a sample that might be representative of the UK biker proﬁle. The participants were sourced through professional contacts or groups of people accessible to the research team and were invited to attend as volunteers. They were paid £50stg for their participation. None of the participants were known to the group facilitator or identiﬁable to the researchers during the process of data analysis, except through a designated identiﬁcation number. Their views provide us with a snapshot of personal opinions and perspectives detailing accounts of their own and others’ experiences and narratives about issues related to speeding. Procedure All discussion from the focus group session was recorded on audio-tape to facilitate subsequent transcription. The session lasted approximately one hour and 45 minutes. Issues for the group discussion were loosely guided by a broad set of themes which were agreed earlier by the research team (see Table 34.1). At the beginning of the session, the facilitator introduced the topic in general and requested that the participants exchange their views and opinions. Agreement or consensus was not a goal of the discussion. The group was encouraged to expand on topics through the use of open questions and validating prompts by the facilitator such as ‘that sounds interesting, can you tell me more about it?’ or ‘does anyone have an alternative view to this?’. The main themes that emerged during the discussions were reﬂected back to the group to check that they had been fully and accurately heard and to provide an opportunity to the participants to either clarify or expand on their views of the topics raised. Method of analysis Transcripts were initially generated from each audio recording and meaning units identiﬁed by four researchers. Emergent categories were then cross-checked. In the next stage, thematic clusters were identiﬁed and subsequently organised in terms of the concepts of the Task-Difﬁculty Homeostasis model. These two steps were carried out by two researchers independently and the results compared. Any discrepancies were resolved through discussion. Thus the data that emerged from the

Driver Behaviour and Training – Volume III

430

Table 34.1 Broad themes discussed during the focus group Interview Themes

Aspects Considered

Speeding adjectives Attitudes to speeding Factors inﬂuencing speed Internal and external Speed reduction inﬂuences Factors relating to speed limit compliance Control and loss of control Road trafﬁc accidents/collisions Speed choice Speed and legal issues

Completely open Positive and negative

Bike/car issues/ Causes of speeding

Effective and ineffective Assist/ reduce compliance Internal/external factors Personal experiences/impact Different contexts Penalties/issue of being caught/ appropriateness of limits Similarities and differences Mood and emotion

Causes of slowing Speeding stereotypes Consequences of speeding Spontaneous recommendations from group General driving behaviours inﬂuencing speed

Internal and external Gender Legal/ personal / Risk/safety issues Other types of dangerous driving

focus group facilitated the themes to emerge from the data using inductive coding of the phenomena. The step by step process is outlined in Figure 34.3. Results In the results, statements made by individual participants are identiﬁed by a number allocated to each participant during the group process to ensure conﬁdentiality. The views expressed should not therefore be taken as necessarily representative of all group members unless consensual agreement is indicated. Nevertheless, there was frequent open agreement and little dissent occurred. Italicised words in parenthesis are added for clariﬁcation. Emergent themes are organised in relation to key concepts of the Task-Difﬁculty Homeostasis model. Task difﬁculty homeostasis Bikers made some reference to using speed as a means of maintaining a desired level of difﬁculty or safety margin: Well, I, could control the safety margins with the speed. I feel quite happy doing 80–85, but if something, if the weather.., if conditions got worse, if the rain gets heavier, then I would slow down, I would kinda back off. (PTW 2)

Understanding Inappropriate High Speed by Motorcyclists Cross check and support with direct quotes Integrate into template

Report to research team

431

Finally report findings Research team Research concept

Collate audio tape, text and transcribe

Identify units; essence of what was said

Develop categories and verify

Interpretatively determine connections & create clusters

Verify and cross check

Figure 34.3 Schematic illustration of the research process Some bikers referred to this desired state as a ‘comfort zone’: And again it was on the motorway, nobody else about, did it (high speed) for a couple of minutes, stopped whenever there was anything looking like it was getting too close. Just a bit too much sensory input for me, and a little bit too quick, even though feels like an empty road, it doesn’t feel comfy. (PTW 8)

The desired state may be informed by the biker’s sense of control: The fastest I’ve ever been on a bike is 130 and that is 15 miles an hour short of what the current bike can do, but I shut it down for various reasons: one, I suddenly realise what the hell happens if someone catches me; but two, it was starting to get busy in front of me as well, trafﬁc was coming towards me at a hell of a fast rate, and I just wasn’t prepared to take a chance. Umm, if somebody gave me carte blanche and a fairly empty bit of motorway, sure, I would wind it on and see: a) what the bike could do; and b) what it felt like. But, to be frank, I don’t ﬁnd it that enjoyable, because I wasn’t overcoming any

432

Driver Behaviour and Training – Volume III problems. As it’s been said, it was so easy. You just sit there, wind it on and just bend down behind the faring, you know. Whereas, you know, it’s just been said, take it round some challenging corners and get it neat and precise and accurate and do it safely. Under control, that is much more satisfying. (PTW 6)

Human factor variables that may inﬂuence capability Bikers mentioned the importance of loss of concentration as contributing to a collision: The only time I had a crash on the bike, in sort of twenty nine years of riding, was I had taken the bike out, err, I had been setting up the suspension, after being down to the bike show, late in the afternoon, slowed down to 60, having been going a little bit faster, testing things out, and I just lost concentration and went into a corner, woke up about ten, twelve yards from the corner and went shit, going far too fast, hit the brakes, drifted wide, and ended up in a ditch. Luckily I walked away from it with just cuts and bruises. OK, that was because my concentration had switched off because there was not enough to think about, having been going quickly for two or three hours. (PTW 1)

They also mentioned that concentration may suffer at low speeds and that recovery from fatigue after a break enables riding at higher speeds. Factors that may inﬂuence task demand All groups referred to road design and other road characteristics and to trafﬁc conditions as inﬂuencing speed choice: I always ride at a speed appropriate for conditions, and it will not necessarily be the speed posted on the signpost, there is so much disparity between 30 limits, 40 limits. Well 30 limits could well be 40 limits and on some roads 30 is ﬁne, it is not a problem. On other roads it depends on the conditions at the time, and they’re not always necessarily appropriate. I think they should be more variable… (PTW 1)

…Depends on the road conditions, and where you are on the road – schools and the likes of that, where there’s possibility of lots of pedestrians, or compared to urban motorways/ dual carriageway, more likely to speed on the open road… (PTW 4)

Related to this, bikers indicated that familiarity with the road was also a factor: There is also a factor as well with roads that you are familiar with; you are probably going to drive, or ride, quicker on them than on roads that you don’t know. It’s just because you have a reasonable expectation of driving down this road, if you do it this sort of speed most of the time you will have no problems. You try doing it on, probably an identical piece of road that you are not familiar with and its like, ‘what’s round this corner’, ‘what’s happening here’ – so there is an element of that. (PTW 1)

Bikers also referred to the characteristics of their bike:

Understanding Inappropriate High Speed by Motorcyclists

433

If you are in a car and you want to accelerate, you got to be in the right gear and it takes time, on a bike its just, boom, and you’re there. You set the speed with your wrists. Not without, with hardly any time delay, brakes equally well. It’s highly effective and almost instantaneous. If you want to be there to there, you can do it in almost instantly, and it’s often said in the bike press, you think where you want to be and the bike goes there. And they often say if the bike handles really well, you just have to think it round the corner. You constantly read that in the press, and that is part of it. There is no, there is very little constraints on what you want to do from the vehicle. (PTW 6)

They also noted the inﬂuence of weather conditions: If it’s in the summer, I may well ride faster than I would in the winter. In the winter time, certainly when its getting round about the freezing, I will be concerned about the possibility of ice and slowing down. (PTW 1)

and the potentially dangerous behaviour of other drivers: I wouldn’t be watching the van because there’s a vehicle, a couple of vehicles in front of the van and one of those might do something, might make the van swerve into your lane and that’s what I would be, I would probably, in fact, possibly slow down until I got past the ﬁrst car. And I could eyeball that group of vehicles up there before shooting past them. (PTW 6)

Distal and proximal factors which affect range of acceptable task difﬁculty and risk threshold At a distal level, one participant stressed the importance of experience in moderating his range of acceptable task difﬁculty and risk threshold: But the one time that I was on a fairly clear stretch of road, it was a motorway with not much trafﬁc, and I thought I would just see if I could get it up to a hundred, just to see, because I never did it before and I thought that I would give it a go, that’s what I thought, the roads quite clear, I know this road, it’s ﬁne, and I just about to get to one hundred, ninety-ﬁve, and a hawk ﬂew out from the side of the road, and I slowed down and I just wing clipped the front of my bike as it went, and I thought well, is that someone trying to tell me that I should not be driving this fast? But I have not tried again since then because my thinking was, well, there was no way to predict that ever happening at that exact moment. (PTW 9)

At a proximal level, the participants experienced high speed as a great feeling: Yeah, it is, it’s a great feeling. Your head feels empty, you’re just scooting along and you’re going ‘this is the business’. You know a bit of speed and the ﬁrst time you do it, Wohoh, look at me! You know… (PTW 7) That is the biggest thing that I have noticed. Cause it is, you’ve got that exhilaration feeling, you’ve got wind rushing past you. It’s great, isn’t it? You don’t experience that in a car. (PTW 7)

They described the equivalent of their risk threshold:

434

Driver Behaviour and Training – Volume III So, I think the conclusion that we are coming to is, that it is not that the actual number, it’s the relative speed compared with, emm, what is possible, the feeling that you are making, you are pushing the bike somewhere close to your own limits and learning and developing that. But then, comes the ceiling that’s set individually by self-preservation, of yourself and of your licence, and also the effect you are having on everybody else in the world. (PTW 6)

and the desire to experience a feeling of power: …And there is a certain amount feeling of, I don’t know, bravado, perhaps, and a certain macho type of thing… There’s a sort of thing about being on the bike, rather than in a car, and you feel that you are a bit more at one with the world. (PTW 6)

Other inﬂuences which tended to raise risk threshold for all participants were to see how fast the vehicle could go: Every bike I have had I’ve taken it to the limit at some point. (PTW 2) the expression of frustration or annoyance: Frustrations and that, getting held up by cars, trucks and buses, want to get past them to a bit of clear road and will do the speed needed to get past them. (PTW 4)

pressure from other drivers/riders: You normally get, it normally works out that the person who is going the fastest ends up at the front and everyone else is playing catch-up, and it takes a certain amount of self discipline to say ‘No, I don’t want to go that fast’, and a lot of young, younger people become croppers without the experience, they get into a group of people and think that they have got to keep up, and nobody said to them ‘Look, we’ll wait for you’. (PTW 6)

competitiveness and a desire to impress: I would say that a lot of the, a lot of the fun for me, is trying to get by on old, or substandard, or, you know, antique transport. You get a lot of satisfaction from, you know, from keeping up with modern trafﬁc on something 30 years old, ok it’s … leaking again, ok, … but, I’ve a friend who is constantly rebuilding a three-cylinder Triumph cause he tries to drive it like a sports bike, cause all his mates have got sports bikes. But he keeps up with them, only for three months of the year, then its in pieces again. (PTW 8)

…You know the kind into speeding, at the top that’s going to be the kind of weekend warrior types with the one-piece leathers and the state-of-the art sports bike. And the pressure is on them cause they’ve got a point to prove. (PTW 5)

Bikers pointed out, however, that their behaviour often depends on the prevailing mood they happen to be in:

Understanding Inappropriate High Speed by Motorcyclists

435

But then again to be honest, it depends how I felt, if I was in the mood, I would maybe take it up a bit faster, then again depending how I feel I might go a bit slower, you know, just how I felt on the day. (PTW 3)

Factors affecting journey goals There was agreement about the pressure of being late: Part of the business is if you are trying to get to somewhere for a certain speciﬁc time, and you have been held up or whatever, its like yeh, I am going to try and get there, ummm, not always a sensible decision, but it happens, so you may have the pressures of making an appointment, or whatever, and I’ll, you know, drive faster than I probably really want to. (PTW 1)

And keeping with the trafﬁc ﬂow is a common immediate inﬂuence on driving goals: …but certainly at the bottom end of the M1 or something, in a car everybody, and someone was saying just now, is doing 85, 90 miles an hour, in a, like a train, and if you do less then you are getting in everybody’s way, and it is, in my mind, becoming more dangerous. I wouldn’t actually want to be on a bike in that sort of a train, because I would feel that if anything went wrong, I am meat in a sandwich. (PTW 6)

For bikers a further goal may be to test one’s skill: There are occasions when… when you go out… to, sort of, test yourself, almost, and you sort of say I am just going to go a wee bit faster around that bend. Biking is all about going around bends. (PTW 6)

Factors inﬂuencing compliance with speed limits Dispositional inﬂuences Bikers generally agreed they should comply with speed limits and that non-compliance could lead to a dangerous loss of control: Awareness of what can go wrong, the penalties of getting it wrong, you are much more aware of what is going on around you, because on a bike if you miss a potential hazard it can kill you, if you miss it in the car it will dent the car and maybe knock you out, there is a signiﬁcant difference in the penalty for getting it wrong on a bike, and you tend to be that much more aware. (PTW 6)

Enforcement is commonly regarded as encouraging levels of compliance: I am aware that if I was caught speeding then it involves penalty points and I drive for a living so that keeps me on a much more sort of even keel, you know… (PTW 7)

Bikers also indicated that compliance may be induced by feelings of shame. Hence speed violations are more likely to occur when there is no-one else looking:

436

Driver Behaviour and Training – Volume III …on the open road I will do what I think is acceptable, I am always concerned about the image, impression I give to other people about the rate at which I go, so I am more likely to speed when there’s nobody around. (PTW 6)

Training and age were reported to have inﬂuenced levels of compliance with speed limits: When I started riding, about 20 years ago, I automatically thought that 40 meant 50 and 30 meant 40 and 70 meant as fast as you could go on a small and knackered old bike, and now I am pretty strict in town and… I think that comes with doing an advanced driving course. (PTW 8) I concur with that a lot, I always thought of speed limits being ugly – 10 miles over – it was my standard. I now don’t exceed 30 in a 30, I normally try to stay down at 20 in 20s while going down my high street for example, and totally ignore the people behind me who want to go faster… I have slowed down and accepted the speed limits as being what I should be doing, probably only in the last 5 years possible even really, and part of that is the result of doing an Institute of Advanced Motorists (IAM) training and showing a commitment to that, and just getting older. (PTW 6)

All groups also agree that they may not comply when the speed limit is perceived to be too low for the road characteristics – the speed limit needs to make sense: …when it was busy, forty limit, past road-works… even when maybe there weren’t many people working there, most people were respecting the speed limit. Seven o’clock at night, quiet, no not much trafﬁc, nobody around, everybody was thinking, ‘what the hell are we doing forty for?’, and they are all doing ﬁfty, and it was perfectly safe, umm. (PTW 6)

They also indicate that it can be difﬁcult to recalibrate one’s speed when entering a lower speed zone and that the distraction of frequent checking of the speedometer can be dangerous: I would be concentrating on the road. As it’s been said there is no point in taking your eyes off what could happen because somebody could pull out in front of you or swerve from the hard shoulder into the central reservation and you want that extra tenth of a second. You don’t care about the speedometer, it’s nonsense. What you are caring about is, you’re caring about getting there safely and avoiding any potential hazards. (PTW 1)

Group comparison Bikers agreed with car drivers in the other focus groups that speeding is easy and can be unintentional. Inﬂuences which raise speed include pressure from others, keeping with the trafﬁc ﬂow, seeking an adrenaline rush and in response to frustration or annoyance. Vehicle size or type and seeing how fast the vehicle will go can also induce increased speed. However, unlike the bikers, drivers reported that speeding is common – everyone is in a hurry. Bikers see most riders as sticking to the speed limit. Like the professional drivers, they refer to riding in a ‘comfort zone’, but also talk about ‘getting into a groove’ and being ‘at one’ with the bike. Speeds which induce this state may be over the limit:

Understanding Inappropriate High Speed by Motorcyclists

437

…you get into, emm, a sort of close relationship with the bike and what you are doing. And everything becomes sort of semi-automatic and it’s going well and that can make you forget about the speed limit because you are just enjoying it so much and the experience that you just ride the bike and you suddenly wake up, oh, what was that, I’m going fast. And you’re going fast against the speed limit, not against what’s possible. All right, but, the, ehh, ehh, what keeps you doing that is its just bloody enjoyable. You know you get a great deal of pleasure out of doing that. And you get a sense of satisfaction and achievement as well. Emm, which you don’t normally get in a car. Because it is not so controlled, not so adaptable and not so ﬂexible, emm, as, as a bike is. (PTW 6) It’s like the sweet spot on a golf club or a tennis racket (PTW1)

Like the drivers on the speed awareness course, bikers think that exceeding the speed limit is not necessarily unsafe. They are more likely to comply when the limit makes sense. They see as general inﬂuences on speed ‘common sense’, the time available to respond, weather conditions and fear of mechanical failure at high speeds. Avoidance of feelings of guilt and shame similarly helps motivate compliance. Also in similar vein to those drivers, bikers may be inﬂuenced to go faster by others and the desire to impress with a demonstration of skill. Speeding can be fun but is also mood dependent. Bikers say that low speeds can impair concentration. They recognise that the safety margin is up to the rider and leave space to enable avoidance actions if necessary, try to keep braking to a minimum and slow down to deal with the dangerous behaviour of other road users. They are aware that bikes are less stable than other vehicles, are not so easily seen and that riders are particularly vulnerable. Discussion From the main content of participants’ comments it may be seen that the themes can be readily mapped onto the conceptual framework provided by the Task-Difﬁculty Homeostasis model and no theme emerged that could not be accommodated by the model. For this reason the results have been presented using the model’s framework. Inﬂuences on speed choice are easily translated into the guiding theoretical concepts of driver capability, driving task demand, driver task difﬁculty, risk threshold and disposition to comply with speed limits. In fact the only dimension of the model not speciﬁcally identiﬁed by participants’ comments was that of effort motivation. Bikers reported using their speed in order to control safety margins and to obtain a ‘comfortable’ state. Rider capability was regarded as being affected by experience, training, stress, age, concentration and fatigue. The demands of the task of riding a powered-two-wheeler were seen to be inﬂuenced by road and trafﬁc conditions, including familiarity with the road, bike characteristics, weather conditions and the behaviour of other road users. There was awareness that high speeds reduced the time available to deal with contingencies and riders wanted to keep braking to a minimum, particularly sudden hard breaking, to avoid loss of control. On the other hand there was general agreement that speed could be pleasurable and could deliver an adrenaline rush and a feeling of power. Speed increases could also be triggered

438

Driver Behaviour and Training – Volume III

by pressure from others and by competitiveness. Sometimes bikers used speed to test their skill and the capacity of their bike, particularly when it was a new machine. Others felt that the need for this changed with age and maturity. Non-compliance with the speed limit was generally regarded as potentially dangerous, but not necessarily so where road and trafﬁc conditions permitted. The margin above the limit, allegedly ‘permitted’ by enforcing agencies, is regarded as authorising minor levels of infringement and there is no shame felt in engaging in this behaviour. Going over the limit can also happen unintentionally, especially where signs are not detected: compliance is much easier when the limit makes sense. There was a greater tendency to speed when there was a perceived reduction of risk such as on non-urban roads with low trafﬁc or when there was a lack of potential witnesses. Thus the conclusion has to be that the evidence arising from the focus group content is adequately captured by the model and is within its range of convenience. Riders’ own perceptions and experiences regarding speed choice are entirely consistent with the theoretical formulation and provide for a rich elaboration of instances of its key concepts. The focus group method in addition enabled capture of the narration of dynamic episodes, in contrast to the unitary responses typical of questionnaire surveys. For example, consider this narrative about ‘putting down’ other motorcyclists: I can think of two or three friends, I’m not actually a good enough driver to do this very much I don’t think, but I certainly got two friends who got old, old and British bikes and trigger for them is someone with something newer, faster, bigger, more modern, and they go ‘ah, look at that’, and especially if it is someone who has fairly obviously recently just passed their direct test, and got a big sporty bike and you can just see that they are a little bit wobbly on the corners. It’s let’s just show them what they have wasted seven or eight thousand pounds on. And its not clinical speeding, this will tend to be, single track roads… ummm and you can just see them (the two friends) disappear into the distance, and you see them ﬁfteen miles away with a big smirk on their face, and I’ll kinda try and maybe stay with these bikes but, but they will be determined to overtake, and wave goodbye to them. That’s just a statement of fact, which is kind of childish but, there is a lot of smirking going on in this room. (PTW 8)

Comparing groups on the themes they identiﬁed, bikers talk about very similar themes as drivers. However they additionally identify desirable states when motorcycling, such as being in ‘the groove’. Furthermore, because of their particular vulnerability, they are more expressive of the importance of maintaining concentration and of consciously maintaining an adequate safety margin which, for example, minimises the requirement for excessive braking. What are the implications of these ﬁndings for identiﬁcation of themes for possible media campaigns? Bikers appear to be aware of the relationship between vehicle speed and risk of collision and severity of consequences. They appear to be aware that many factors can increase task difﬁculty or decrease their capability. They seem to be aware that there are many conditions that can raise their risk thresholds, as well as lower them. And they seem to be aware of inﬂuences on their disposition to conform to speed limits. At times they are motivated to exceed these by excessive

Understanding Inappropriate High Speed by Motorcyclists

439

amounts, but exploit this possibility in the context of an otherwise low task difﬁculty and a low probability of being detected and penalised. At other times they marginally exceed limits as a natural adjustment to a limit that may be a legal requirement but is otherwise a rather imprecise guide as to what the safe upper limit actually is at that moment in time. Thus there appears to be little of a knowledge gap that a media campaign might aim to inﬂuence. However, one possibility that might be targeted by a media campaign is the potentially dangerous inﬂuence of mood state on speed choice. We know far more today about how mood state can be managed, and this knowledge might usefully be passed on to bikers to help minimise the negative inﬂuence of mood state on decision making. A second possibility has to do with the establishment and inﬂuence of perceived norms. Inevitably riders’ attention is captured by marked deviations rather than the norm, but such selective attention may lead to the development of an inappropriate bias in the rider’s perceived norm. If riders are inﬂuenced in the direction of conforming to perceived norms (and there is evidence that drivers are so inﬂuenced; for example, Åberg et al., 1997), this bias can lower the quality of the rider’s behaviour and that in turn may feedback into contributing to a further decline in the actual norm. The consequence is a slow deterioration in the underlying riding culture of the community. Media campaign presentation of appropriate actual norms might help counteract this process, indicating for example that by far and away the majority of riders do not speed excessively, do not overtake dangerously and are not aggressive or competitive in their interaction with other road users. All groups identiﬁed time pressure as a key motive inﬂuencing increased speeds. Providing advice on time management to avoid this inﬂuence might be productive. On the other hand, counteracting the use of speed to obtain an adrenaline high may be much more intractable. An emergent theme that was frequently visited was the inherently social nature of riding: social in terms of perceived norms of behaviour, including speeding; social in terms of temporary ‘skirmishes’ and social in terms of the opportunities for status in bike ownership, style of riding and competitive engagement. These social variables are of interest because they not only drive up speed and risk thresholds but they are essentially independent of rider mobility goals. Perhaps then it should be in this area of inculcating more socially acceptable behaviour that a media campaign might focus. Acknowledgements This work was completed in part fulﬁlment of Department for Transport contract Number PPRO 4/001/015 (‘Improved Driver Information on Speed and Accident Risk’) and does not necessarily reﬂect the views of the Department. The researchers are indebted to the very cooperative participation of the focus group members and would like also to express their thanks to Alan Fisher and his staff for their generous assistance in enabling access to the Speed Awareness Course participants.

440

Driver Behaviour and Training – Volume III

References Aarts, L., and Van Schagen, I. (2006). ‘Driving speed and the risk of road crashes: a review.’ Accident Analysis and Prevention, 38, 215–24. Åberg, L., Larsen, L., Glad, A. and Beilinsson, L. (1997). ‘Observed vehicle speed and drivers’ perceived speed of others.’ Applied Psychology: an International Review, 46, 287–302. Aroni, R., Goeman, D., Stewart, K., Sawyer, S., Abramson, M. and Thien, F. (1999). Concepts of Rigour: When Methodological, Clinical and Ethical Interests Interact. Paper presented at the First AQR Conference, Melbourne, Australia, Retrieved February 9, 2007, from http://wwwlatrobe.edu.au/aqr/offer/papers/RAroni.htm. Baughan, C., Sexton, B. and Elliott, M. (2004). ‘Motorcyclists’ accident risk: results from a new survey.’ In Behavioural Research in Road Safety XIV. Department for Transport, London, 21–37. Bina, M., Graziano, F. and Bonino, S. (2006). ‘Risky driving and lifestyles in adolescence.’ Accident Analysis and Prevention, 38, 472–81. Clarke, D.D., Ward, P., Truman, W. and Bartle, C. (2004). ‘An in-depth case study of motorcycle accidents using police road accident ﬁles.’ In Behavioural Research in Road Safety XIV, Department for Transport, London, 5–20. Daly, J., Kelleher, A. and Gliksman, M. (1997). The Public Health Researcher: A Methodological Approach. Oxford University Press, Melbourne, Australia. Denzin, N.K. and Lincoln, Y.S. (1994). Handbook of Qualitative Research. Sage, Thousand Oaks, CA. Elliott, M.A., Sexton, B. and Keating, S. (2003). ‘Motorcyclists’ behaviour and accidents.’ In Behavioural Research in Road Safety XIII. Department for Transport, London, 139–52. Federal Highway Administration (1998). Synthesis of Safety Research Related to Speed. Publication No. FHWA-RD-98-154, Federal Highway Administration, Washington, DC. Fuller, R., Bates, H., Gormley, M., Hannigan, B., Stradling, S., Broughton, P., Kinnear, N. and O’Dolan, C. (2006). ‘Inappropriate high speed: who does it and why?’ In Behavioural Research in Road Safety 2006. Sixteenth Seminar. Department for Transport, London, 70–84. Fuller, R., Bates, H., Gormley, M., Hannigan, B., Stradling, S., Broughton, P., Kinnear, N. and O’Dolan, C. (2007). The Conditions for Inappropriate High Speed: A Review of the Research Literature from 1995 to 2006. Report under Contract Number PPRO 4/001/015 Improved Driver Information on Speed / Accident Risk (T201G). Department for Transport, London. Fuller, R. and Santos, J.A. (2002). Human Factors for Highway Engineers. Pergamon, Oxford. Koch, T. (1994). ‘Establishing rigour in qualitative research: The decision trail.’ Journal of Advanced Nursing, 19, 976–86. Morgan, D.L. (1997). Focus Groups as Qualitative Research (2nd ed.). Sage, Newbury Park, CA. Rice, P. and Ezzy, D. (1999). Qualitative research and evaluation methods (3rd ed.)

Understanding Inappropriate High Speed by Motorcyclists

441

Sage, Thousand Oaks, CA. Stradling, S., Gormley, M., Fuller, R., Broughton, P., Kinnear, N., O’Dolan, C. and Hannigan, B. (2007). ‘A typology of speeding drivers: extent of and motives for exceeding the speed limit.’ In Behavioural Research in Road Safety 2007. Seventeenth Seminar. Department for Transport, London, in press.

This page intentionally left blank

Index

Note: Bold page numbers indicate illustrations; italic page numbers indicate tables and graphs. a2om Ltd 4, 32, 35 AAC see Early Driver Training Aberg, L. 202 ABS 291, 381 absenteeism 167-172 accelerometer 19 accident risk 76-77 age factor in 133, 277 and cognitive competence 253 and physical activity 167-172 self-reporting of 168 study method 168-169 study results 169-170, 170 action selection 252-253 ADHD 99, 102 Adolescent Road User Behaviour Questionnaire (ARBQ) 67 adolescents and PTWs 399-401, 425 advanced driving training 17, 190, 436 advertising campaigns 228 age factors 76, 108, 114 in accidents/speeding 133, 277 in ﬂeet drivers 237, 242-243, 242 and hazard perception 132, 135, 136 in navigation system usage 292, 303, 306 older drivers 129-137 in accidents 129 cognitive abilities of 129-130 driving performance of 131 self-reporting by 130 and passengers 3 and self-reporting 131-132 and stress 129-137 coping strategies 133-137, 134, 135 study method for 131-132 study results for 132-134 transactional model for 137 in trafﬁc offences 129, 132 in vigilance 373, 374, 374

aggression 3, 8, 10-11, 20, 77, 94-96, 117126 age/gender variables in 122, 123, 126, 129, 132, 135, 136 in DBQ 202 deﬁned 94 in ﬂeet drivers 236, 240, 240, 242, 242 and intent 118, 121, 122, 124-125, 125 in personality studies 52, 53, 57, 66, 91, 93, 94, 99 self-reporting of 123-124, 132 and speeding 124, 124 study method for 119-120 study results for 121-126 typology of 117, 121, 122, 124-125, 124, 125, 126 see also stress airbags 382-383 AISS (Arnett’s Inventory of Sensation Seeking) 68-70 alcohol see drink driving Alexander, J.L. 330 Allen, R.W. 265 Alverson-Eiland, L.G. 93, 99, 101 anger 91, 93, 94-96 measurement of see DAS; PADS and risky behaviour 107, 344, 427 trait see trait anger anticipatory driving 340-341 António, P. 401 anxiety 92, 93, 97, 99, 417 Appel, C. 93 appraisal theory 100 ARBQ (Adolescent Road User Behaviour Questionnaire) 67 ARCES (attention-related cognitive errors scale) 304 Arnett, J. 66, 68, 94, 102 assaultiveness 56, 57 attention, sustained 301-302, 304

444

Driver Behaviour and Training – Volume III

attitude see personality/attitude attributional theory 118 Australia 51-60, 75, 132 crash factors in 230 crash investigation in 379-388 DBQ study in 203-212 driver education study in 155-164 ﬂeet safety study in 227-233 PBT study in 66 risky behaviour study in 175-185 rural roads in 369-370, 370 vigilance study in 367-375 work-related trafﬁc injuries in 227-228 Austria, ISA study in 311-320 automation 252, 291 reliance/complacency 302-303, 304-305 see also navigation systems BAC (blood alcohol concentration) 327, 341, 380 Bañuls Egeda, R. 97 Barkley adult attention scale 79 Barkley, R.A. 79, 83 Bartle, C. 419 Bartlett’s sphericity test 240 Baughan, C.J. 67 Baxter, J. 129, 144, 327 Beirness, D.J. 52 Bell, P. 95 Ben-Yaacov, A. 326 benchmarking 227 Berg, H.Y. 60 Bergeron, J. 118 Biggs, H.C. 176 Bisantz, A.M. 304-305 Björnstig, U. 143 Blanchard, E.B. 118 Blomkvist, A. 93 blood alcohol concentration (BAC) 327, 341, 380 booster seats 393, 394, 396 Boud, A. 34 brake lights 326 braking 18, 20, 22, 27, 291 Britain 94, 311, 419, 429 driving instructors in 31-32 driving test in 31 injury/death rates for 391, 399, 415 Brno 312, 313

Brown, K.W. 304 Brown, L.B. 370 Burnett, G.E. 293-294 Burns, N. 66, 72 bus drivers 167-172, 240, 257 celeration behaviour of 190-196 Bylund, P.O. 143 caffeine 330 Caird, K.J. 143 Cairns, David 390 calibration of skills 38 Campbell, K. 129, 144, 327 Canada 75-85, 271-272, 295 capability 418-419, 418, 420-421 perceived 337, 338-339, 338, 342, 345346, 426, 427 car ownership 394-395, 394 Carbonell Vaya, E. 97 Carcary, B. 132, 137 Carmichael, S. 8 Carney, C. 20, 27 CARRS-Q 229, 233 Casanoves, M. 97 CBR (Dutch driving test organisation) 37 CBT (computer-based training/tests) 31, 252, 258, 259-262 CDC (Center for Disease Control) 266-267 CFQ (cognitive failure questionnaire) 304 chain of responsibility (COR) 228 challenge based enjoyment 415-416, 419, 420-421 change see stages of change change detection 252-253 Chapman, P. 96-97, 99, 102, 288 Charlton, S.G. 351 children and aggression 118 booster seats for 393, 394 and road safety 65-66 seat belt laws for 389-396 need for 391, 395 objective of 390-391 study method for 392 study results for 392-394, 392-394 Chisvert, M. 97 Christmas, S. 11 Cito Drive 252, 256, 260 Clarke, D.D. 419

Index Clarke, S. 144 classroom based training 18 Clegg, B.A. 33 close following 68, 179, 180, 180, 182, 183, 211, 327 countermeasures for 326 and other risk factors 331, 331 coach drivers 167-168 see also bus drivers cognitive failure questionnaire (CFQ) 304 cognitive functions 18, 38, 46, 76, 77, 97, 235-236 age factor in 129, 130, 277 and ﬂow 417-418 and GPS usage 303, 304 and Groeger’s model 253-256, 254 and training 252-253 see also neurological factors Cohen, J. 54 commuter drivers 239, 242, 245 complacency 302-303, 304, 305 computer-based training/tests see CBT concentration, loss of 107, 108, 110, 111, 112, 113, 432, 438 Connors’ Continuous Performance Test (CPT-II) 78 control, loss of 107, 108, 110, 111, 112, 113 Cook, M.L. 265 COR (chain of responsibility) 228 Craen, S. 408 crash investigation 379-388 case study 382-383 computer modelling in 381 contributary factors data 383-385, 385 driver interviews 380, 383 local review panels 379-380, 381-382, 383, 385, 387 recommendations of 386, 388 site inspections 381, 383, 384, 386 study method for 380-383 study objectives 386-388 study outcomes 383-388 study participants 380, 382 vehicle inspection 380-381 crash records 17, 27, 51, 76, 114-115, 311 crash risk 3, 17 crashes costs of 175, 203, 227 factors in 325-332, 383-385, 385

445

alcohol 327, 331, 331, 332 close following 326, 331, 331 correlations between 330-332, 331 fatigue 328-331, 332 hazard perception 330, 331, 332 speed 325-326, 331, 331, 332 violations 327-328, 331, 332 work-related 329 and human factor research 76-77 injuries/deaths in 65, 391, 415 novice drivers in 35, 37, 40, 51, 75 and driving exposure 51, 52-53, 60 environmental factor 51 inadequate skills factor 37, 38-39, 252-253 personality/attitudinal factor 51-60, 76-77 situation awareness factor 255-256 in post-training performance 20-22, 22, 23, 26 predictability of 75-76 self-reporting 52 work-related factors in, see also professional drivers crashes novice drivers in, age-related factor in 37-38, 39 Crick, J. 17 cruise control 291 Crundall, D. 96-97, 99, 102 Csikszentmihalyi, M. 416-418 cyclists 65-66, 176 Czech Republic, ISA study in 311-320 Dahlen, E.R. 107, 114 DAQ (Driver Attitudes Questionnaire) 67, 68, 175, 176, 177, 178, 180, 181182 DAS (Driving Anger Scale) 107-115, 175 reliability of 107-108, 110, 114 study method for 109-110 study results for 110-111 variables in 108, 111, 111, 112 Dawson, D. 371 DAX (Driving Anger Expression Inventory) 120, 123 DBI (Driver Behaviour Inventory) 239 DBQ (Driving Behaviour Questionnaire) 18, 109, 131, 135, 144, 145-146, 175176, 177, 178, 180, 181-182

446

Driver Behaviour and Training – Volume III

aggressive violation in 202, 211 errors/violations in 201, 202, 211, 314 fatigue in 203, 204, 211-212 modiﬁcations for ﬂeets 201-212 factor structure/reliability 205-208, 205, 208, 210-211 list of items 204-205 predictive value 209 study method for 203-205 study results for 205-209 Positive Driver Behaviours in 145 range of uses for 201, 202 DCQ (Driving Coping Questionnaire) 131, 132, 236, 238 de Craen, S. 420 Deane, F.P. 98-99 defensive driving 17, 267 Deffenbacher, D.M. 126 Deffenbacher, J.L. 94-95, 99-100, 109-110, 114, 126 Dehaene, S. 371 DeKock, A.R. 92 Demick, J. 93, 101 DePasquale, J.P. 108 Desmond, P. 132, 137, 302 DfT (Dept for Transport) 291-292, 300, 390, 415 Dietrich, A. 417 Dimmer, A.R. 203 Dodge, K.A. 118 Dodson, J.D. 342 Dorn, L. 100, 101, 167, 171 Dorrian, J. 371 drink driving 3, 9, 13, 20, 54, 66, 75, 76, 80, 84, 176, 182, 341 blood alcohol concentration (BAC) 327, 341, 380 countermeasures for 327 and moped riders 404, 405 and other risk factors 331, 331, 332 and pre-drivers 68, 69-70, 72 strategies/interventions for 228 Driver Attitudes Questionnaire see DAQ Driver Behaviour Inventory (DBI) 239 driver celeration 189-196 habitual nature of 190, 194, 195 measurement of 191-192 monitoring required for 196 study method for 191-193, 195

study results for 193-194, 193, 194, 195-196 driver conﬁdence 241, 241, 242, 243-244, 245 driver education see under professional drivers driver errors 17-18, 92, 93-94, 132, 135-136 distinct from violations 144, 145, 201, 327 of ﬂeet drivers 179, 180, 181, 182, 184, 212 and vigilance 371, 372, 374, 375 driver improvement training 17 Driver Intervention Programme 53 Driver Social Desirability Scale (DSDS) 237, 243 Driver Stress Inventory see DSI driver training and celeration see driver celeration initial/post-licence 251-252 and predicting safe driving 189-196 simulation see simulation training/testing see also Task-Difﬁculty Homeostasis Model driver training intervention 231 evaluating 17-28 errors/weakness in 17-18 see also telemetric data tracking holistic 18 video trial 20, 27 DRIVERS database 54 Driving Anger Expression Inventory (DAX) 120, 123 Driving Anger Scale see DAS driving aspiration 6, 10-11, 13 Driving Behaviour Inventory (DBI) 98 Driving Behaviour Questionnaire see DBQ driving distance 24, 27 driving experience 38, 39 driving exposure 51, 52-53, 55, 56-58, 5960, 59, 71 age factor in 130 and anger 110, 111 of professional drivers 181, 182, 182, 183-184, 207-209, 212 driving, fear/dislike of 99, 100, 132, 134, 134, 236, 245 Driving Instructor Checklist 78-79

Index driving instructors 33-35 core competencies of 31-32 as evaluators 78-79, 80, 84-85, 101, 189-190 and independent driving 45 and self-assessment 35 Driving Instructors’ Handbook (2006) 33 driving lessons 32-36 and hazard perception 47 and independent driving 45 learning curve in 33-34, 34 not true driving environment 32-33, 39 and self-assessment 34-35 see also driver training driving licences 53 driving logs 96, 97 driving simulators 101, 368 Driving Skills Inventory (DSI) 175, 216, 217, 220 driving speed 23, 24, 26 see also speeding Driving Standards Agency (DSA) 31 driving tests 31, 39-40 fault-based assessment in 31-32, 33, 35 multimedia 251-262 practical 37-48 changes to 37, 48 examiners’ questions in 47, 48 hazard perception/management element 40, 42, 44, 47 independent driving element 40-41, 44-46, 45, 47-48 Learner Interim 43(n2), 44, 45 ‘productive’ special manoeuvres element 40, 41-42, 45, 46, 48 self reﬂection element 40, 42-43, 44, 47, 48 study method 43-44 study participants 44 study results 44-47 as predictors of safe driving 189-191 shortcomings of 189-190 three testing methods 189 Wiener Fahrprobe 311, 312-314 see also responsive driving; Task-Difﬁculty Homeostasis Model drugs 66, 81-82, 82-83, 84, 341, 386, 400 Drummond, A.E. 252 Drury, C.G. 304-305

447

DSDS (Driver Social Desirability Scale) 237, 243 DSI (Driver Stress Inventory) 131, 132, 134, 135, 136, 236, 238 Early Driver Training (AAC) 277-278 eating while driving 204 EcoDriving 191 educational factors 76, 408 Edwards, I. 32 Ellingstad, V.S. 92 Elliott, M.A. 67 Ellison-Potter, P. 95 emotions and road user behaviour 91-102, 331, 331, 341 aggression/risky behaviour studies 92, 94-96 concepts/frameworks in studies 99-100, 102 and driver training 342 errors/violations studies 92, 93-94 and fatigue 332 key terms 91 measurement of 101, 102 personality studies 98-99 road safety/accidents studies 92, 96-97, 101 self-reporting in studies 94, 102 situational characteristics studies 98, 99 speeding 342, 343-344 task performance studies 92-93 see also mood Endsley, M.R. 255-256 Engström, I. 31 enjoyment 415-418 challenge based 415-416, 419, 420-421 and ﬂow 416-418, 417 rush based 415, 416, 419, 420, 421, 433-434 see also sensation seeking EPI (Eysenck Personality Inventory) 98 EPQR (Eysenck Personality Questionnaire) 370, 375 Evans, B.L. 351 Evans, J. 99 Evans, L. 215, 326 eye scanning 18 Eysenck Personality Inventory (EPI) 98

448

Driver Behaviour and Training – Volume III

Eysenck Personality Questionnaire (EPQR) 370, 375 FA (factor analysis) 238-239 fathers, and sons’ driving behaviour 3, 6-8 fatigue 32, 132, 134, 134, 184, 328-330, 383, 432, 437 countermeasures for 330 in DBQ 203, 204, 211-212 and emotion 332 of ﬂeet drivers 229, 230, 235, 236, 237, 240, 243, 245 measuring 328-330 need for recovery from 167-172 fault-based assessment 31-32, 33, 35 FDRI (ﬂeet driver risk index) 241, 242, 244, 246 Field, A. 239 Fine, M.A. 94 Finland 42, 94, 143 Fisher, D.A. 288 FJC programme see moped riders Flannagan, M. 326 ﬂeets/ﬂeet drivers 175-185 age factors 237, 242-243, 242 aggression in 236, 240, 240, 242, 242 and BDQ 202-203 crash factors in 230, 235 crash types in 229-230 data collection approaches in 229-231 driver behaviour factors 229, 229 gender differences in 242, 242 OHS in 228 risk assessment of 235-246 factor analysis 237-241, 240 FDRI 241, 242, 244, 246 validity analysis 237, 241-245 violations/crash involvement 244245 safety culture of 180, 181, 183, 184-185, 227-233 baseline measures 232-233 driver training /education 231, 236-237 needs analysis 232 safety climate 231-232 self-assessment by 237 and speeding 176, 179, 180, 182, 183, 210, 232

and stress 235-237 coping strategies 236, 237, 240-241, 241, 242-243, 244, 245 time pressure 235, 240 thrill seeking in 236, 240, 240, 242 Fletcher, J. 419 ﬂow theory 416-418, 417 following distance 326, 403 Forbes, N. 293-294 Ford, F.H. 93, 99, 101 Forward, S.E. 4 France 399 Early Driver Training in 277-278 Francis, L.J. 370 Friedenberg, B.M. 118 friends, inﬂuence driving behaviour 3, 8-9 Frings-Dresen, M.H. 167-168, 171 frustration-aggression hypothesis 100 fuel-efﬁcient driving 191 Fuller, R. 349, 352, 353, 362, 363, 418 g-force 19, 20, 21-22, 27 GADGET project 39 Garrity, R.D. 93, 101 GDE (Goals for Driver Education) 14, 18, 37 hierarchical matrix in 31, 33, 34, 35, 3940, 47, 251-252 GDL (graduated driver licencing) 266 gear-changing 287 gender factors in aggression 122, 123, 126, 129, 132, 135, 136 in coping strategies 133 in ﬂeet drivers 242, 242 in lapses/errors 132, 135-136, 374-375 in navigation system use 292, 300-301 in novice drivers 38, 52, 55, 59, 60, 69, 76, 114 accident rates 277 and passengers 3 genetics of driving behaviour 176 Gerbers, M.A. 327 GHQ (General Health Questionnaire) 168, 169 Gilliland, K. 132, 137 Glendon, A.I. 12 Goals for Driver Education see GDE Goldberg, A.I. 151

Index Goldenbeld, C. 408, 420 Goodman, M. 305 GPS see navigation systems graduated driver licencing (GDL) 266 Greenhouse-Geisser statistic 360 Gregersen, N.P. 60, 236-237, 330 Gregor, S. 304 Groeger, J.A. 33, 92 Groeger’s four facet model 253-256, 254, 259 GRSW (Global Road Safety Week) 65 Guadagnoli, E. 238 guilt/shame 435-436, 437 Gulian, E. 100, 137 Haigh, K.Z. 303 Haigney, D. 129 Hall, J. 78 Hamilton, K. 419 hard braking 20, 22, 27 Harvey, C.F. 326 Hattaka, M. 31 hazard perception 18, 27, 31, 37, 40, 42, 47, 132, 135, 136, 236 countermeasures for 330 driver education and 155 and other risk factors 331, 332 and simulators 265 health & safety see OHS Heavy EcoDriving 191 Heimstra, N.W. 92 Hewson, C. 300 higher order skills 35, 37, 38, 39, 44 Hiles, J. 371 Ho, G. 303 Hobbes, G.E. 77 Hofmann, D.A. 151 horn-honking 91 Horne, J. 328 Horswill, M. 330 Horta, M. 401 Huberman, A.M. 158 Human Information Processing Model 371 impression management 237, 241, 241, 243244, 245 in-car devices see ISA; navigation systems Inagaki, T. 305 inattention 77, 78

449

independent driving 40-41, 44-46, 45, 47-48 deﬁned 41 information processing 349 see also cognitive functions; neurological factors insurance 175, 177, 231 Internet questionnaires 294, 300-301 IPAQ (International Physical Activity Questionnaire) 168, 169, 170, 172 IRT (item response theory) models 261-262 ISA (intelligent speed adaptation) 311-320, 326, 386 attitudes towards 317-319, 319, 320 and pedestrians 311 and speeding 312, 317, 318, 319, 345 study method for 311-315 ISA questionnaire 314-315 psychological group training 315, 320 use of Wiener Fahrprobe 311, 312314 study, participants in 312, 319-320 study results for 315-319 self-reported 315-317, 316 ISH (inappropriate high speed) see speeding Italy 399 item response theory, see also IRT Iversen, H. 72 Janke, M.K. 271, 271-274 Japan 195 J.D.Power survey 292-293, 300 JDQ (Jerome driving questionnaire) 79, 81 Jenkins, D. 326 Jerome, L. 77, 80 Jessor, R. 66, 72 Jian, J.J. 304-305 Joint, M. 100 Joyner, L. 132, 137, 302 Kaiser-Meyer-Olkin test 240 Kiff, L.M. 303 Kipling, J. 420 Kirschenbaum, A. 151 Kline, P. 239 Kline, T.J. 143 Knee, C.R. 95-96, 100 Kölner Fahrverhaltens test 313 Krupat, E. 330

450

Driver Behaviour and Training – Volume III

Ladyman, Stephen 390 Lähdeniemi, E. 143 Lajunen, T. 94, 130, 202 Lamond, N. 371 Larsson, T.J. 143 Lawton, R. 93-94, 99, 100, 201 learning 3-4 situational 3 and social processes 4 Lee, J.D. 20, 27 Lee, T.R. 144 Levelt, P.B.M. 97 log books 17 Loh, S. 371 lorry drivers 97, 257 Lynch, R.S. 94-95, 114, 126 MAAS (mindful attention awareness scale) 304 Macdonald, L. 130 McGehee, D.V. 20, 27 McKenna, F.P. 17, 328, 330, 332 Mackworth, N.H. 371 Madsen, M. 304 Malta, L.S. 98, 117-118 Malz, M. 326 Manchester Driving Behaviour Questionnaire see DBQ Mann, R.E. 327 manoeuvring 18, 31 Manstead, A.S.R. 68, 93, 129, 144, 201, 327 Masten, S.V. 271, 271-274 Matos, M. 401 Matthews, G. 100, 101, 132, 137, 302 Mattila, M. 144 Mayhew, D.R. 271-272, 271-274 Meadows, M. 202 Mesken, T. 202 Mesken, J. 253 meta-cognitive skills 38 Meuter, R.F. 371 Miles, M.B. 158 Miller, J. 33 mindful attention awareness scale (MAAS) 304 Mitsopoulos, E. 3 mobile phones 76, 203, 204, 205, 328 modelling behaviour 6-9 Molloy, R. 305

Monash University Accident Research Centre see MUARC Monteagudo, M.J. 97 mood and driving performance 92-93, 98, 102, 404, 405, 410 see also emotion moped riders 399-410 accident rates of 399, 404-406, 405, 407, 408 socio-economic factors in 407, 408, 409 adolescents as 399-401 family background 406, 407, 408, 409 study method for 401-402 study results for 402-408 and suicide risk 399, 400, 401, 402, 406-409, 407, 410 Moray, N. 305 Morgan, D.L. 429 mothers, and children’s driving 7, 8 motorcyclists 93, 257, 312, 340, 353, 381 in crashes 425 on bends 419, 425 mortality rates 399, 404-406, 405, 415 while overtaking 419-420 rider motivation 415-418, 437 enjoyment and ﬂow 416-418, 417 enjoyment types 415-416, 419, 420, 421, 431-432, 433-434 and speed limits 435-437 speeding and 425-440 and bike characteristics 432-433, 434 competitiveness factor 434, 438 countermeasures for 438-439 discussion themes in study 429-430, 430 distal/proximal factors 433-435 research process in study 429-430, 431 and road characteristics 432 study aims 428 study method 428-430 study results 430-437 time-saving factor 435 and weather 433, 437

Index and task demand/difﬁculty 417, 418421, 418 factors inﬂuencing 432-435 task-difﬁculty homeostasis 426-427, 426, 427, 430-432, 437 training 415-421 skills/behavioural 420-421 see also mopeds Mouloua, M. 303 MUARC (Monash University Accident Research Centre) 379, 380, 381, 382, 388 Mulder, M. 256-257 multiple-vehicle crashes 129 multitasking 76, 203, 204-205, 210 Murray, W. 176, 227 Nada-Raja, S. 83 Nagin, D.S. 328 navigation systems 19, 27, 44, 45, 45, 291306 distraction caused by 292 future research on 303-306 attenion/trust/complacency tests 304-305 map use/updating on 297, 297, 300, 301 misdirection by 293, 295-297, 296, 298299, 301-303 and sustained attention 301-302, 304 over-reliance on 301-303, 304-305 study method for 294 study results for 295-299, 295-300 surveys of 291-294, 300 user demographics 292, 295, 295, 300301, 306 near accidents 92, 96-97, 101, 130 and anger 107, 108, 110, 111, 112, 113 Neighbors, C. 95-96, 100 Netherlands 94 periodic Road Safety Surveys in 39 practical driving test in 37-48 responsive driving tests in 251-262 neurological factors 349, 363, 417-418 see also cognitive functions New Zealand 67-72, 75, 156, 351 angry driving study in 109 telemetric data tracking study in 17-28 Newman, S. 176

451

NFOVD simulator 265, 266, 269, 271-274, 273 night cruises 8-9 nightshift working 329, 332 non-driving road users 65-67 vulnerability of 65-66 Northern Ireland 389-396 injury/death ﬁgures for 391 novice drivers see young drivers Nutter, A. 99 oblique rotation 239 occupational factors 76 OECD Young Driver report (2006) 3, 32 Offer, D. 94 OHS (occupational health & safety) 176177, 184-185, 227 Oigenblick, L. 151 one-parameter logistic model (OPLM) 261 Ontario 75-85 OPLM see one-parameter logistic model Osberg, T.M. 237 other-blame 100 overtaking 68, 176, 180-181, 180, 182, 183, 236 and ISA 317 in simulator training 283, 283, 286 PADS (Propensity for Angry Driving Scale) 107-115 reliability of 107-108, 110, 114 study method for 109-110 study results for 110-111 variables in 108, 111, 111, 112 Pajo, K.B. 202 Pak, A. 271-272, 271-274 Parasuraman, R. 302, 305 parents 400, 406 and child seat belt laws 392-394, 392394, 395 and drink driving 9 fathers/sons’ driving behaviour 3, 6-8 inﬂuence on children’s driving 3, 6-9, 13, 77 distancing behaviour 10 mothers and children’s driving 7, 8 and road safety 66 unsafe driving by 13 Park, G.D. 265

452

Driver Behaviour and Training – Volume III

Parker, D. 32, 68, 93, 94, 100, 130, 132, 201, 202, 203, 327 parking 18, 41-42, 46, 46 Parkinson, B. 99, 100 passengers inﬂuence driving behaviour 3, 12, 117, 344 see also children Pastor, G. 97 PBT (Problem Behaviour Theory) 66, 72, 77 PCA (principal components analysis) 238240, 243 Peck, R.C. 327 pedestrians 66, 67, 95 and ISA 311 in simulator training 265, 268, 279, 280, 281, 286, 287 peer groups 3, 8-9, 38, 77 and road safety 66 peer pressure 11-12, 32 Pelz, D.C. 330 personality proﬁles 370 personality/attitude 51-60 and aggression 117-118 and non-driving road users 66 of pre-drivers 67 in predicting risk 75-77 and trafﬁc offences 51-60 and driving exposure 51, 52-53, 55, 56-58, 59-60, 59 limitations of research 52 statistical analysis 54 study method 53-54 study, participants in 53 study results 54-58 types 52, 66, 68 see also aggression; psychopathology Persson, L. 93 Philipchalk, R. 370 planned behaviour theory see TPB Plocher, T. 303 Podd, J.V. 98-99 Pogarsky, G. 328 police drivers 240 POMS (Proﬁle of Mood States) 93 Portugal, moped study in 399-410 pre-driving attitudes 65, 67 interventions in 71 study method 67-69

study results 69-71 pre-licence training 17 principal components analysis see PCA Privilege Insurance 292 Problem Behaviour Theory see PBT problem driving risk 75-85 predictability of 75-77, 83-85 screening instruments 78-79, 85 study method 77-78 study results 80-83 professional drivers 97, 143-151 accident risk of 143, 149-151, 167-168, 169, 170, 175, 202-203, 227-228 costs of crashes by 175, 203, 227 DBQ and 202-203 driver behaviours of 144-145 error/violation rates 146-149 positive 145, 146, 148, 149 risky see under risky behaviour driver education for 155-164 effectiveness of 155 and stages of change model 155157, 162-163 study method 157-158 study results 158-162 and physical activity 167-172 self-reporting of 168 study method 168-169 study results 169-170, 170 safety culture of 144, 149-151 measurement scale 146, 147 regression analysis 149, 150 study method 145-146 study results 146-149, 147-148 see also ﬂeets/ﬂeet drivers Proﬁle of Mood States (POMS) 93 Propensity for Angry Driving Scale see PADS psychometrics 32, 107-108, 314 psychopathology 399, 400, 401, 402, 406409, 407, 410 PTW (powered two-wheelers) see moped riders; motorcyclists PVT (Psychometer Vigilance Task) 367, 368-369, 371 qualitative/quantative research 428 Quenault’s driving test 313 Quimby, A.R. 330

Index Rabbitt, P. 129, 135 Raby, M. 20, 27 Rachman theory 100 Ragan, K.M. 107, 114 RBQ (Road Behaviour Questionnaire) 401, 402 reaction speed 254 reactive driving 340-341 rear-view mirror 287, 403, 404 Reason, J.T. 93, 144, 149, 201, 327 Regan, M.A. 3 Regional Development Strategy 395 reliance 302 responsive driving, training/testing 251-262 CBT in 252, 258, 259-262 change detection/action selection 252253, 256, 259 competence assessment 256-258 constituents of 256-257 eclectic model of 257, 258 methods of 258 purposes of 257-258 data analysis in 258, 261-262 and Groeger’s model 253-256, 254 instruments used in 260-262 item selection 260-261 participants in 261 and situation awareness 255-256 Reyes, M.L. 20, 27 Reyner, L. 328 Richards, T.L. 94-95, 126 Rimmo, P. 202 risk awareness/perception 251, 352, 353 driver’s feeling of 349-363 and loss of control probability 361, 361, 363 and maximum speed 362 safety margin 350, 363 study method for 354-356 study, participants in 355 study results for 356-362 and task difﬁculty 350-353, 351, 352, 356-357, 357, 358, 359360, 359, 362 and TCI 349, 350, 352 management 18, 158, 159, 161, 162-163 threshold 426-427, 426, 427, 433-434 risky behaviour 3, 8-9, 12, 18, 27

453

and adolescents 400, 403, 403, 404 and anger 107, 113, 118, 344 countermeasures for 345 deﬁned 94-95 of non-driving road users 66 and norms 4, 12-13, 56 personality/attitudinal factor in 52, 54, 56, 56, 57, 344-345 and pre-drivers 67-68, 72 of professional drivers 175-185, 344 measurement tools for 175-177, 178, 179, 180-182, 180 self-reporting by 175, 176, 177, 181, 185 study implications/difﬁculties 184185 study method for 177-179 study results for 179-182 psychosocial models of 77 see also drink driving; speeding Roach, G. 371 Road Behaviour Questionnaire see RBQ road lighting 351, 385 road rage 99, 100 road safety 65-66, 67, 71, 96-97, 117 and ISA 312, 315, 318, 319 strategic partnerships in 387 work-related 180 change in 155-164 Road Safety Strategy 2004-6 345 road width 351 Roberts, K. 288 Roelofs, E.C. 257 Rosenthal, T.J. 265 Rothengatter, T. 295 rumble strips 330 Rundmo, T. 72 rural driving 26-27, 69, 178 and vigilance 369-370 rush based enjoyment 415, 416, 419, 420, 421, 433-434 rush hour 117 Ryan, R.M. 304 Safety Climate Questionnaire see SCQ-MD Salminen, S. 143 Sanders, P. 257 SART (sustained attention to response task) 302, 304

454

Driver Behaviour and Training – Volume III

satnav see navigation systems SCQ-MD (Safety Climate QuestionnaireMD) 175, 176, 177, 178, 180, 181-182 seat belts 67, 68, 69, 72, 76, 204-205 and booster seats 393, 394, 396 hazardous to children 395 laws 215, 224, 389-396 attitudes towards 394, 394 need for 391 objective of 390-391 study method for 392 study results for 392-394, 392-394 see also under taxi drivers Segal, A.U. 77, 80 self assessment 20, 21, 34-35, 39, 42-43, 47, 237, 251 self deception 237 see also driver conﬁdence self-image 4, 18, 27, 32 self-reporting 52, 59, 60, 69, 72, 79, 83, 114, 123-124 Senior, V. 68 sensation seeking 52, 53, 66, 72, 76, 132, 133-134, 136, 327 in ﬂeet drivers 236, 240, 240, 242 inventory of (AISS) 68-70 in motorcyclists 415-418, 417 in non-driving road users 66 and other risk factors 331, 331, 332 sex differences in 69, 71 see also enjoyment Sexton, B. 419 Shaffer, J.P. 239 Sheridan, T. 302 shift work 329, 332 Shinar, D. 326 Shrauger, S.J. 237 siblings, inﬂuence driving behaviour 8-9, 10-11 Sigman, N. 371 signiﬁcant others and young drivers 3-14, 344 distancing behaviour 10-13 and drink driving 9 and driver education 13-14 modelling behaviour 6-9, 13 and risk taking 12, 13 as role models 4

study methods for 4-6 Simpson, H.M. 271-272, 271-274 simulation training/testing 265-275, 266, 351 and accident rates 265 of accident scenarios 277-288 response times 281-286, 281-284 study method for 278-280, 286 study results 280-286 driving scenarios in 267-268, 279-280 efﬁcacy of 240, 266-267, 273-275, 287288, 305 and ISA 311 pedestrians in 265, 268, 279, 280, 281, 286, 287 performance measures in 268 study method 267-268 study, participants in 268, 269 study results 269-273, 270 trafﬁc lights in 265, 267, 281, 282, 286, 287 vigilance test 368-375 simulator sickness 305 Singh, I.L. 305 Sinnett, E.R. 239 situation awareness 255-256, 304 situational learning 3 Sivak, M. 326 Sjöberg, L. 93 skid training 18, 93 sleep 328-330, 331 countermeasures for 330 disorders 328 see also fatigue sleepy driving 329, 330-331, 332 Sluiter, J.K. 167-168, 171 smoking 169 social cognition 4 social information processing theory 100 speed cameras 326, 328, 339 speed choice 351-353 and loss of control probability 361, 361, 363 and road width 351 safety margin 350, 363 study method for 354-356 study, participants in 355 study results for 356-362

Index and task difﬁculty 349, 351, 351, 356357, 357, 358, 359-360, 359, 362, 419-420 speed humps 326 speed limits 20, 23, 318, 319, 435-437 speeding 3, 6-8, 12, 37, 76, 190 and aggression 124, 124 countermeasures for 325-326 and emotion 94, 110, 332, 342, 343-344 and ﬂeet drivers 176, 179, 180, 182, 183, 210, 232 and guilt/shame 435-436, 437 as immature behaviour 344-345, 425, 438 and ISA 312, 318, 319317 measuring 325 and norms 56, 57, 68, 70, 183 and other risk factors 331, 331, 332 in personality study 53, 56, 66, 68 and task-difﬁculty homeostasis 337-338, 338, 342 telemetric data tracking of 20, 22, 25, 27 and time-saving 343 TPB- 67, 68, 69, 70 see also driver celeration; and see under motorcyclists Spielberger, C. 99-100 Stacey, S. 33 stages of change model 155-157 ﬁve stages in 162-163 State Trait Anxiety Inventory 93 Stephens, A.N. 92 Stetzer, A. 151 stop-signal paradigm test 78 stopping manoeuvre 42 Stork, J. 401, 409 Stradling, S. 32, 68, 93, 94, 100, 129, 144, 201, 327, 343, 344 stress 32, 52, 91 age factor in see under age factors coping with 131, 132-137, 236 measurement of 132 strategies 133-137 training in 245 measuring 132 in professional/ﬂeet drivers see under ﬂeet drivers transactional models of 100, 235-236

455

suicide risk 399, 400, 401, 402, 406-409, 407, 410 Sullman, M.J. 202, 203 Summala, H. 94, 130, 202 Sumner, R. 326 sustained attention 301-302, 304 Sutcliffe, P. 130 Svahn, F. 292, 293, 300 Sweden 42, 132, 143, 311 celeration study in 189-196 SWOV (Scientiﬁc Research of Trafﬁc Safety) 252 Taieb, M. 326 tailgating 2, 120, 125, 236 task demand/difﬁculty 349-353, 350, 415416, 418 and motorcyclists 417, 418-421, 426427, 426, 427 Task-Difﬁculty Homeostasis Model 337346, 350-353 acceptable task difﬁculty in 338, 343 calibration problem 339-340 comparator in 351, 352, 353, 426 determinants in 338 driver competence in 338-339 human factor variables in 338, 341-342, 346 model described 337-339, 338 and motorcyclists 425, 426-427, 426427, 429, 430-432, 437 perceived capability in 337, 338-339, 338, 342, 345-346 perceived demands in 337, 338, 338, 339 and reactive/anticipatory driving 340341 risk threshold in 337, 338, 339, 343-345, 346 and simulator driving 340 and speed 337-338, 338, 342, 343 taxi drivers/taxis 97, 145 seat belt laws for 389, 390 seat belt study 215-224 private car use 217, 222, 222 reasons for not using 215, 216, 217220, 218-219, 221, 222-223 self-reporting in 224 study method 216-217

456

Driver Behaviour and Training – Volume III

study results 217-223 Taylor, A.H. 167, 171 Taylor, J.E. 98-99 TCI (Task Capability Interface) 349, 350, 352, 362-363 TCI (temperament & character inventory) 79 telemetric data tracking 18-28 map-based function 23, 27 participants in study 18-19 results 20-23, 24-26 study method 18-20 tracking system 19-20, 19, 23 tension 66, 77, 98 thematic analysis 428-429 three pathways theory 100 thrill seeking see sensation seeking Tillman, W.A. 77 time management 343, 439 time pressure 235, 240, 343, 435, 439 TNO 252 TPB (Theory of Planned Behaviour) 100, 120, 124 TPB-speeding 67, 68, 69, 70 trafﬁc control devices 265, 267, 281, 282, 286, 287 trafﬁc jams 97 trafﬁc violations 109, 110, 331 age factor in 129, 132 and aggression 124, 124 countermeasures for 327-328 and driving exposure 55 and emotion/mood 93-94, 96, 101, 111, 111, 112, 114 and other risk factors 331, 331, 332 personality/attitudinal factors in 51-60, 55, 56-58, 57 and social norms 2 see also speeding trait anger 99-100, 102, 110 Scale (TAS) 109, 111, 111, 112, 114 Transport, Dept of, study (2007) 11 Transport Research Laboratory 39-40 Truman, W. 419 trust/over-trust 302-303, 304-305 Turkey, seat belt study in 215-224 turning manoeuvre 41, 46, 46 Twisk, D. 408, 420

Ulleberg, P. 72 UN 65 Underwood, D. 96-97 Underwood, G. 99, 102, 288 United States 75, 143, 295 GDL in 266 seat belt use in 215 simulator training in 265-275 Vallerand, R.J., 118 Vallières, E.F. 118 van der Beek, A.J. 167-168, 171 Varimax analysis 239, 240 Varonen, U. 144 vehicle control skills 18 Velicer, W.F. 238 Vienna 312, 313 Vietor, N.A. 95-96 vigilance 367-375 age factor in 370, 372, 372, 373-374, 374 and BMI 370, 372, 375 gender factor in 370, 372, 372, 373, 373, 374-375 low, impact of 371, 371 and performance metrics 371, 372-373, 374 reaction times, analysis of 372-373, 372, 373 study method 368-372 study results 372-374 subjective factors and 367, 370, 371, 373, 374 test (PVT) 301-302, 367, 368-369, 371 three factors of 367 Vincenzi, C. 303 Von Thaden, T.L. 144 Vuuren, W. 151 Ward, P. 419 Wasielewski, P. 326 Watson, B. 176 Watts, G.R. 330 weather conditions 51, 433, 437 Weiner, E.L. 302 Wellbrink, J. 371 West, R. 66, 72, 78 Westerman, S.J. 129

Index WFOVC/WFOVD simulator 265, 266, 269, 271, 271-274, 273-274 White, G.S. 114 Whitmore, 34 Wiegmann, D.A. 144 Wiener Fahrprobe method 311, 312-314 observation variables in 313-314 Wilde, Gerald 76 Williamson, A. 371 Wills, A.R. 176 work-related driving see professional drivers World Health Day (WHD) 65 World Health Organisation (WHO) 65 Wright, S. 96-97, 102 Wundersitz, L. 66, 72 Xie, C. 203 Yagil, D. 96 Yerkes, R.M. 342 Yerkes-Dodson Law 342 young drivers accident rates of 269-274, 270-274 coping strategies of 133-137, 135 in crashes 20-22, 27, 35, 37, 51, 265, 266-267

457

deaths 65, 75 and Early Driver Training 277-278 gender factors see under gender factors and GRSW 65 post-training performance of 17-28 study method 18-20 study results 20-21, 24-26 tracking system 19-20, 19, 23 reactive responses of 340-341 risk-taking by 400, 403, 403, 404 and signiﬁcant others see signiﬁcant others in simulation training see simulation training/testing speeding by 27 see also mopeds YRBS (Youth risk behaviour surveillance system) 79, 81, 84 Zhang, H. 144 Zohar, D. 144 Zyda, M. 371 Zylman, R. 59