European Review of Social Psychology (Volume 12)

EUROPEAN REVIEW OF SOCIAL PSYCHOLOGY VOLUME 12

EUROPEAN REVIEW OF SOCIAL PSYCHOLOGY Editorial Board Dominic Abrams Michael Diehl AliceEagly Klaus Fiedler Kenneth Gergen Michael A. Hogg Charles M. Judd Arie W. Kruglanski Anne Maas Antony Manstead David Mesick Gerold Mikula John Rijsman BemardRime Fritz Strack Thomas A. Wills

University of Kent, UK Universitiit Tiibingen, Germany Northwestern University, USA Universitiit Heidelberg, Germany Swarthmore College, USA University of Queensland, Australia University of Colorado, Boulder, USA University of Maryland, USA Universita di Padova, Italy Universiteit van Amsterdam, The Netherlands Northwestern University, USA Universitiit Graz, Austria Tilburg University, The Netherlands Universite Catholique de Louvain, Belgium Universitiit Wiirzburg, Germany Ferkauf Graduate School, USA

Editors

Wolfgang Stroebe Miles Ht!WStone

Utrecht University, The Netherlands Oxford University, UK

About the editors

Wolfgang Stroebe has published widely on the topics of attitudes, group processes and health psychology. A former President of the European Asation of Experimental Social Psychology and a fellow of the British Psychological Society, the Society for the Psychological Study of Social Issues, and the Society of Personality and Social Psychol­ ogy, he is currently Professor of Social, Organizational and Health Psychology at the University of Utrecht (The Netherlands). Miles Hewstone has published widely on the topics of social cognition and intergroup relations. He was awarded the British Psychological Society's Spearman Medal in 1987 and its Presidents' Award for Distinguished Contributions to Research in 2001 . He was a Fellow at the Center for Advanced Study in the Behavioral Sciences, Stanford, California from 1987-1988 and 199200, and is an Academician of the Academy of Learned Societies in the Social Sciences. Having held Chairs at the Universities of Bristol, Mannheim and Cardiff, he has recently joined the Department of Experimen­ tal Psychology at the University of Oxford and is a Fellow of New College.

Copyright © 200 2 by John Wiley & Sons Ltd, Baffins Lane, Chichester, West Sussex P01 91 UD, England National 01 243 77 9777 International (+4) 1 243 77 97 77 e-mail (for orders and customer service enquiries): cs-book @wiley.co.uk Visit our Home Page on http:Hwww.wiley.co.uk or http:Hwww.wiley.com Al rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except under the terms of the Copyright, Designs and Patents Act 1 988 or under the terms of a licence issued by the Copyright Licensing Agency, 90 Tottenham Court Road, London W1P 9HE, U K, without the permission in writing of the Publisher. European Review of Social Psychology

Published annually by John Wiley & Sons

British Library Cataloguing in Publication Data:

European review of social psychology. Vol 1 2, 200 2. - (European review of social psychology) I. Stroebe, Wolfgang II. Hewstone Miles Ill. Series 30 2.094 ISBN 0-471-48675- 2 Typeset in 10/1 2pt Times by Dorwyn Ltd, Rowlands Castle, Hants Printed and bound in Great Britain by Antony Rowe Ltd, Chippenham This book is printed on acid-free paper responsibly manufactured from sustainable forestry, in which at least two trees are planted for each one used for paper production.

vi

Chapter 8

CONTENTS The Self in Relationships: Whether, How and When Close Others Put the Self "in Its Place"

Constantine Sedikides, W. Keith Campbell, Glenn D. Reeder and Andrew J. Elliot Chapter 9

231

Prospects, Pitfalls and Plans: A Proactive Perspective on Social Comparison Activity

Lisa G. Aspinwall, Douglas L. Hill and Samantha L. Leaf Chapter 10

261

Social and Cognitive Influences in Group Brainstorming: Predicting Production Gains and Losses

Paul B. Paulus, Karen Leggett Dugosh, Mary T. Dzindolet, Hamit Coskun and Vicky L. Putman Author Index Subject Index

299 327 341

Contributors

G. AsPINWALL, Department of Psychology, University of Utah, 380 South 1530 East, Room 502, Salt Lake City, UT 84112-0251, USA

LISA

Department of Psychology, 1415 Jayhawk Blvd, University of Kansas, Lawrence, KS 66045-7556, USA

MoNICA BIERNAT,

R. BRANSCOMBE, Department of Psychology, 1415 Jayhawk Blvd, Uni­ versity of Kansas, Lawrence, KS 66045-7556, USA

NYLA

W. KEITH CAMPBELL, Department of Psychology, University of Georgia, Uni­ versity Plaza, A tlanta, GA 30303, USA Department of Psychology, University of Padova, Via Vene­ zia 8, Padova 35131, Italy

LUIGI CASTELLI,

MARK CoNNER,

School of Psychology, University of Leeds, Leeds LS2 9JT,

UK HAMIT CosKUN,

Department of Psychology, Abant Izzet Raysal University,

Bolu, Turkey Department of Psychology and Human Ecology, Cameron University, Lawton, 0K 73505, USA

KAREN LEGGETT DuGOSH,

T. DziNDOLET, Department of Psychology, Box 19528, University of Texas at Arlington, A rlington, TX 76019, USA

MARY

viii

CONTRIBUTORS

Department of Psychology, University of Rochester, Wilson Boulevard, Rochester, NY 14627, USA

ANDREW J. ELLIOT,

EssEs, Department of Psychology, University of Western Ontario, London, Ontario, Canada N6A 5 C2

VICTORIA M.

Department of Psychology, University of California, Santa Barbara, CA 93106, USA

DAVID L. HAMILTON,

Department of Psychology, University of Maryland, College Park, MD 20742-4411, USA

DouGLAS L. HILL,

Department of Psychology, University of Utah, 380 South 1530£, Room 502, Salt Lake City, UT 84112-0251, USA

SAMANTHA L. LEAF,

School of Psychology, Cardiff University, PO Box 901, Cardiff CFJ 3YG, UK

GREGORY R. MAIO,

School of Psychology, Victoria University of Wellington, PO Box 600, Wellington, New Zealand

JoHN McCLURE,

Department of Psychology, Box 19528, University of Texas at Arlington, Arlington, TX 76019, USA

PAUL B. PAULUS,

Department of Psychology, Box 19528, University of Texas at Arlington, Arlington, TX 76019, USA

VICKY L. PuTMAN,

Department of Psychology, Illinois State University, Urbana, IL 61801, USA

GLENN D. REEDER,

Department of Psychology, 1415 Jayhawk Blvd, Uni­ versity of Kansas, Lawrence, KS 66045-7556, USA

MICHAEL T. ScHMITT,

Department of Psychology, University of South­ ampton, Highfield Campus, Southampton SOJ7 JBJ, UK

CoNSTANTINE SEDIKIDES,

Department of Psychology, University of Sheffield, Sheffield SJO 2TN, UK

PASCHAL SHEERAN,

Department of Psychology, Indiana University, Terre Haute, Indiana, IN 47809, USA

STEVEN J. SHERMAN,

Acknowledgments

We would like to thank the following reviewers who helped us and the authors to shape these chapters into their final versions; G. Abakoumkin, D. Abrams, C. Armitage, D. de Ridder, N. de Vries, M. Diehl, A. Eagly, K. Fiedler, F. Forsterling, A. Haslam, D. Hilton, M. Hogg,

K. Jonas, M. Lalljee, H. Lodewijkx, A. Maass, B. Major, A. S. R. Manstead, B. Nijstad, and T. A. Wills.

Chapter 1

Intention-Behavior Relations: A Conceptual and Empirical Review Paschal Sheeran University of Sheffield, UK

ABSTRACT This chapter addresses two questions; how big is the "gap" between intentions and behavior, and what psychological variables might be able to "bridge" the intention-behavior gap? A meta-analysis of meta-analyses is used to quantify the gap and a conceptual analysis of intention-behavior discrepancies is pre­ sented. Research is described on the extent to which four groups of variables­ behavior type, intention type, properties of intention, and cognitive and person­ ality variables-moderate intention-behavior relations. Finally, the scope of the intention construct is discussed in the light of recent evidence concerning the role of habits and automaticity in human behavior.

Several social psychological models, including the theory of reasoned action (Fishbein, 1980; Fishbein & Ajzen, 1975), the theory of planned behavior (Ajzen, 1985, 1991), Triandis's (1980) attitude-behavior theory, and protec­ tion motivation theory (Rogers, 1983) converge on the proposal that the most immediate and important predictor of a person's behavior is his/her intention to perform it. The assumption is that people do what they intend to do and do not do what they do not intend. This raises two questions: (a) how well do intentions predict behavior? and, (b) what determines how well intentions Correspondence regarding this chapter should be addressed to: Paschal Sheeran, Department of Psychology, University of Sheffield, Sheffield SlO 2TN, UK. E mail: [email protected] European Review of Social Psychology, Volume

© 2002 John Wiley &

Sons Ltd.

12. Edited by Wolfgang Stroebe

and Miles Hewstone.

2

PASCHAL SHEERAN

predict behavior? The purpose of this chapter is to review recent research that has attempted to answer these questions. The chapter is divided into five sections and begins by defining behavioral intentions and briefly describing their status in models of attitude-behavior relations and theories of health-related behavior. Next, the strength of the intention-behavior relation is quantified in a meta-analytic synthesis and the role of measurement artifacts in attenuating observed associations is dis­ cussed. The third section presents a conceptual analysis of the intention­ behavior "gap". Evidence concerning four factors that seem to influence the degree of consistency between intentions and behavior is then reviewed. These factors are behavior type, intention type, properties of behavioral in­ tentions, and personality and cognitive variables. Finally, the scope of the intention construct is discussed in the light of recent research on automaticity and habits.

BEHAVIORAL INTENTIONS AND SOCIAL PSYCHOLOGICAL MODELS OF BEHAVIOR

"Behavioral intentions are instructions that people give to themselves to be­ have in certain ways" (Triandis, 1 980, p. 203). They are people's decisions to perform particular actions. Intentions can be inferred from participants' re­ sponses that have the form, "I intend to do X", "I plan to do X", or "I wil do X". In psychological terms, a behavioral intention indexes a person's motivation to perform a behavior. That is, behavioral intentions encompass both the dir­ ection (to do X vs. not to do X) and the intensity (e.g., how much time and effort the person is prepared to expend in order to do X) of a decision. The intention construct lies at the heart of theories of attitude-behavior relations and models of health-related behavior (see e.g., Abraham, Sheeran, & Johnson, 1 998, for review). According to the theory of reasoned action, intentions are the proximal predictors of behavior and mediate the influence of both the theory's predictors (attitudes and subjective norms) and extra­ neous variables (e.g., personality) on behavior. Likewise, protection motiva­ tion theory assumes that behavioral intentions mediate the relationship between the putative predictors (threat appraisal and coping appraisal) and behavior. The theory of planned behavior also assumes that intentions are the most important predictor of behavior but acknowledges that people may not always have sufficient control over performing the behavior to actually enact their intentions (Ajzen, 1 985, 1 991 ). Similarly, Triandis (1980) suggests that "low facilitating conditions" could prevent performance of an intended be­ havior. Triandis also points out that certain behaviors are more likely to be controlled by "habits" than by conscious intentions-an issue that is con­ sidered in the final section. For present purposes, it is sufficient to say that all

THE INTENTION-BEHAVIOR GAP

3

of these models agree that if a researcher wants to know how people will behave, then the best way to find out is to ask people how they intend to behave. HOW WELL DO INTENTIONS PREDICT BEHAVIOR?

Intentions have been used to predict an impressive range of behaviors, includ­ ing consumer and leisure decisions (e.g., Warshaw & Davis, 1 984), diet (e.g., Conner & Sparks, 1 996), physical activity (e.g., Norman & Smith, 1995; Sheeran & Orbell, 2000a), weight loss (e.g., Bagozzi & Warshaw, 1990), smo­ king (Norman, Conner, & Bell, 1999), academic activities and achievement (e.g., Manstead & van Eekelen, 1998; Sheeran, Orbell, & Trafimow, 1999), sex, contraception and abortion (e.g., Davidson & Morrison, 1983; Smetana & Adler, 1980; Warshaw & Davis, 1984), illicit drug use (e.g., Conner, Sherlock, & Orbell, 1998), gambling (e.g., Sheeran & Orbell, 1999b), voting (e.g., Bassili, 1995), occupational choice (LaRocco, 1983), driver behavior (Parker, 1997), attendance behaviors (e.g., Sheeran & Orbell, 2000b), whether or not to have a child (e.g., Vinokur-Kaplan, 1978), and prosocial behaviors such as blood donation (e.g., Warshaw, Calantone, & Joyce, 1986). The intention-behavior relationship has been the subject of both narrative (e.g., Ajzen, 1991) and meta-analytic (e.g., Sheeran & Orbell, 1 998) reviews. A literature search (January 2000) using computer databases and informal sources (see Mullen, 1989) revealed 10 meta-analyses that reported the total number of participants included in the analyses (n ), the number of hypotheses studied (k), and the sample-weighted average correlation (r+ ) (five meta­ analyses could not be included because n was not reported: Godin & Kok, 1996; Notani, 1998; Ouelette & Wood, 1998; Ryan & Bonfield, 1975, and Zimmerman & Vernberg, 1994). The 10 meta-analyses include reviews of the theory of reasoned action (e.g., Sheppard, Hartwick, & Warshaw, 1988), the theory of planned behavior (Armitage & Conner, in press) and protection motivation theory (Milne, Sheeran, & Orbell, 2000), as well as reviews of specific behaviors (e.g., condom use; Sheeran & Orbell, 1998), and all involve prospective measures of behavior. Since these meta-analyses overlap quite modestly, it is possible to conduct a meta-analysis of meta-analyses in order to provide a precise answer to the question of how well intentions predict be­ havior in previous research. Table 1 . 1 presents findings from each of the meta-analyses included in the review. Correlations ranged from 0.40 to 0.82 and had a standard deviation of 0.12. The sample-weighted average correlation derived from these studies was 0.53 with a 95% confidence interval from 0.52 to 0.53 based on 422 hypotheses and a total sample size of n 82,107. This means that intentions account for 28% of the variance, on average, in prospective studies. =

4

PASCHAL SHEERAN

Table Ll

A meta-analysis of meta-analyses of the intention-behavior relationship n

Authors

k

R2

r

48 32 18 4 98

0.47 0.47 0.82 0.40 0.45 0.44 0.51 0.53 0.57 0.62 0.53

Armitage & Conner (in press) Hausenblas et al. (1997) Kim & Hunter (1993) Milne, Sheeran, & Orbell (200) Randall & Wolff (1994) Sheeran & OrbeD (1998) Sheeran & Sutton (1999) Sheppard, Hartwick, & Warshaw (1988) Trafimow et al. (1999) Van den Putte (1993)

7318 10621 5216 432 26906 2532 6301 1 1 566 1475 9740

40 87 9 58

0.22 0.22 0.67 0.16 0.20 0.19 0.26 0.28 0.33 0.38

Overall

82107

422

0.28

28

But is explaining 28°k of the variance in behavior good or bad (cf. Sutton, 1998}? It may appear bad-after all, 72°k of the variance has not been ex­ plained. However, as Hunter and Schmidt (1990) pointed out: The "percent variance account[ed] for" is statistically correct but substantively erroneous. It leads to severe underestimates of the practical and theoretical significance of relationships between variables. This is because R2 (and al other indices of percent variance accounted for) are related only in a very non-linear way to the magnitudes of effect sizes that determine their impact in the real world. (p. 199).

There are alternatives to percentage variance accounted for as a measure of effect size which provide for more accurate interpretations regarding how strongly intentions are related to behavior. As an illustration, consider the example (from Sutton's, 1998, seminal review of the intention-behavior rel­ ationship) of an intervention and a control group (n 100 in both cases) who have success rates of 70% and 30o/o, respectively, at giving up smoking. The difference in success rates for the two groups is 40°/o, the relative success rate is 2.3, and the odds rate is 5.4. These indices all suggest that the intervention was very successful in helping people to quit smoking (see also Abraham, Sheeran, & Orbell, 1998, for a discussion of Rosenthal & Rubin's, 1979, Binomial Effect Size Display). However, when these findings are considered in terms of percentage variance accounted for, the result is only R2 0.16-­ this value appears much less impressive. Given such difficulties in interpreting percentage variance, it would be valuable to have standards that one could use to judge how big are particular effects. Fortunately, Cohen's (1992) power primer provides a useful guide to interpreting correlations and R2• According to Cohen's analysis, r+ 0.10 is a "small" effect size, r+ 0.30 is "medium", and r+ 0.50 is "large". The =

=

=

=

=

5

THE INTENTION-BEHAVIOR GAP

sample-weighted average correlation of 0.53 obtained here is, therefore, a "large" effect size. Consequently, explaining 28% of the variance should probably be considered "good". Measurement Artifacts and Attenuation of Intention-Behavior Correlations

It is also the case that r+ 0.53 underestimates the "true" relationship be­ tween intentions and behavior. This is because a variety of measurement artifacts attenuate the observed association. Kim and Hunter ( 1 993) provided a dramatic illustration of how serious the attenuation produced by just two such artifacts could be. The uncorrected average correlation between inten­ tions and behavior in their meta-analysis was r+ 0.46 (only slightly lower than the value obtained here). After correction due to (artificial) dichotomiz­ ation of measures of intentions and/or behavior, the average correlation in­ creased to 0.54. However, when the effects of both dichotomization and measurement error (lack of reliability in measures of intentions and behavior) were corrected, r+ increased to 0.82. Dichotomization and measurement error are just two of the artifacts that reduce intention-behavior correlations. There have also been empirical demon­ strations of attenuation due to: (a) lack of compatibility between intentions and behavior in terms of the action, target, context, and time specified in the two measures (Ajzen & Fishbein, 1973, 1974); (b) lack of correspondence between the scales used to measure intentions and behavior [e.g., when a dichotomously­ worded intention measure ("Do you intend to use a condom in the next month?") is used to predict a frequency measure of behavior ("How often did you use a condom in the last month?"), see Coumeya, 1994]; and (c) the marginal distributions of the intention and behavior measures do not match (see Sutton, 1998). In sum, researchers should carefully scrutinize their data for measurement artifacts before concluding that small or non-significant correla­ tions indicate a lack of relationship between intentions and behavior. =

=

WHY IS THERE A "GAP" BETWEEN INTENTIONS AND BEHAVIOR? A CONCEPTUAL ANALYSIS

Notwithstanding attenuation due to dichotomization, measurement error, compatibility, scale correspondence, and mismatched marginal distributions, even the most optimistic estimates of intention-behavior consistency are far from perfect (r 1 .00). This suggests that there is more than sub-optimal measurement involved in discrepancies between intentions and behavior. It can also be argued that correlations tend to obscure the nature of the intention-behavior "gap" (Orbell & Sheeran, 1998). This is because correla­ tional (and regression) analyses really only tell us that positive intention =

6

PASCHAL SHEERAN

scores are related to greater likelihood of behavioral performance than nega­ tive intention scores (and the converse; that negative intention scores are related to less likelihood of behavioral performance than positive intention scores). However, it is helpful to decompose intention-consistency into a 2 (positive intention vs. negative intention) x 2 (performance vs. non­ performance of the behavior) matrix (McBroom & Reid, 1 992; Orbell & Sheeran, 1998). So doing means that it is possible to discern the sources of both consistency and discrepancy between intentions and subsequent action (see Table 1.2). Table 1.2

Decomposition of the intention-behavior relationship

Subsequent behavior

Intention Positive

Acted Did not act

Inclined actor Inclined abstainer

Negative Disinclined actor Disinclined abstainer

A "positive" intention or being "inclined" can refer both to an intention to perform a behavior and to an intention not to perform a behavior. The issue here is whether or not the person intends to act, not whether the action to be performed involves an approach versus an avoidance response. Relatedly, "actor" and "ab stainer" refer to performance versus non-performance of the focal behavior, re­ gardless of whether that behavior involves the person doing something or avoiding doing something. Note.

According to this analysis, intention-behavior consistency can be attributed to two groups of participants, namely, those participants with positive inten­ tions who subsequently act (designated "inclined actors") and participants with negative intentions who do not act ("disinclined abstainers"). Import­ antly, it is also possible to see who is responsible for the gap-the two groups of participants who do not act according to their intentions. These are the participants with positive intentions who fail to act (inclined abstainers) and the participants who perform the behavior despite negative intentions to do so (disinclined actors). But how general is this analysis? And are inclined abstainers or disinclined actors predominantly responsible for the "gap"? Table 1 .3 presents findings from selected studies of intention-behavior rel­ ations indicating the percentage of participants with positive intentions (the inclined) who acted or did not and the percentage of participants with nega­ tive intentions (the disinclined) who acted or did not. Findings seem to indi­ cate that analysing intention-behavior consistency in terms of inclined actors, inclined abstainers, disinclined abstainers and disinclined actors has some generality. In four out of the six cases, all four cells contained entries (in fact, the two cases where this was not true both involved just a single participant who was "disinclined"; Sheeran & Orbell, 200b; Stanton et al., 1 996 ). More

7

THE INTENTION-BEHAVIOR GAP

importantly, Table 1.3 indicates that the lack of consistency between inten­ tions and behavior is mainly due to inclined abstainers rather than disinclined actors. The median percentage of intenders who failed to enact their inten­ tions was 47% whereas the median percentage of non-intenders who subse­ quently performed the behavior was 7% . Thus, it is those participants who fail to act upon their positive intentions who are mainly responsible for the intention-behavior gap. Percentages of participants with positive versus negative intentions who subsequently acted versus did not act in selected studies of intention-behavior relations

Table 1.3

Authors

Behavior

Actors

Abstainers

Actors

Abstainers

43 43 46 70 74 61

57 57 54 30 26 39

10 12 3 0 35 0

90 88 97 100 65 100

(%)

Gallois etal. (1992) Orbell & Sheeran (1998) Sheeran & OrbeU (200a) Sheeran & Orbell (200b) Sutton etal. (1994) Stanton eta/. (1996)

Condom use Cancer screening Exercise Cancer screening Cancer screening Condom use

Disinclined

Inclined

(%)

(%)

(%)

Two further points are worth noting regarding Table 1 .3. The first is that, although this analysis involves a relatively small number of studies-and fur­ ther tests are required to determine generality-there do not appear to be strong grounds for assuming that these findings are atypical, at least in terms of the strength of the intention-behavior correlations. For example, the relev­ ant correlations were 0.44 in the case of Gallois et al. (1992) and 0.51 in the case of Sheeran and Orbell (2000a). The second point is that Table 1 .3 seems to show how well-paved with good intentions is the proverbial road to hell; it is disconcerting to consider that almost one-half of participants who intended to use a condom, attend for cancer screening or exercise at a particular level subsequently failed to do so. Can Traditional Variables Discriminate between Patterns of Intention­ Behavior Consistency?

The present analysis of patterns of intention-behavior consistency begs an important question, namely, can traditional social psychological models of behavior discriminate between inclined actors and inclined abstainers? It could be the case that inclined abstainers have lower perceived behavioral control than inclined actors, which would explain why they do not translate their intentions into action ( cf. Ajzen, 1991 ). In fact, Orbell and Sheeran (1998) used variables specified by protection motivation theory (perceived

8

PASCHAL SHEERAN

susceptibility, perceived severity, worry, response efficacy, and self-efficacy) to try to distinguish between all four groups and were unable to discriminate inclined abstainers from the other groups. However, two aspects of their analyses bear comment. First, protection motivation theory has weaker pre­ dictive validity than the theory of planned behavior (Milne et a/., 200), so perhaps findings would have been different if the theory of planned behavior had been employed. Second, Orbell and Sheeran (1998) did not employ a measure of strength of intention. Thus, their study leaves open the possibility that inclined actors simply had stronger intentions to perform the behavior than did inclined abstainers. To provide a sterner test of the ability of cognitive variables to discrimi­ nate inclined abstainers from the other groups, I reanalysed data from Sheeran and Orbell's (200a) study of exercise behavior (note that this study found a "strong" correlation between intentions and behavior, r = 0.51 ). Participants whose intention scores were on the scale midpoint (4) were excluded from the analysis and the remaining sample was partitioned into disinclined ( 1-3) and inclined (5-7). Decomposition showed that there were n = 31 inclined actors, n = 36 inclined abstainers, n = 2 disinclined actors, and n = 69 disinclined abstainers. Because there were so few disin­ clined actors, these participants were excluded from further analysis. Dis­ criminant function analysis was then employed to determine whether intentions (the continuous measure), attitudes, subjective norms, and per­ ceived behavioral control could distinguish between inclined actors, inclined abstainers, and disinclined abstainers. Findings showed that the first discriminant function was statistically rel­ iable but that the second derived function was not. To aid interpretation, the group centroids were plotted and analysis of variance was used to compare the means for theory of planned behavior variables for the three groups. As Figure 1 . 1 shows, the first discriminant function clearly distinguished the disinclined abstainers from the two inclined groups. ANOVAs confirmed that disinclined abstainers had lower scores on each of the predictors than both inclined actors and inclined abstainers. However, Figure 1 . 1 also shows that the second discriminant function was not capable of distinguishing be­ tween inclined actors and inclined abstainers. ANOVAs confirmed that in­ clined actors and inclined abstainers did not differ on intentions, attitudes, subjective norms, or perceived behavioral control (see Table 1 .4). Thus, despite the fact that the theory of planned behavior would suggest that perceived behavioral control in particular should be capable of distinguish­ ing between inclined actors and inclined abstainers, this was not the case here. I suspect that most applications of the theory of planned behavior would reveal a sizable number of participants with equivalent intentions, perceived behavioral control, etc., who nonetheless differ in their subse­ quent behavior.

10

PASCHAL SHEERAN

behavior is whether the behavior being predicted is a single action or a goal (an outcome that can be achieved by performing a variety of single actions). Intentions are likely to be superior predictors of single actions, such as read­ ing a textbook or attending one's lectures, than the goal of getting an "A" in an examination. This is because getting an ''A" is dependent upon reading the textbook and attending one's lectures and other factors. In fact, the extent to which the performance of particular behaviors actually controls whether or not a goal will be achieved is the critical factor. For example, a person may have strong intentions to get an "A", and perform a range of study behaviors, but still not achieve his/her goal if the study behaviors he/she has performed do not control the outcome (e.g., because the lecturer jokingly decides to fail everyone regardless of their examination answers). There is also another sense in which control is important in determining the strength of intention-behavior relations-the person must have control over performing a behavior if the intention to perform that behavior is to be realized. Several control factors are important in determining whether an intention is translated into action. These factors include knowledge (e.g., in order to realize my intention to use a condom, I must know what a "condom" is and what to do with one), ability (e.g., my partner or I must be able to put on a condom in order to use one), resources (e.g., one needs money to buy a condom so it can be used), opportunity (e.g., one needs a sexual partner to realize the intention to use a condom), availability (e.g., a condom must be available for the intention to use one to be realized), cooperation (e.g., my partner must agree to using the condom) and unexpected situations (e.g., I find myself in a sexual situation that I had not anticipated). In sum, these factors determine the amount of control a person actually possesses over performing the behavior and should moderate the intention-behavior relationship. Very little research has been specifically concerned with demonstrating that actual control determines the extent of consistency between intentions and behavior. I suspect that this is the case for two reasons. First, it is logically impossible for certain behaviors to be performed in the absence of particular abilities, resources, and so on, and demonstrations of this logical impossibility are likely to be trivial (a possible exception is examination of the effects of unexpected situations on intention-behavior consistency where it could be valuable to know from, e.g., a service delivery viewpoint, the extent to which uncontrollable factors prevented performance of the behavior; cf. Cote, Mc­ Cullough, & Reilly, 1985). The second reason why the effects of actual control have not been studied is that researchers generally study controllable be­ haviors (cf. Fishbein & Ajzen, 1975). That is, researchers are rarely interested in behaviors that are actually impossible for their participants to perform. Indeed, it seems unlikely that many people would intend to perform be­ haviors that, in reality, they cannot perform.

THE INTENTION-BEHAVIOR GAP

11

In field studies of the theory of planned behavior, it is not generally feasible for researchers to determine how much actual control participants possess over performance of the behavior. For this reason, researchers have relied on participants' perceived behavioral control (PBC) on the assump­ tion that perceptions of control are reasonably accurate reflections of actual control (Ajzen, 1991). Evidence concerning whether perceived behavioral control moderates the intention-behavior relationship is equivocal. For ex­ ample, Armitage and Conner (in press) reviewed 19 studies and found that only nine ( 47% ) obtained a significant interaction between perceived be­ havioral control and intention (in all cases higher perceived behavioral con­ trol was associated with greater consistency between intentions and behavior). To provide a stronger test of this relationship, I reanalysed data from a theory of planned behavior study of 30 behaviors by Sheeran, Trafimow, Finlay, Norman and Orbell (1999). Between-participants moderated regres­ sion analyses showed that only seven (23% ) interactions between perceived behavioral control (PBC) and intentions were significant (greater PBC -stronger intention-behavior relations). Moreover, the stronger test afforded by within-participants regressions of behavior on intentions, perceived be­ havioral control, and their interaction failed to provide evidence of modera­ tion. Of course, statistical explanations for these findings can be adduced, such as the difficulty of detecting moderation in field studies (McClelland & Judd, 1993) and the consideration that linear models generally explain psychologi­ cal data quite well, even when the data were generated by a multiplicative model (e.g., Busemeyer & Jones, 1983). However, the data also revealed evidence related to my previous points about researchers being mainly interested in controllable behaviors and the effects of lack of perceived behavioral control on intentions to perform the behavior. Out of 5580 responses to measures of perceived behavioral control (30 behaviors by 186 participants), less than 1% (i.e., 52 responses) had a mean equal to 1.00 on a 1-7 scale. Moreover, of these 52 responses, in only two cases was the associated intention above the scale midpoint (participants do not generally intend to perform behaviors they perceive to be outside their control). Interestingly, there was a significant relationship between the pro­ portion of the sample below the mean for perceived behavioral control and whether or not the interaction between perceived behavioral control and intention was statistically reliable ( r = 0.45). This finding suggests that it may be difficult to discern moderation of the intention-behavior relation by per­ ceived behavioral control unless a substantial proportion of the sample have low perceived behavioral control. Clearly, however, further research is re­ quired to test the hypothesis that low levels of perceived behavioral control (in the absolute rather than relative sense) reduce the consistency between intentions and behavior.

12

PASCHAL SHEERAN

Intention Type: Intentions versos Expectations

Warshaw and Davis (1985) drew a distinction between behavioral intentions and behavioral expectations. Whereas intentions are typically measured by responses to statements such as, "I intend to do X", expectations are mea­ sured by estimates of the likelihood of performance, such as "How likely are you to do X?" Expectations are assumed to capture "various factors that could cause [participants] to be unsuccessful in their attempt to carry out [their] intentions" (Sheppard et al., 1988, p. 328) and should, therefore, pos­ sess greater predictive validity than intentions. Findings from Sheppard et al. 's (1988) meta-analysis supported this view, where the average correlations were 0.57 and 0.49 for expectations and intentions, respectively. However, more recent meta-analyses by Armitage and Conner (in press) and Sheeran and Orbell (1998) both failed to obtain reliable differences in the average correla­ tions obtained for intentions versus expectations. More seriously, it is debata­ ble whether behavioral expectation qualifies in causal models of behavior (can the expectation of performing a behavior be said to cause that behavior? cf., Ajzen, 1985). That is, behavioral expectation may predict behavior but this construct might not explain behavior (see Sutton, 1998, for discussion of the distinction between prediction and explanation). Another difficulty is that it is unclear how well expectations capture control factors such as ability and resources, or whether using a measure of perceived behavioral control alongside a traditional intention measure makes employment of an expecta­ tion measure redundant. Future research on behavioral expectations should be directed towards determining its construct validity and explanatory value. However, at present, it appears that whether the type of intention measure is a traditional one or an estimate does not influence how well intentions predict behavior. Intention Type: Implementation Intentions

Perhaps the most exciting development in intention-behavior relations in recent years is Gollwitzer's (1993, 1996, 1999) concept of implementation intentions. Whereas behavioral intentions have the form "I intend to do X", implementation intentions involve propositions of the form, "I intend to do X in situation Y". The usual paradigm in studies of implementation intentions has been to measure intentions (and other theory of planned behavior or protection motivation theory constructs) prior to a manipulation that asks one half of the sample to nominate the time and place for performing the be­ havior. Despite equivalent intentions to perform the behavior, participants who specify the situation for performing an intended action in this manner (form implementation intentions) seem invariably to demonstrate increased likelihood of performance compared to controls. Findings have shown that

13

THE INTENTION-BEHAVIOR GAP

participants who form implementation intentions are more likely to attend for cervical cancer screening (Sheeran & Orbell, 2000b) , take vitamin supple­ ments (Sheeran & Orbell, 1 999a), perform a breast self-examination (Orbell, Hodgkins, & Sheeran, 1 997), eat healthily (Verplanken & Faes, 1 999), ex­ ercise (Milne, Orbell, & Sheeran, in press), resist smoking (Stephens & Con­ ner, 1999), resist alcohol consumption (Murgraff, White, & Phillips, 1996), and complete their work on time (Gollwitzer & Brandstatter, 1997, Study 1, inter alia) (see Table 1.5). Meta-analysis of the effects of implementation intentions on behavioral performance

Table 1.5

Authors

Behavior

Aarts, Dijksterhuis, & Midden (1999) Gillholm etal. (1999) Study 1 Gillholm eta!. (1999) Study 2 Gollwitzer & BrandsHitter (1997) Study 1 Milne, Orbell, & Sheeran (in press) Murgraff, White, & Phillips (1996) Orbell, Hodgkins, & Sheeran (1997) Orbell & Sheeran (2000) Sheeran {1999a) Sheeran & Orbell (2000a) Sheeran & Orbell (1999a) Study 1 Sheeran & Orbell (1999a) Study 2 Silverman & Sheeran (1999)

Collect a coupon Mundane activities Mail response forms Complete a written report

Stephens & Conner {1999) Verplanken & Faes (1 999) Overall

Exercise Binge drinking Breast self-examination Recovery of functional activities Independent study Cervical cancer screening Vitamin supplement use Vitamin supplement use Attendance at workplace fire training Resist taking up smoking Healthy eating

d

r

40 28 50 36

0.63 1 .01 0.02 0.85

0.30 0.45 0.01 0.39

248 102 155 64 172 104 78 37 271

1 .25 0.67 1 .22 0.70 0.60 0.58 0.43 0.60 0.52

0.53 0.32 0.52 0.33 0.29 0.28 0.21 0.29 0.25

124 102

0.37 0.47

0.1 8 0.23

1611

0.70

0.33

n

Implementation intentions have been found to increase both the likelihood of performing a behavior [e.g., 92% of participants who formed implementa­ tion intentions attended for cancer screening compared to 69% of controls in Sheeran & Orbell's (2000b) study] and the speed of action initiation (e.g., patients undergoing joint replacement surgery who formed implementation intentions recovered functional activities 2.45 weeks sooner, on average, than control participants in Orbell & Sheeran's, 2000b, study). Implementation intentions are effective for both discrete "one-off" behaviors, such as annual attendance at fire training (Silverman & Sheeran, 1999), and for repeated behaviors that are performed daily, such as vitamin supplement use (Sheeran & Orbell, 1999a). Moreover, effectiveness has been demonstrated for student (e.g., Sheeran & Orbell, 1999a), general population (Sheeran & Orbell, 2000b ), and clinical samples (e.g., Lengfelder & Gollwitzer, 1998, cited in Gollwitzer, 1999) and for both self-report (e.g., Orbell et al., 1997) and less

14

PASCHAL SHEERAN

subjective behavioral measures, such as medical attendance records (Sheeran & Orbell, 200b) and pill counts (Sheeran & Orbell, 199a). In sum, forming an implementation intention seems to be very effective at turning inclined abstainers into inclined actors. To gain insight into the overall effectiveness of implementation intentions in promoting behavior, I conducted a meta-analysis of studies to date. Table 1 .5 describes characteristics of the 15 studies that could be included in the review. Because Pearson correlations are probably more familiar estimates of effect size than (the technically correct) statistic d, both indices are presented. Correlations from individual studies ranged from 0.01 to 0.53 and had an SD of 0.13. Not surprisingly, given the range of different behaviors examined, the sample correlations were heterogeneous, x? (14) 36.59, p < 0.01. Import­ antly, however, the Fail Safe N (FSN) criterion (5k + 10; see Rosenthal, 1984) was reached (FSN = 85, tolerance 85), indicating that the average correla­ tion is robust (i.e., likely to be resistant to future null findings). The overall sample weighted average correlation was r+ = 0.33 and the 95°/o confidence interval was 0.29--0.38. This is a "medium" effect size according to Cohen's (1992) power primer. This is an impressive finding when one considers that r+ = 0.33 represents the magnitude of the difference in subsequent behavior for participants who formed an implementation intention compared to particip­ ants with equivalent behavioral intentions who did not form an implementa­ tion intention. =

=

Why Are Implementation Intentions Effective? Gollwitzer (1993, 1999) has presented several lines of evidence that suggest that implementation intentions are effective because the process of specifying where and when one will perform the behavior means that the environmental cues one has specified elicit the behavior once those cues are encountered. In effect, the person passes the burden of remembering what he/she is supposed to do to a specified situational context, which then automatically prompts the behavior. A number of findings from field studies of implementation inten­ tions have been consistent with this proposal. First, several studies have dem­ onstrated that forming implementation intentions does not affect intentions or other theory of planned behavior or protection motivation theory variables (e.g., Milne et al., in press; Sheeran and Orbell, 1999a). Thus, the effects of implementation intentions cannot be attributed to changes in participants' motivation. Second, consistent with the idea that specified environmental cues elicit the behavior, participants who form implementation intentions rarely report "forgetting" to perform the behavior whereas control participants often seem to forget (e.g., 43o/o "forgot" in Sheeran & Orbell, 1 999a, and 70o/o "forgot" in Orbell et al., 1997). Third, participants who form implementation intentions demonstrate remarkable consistency between the time and place

THE INTENTION-BEHAVIOR GAP

15

specified i n their implementation intentions and the time and place that they perform the behavior-reported by participants to be several weeks later (Milne et al., in press; Orbell et al., 1997; Sheeran & Orbell, 1999a). Moreover, these reports appear to have some validity. A re-analysis of data from Sheeran and Orbell (2000b) for the present chapter showed that medical center records indicated that 74% of participants made their appointment for cervical cancer screening on the date that they had specified in their imple­ mentation intention. A fourth piece of evidence consistent with Gollwitzer's proposal is that the effects of implementation intentions do not diminish over time in the same manner exhibited by behavioral intentions (see Sheeran & Orbell, 1998). For example, Silverman and Sheeran (1 999) found that implementa­ tion intentions were just as effective at increasing attendance at fire training when participants specified a course in the next 5-8 weeks as when particip­ ants specified a course in the next 1-4 weeks (according to attendance re­ cords). Finally, and perhaps most impressively, implementation intentions appear to be able to replace previous action tendencies with new behaviors that participants intend to perform. The idea here is that implementation intentions involve a cognitive rehearsal of the link between a specific action and specific environmental cues. This cognitive rehearsal parallels the be­ havioral rehearsal of the link between an action and a situation which is characteristic of the development of a habit. Thus, cognitive rehearsal can attach the environmental cues previously associated with another behavior to a new behavior whose performance is desired by the actor. Consistent with this reasoning, Orbell et al.'s ( 1 997) study of breast self-examination showed that past behavior was the best predictor of future behavior among controls, but past behavior did not predict future behavior among particip­ ants who formed implementation intentions. Similarly, Sheeran and Orbell (2000b) found that forming an implementation intention significantly at­ tenuated the correlation between previous delay behavior and subsequent attendance ( r = -0.29 vs. r = 0.70). In sum, implementation intentions seem to be effective because the contextual cues specified in those intentions are effective in eliciting an intended behavior. The Issue of Specifying the Behavior in the Implementation Intention

A neglected issue regarding implementation intentions concerns what be­ havior should be specified in the implementation intention. When a single action is to be performed, then things are simple; the intention to do X is simply supplemented by the implementation intention, "I intend to do X in situation Y". Indeed, single actions have mainly been studied in applications of implementation intentions (e.g., Orbell et al., 1 997; Sheeran & Orbell, 1999a; Silverman & Sheeran, 1999). However, what if the person intends to

16

PASCHAL SHEERAN

achieve a goal or outcome? Here, the behavior to be performed in the service of the goal must be specified. In other words, "I intend to do X" must be supplemented by an implementation intention of the form "To achieve X, I intend to do W in situation Y". Of course, the problem then is what is the behavior W that wil achieve goal X. Sheeran and Orbell (200b) found that forming an implementation intention to make an appointment to attend for screening (behavior W) was very successful in achieving the goal of attending for cervical cancer screening. This was because there was a perfect relationship between behavior W and goal X in this case-all of the women who made an appointment subsequently attended for screening. However, a potentially serious limit to the utility of implementation intentions in promoting performance is apparent here. The effectiveness of implementation inten­ tions in promoting goal achievement is directly related to the strength of both (a) the relationship between implementation intentions and perfor­ mance of the behavior in the service of the goal, and (b) the relationship between performance of the behavior and the achievement of the goal. It is, therefore, very important to specify a behavior that is strongly related to the focal goal. Such specification might not always be easy. For example, is forming implementation intentions to resist distractions likely to be more effective in promoting performance than forming implementation intentions to increase persistence (cf. Schaal & Gollwitzer, 1999, cited in Gollwitzer, 1999)? What are the cognitive or behavioral strategies that are likely to be most effective for resisting distraction or for promoting persistence? Rel­ atedly, what sorts of behaviors should be specified in order to achieve goals such as preventing infection by HIVIAIDS (cf. Sheeran, Abraham, & Or­ bell, 1999), quitting smoking or other addictive substances, or improving the quality of the relationship with one's partner? Further research is required to test the utility of implementation intentions in the case of these complex goals. Properties of Behavioral Intentions

Another line of research has addressed properties of behavioral intentions that might determine how well intentions predict behavior. Characteristic measures of "strength" of intention generally capture the direction and inten­ sity of a decision only (i.e., to what extent a person intends to do X/intends not to do X). However, two people might have identical scores on a characteristic 1-7 intention scale but still differ in the quality of their motivation. This is because people's intentions possess other dimensions or properties that can affect their predictive validity. Important among these properties of intention are temporal stability, degree of intention formation, attitudinal versus nor­ mative control, and certainty and accessibility.

THE INTENTION-BEHAVIOR GAP

17

Temporal Stability of Intentions

The theories of reasoned action and planned behavior both acknowledge that participants' intentions can change prior to performance of a behavior (e.g., as a result of new information), and that, consequently, the original intention measure may not provide accurate prediction of behavior (Ajzen, 1985, 1991; Ajzen & Fishbein, 1980; Fishbein, 1980; Fishbein & Ajzen, 1975). Fishbein and Ajzen, therefore, repeatedly recommended that intentions should be measured as closely as possible to the time at which the behavior will be performed. Interestingly, however, Randall and Wolff's (1994) meta-analysis did not support this proposal; they found that the time interval between measurements of intentions and behavior was not related to the strength of intention-behavior correlations. Sheeran and Orbell ( 1998) criticized the database employed in Randall and Wolff's study on the grounds that there were insufficient studies involving the same behavior type to draw firm con­ clusions. Sheeran and Orbell conducted equivalent, but more powerful, analyses in the context of a single behavior (condom use) and found a nega­ tive relationship between the time interval and intention-behavior consis­ tency (r = -0.59). Although this finding is consistent with the hypothesis that the temporal stability of behavioral intentions moderates intention-behavior relations, this does not consititute a direct test of the moderator hypothesis (because inten­ tion stability was not measured). Direct tests were conducted by Bagozzi and Yi (1989) and Doll and Ajzen ( 1992), who computed measures of intention stability using within-participants correlations between intention scores taken at two time points. Findings were contradictory, however; Doll and Ajzen obtained significant moderation, but Bagozzi and Yi did not. Sheeran, Orbell, & Trafimow (1999) pointed out that one difficulty with these studies might be that intention stability was computed from measures taken before and after the behavioral performance. Thus, the intention stability measure may have been subject to self-presentational or consistency biases. Sheeran et al. (1999) computed an intention stability measure using within­ participants correlations between measures of intention taken at two time points prior to the performance of the behavior (studying over the winter vacation). Findings indicated that when intentions were unstable, they did not predict behavior. Rather, previous study behavior was the best predictor of future behavior. In contrast, when intentions were stable, past behavior did not predict future behavior and intentions were strong predictors of behavior. One difficulty with Sheeran et al. 's (1999) study is that the focal behavior may not constitute the most stringent test of the capacity of temporal stability to moderate intention-behavior relations. This is because studying over the winter vacation involves an unstable context (i.e., students will be studying in their family home rather than at university, where they usually study); and

18

PASCHAL SHEERAN

intentions should be particularly good predictors of behaviors performed in unstable contexts because habits are less likely to develop in unstable contexts (see Ouelette & Wood, 1998). Conner, Sheeran, Norman, and Armitage (200), therefore, examined two behaviors that involved stable contexts but differed in terms of their frequency of performance (annual attendance for health screening vs. eating a low-fat diet daily). Intention stability was com­ puted using a four-item index that was strongly associated with the within­ participants correlation measure. Findings confirmed those obtained by Sheeran et al. (199). Intention stability and PBC stability moderated past behavior-future behavior relations for health check attendance and eating a low-fat diet, respectively. More importantly, stable intentions were much bet­ ter predictors of both behaviors than were unstable intentions. In sum, ac­ cumulated evidence indicates that intention stability strongly determines how well intentions predict behavior.

Degree of Intention Formation Degree of intention formation refers to the extent to which participants have thought through the consequences of their decision to act in a particular manner. Thinking through the basis of one's intention should influence how well intentions predict behavior because participants who have not thought the decision through (i.e., possess poorly formed intentions) will be more likely to encounter unforeseen disadvantages or difficulties with performing the behavior and should be more likely to change their intention. Participants who possess well-formed intentions, on the other hand, anticipate possible outcomes and obstacles and should be more likely to try to continue to enact their intentions. Consistent with this reasoning, evidence suggests that well­ formed intentions exhibit greater temporal stability than poorly formed inten­ tions (Bagozzi & Yi, 1989). Bagozzi and Yi (1989) were the first to demonstrate that degree of intention formation moderates intention-behavior consistency. They manipulated the degree of intention formation by giving one half of the sample the oppor­ tunity to consider the consequences of performing a novel behavior (well­ formed intentions condition), whereas the other half of the sample were assigned a distractor task. The findings showed a significantly stronger intention-behavior relationship for participants with well-formed intentions (see also Verplanken, Aarts, van Knippenberg, & Moonen, 1998). Sheeran, Norman, Trafimow, Armitage, Conner, and Orbell (199) at­ tempted to extend this analysis to behaviors that are known to participants (as opposed to the novel behavior studied by Bagozzi and Yi, 1989). They argued that the coefficient of determination (R2) between intention and attitude, subjective norm, and perceived behavioral control could be used to indicate the extent to which participants had thought about the behavioral, normative,

20

PASCHAL SHEERAN

Sheeran, Norman, and Orbell (1999) tested the effects of attitudinal versus normative control of intentions on the intention-behavior relationship. Within-participants analyses showed that participants with attitudinally con­ trolled intentions had stronger intention-behavior correlations than did parti­ cipants with normatively controlled intentions. Between-participants analyses showed that intentions to perform 21 out of the 30 behaviors studied were more strongly predicted by attitudes than by subjective norms. Importantly, these attitudinally controlled intentions were associated with greater likeli­ hood of performance than normatively controlled intentions-even when strength of intention was statistically controlled. In sum, whether behavioral intentions are attitudinally versus normatively controlled affects how well intentions determine behavior.

Certainty and Accessibility of Intentions Two other properties of behavioral intentions that have received attention are certainty (measured by participants' ratings of how certain they are about their intention) and accessibility (measured by response latencies to intention items). Several authors have demonstrated that intentions held with greater certainty or confidence are associated with greater intention-behavior consis­ tency (Bagozzi & Yi, 1989; Bassili, 1993; Pieters & Verplanken, 1995). Sheeran (1999a) computed a certainty index based on participants' ratings of the strength of the focal intention compared to other (own) intentions and other participants' intentions. Findings showed that participants with more certain intentions exhibited stronger correlations between their intentions and subsequent examination performance than did participants with less cer­ tain intentions (r = 0.59 vs. r = 0.15, respectively). Findings concerning the accessibility of intentions have been more contra­ dictory. Bassili (1993, 1995) showed that participants with highly accessible voting intentions were more likely to vote for their intended candidate than were participants with less accessible intentions; indeed, greater accessibility was related to a perfect intention-behavior relationship in Bassili's (1995) study and accessibility better moderated the intention-behavior relation than certainty in Bassili's (1993) study. However, Doll and Ajzen (1992) failed to obtain evidence of moderation of the intention-behavior relationship and concluded that intention stability had greater influence on intention-behavior relationships. Further research is required to test the role of accessibility in determining intention-behavior consistency.

Personality and Cognitive Variables The final category of variables that affects how well intentions determine behavior examined here are personality and cognitive variables. Personality

THE INTENTION-BEHAVIOR GAP

21

variables involve people's general tendencies to think and behave in certain ways and their importance has rarely been examined in the context of intention-behavior relations. However, Pieters and Verplanken (1995) showed that individual differences in the tendency to engage in reasoning (indexed by the need for cognition scale; Cacioppo & Petty, 1982) were related to amount of reasoning about the behavior, which in turn was associ­ ated with greater consistency between intentions and behavior (a finding that seems to be consistent with the analysis of degree of intention formation presented earlier). Action Control

Greater attention has been devoted to Kuhl's (1985; Kuhl & Beckmann, 1994) action-control scale. According to Kuhl's analysis, participants who are action-orientated have a tendency to focus on the actions required to reduce the discrepancy between their current and future intended state; unlike state­ orientated participants, who tend to focus exclusively on their current state or the intended state and do not consider alternative actions required for goal achievement. Consequently, greater action control should produce stronger intention-behavior relations. In fact, tests of this hypothesis have been equivocal. Kuhl (1982) and Schif­ ter and Ajzen (1985) both obtained significant moderation of the intention­ behavior relation, whereas Kendzierski's (1990a) and Bagozzi, Baumgartner, and Yi's (1992) findings did not support moderation. One explanation for this discrepancy could be that most of the studies involved single behaviors and it is generally difficult to find relationships between broad personality variables and single actions (cf. Ajzen, 1 988). Norman, Sheeran, and Orbell (in press) conducted a sterner test using within-participants analyses of the intention­ behavior relation for 30 behaviors. Overall findings did not support the view that greater action-orientation is related to greater intention-behavior consis­ tency. However, subsample analyses indicated that state-orientated particip­ ants had stronger intention-behavior relations for normatively controlled behaviors than did action-orientated participants. These findings are consis­ tent with Kuhl's theory (e.g., Fuhrmann & Kuhl, 1998), which posits complex interactions between action and state orientation, attitudinal versus norma­ tive control of intentions (whether reward or punishment systems are acti­ vated), goal difficulty, and participants' use of strategies to control attention, motivation, and impulses, inter alia (cf. Kuhl & Fuhrmann, 1 999). A great deal more research is required that tests the variety of interactions proposed by this intriguing theory. The other line of research has tested whether particular cognitions about behaviors moderate intention-behavior relations. Here, moderation has been demonstrated for anticipated regret, self-schemas, and conflicting intentions.

22

PASCHAL SHEERAN

Anticipated Regret Anticipated regret refers to participants' beliefs about the extent of regret, tension, or upset they would feel if they did not perform a particular behavior. Evidence suggests that anticipated regret is not the same construct as attitude towards the behavior and that anticipated regret has strong associations with intentions to perform behaviors after other predictors have been controlled (e.g., Parker, Manstead, & Stradling, 1995; Richard, van der Pligt, & de Vries, 1995, 1 996). Sheeran and Orbell (1999b) proposed that anticipating regret about performing a behavior might bind participants to their intentions, such that participants who both intend to perform a behavior and anticipate con­ siderable regret if they do not perform it, should exhibit greater intention­ behavior consistency than participants with equivalent intentions who do not anticipate regret. Findings from a study of lottery play confirmed that antici­ pated regret moderated intentions in this manner; lottery play was greatest when participants both intended to play and anticipated regretting not play­ ing. More recently, Abraham and Sheeran (200) tested whether anticipated regret moderated intentions to exercise (using both correlational and experi­ mental designs) and found an equivalent effect of the interaction between anticipated regret and behavioral intentions. Abraham and Sheeran also con­ firmed that moderation of the intention-behavior relation by anticipated re­ gret was independent of participants' past exercise behavior.

Self-schemas Self-schemas refer to people's self-definitions in domains of enduring invest­ ment and concern (e.g., Markus, 1977) and are characteristically measured by ratings of the descriptiveness and importance of traits related to the be­ havioral domain (e.g., ratings of "physically active" and "keeps in shape" have been used to measure exercise self-schemas, Kendzierski, 1990b ) Self­ schemas systematically affect information processing (e.g., Markus & Wurf, 1 987) and should affect intention-behavior relations in the schema-relevant domain. This is because schematics (participants for whom the domain is highly self-descriptive and highly important) have more stable intentions compared to aschematic participants (Kendzierski & Whitaker, 1997). Ken­ dzierski and Whitaker also observed a (non-significant) tendency for schema­ tics to exhibit stronger intention-behavior relationships. Sheeran and Orbell (200a) used slightly different procedures to test whether self-schemas mod­ erated the relationship between intentions and exercise behavior and found that intentions provided significantly better prediction of behavior for parti­ cipants who were schematic for exercise. A second study also showed that self-schemas moderated the relationship between dieting intentions and be­ havior (Sheeran, 1999b ) .

.

24

PASCHAL SHEERAN

Relationships between Determinants of Intention-Behavior Consistency An important outstanding issue that will need to be addressed by future research concerns the relationships between variables in each of the catego­ ries (behavior type, intention type, properties of intentions, and cognitive and personality variables) described above. For example, degree of intention formation, anticipated regret, and self-schemas are all assumed to affect intention-behavior consistency because they affect the temporal stability of behavioral intentions (the grounds for this hypothesis being that temporal stability is a defining and antecedent property of intentions; see Sheeran

1999).

et a/.,

Although there is evidence demonstrating that, for example, the be­

havioral intentions of schematics exhibit greater stability than those of on­ schematics (Kendzierski

&

Whitaker,

1997),

no formal test of whether

intention stability mediates the effects of self-schemas (or other variables) on the intention-behavior relationship appears to have been conducted. Such tests are warranted in future studies. Several other interrelationships between variables also merit investigation. For example, does being schematic in the behavioral domain or possessing greater action orientation cause people to

spontaneously

form implementa­

tion intentions? Do more stable or well-formed intentions mean that particip­ ants can identify behaviors that are effective in achieving particular goals or mean that participants can shield their current intention from conflicting in­ tentions? Relatedly, it remains to be determined which of the properties of behavioral intentions identified here is best capable of turning inclined ab­ stainers into inclined actors. Tests also need to be conducted on the volitional strategies identified by Kohl and Fuhrmann

(1999)

to determine their effec­

tiveness in translating intentions into action and to examine their relation­ ships with other moderators. In sum, although there has been a good deal of progress in identifying determinants of how well intentions predict behavior, much more research is required to understand the mechanisms that underlie their effectiveness.

THE SCOPE OF THE INTENTION CONSTRUCT: INTENTIONS, AUTOMATICITY, AND HABITS This final section of this chapter considers the scope of the intention construct in the light of several recent lines of research that suggest that intentional control of behavior may be a great deal more limited than, for example, the theories of reasoned action and planned behavior appear to assume. For example, Wegner and Wheatley

( 1999)

proposed that " . . . the real causal

mechanisms underlying behavior are never present in consciousness. Rather, the engines of causation are unconscious mechanisms of the mind" (p. 490).

THE INTENTION-BEHAVIOR GAP

25

According to this analysis, the idea that intentions cause behavior is in fact an illusion. Wegner and Wheatley suggested that the experience of intentionality can be analysed from an attributional perspective and cite an experiment that varied the priority of a thought about an action over the subsequent action (i.e., how soon before the action participants were induced to think about that action). Despite the fact that participants had no control over the action, participants were more likely to report that they had intended to perform it if the thought occurred 5 seconds or 1 second before the action than if the thought occurred 30 seconds before the action or 1 second after the action. Thus, participants can be led to believe that their intention caused a particular action when the experimental situation rules out the possibility that intention was the causal factor. A related line of research by Bargh (e.g., Bargh, 1 997, Bargh & Chartrand, 1 999; Bargh, Chen, & Burrows, 1996) suggests that behavior may be governed either by intentions or by automatic processes. However, the assumption is that most human behavior is governed by automatic processes rather than by intentions ("99 and 44/loo% automatic" as Bargh, 1 997, p. 243, polemically suggested). Bargh's analysis also sheds light on what are the "unconscious mechanisms of the mind'' that actually guide behavior. First, there is a direct link between perception and behavior, such that priming stereotypes or traits outside participants' awareness automatically produces behavior consistent with those constructs. Thus, for example, participants primed with traits re­ lated to rudeness were more likely to interrupt an experimenter during a subsequent encounter, and participants primed with traits related to polite­ ness were less likely to interrupt the experimenter, compared to controls (Bargh, Chen, & Burrows, 1 996, Study 1 ; see also Dijksterhuis et al., 1 998, for examples of behavioral contrast following priming of exemplars rather than stereotypes or traits). Second, there is a direct link between goals and be­ havior, such that goals activated outside of participants' conscious awareness produce predictable shifts in their behavior. Thus, for example, participants primed with the achievement goal were more likely to continue working on a verbal task (surreptitiously) when they had been told to stop, compared to controls who had not been primed (Bargh, Gollwitzer, & Lee-Chai, 1999, cited in Bargh & Chartrand, 1999). Bargh, Raymond, Pryor, and Strack (1995) have also demonstrated that situational features can activate goals without participants being aware that those goals have been activated. A third uncon­ scious process with behavioral consequences is automatic evaluation. Chen and Bargh (1999) found that participants automatically demonstrated an ap­ proach response for liked stimuli but demonstrated an automatic avoidance response for disliked stimuli, using a reaction time task that involved pushing (avoid response) versus pulling (approach response) a lever. The third-related-line of research that challenges the role of intentions as predictors of behavior concerns the oft-cited finding in studies of the

26

PASCHAL SHEERAN

theories of reasoned action and planned behavior that a measure of past behavior often predicts future behavior better than a measure of intention. According to certain analyses of this issue, correlations between past behavior and future behavior may indicate the operation of habits (e.g., Ouelette & Wood, 1998), where "habits" refers to behaviors that are automatically eli­ cited by situational cues. This automatic elicitation occurs because of the strong cue-response links produced by repeated performance of a behavior in a particular context. Consistent with this reasoning, Ouelette and Wood's (1998) meta-analysis showed that for behaviors performed frequently in sta­ ble contexts (i.e., contexts conducive to habit formation), past behavior pre­ dicts future behavior better than do intentions. In contrast, for behaviors that are infrequently performed in unstable contexts (contexts less conducive to habit formation), intentions predict future behavior better than does past behavior. This suggests that another "behavior type" distinction should be made on the basis of the frequency and stability of the context of performance of behaviors, since this constitutes another determinant of how well intentions predict behavior. These analyses by Wegner and Wheatley (199), Bargh (1997), and Ouelette and Wood (1998) are both compelling and thought-provoking. Most re­ searchers are likely to agree that people sometimes see intentions as causes of their behavior when intentions are not the cause, and that automatic processes have an important role in understanding behavior. However, there is currently too little evidence to warrant the conclusion that intentions cause 0% or only 0.56% of human behavior. The fact that certain behaviors can be primed in the laboratory without participants being aware of it does not tell us how often such priming is responsible for human behavior outside the laboratory. Moreover, it is unclear whether priming studies really bypass participants' intentions. For example, participants in Bargh et al.'s (1996) Study 1 were induced to interrupt or not interrupt an experimenter without their being aware of it. However, it seems safe to assume that most participants formed an intention about whether or not to interrupt the experimenter. It also seems likely that if participants had been asked their intentions, then those inten­ tions would have been affected by the manipulation (e.g., Sheeran, 199c, showed that participants primed with an image of Princess Diana had stronger intentions to give to charity than controls). It may be that the reasons for these intentions are not available to conscious scrutiny, but there are probably a wide variety of influences on intentions of which we are not aware. However, this does not rule out the possibility that, in many instances at least, particip­ ants' conscious intentions may mediate the effects of unconscious primes on behavior. Macrae and Johnston's (1998) demonstration, that manipulating either the attractiveness of the target behavior or the presence of competing goals eliminated the effect of a "helpfulness" prime on subsequent behavior, seems to be consistent with this idea.

THE INTENTION-BEllAVIOR GAP

27

Applied social psychologists are also likely to be concerned about the gen­ erality of these analyses. For example, the sorts of behaviors that have been examined in priming studies are much less involving for participants than are the behaviors that have characteristically been examined in applications of the theories of reasoned action and planned behavior (see van der Pligt, de Vries, Manstead, & van Harreveld, 2000, for discussion of this point in rela­ tion to deliberative vs. automatic analyses of attitudes). To be entirely con­ vinced by this approach, applied researchers will want to see demonstrations that the sorts of behaviors that have been predicted by intentions in their research (e.g., condom use, exercise) are better predicted by priming traits, stereotypes, or goals outside participants' conscious awareness. In sum, much more research is required on the scope of automatic processes in order to warrant the conclusion that they provide better prediction of behavior than do behavioral intentions. Similarly, the past behavior-intention-future behavior relationship re­ quires a great deal more attention before clear conclusions can be drawn about the relative importance of past behavior versus intentions as predictors of future behavior. For example, although Ouelette and Wood's (1998) meta­ analysis is compelling, only eight studies of frequently performed behaviors and six studies of infrequently performed behaviors were included in their analyses. Verplanken and Aarts (1999) have also pointed out that examina­ tions of correlations between these three variables may not be terribly il­ luminating because of problems of shared method variance (see, however, Conner, Warren, Close, & Sparks, 1 999, for an exception) and because of the possibility that unmeasured variables are responsible for the associations. The recent development of a script-based measure of habit (e.g., Aarts, Ver­ planken, & van Knippenberg, 1997; Verplanken, Aarts, van Knippenberg, & van Knippenberg, 1994; Verplanken et al., 1998) that eschews frequency of past behavior represents an important methodological advance in this area and deserves more widespread use. Care also needs to be taken when attributing variance to "habit" in analyses of past behavior-intention-future behavior relations. For example, the fact that someone has not performed a behavior in the past and subse­ quently does not perform the behavior does not constitute sufficient evidence to designate this sequence a "habit". If sequences of non-performance of a behavior constituted habits, then every behavior that one has not performed over the course of one's lifetime could be characterized as a "habitual" be­ havior. In fact, much of the strength of past behavior-future behavior correla­ tions might be attributable to such sequences of non-performance. Adapting the analysis presented in Table 1 .6 (see also McBroom & Reid, 1992) in order to decompose the relationship between past behavior, inten­ tions, and future behavior may provide greater insight into whether past behavior (possibly "habit") or intentions better predict future behavior.

30

PASCHAL SHEERAN

Kevin McKee, Paul Norman, Sheina Orbell, Diane Parker, & Marco Peru gini) for discussions on several issues related to this chapter. Particular thanks are due to Chris Armitage and Mark Conner for providing additional information about their research (on several occasions and at short notice ) .

REFERENCES Aarts, H., Dijksterhuis, A., & Midden, C. ( 199). To plan or not to plan? Goal achieve­ ment or interrupting the performance of mundane behaviours. European Journal of Social Psychology, 29, 971-9. Aarts, H., Verplanken, B., & van Knippenberg, A. (1997). Habit and information use in travel mode choices. Acta Psychologica, 96, 1-14. Abraham, C. & Sheeran, P. (200). "How bad will I feel if I don't?" Anticipated regret, the theory of planned behavior, and exercise behavior (manuscript under review). Abraham, C., Sheeran, P., & Johnson, M. (1998). From health beliefs to self­ regulation: Theoretical advances in the psychology of action control. Psychology and Health, 13, 569-92. Abraham, C., Sheeran, P., Norman, P., Conner, M., Otten, W., & de Vries, N. ( 199). When good intentions are not enough: Modelling post-intention cognitive correlates of condom use. Journal ofApplied Social Psychology, 29, 2591-612. Abraham, C., Sheeran, P., & Orbell, S. (1998). Can social cognition models contribute to the effectiveness of HIV -preventive behavioural interventions? A brief review of the literature and a reply to Joffe (1996, 1997) and Fife-Schaw (1997). British Journal of Medical Psychology, 71, 297-310. Ajzen, I. (1985). From intentions to actions: A theory of planned behavior. In J. Kuhl & J. Beckmann (Eds), Action Control: From Cognition to Behavior (pp. 1 1-39). Berlin: Springer-Verlag. Ajzen, I. (1988). Attitudes, Personality, and Behavior. Milton Keynes: Open University Press. Ajzen, I. (1991 ). The theory of planned behavior. Organizational Behavior and Human Decision Processes, SO, 1 79-21 1 . Ajzen, 1 . , & Fishbein, M . (1973). Attitudinal and normative variables as predictors of specific behaviors. Journal of Personality and Social Psychology, 27, 41-57. Ajzen, 1., & Fishbein, M. (1974). Factors influencing intentions and the intention­ behavior relation. Human Relations, 27, 1-15. Ajzen, 1., & Fishbein, M. (1980). Understanding Attitudes and Predicting Social Be­ havior. Englewood-Cliffs, NJ: Prentice-Hall. Armitage, C. J., & Conner, M. (in press). Efficacy of the theory of planned behaviour: A meta-analytic review. British Journal of Social Psychology. Bagozzi, R. P., Baumgartner, H., & Yi, Y. (1992). State versus action orientation and the theory of reasoned action: An application to coupon usage. Journal ofConsumer Research, 18, 505-18. Bagozzi, R. P., & Warshaw, R. P. (1990). Trying to consume. Journal of Consumer Research, 17, 127-40. Bagozzi, R. P., & Yi, Y. (1989). The degree of intention formation as a moderator of the attitude-behavior relationship. Social Psychology Quarterly, 52, 266-79. Bargh, J. A. (1982). Attention and automaticity in the processing of self-relevant information. Journal of Personality and Social Psychology, 43, 425-36.

32

PASCHAL SHEERAN

Coumeya, K. S. (1994). Predicting repeated behavior from intention: The issue of scale correspondence. Journal of Applied Social Psychology, 24, 580-94. Davidson, A. R., & Morrison, D. M. (1983). Predicting contraceptive behavior from attitudes: A comparison of within- versus across-subjects procedures. Journal of Personality and Social Psychology, 45, 997-1009. Deci, E. L., & Ryan, R. M. (1990). A motivational approach to self: Integration in personality. In R. D. Dienstbier (Ed.), Nebraska Symposium on Motivation (Vol. 38, pp. 237-88). Lincoln, NE: University of Nebraska Press. Dijksterhuis, A., Spears, R., Postmes, T., Stapel, D. A., Koomen, W., van Knippen­ berg, A., & Scheepers, D. (1998). Seeing one thing and doing another: Contrast effects in automatic behaviour. Journal of Personality and Social Psychology, 75, 862-71 . Doll, J., & Ajzen, I . (1992). Accessibility and stability of predictors i n the theory of planned behaviour. Journal of Personality and Social Psychology, 63, 754-65. Fishbein, M. (1980). A theory of reasoned action: Some applications and implications. In H. Howe & M. Page (Eds.), Nebraska Symposium on Motivation (Vol. 27, pp. 651 16). Lincoln, NB: University of Nebraska Press. Fishbein, M., & Ajzen, I. (1975). Belief, Attitude, Intention and Behavior: An Introduc­ tion to Theory and Research. Reading, MA: Addison-Wesley. Fuhrmann, A., & Kuhl, J. (1998). Maintaining a health diet: Effects of personality and self-regard versus self-punishment on to and enactment of self-chosen and assigned goals. Psychology and Health, 13, 651-86. Gallois, C., Kashima, Y., Terry, D., McCamish, M., Timmins, P., & Chauvin, A. (1992). Safe and unsafe sexual intentions and behavior: The effects of norms and attitudes. Journal ofApplied Social Psychology, 22, 1521-45. Gillholm, R., Ettema, D., Selart, M., & Garling, T. (199). The role of planning for intention-behavior consistency. Scandinavian Journal of Psychology, 40, 241-50. Godin, G., & Kok, G. (1996). The theory of planned behavior: A review of its applica­ tions to health-related behaviors. American Journal of Health Promotion, 11, 97-8. Gollwitzer, P. M. (1993). Goal achievement: The role of intentions. In W. Stroebe & M. Hewstone (Eds), European Review of Social Psychology (Vol. 4, pp. 141-85). Chichester: Wiley. Gollwitzer, P. M. (1996). The volitional benefits of planning. In P. M. Gollwitzer & J. A. Bargh (Eds), The Psychology ofAction (pp. 287-312), New York: Guilford. Gollwitzer, P. M. (1999). Implementation intentions: Strong effects of simple plans. American Psychologist, 54, 493-503. Gollwitzer, P. M., & Brandstatter, V. (1997). Implementation intentions and effective goal pursuit. Journal of Personality and Social Psychology, 73, 186-99. Hausenblas, H. A., Carron, A. V., & Mack, D. E. (1997). Application of the theories of reasoned action and planned behaviour to exercise behaviour: A meta-analysis. Journal of Sport and Exercise Psychology, 19, 36-5 1. Hunter, J. E., & Schmidt, F. L. (1990). Methods of Meta-analysis: Correcting Error and Bias in Research Findings. Newbury Park, CA: Sage. Kashima, Y., Gallois, C., & McCamish, M. (1993). The theory of reasoned action and co-operative behaviour: It takes two to use a condom. British Journal of Social Psychology, 32, 227-39. Kendzierski, D. (1990a). Decision making vs decision implementation: An action con­ trol approach to exercise adoption and adherence. Journal of Applied Social Psy­ chology, 20, 27-45. Kendzierski, D. (1990b). Exercise self-schemata: Cognitive and behavioural correlates. Health Psychology, 9, 69-82. Kendzierski, D., & Whitaker, D. J. (1997). The role of self-schema in linking intentions with behavior. Personality and Social Psychology Bulletin, 23, 139-47.

34

PASCHAL SHEERAN

Orbell, S., & Sheeran, P. (200). Motivational and volitional processes in action initia­ tion: A field study of the role of implementation intentions. Journal of Applied Social Psychology, 30, 780-97. Orbell, S., & Sheeran, P. (1998). "Inclined abstainers": A problem for predicting health-related behaviour. British Journal of Social Psychology, 37, 151-65. Ouelette, J. A., & Wood, W. (1998). Habit and intention in everyday life: The multiple processes by which past behaviour predicts future behaviour. Psychological Bulletin, 124, 54-74. Parker, D. (1997). The relationship between speeding attitudes and speeding be­ haviour. Paper presented to the Department of Transport, Behavioural Studies Seminar, Esher, UK. Parker, D., Manstead, A. S. R., & Stradling, S. G. (1995). Extending the theory of planned behavior: The role of personal norm. British Journal of Social Psychology, 34, 127-37. Pieters, R. G. M., & Verplanken, B. (1995). Intention-behavior consistency: Effects of consideration set size, involvement and need for cognition. European Journal of Social Psychology, 25, 531-43. Randall, D. M., & Wolff, J. A. (1994). The time interval in the intention-behaviour relationship: Meta-analysis. British Journal of Social Psychology, 33, 405-18. Richard, R., van der Pligt, J., & de Vries, N. K. (1995). Anticipated affective reaction and prevention of AIDS. British Journal of Social Psychology, 34, 9-21. Richard, R., van der Pligt, J., & de Vries, N. K. (1996). Anticipated affect and be­ havioral choice. Basic and Applied Social Psychology, 18, 1 1 1-29. Rogers, R. W. (1983). Cognitive and physiological processes in fear appeals and atti­ tude change: A revised theory of protection motivation. In B. L. Cacioppo & L. L. Petty (Eds), Social Psychophysiology: A Sourcebook. London: Guilford. Rosenthal, R. (1984). Meta-analytic Procedures for Social Research. Beverley Hills, CA: Sage. Rosenthal, R., & Rubin, D. B. (1979). A note on percent variance explained as a measure of the importance of effects. Journal of Applied Social Psychology, 9, 395-6. Ryan, M. J., & Bonfield, E. (1975). The Fishbein extended model and consumer behavior. Journal of Consumer Research, 2, 1 18-36. Ryan, R. M., Sheldon, K. M., Kasser, T., & Deci, E. L. (1996). All goals are not created equal: An organismic perspective on the nature of goals and their regulation. In P. M. Gollwitzer & J. A. Bargh (Eds), The Psychology of Action (pp. 7-26). London: Guilford. Schifter, D. E., & Ajzen, I. (1985). Intention, perceived control, and weight loss: An application of the theory of planned behavior. Journal of Personality and Social Psychology, 49, 843-51. Sheeran, P. ( 199a). Strength of intention and intention-behaviour relations (un­ published raw data). Department of Psychology, University of Sheffield, UK. Sheeran, P. (1999b). Self-schemas. intentions and dieting behaviour (unpublished raw data). Department of Psychology, University of Sheffield, UK, Sheeran, P. (1999c). Doing and Di: Effects of priming Princess Diana on intentions to give to charity (unpublished raw data). Department of Psychology, University of Sheffield, UK, Sheeran, P., Abraham, C., & Orbell, S. (199). Psychosocial correlates of heterosexual condom use: A meta-analysis. Psychological Bulletin, 125, 90-132. Sheeran, P., Norman, P., & Orbell, S. (199). Evidence that intentions based on atti­ tudes better predict behaviour than intentions based on subjective norms. European Journal of Social Psychology, 29, 403-6.

THE INTENTION-BEHAVIOR GAP

35

Sheeran, P., Norman, P., Trafimow, D., Armitage, C. J., Conner, M., & Orbell, S. (1999). Why reason one's actions or plan one's behaviour: Degree of intention formation and intention-behavior relations (manuscript under review). Sheeran, P., & Orbell, S. ( 1998). Do intentions predict condom use? Meta-analysis and examination of six moderator variables. British Journal of Social Psychology, 37, 231-50. Sheeran, P., & Orbell, S. ( 1999a). Implementation intentions and repeated behaviour: Augmenting the predictive validity of the theory of planned behaviour. European Journal of Social Psychology, 29, 349-69. Sheeran, P., & Orbell, S. (1999b) . Augmenting the theory of planned behavior: Roles for anticipated regret and descriptive norms. Journal of Applied Social Psychology, 29, 2107-42. Sheeran, P., & Orbell, S. (2000a). Self-schemas and the theory of planned behaviour. European Journal of Social Psychology, 30, 533-50. Sheeran, P., & Orbell, S. (2000b). Using implementation intentions to increase attend­ ance for cervical cancer screening. Health Psychology, 18, 283-9. Sheeran, P., Orbell, S., & Norman, P., (1998). The theory of planned behaviour and intention-behaviour relations: Significance of conflicting and complementary inten­ tions. Unpublished manuscript. University of Sheffield, UK. Sheeran, P., Orbell, S., & Trafimow, D. (1999). Does the temporal stability of be­ havioral intentions moderate intention-behavior and past behavior-future behavior relations? Personality and Social Psychology Bulletin, 25, 721-30. Sheeran, P., & Sutton, S. (1999). Past behaviour and the theory of planned behaviour: A meta-analytic review (unpublished raw data). University of Sheffield, UK. Sheeran, P., Trafimow, D., Finlay, K. A. & Norman, P. (in press). Evidence that the type of person affects the perceived behavioral control-intention relationship. British Journal ofSocial Psychology. Sheppard, B. H., Hartwick, J., & Warshaw, P. R. (1988). The theory of reasoned action: A meta-analysis of past research with recommendations for modifications and future research. Journal of Consumer Research, 15, 325-42. Silverman, M., & Sheeran, P. (1999). Using implementation intentions to increase attendance at fire-training (unpublished raw data). Department of Psychology, Uni­ versity of Sheffield, UK, Smetana, J. G., & Adler, N. E. (1980). Fishbein's valence X expectancy model: An examination of some assumptions. Personality and Social Psychology Bulletin, 6, 8996. Stanton, B. F., Li, X., Black, M. M., Ricardo, I., Galbraith, J., Feigelman, S., & Kaljee, L. (1996). Longitudinal stability and predictability of sexual perceptions, intentions, and behaviors among early adolescent African-Americans. Journal of Adolescent Health , 18, 1�19. Stephens, A., & Conner, M. (1999). A smoking prevention intervention in adolescents. Paper presented at the British Psychological Society Division of Health Psychology Annual Conference, University of Leeds, September. Sutton, S. (1998). Predicting and explaining intentions and behaviour: How well are we doing? Journal of Applied Social Psychology, 28, 1317-38. Sutton, S., Bickler, G., Sancho-Aldridge, J., & Saidi, G. (1994). Prospective study of predictors of attendance for breast screening in inner London. Journal of Epidemiol­ ogy and Community Health, 48, 65-73. Trafimow, D., & Finlay, K. (1996). The importance of subjective norms for a minority of people. Personality and Social Psychology Bulletin, 22, 82�28. Trafimow, D., Sheeran, P., Conner, M., & Finlay, K. A. (in press). Is perceived be­ havioural control a multidimensional construct? Perceived control and perceived difficulty. British Journal of Social Psychology.

36

PASCHAL SHEERAN

Triandis, H. C. (1980). Values, attitudes, and interpersonal behavior. In H. Howe & M. Page (Eds), Nebraska Symposium on Motivation (Vol. 27, pp. 195-259). Lincoln, NB: University of Nebraska Press. van den Putte, B. (1993). On the theory of reasoned action (unpublished doctoral dissertation). University of Amsterdam, The Netherlands. Verplanken, B., & Aarts, H. ( 199). Habit, attitude, and planned behaviour: Is habit an empty construct or an interesting case of automaticity? In W. Stroebe & M. Hewstone (Eds), European Review of Social Psychology (Vol. 10). Chichester: Wiley. Verplanken, B., Aarts, H., van Knippenberg, A., & Moonen, A. (1998). Habit versus planned behaviour: A field experiment. British Journal of Social Psychology, 37,

1 1 1-28. Verplanken, B., Aarts, H., van Knippenberg, A., & van Knippenberg, C. (1994). Atti­ tude versus general habit: Antecedants of travel mode choice. Journal of Applied

Social Psychology, 24, 285-300. Verplanken, B., & Faes, S. ( 199). Good intentions, bad habits, and effects of forming implementation intentions on healthy eating. European Journal of Social Psychol­

ogy, 29, 591-604. van der Pligt, J., de Vries, N. K., Manstead, A. S. R., & van Harreveld, F. (200). The importance of being selective: Weighing the role of attribute importance in attitudi­ nal judgment. In M. P. Zanna (Ed.). Advances in Experimental Social Psychology (Vol. 32, pp. 135-200). New York: Academic Press. Vinokur-Kaplan, D. (1978). To have-or not to have-another child: Family planning attitudes, intentions, and behavior. Journal of Applied Social Psychology, 8, 29-46. Warshaw, P. R., Calantone, R., & Joyce, M. (1986). A field application of the Fishbein and Ajzen intention model. Journal of Social Psychology, 126, 13�. Warshaw, P. R., & Davis, F. D. (1984). Self-understanding and the accuracy of be­ havioral expectations. Personality and Social Psychology Bulletin, 10, 1 1 1- 18. Warshaw, P. R., & Davis, F. D. (1985). Disentangling behavioral intention and be­ havioral expectation. Journal of Experimental Social Psychology, 21, 213-28. Wegner, D. M., & Wheatley, T. ( 199). Apparent mental causation: Sources of the experience of the will. American Psychologist, 54, 480-92. Zimmerman, R. S., & Vemberg, D. (1994). Models of preventive health behavior: Comparison, critique, and meta-analysis. Advances in Medical Sociology, 4, 4�7.

Chapter 2

Ambivalence and Attitudes Mark Conner University of Leeds, UK

Paul Sparks

University of Sussex, UK

ABSTRACT This chapter explores the concept of ambivalence and its relationship to atti­ tudes. Definitions and different measures of ambivalence are reviewed. We present three dimensions on which measures of ambivalence can be classified. We then go on to examine some of the elicitors of ambivalence and present some preliminary findings on attitude objects that generate high levels of ambivalence. The major section of the chapter then reviews findings related to treating the ambivalence construct as an aspect of attitude strength. Research is presented in relation to four major predictions: (a) temporal stability of attitudes; (b) impact of attitudes on intentions and behavior, (c) pliability of attitudes; and (d) impacts on information processing. The final section offers suggestions for further re­ search in this area.

Attitude has retained its position in social psychology as the most distinctive and indispensable concept for well over 50 years (Allport, 1935). Interest remains high, with significant advances in our understanding of the antece­ dents and consequences of attitudes (see Eagly & Chaiken, 1993, for an extensive review). A particular focus of research interest in recent years has been on aspects of the structure (e.g., Pratkanis, Breckler, & Greenwald, 1 989) and strength (e.g., Petty & Krosnick, 1 995) of attitudes. The recent Correspondence regarding this chapter should be addressed to: Mark Conner, School of Psychol ogy, University of Leeds, Leeds LS2 9JT, UK. E mail: [email protected] European Review of Social Psychology, Volume Sons Ltd.

© 2002 John Wiley &

12. Edited by Wolfgang Stroebe

and Miles Hewstone.

38

MARK CONNER AND PAUL SPARKS

interest in attitudinal ambivalence can be partly attributed to the resurgence of interest in the dimensions of attitude structure and strength within which ambivalence fits. Attitudinal ambivalence may be defined as a psychological state in which ". . . a person holds mixed feelings (positive and negative) towards some psychological object" (Gardner, 1987, p. 241 ). Ambivalence thus is at odds with traditional views of the attitude concept as a unidimensio­ nal evaluation (Eagly & Chaiken, 1993). This chapter explores the am­ bivalence concept through a review of research on its definition, measurement, antecedents, and consequences. The ambivalence concept was introduced into social psychology by Scott (1966, 1968, 1969), followed by other important contributions by Kaplan (1972), and Katz and Hass (1988). Nevertheless, the resurgence of interest in the con­ cept has occurred principally in the 1990s (see Jonas, Bromer, & Diehl, 200; Priester & Petty, 1996; Thompson, Zanna, & Grif, 1995, for reviews of earlier research with the ambivalence concept). Research with the ambivalence con­ cept since 1990 has principally focused on predictions for the general concept of ambivalence, although there has also been interest in ambivalence in specific domains, such as racist attitudes (e.g., Katz & Hass, 1988), sexist attitudes (e.g., Glick & Fiske, 1996), and consumer attitudes (Otnes, Lowrey, & Shrum, 1997) (see Jonas et al., 200, for a useful review of this research). The present chapter will focus on the general concept of ambivalence. Jonas et a/. (200) provided an earlier review of the literature on ambivalence. That chapter provided an excellent review of the origins and history of the ambivalence concept and of the relative utility of different formulae employed to tap measures of ambivalence. Neither of these issues are the focus of attention of the present chapter. Both chapters consider the implications of treating am­ bivalence as a measure of attitude strength. Jonas et al. focus on distinguishing ambivalence from attitude strength and give relatively brief attention to empirical data supporting the presumed implications of treating ambivalence as a measure of attitude strength. In contrast, the focus in the present chapter is reversed. The first section of the present chapter explores various definitions of the ambivalence concept that have been offered. The differing definitions suggest a number of different forms of ambivalence. In considering the variously used measures of ambivalence, we present three dimensions along which different measures of ambivalence can be clasified. We then present some preliminary data on the interrelationship of these different measures of ambivalence and show that the measures posss discrimiant validity, but suggest that further work on this issue is required. We then go on to explore some of the elicitors of ambivalence. After presenting some preliminary findings on the types of attitude objects that gener­ ate high levels of ambivalence, we examine the relationship between various individual difference variables and ambivalence. The largest section of the present chapter then explores the consequences of holding ambivalent attitudes. Based on treating ambivalence as a dimen-

AMBIVALENCE AND ATTITUDES

39

sian of attitude strength, we examine the evidence to support four predicted consequences of ambivalence-its impacts on: (a) the temporal stability of attitudes; (b) the relationship of attitudes to intentions and behavior; (c) the pliability of attitudes; and (d) the processing of information (Krosnick & Petty, 1995) . The final section of our chapter offers a number of suggestions for further research with the concept of attitudinal ambivalence.

DEFINITIONS OF AMBIVALENCE

Ambivalence has been defined in various ways. Given our belief in the im­ portance of ambivalence to attitude research, we should spend some time on an assessment of different approaches to the concept. Most definitions in the psychological literature make reference to the simultaneous existence ofposi­ tive and negative evaluations of an attitude object (see Table 2. 1 for a selec­ tion). However, there are important differences between the various definitions of the concept that merit some attention. Table 2.1

Interpretations and definitions of ambivalence

"The existence of simultaneous or rapidly interchangeable positive and negative feelings toward the same object or activity, with the added proviso that both the positive and negative feelings be strong" (Meehl, 1964, quoted in Emmons, 1 996) "A psychological state in which a person holds mixed feelings (positive and negative) towards some psychological object" (Gardner, 1987, p. 241 ) "The extent of beliefs' evaluative dissimilarity (or inconsistency)" (Eagly & Chaiken, 1993, p. 123) "A conflict aroused by competing evaluative predispositions" (Breckler, 1994, p. 359) " . . . the extent to which one's reactions to an attitude object are evaluatively mixed in that both positive (favorable) and negative (unfavorable) elements are included" (Wegener, Downing, Krosnick, & Petty, 1995, p. 460) An inclination " . . . to give it [an attitude object] equivalently strong positive or negative evaluations" (Thompson et al., 1995, p. 367). "An approach-avoidance conflict-wanting but at the same time not wanting the same goal object" (Emmons, 1996, p. 326)

First, the idea that positive and negative evaluations need to be simultaneous is not a universal feature of orientations to the ambivalence concept. Ainslie (1986), for example, with his focus on temporal dimensions of choice, suggests that long-term interests (e.g., health) and short-term inter­ ests (e.g., hedonic pleasure) are not invariably evaluated in conjunction with one another to form a consistent choice because each may be "dominant" at

40

MARK CONNER AND PAUL SPARKS

different times. From Ainslie's decision-making perspective on competing choice options, ambivalence is defined as an "inability to make a stable choice between two alternatives despite familiarity with their consequences" (Ainslie, 1992, p. 370). Moreover, a distinction has been made between dia­ chronic ambivalence and synchronic ambivalence (see Stocker, 1990) to re­ flect the distinction between "inconstancy" and "inconsistency" in evaluations. Like diachronic ambivalence, the notion of "attitude variability" (e.g., Sparks, Hedderley, & Shepherd, 1992) focuses on vacillating attitudes over time, whereas the simultaneous experience of positive and negative eval­ uations is central to the (synchronic) ambivalence, which is of predominant current interest to social psychologists researching the relationship between attitudes and ambivalence. However, lay people's responses to, for example, direct measures of ambivalence in research work (see below) will of course reflect their own interpretations of the questions put to them. Second, attitude objects may be, for example, behaviors, targets, events, or states of affairs. People may be ambivalent towards behaviors (e.g., reducing fat consumption), towards goals (e.g., losing weight), towards events (e.g., Millennium celebrations) or states of affairs [e.g., the existence of 70 mph ( 112 kph) motorway speed limits]. They may feel ambivalent about a particular behavior because the behavior has both negative and positive attributes, but there may also be a fundamental ambivalence about any one attribute. In this chapter, we are principally concerned with ambivalence towards behaviors, since most contemporary attitude research also displays this orientation and because we believe that closely specified personal behaviors are an appropri­ ate focus for the study of motivational conflicts. Third, some approaches to ambivalence focus more on people's beliefs, while others more directly on their attitudes. Given that the relationship between beliefs and attitudes may be complex, we have a general preference for the more direct focus on people's attitudes (although exploratory work on both is of great interest), since these are central to the evaluative issues at stake in the notion of ambivalence. While outcome evaluations may be as­ sessed in conjunction with behavioral beliefs within the Theory of Reasoned Action (TRA; Fishbein & Ajzen, 1975) and the Theory of Planned Behavior (TPB; Ajzen, 1991), it is the specific evaluation of behaviors on which we are concentrating here for our assessment of ambivalence. Fourth, it is likely to be fruitful to examine other approaches to am­ bivalence, given that this concept attracts attention across different disci­ plines. For example, ambivalence is implicated in political science perspectives on people's conflicting values (Zaller & Feldman, 1992), so­ ciological work on conflicting role demands (Merton, 1976), and psychological approaches to both the mixed feelings that people may have about the objects of their attitudes (Tourangeau, Rasinski, Bradburn, & D' Andrade, 1989) and to conflicting parts of their personality (Freud, 1962). It is expected that

44

MARK CONNER AND PAUL SPARKS

a formula based on the number of positive and negative attributes selected as being important (see Van der Pligt, De Vries, Manstead & Van Harreveld, 200, for a review of this research). Open-ended belief-based measures em­ ploy a similar approach, but require respondents to self-generate their beliefs in response to a prompt (Bell, Esses, & Maio, 1996). For example, Conner et al. (1999c ), in a study of ambivalence towards eating a low-fat diet, used the prompt, "The thoughts and feelings I associate with eating a low fat diet are . . . " to elicit such beliefs. Participants can also be requested to rate each belief in terms of how positive or negative it is (alternatively, independent judges can rate each belief as positive or negative). The numbers of positive and negative beliefs or the sum of the ratings of the positive and negative beliefs can then be entered into an equation to compute a measure of am­ bivalence (see Erber, Hodges, & Wilson, 1995 for a similar "belief homoge­ neity" measure). The third dimension along which measures of ambivalence can be classified is in terms of the content of ambivalence judgments (focusing on cognitive conflicts, affective conflicts, cognitive-affective conflicts). So, for example, a direct ambivalence measure can refer to mixed cognitions or thoughts (e.g., "I have conflicting thoughts about X", strongly agree-strongly disagree), mixed affect or feelings [e.g., "I have mixed (both positive and negative) feelings about X", strongly agree-strongly disagree) or a conflict between thoughts and feelings (e.g., "I have positive thoughts but negative feelings about X", strongly agree-strongly disagree). As Thompson et al. (1995) make clear, this latter form of ambivalence measure (i.e., the heart vs. the mind) can take two distinct forms: pro-cognition and con-affect versus con-cognition and pro­ affect. For example, an individual may like the taste of one particular food, but dislike the high number of calories that the food contains (pro-affect, con­ cognition) or dislike the taste of another food, but like the low number of calories the food contains (con-affect, pro-cognition) (Grogan, Bell, & Con­ ner, 1 997). These different contents of ambivalence measures may be dif­ ferentially important in different areas (see also MacDonald & Zanna, 1998). For example, while thinking about an event such as the introduction of a new social policy may arouse principally cognitive ambivalence, thinking about a particular politician (an object) may arouse principally affective ambivalence. Similarly, thinking about an action or behavior (e.g., using a condom with a new sexual partner) may arouse conflicting thoughts and feelings. Ortony, Clore, and Collins (1988) similarly propose three aspects of the world that we can appraise emotionally: events, actions, and objects. However, they further suggest that, within these categories, it is only appraisals of events and actions that are likely to produce conflicting emotions (and ambivalence) because these appraisals can focus on consequences for self and other or on the immediate future and the more distant future (pp. 51-52). In this scheme, objects only produce positive or negative reactions for the self in the immedi-

45

AMBIVALENCE AND ATTITUDES

ate future. However, appraisal of objects and events/actions could produce between-category ambivalence. For example, a disliked person (object) per­ forming a kindness (action) might produce such conflicting emotions. The best way to combine separately obtained positive and negative judg­ ments of an attitude object into an overall measure of ambivalence has re­ ceived attention in a number of studies (Breckler, 1 994; Jonas et a/., 2000; Priester & Petty, 1996; Thompson et al., 1995) . Kaplan (1972) suggested that in a measure of ambivalence, scores should increase as the positive and nega­ tive judgments become more polarized (both larger in value) and the more equal they are in absolute value. A variety of measures of ambivalence have been proposed to capture these conditions in a measure of ambivalence (see Jonas et al., 2000, for a recent review). Perhaps the most widely employed such equation is that devised by Griffin and presented in Thompson et al. (1995) : ambivalence is calculated as half the polarization of the positive (P) and negative (N) judgments, minus the absolute difference between the two. That is, ambivalence is denoted by the equation: Ambivalence

=

( P +N)/2 - IP - N1

where P and N are measured on unipolar scales (in absolute values) in two separate questions. So, for example, an individual who rates an attitude object as both moderately positive (2) and moderately negative (2) would be as­ signed an ambivalence score of [(2 + 2)/2 - 12 - 21 = 2], while an individual who rates the same attitude object as extremely positive (3) and only mildly nega­ tive ( 1 ) would be assigned an ambivalence score of [(3 + 1 )/2 - 13 - 1 I = 0]. This equation is the one employed in all the analyses reported in the present chapter where separate positive and negative ratings are combined into a single index of overall ambivalence. Table 2.2 presents the different dimensions of ambivalence and provides examples of items used to measure ambivalence. In general, very few studies have reported comparison amongst these different measures of ambivalence. Thompson et al. (1995) report a significant correlation of 0.40 between a direct measure of ambivalence and a multi-item split semantic differential measure (with ambivalence calculated using the Griffin formula; n = 146). Lipkus, et al. (in press) reported a significant correlation of just 0.19 between the same measures in a study of smoking (n == 157). In our own research we have compared levels of ambivalence based on open-ended belief-based measures of ambivalence and split semantic differential-based measures of ambivalence (Conner et al., 1999c) and employed the Griffin equation to compute a mea­ sure of overall ambivalence. For both behaviors examined (eating a low fat diet, n = 140; eating five portions of fruit and vegetables per day, n = 142), the degree of correspondence was only modest ( r = 0.18 and r = 0.27 respectively, both significant). Hence, across the studies that have reported the relevant

46

MARK CONNER AND PAUL SPARKS

data, there appears to be evidence that the different measures of ambivalence do not show strong convergent validity. We believe that an important area for further research with the am­ bivalence construct is a systematic examination of the convergent and dis­ criminant validity of the various ambivalence measures employed across studies (Table 2.2). Without such data it is difficult to offer good advice about the most appropriate means to measure ambivalence. Each of the different measures has potential problems. For example, we have reservations about direct measures of ambivalence because it is not clear to what extent individ­ uals have access to the relevant information on which to base such judgments. In addition, as noted earlier, there may be individual differences in the way ambivalence is interpreted, posing further problems for the potential useful­ ness of direct measures of ambivalence. For example, Priester and Petty (2001) have recently reported data to indicate that direct measures of am­ bivalence tap both intra personal ambivalence and discrepancy between one's own attitude and that of valued others (i.e., interpersonal ambivalence). We also have some concerns about split semantic differential measures of ambivalence because it is not clear whether respondents are able to comply with instructions to ignore judgments of the opposite valence (e.g., ignore negative evaluations when making a positive evaluation) when making a judg­ ment. In addition, there is as yet only limited evidence to support the view that the positive and negative evaluations supposedly assessed by such mea­ sures are both available for retrieval from memory (Cacioppo, Gardner, & Bertson, 1 997). However, we should note that much of our own research employing measures based on obtaining separate positive and negative judg­ ments has produced findings consistent with theoretical predictions. Hence, such measures may have considerable validity. Nevertheless, we also see the value of belief-based measures of am­ bivalence (see Conner et al., 1999c) as have other authors (e.g., Haddock & Zanna, 1998; van der Pligt et al., 200). In fact, we have found both open­ ended and closed-ended belief measures of ambivalence to be useful (see Haddock & Zanna, 1998; Esses & Maio, this volume, for useful discussions of open-ended measures of attitudes). Belief-based measures of attitude are consistent with Erber et al.'s (1995) attitudes-as-constructions model and al­ low one to examine cognitive, affective, and cognitive/affective ambivalence by classifying the different beliefs. However, such measures are also not with­ out problems. For example, it is not clear that individuals can easily report all the information upon which their overall attitude judgments are based (Fish­ bein & Ajzen, 1975; Sparks, 1994). Respondents may also find it easier to generate cognitive correlates of their attitudes compared to other influences such as affective factors (Conner et al., 1999c; Sparks, 1994). This appears to be in stark contrast with evidence that affect-based evaluations are more accessible in memory than cognitive-based evaluations (Verplanken, Hofstee,

AMBIVALENCE AND ATIITUDES

47

& Janssen, 1998). Hence, belief-based measures of ambivalence and split semantic differential measures may be based upon different information, thus accounting for their lack of convergent validity. Nevertheless, our own re­ search (Conner et al., 1999c) demonstrates that despite this lack of convergent validity, belief-based and split semantic differential measures of ambivalence are equally well able to moderate the attitude-behavior relationship (a key prediction for an ambivalence construct).

ELICITORS OF AMBIVALENCE

Comparatively little research has examined the possible elicitors of am­ bivalence. This section reports findings from two approaches to understanding the bases of ambivalence. The first approach examines what things commonly produce higher and lower levels of ambivalence across individuals. The sec­ ond approach examines individual difference variables assumed to be import­ ant in the genesis of ambivalent attitudes. Few studies have investigated what it is that tends to produce considerable degrees of ambivalence. Theoretically we might expect anything that tends to generate both strong positive and negative evaluations to be more likely to produce higher levels of ambivalence. One consequence of the work of Ortony et al. (1988), referred to earlier, is that actions/events likely to produce different emotions when focusing on the immediate future compared to the more distant future should produce greater ambivalence. Our previous re­ search had indicated that this conflict was particularly common for health­ related behaviors, where the conflict was between the immediate pleasantness/unpleasantness of the behavior and the longer-term positive/ negative health outcome of the behavior. Two types of health-related be­ havior can be distinguished: health protective behaviors, which can be un­ pleasant or onerous (e.g., exercising compared to non-exercising), and health­ risking behaviors, which are pleasant (e.g., drinking alcohol compared to not drinking alcohol). We did not expect clear differences in ambivalence be­ tween health-protective and health-risking behaviors. Conner, Povey, Sparks, James, and Shepherd (1998a) identified 12 such health behaviors relevant to students. These behaviors varied in overall evaluation and also in whether engaging in the behavior was health-protective or health-risking. Split seman­ tic differential measures were obtained in relation to these 12 different be­ haviors from a group of 143 students. The positive and negative evaluations were combined into measures of ambivalence using the equation suggested by Griffin (Thompson et al., 1995) and overall evaluation (by summing the two evaluations). The results are shown in Table 2.3. There were clearly consider­ able differences in the extent to which these different behaviors produced feelings of ambivalence. Drinking alcohol produced the highest degree of

50

MARK CONNER AND PAUL SPARKS

and personal fear of invalidity being positively related were only apparent when respondents were highly involved with an issue. Hence, the mechanism by which these personality variables influence ambivalence (i.e., through in­ fluencing propensity to engaging in resolving evaluatively inconsistent reac­ tions) may only operate when individuals are highly involved with an issue. However, this effect may be relatively weak and only be detectable when considering the average ambivalence across several attitude topics. For ex­ ample, Sparks et al. (1995) found no evidence that a split semantic differential measure of ambivalence about eating chocolate or meat was significantly related to need for cognition, need for closure (Kruglanski, Webster, & Klem, 1993), or self-monitoring (Snyder & Gangestad, 1986). In addition, there was no evidence for a moderating effect of involvement with the behavior. Hence, we believe the current evidence that individual difference variables elicit ambivalence is weak. We suspect that more specific research hypotheses are required in this area. One interesting hypothesis is that the consequences of ambivalence may be different for different personality types (e.g., am­ bivalence may be more unpleasant or anxiety-provoking for those intolerant of ambiguity). As the above review makes clear, we still know comparatively little about the elicitors of ambivalence or about how ambivalence might best be manip­ ulated. These are important areas for further research (see below).

CONSEQUENCES OF AMBIVALENCE

If we consider attitudinal ambivalence as a dimension of attitude strength, then we might anticipate four consequences of holding an ambivalent attitude (Converse, 1995, p. xi; Krosnick & Petty, 1995): (a) impacts on the temporal stability of attitudes; (b) impacts on the relationship of attitudes to intentions and behavior; (c) impacts on the pliability of attitudes; and (d) impacts on information processing. A number of authors have investigated the effects of attitudinal ambivalence in relation to each of these consequences, although the evidence in support of each effect varies considerably. In the following sections we review the research in relation to each of these four consequences of having an ambivalent attitude, focusing on the findings from our own research. Temporal Stability

A variety of studies have demonstrated that strong attitudes are more stable over time. Indeed, a number of authors have argued that attitude stability is the factor that all attitude strength measures hold in common ( Ajzen, 1996; Doll & Ajzen, 1992; Erber et al., 1995). Other research with intentions (e.g.,

52

MARK CONNER AND PAUL SPARKS

In a second and more defensible approach, ambivalence has been related to more indirect measures of attitude stability (i.e., operative measures). For example, Bargh, Chaiken, Govender, and Pratto (1992) showed that respond­ ents who gave the same evaluation to an attitude object at two time points tended to have less ambivalent attitudes than those who did not give the same evaluation, (r = 0.27, significant). Bassili (1996) reported contrasting results; ambivalence was unrelated to stability for attitudes towards introducing quotas, r = 0.00, or towards pornography, r = 0.03. In our own research we have examined the stability predictions in one study (Armitage & Conner, 200). These authors (study 1) examined the impact of ambivalence on the stability of attitudes towards eating a low-fat diet over three time points in a sample of 346 hospital workers. Am­ bivalence was assessed using a split semantic differential measure, while attitudes were assessed using a reliable multi-item bipolar semantic differen­ tial measure. As attitudes were measured at three time points, a total of three comparisons were possible, representing time intervals of 3 months, 5 months and 8 months between measurement of attitudes. The extent to which ambivalence moderated the degree of attitude stability was investi­ gated in two ways. First, we computed a within-subjects correlation measure of attitude stability. Across respondents, this measure of the stability of attitudes was unrelated to our measure of ambivalence (over 3 months, r = -0.02; over 5 months r = -0.01; over 8 months, r = -0.01 ). Second, we attempted to discount any problems attributable to error variance by com­ puting measures of attitude stability based on between-subject correlations disattenuated for measurement error (using LISREL 8; Joreskog & Sorbom, 1993). In this analysis correlations were computed separately for the higher and lower ambivalence groups (based on a median split). In general, the degree of attitude stability was high (mean disattenuated r = 0.71), although, as might have been expected, stability was inversely related to the time interval between measurements. In no case did the differences in disattenu­ ated correlations reach significance (over 3 months, Zr = 0.00; over 5 months, Zr = 1 .59; over 8 months, Zr = 1 .24 ), although the difference was in the predicted direction (i.e., less stability in the high ambivalence group) for comparisons over the longer time intervals. Thus, research provides only limited support for the predicted impact of ambivalence on attitude stability. Whilst self-reported stability of attitudes appears to be relatively consistently related to ambivalence, designs employ­ ing prospective measures of stability do not tend to observe this relationship of ambivalence to attitude stability. The studies based on self-report measures of stability are particularly open to consistency biases. The one study report­ ing a significant effect using a more defensible measure of attitude stability (Bargh et al., 1992) employed a relatively small sample (but see Jonas et al., 200, for report of a further study producing supportive results).

AMBIVALENCE AND ATTITUDES

53

Attitude-Intention and Attitude-Behavior Relationships

A key prediction for a measure of attitude strength is that strong attitudes should be more likely to guide behavior than weak attitudes (Converse, 1 995, p. xi; Krosnick & Petty, 1995, p. 3). As Schwartz (1978) noted, an attitude assessed at one time is unlikely to predict behavior at a later time if the attitude does not persist over the intervening time interval. Fishbein and Ajzen (1975) make a similar observation in relation to the impact of inten­ tions on behavior (see Conner et al., 2000}. Some researchers (e.g., Erber. et al., 1 995) have argued that the greater impact of strong attitudes on behavior is solely attributable to strong attitudes being more likely to persist over time. However, other mechanisms may also influence the ability of strong attitudes to influence behavior. For example, Fazio (1986, 1995) has argued that atti­ tudes influence our behavior in part by shaping our perceptions of the world (the automatic activation effect). That is, the capacity of an attitude to predict behavior is partly dependent on the attitudes' ability to bias perceptions of the attitude object and the context in which the behavior is performed. Strong attitudes are assumed to be more readily accessible and so more likely to produce these biasing effects. Thus, one might therefore expect low­ ambivalent attitudes to be more capable of automatic activation and so more predictive of behavior than high-ambivalent attitudes. For many attitude objects that elicit high levels of ambivalence, it is useful to consider consistency of expressed attitudes over context as well as consis­ tency over time. Ambivalence towards a behavior may be associated with a temporal and/or contextual instability in dominant motives (or evaluations) that arises as a consequence of fluctuations in the salience and/or strength of different motives. Ambivalence may not only reflect "mixed feelings" to­ wards an attitude object at a particular moment in time but also changes in the dominance of competing motives at different points in time as a result of variations in internal and external contextual cues ( cf. Ainslie, 1992) . While measured attitudes may show a consistency over time if elicited in the same context, this consistency may not be achieved if variations in context are introduced into the measurement procedure: for example, many people's expressed attitude towards eating fruit as dessert may be stable over time if measured in similar laboratory conditions and may only show variation when measured after dinner in full sight of the chocolate cake alternative ! The strength of the relationship between attitudes, intentions, and behavior may be expected to be influenced by whether or not the same evaluations or motives are dominant when attitudes, intentions, and behavior are measured (cf. Ajzen, 1996; Tesser, 1978). Since ambivalence is associated with mixed or conflicting motives, the strength of measured attitude-intention-behavior linkages may be consequently attenuated at higher levels of ambivalence (compared to lower levels of ambivalence).

54

MARK CONNER AND PAUL SPARKS

In addition, much research in the attitude domain has considered be­ havioral intentions as mediators of attitude-behavior relations (e.g., Ajzen, 1991; Fishbein & Ajzen, 1975) and a considerable body of research supports this relationship (see Armitage & Conner, in press a; Conner & Armitage, 1 998; Conner & Sparks, 1996; Eagly & Chaiken, 1993; Godin & Kok, 1996; Sheppard, Hartwick, & Warshaw, 1988; Van den Putte, 1991). Indeed, in researching the attitude-behavior relationship, many studies have employed measures of behavioral intention as a proxy for actual behavior. Previous work on attitudinal ambivalence has examined the moderation of either attitude-intention relations (e.g., Jonas, Diehl, & Bromer, 1997) or attitude­ behavior relationships (e.g., Lavine et al., 1998). The earliest examination of the moderation effect was by Moore (1973, 1980). His research demonstrated a moderating effect of attitudinal am­ bivalence on the relationship between attitudes and behavior, such that there was lower consistency between attitudes and behavior for those high in at­ titudinal ambivalence. However, measures of attitude and behavior were weak and taken in a cross-sectional manner. Thus, this moderation effect is more accurately described as influencing attitude-past behavior relationships. Sparks et al. (1992) examined the moderating role of a mixed measure of attitude ambivalence and attitude variability on attitude-intention relation­ ships in relation to food choices. Higher levels of ambivalence were associated with an attenuation of the attitude-intention correlation. Norman and Smith {1995) also used a mixed measure of attitude ambivalence and attitude vari­ ability, but found no moderating effects on belief-attitude or attitude­ intention relationships for exercise behavior. However, in comparing dif­ ferences in zero-order correlations they failed to take account of potential diferences in variance in the higher and lower ambivalence groups (Aiken & West, 1991 ; Baron & Kenny, 1986). This problem can be addressed through the use of moderated regression analysis or structural equation modeling. Jonas et al. (1997), in a laboratory study, found evidence of stronger attitude-behavioral intention relations for more ambivalent attitudes. Finally, research by Lavine et al. (1998) reported that attitude-behavior rela­ tions were attenuated by high levels of affective-cognitive inconsistency. Hence, previous studies have produced mixed findings on the moderating impact of ambivalence on attitude-intention and attitude-behavior relation­ ships. However, each of the above studies can be criticized in relation to the design (Moore, 1973, 1980), the mixed measure of ambivalence employed (Lavine et al., 1998; Norman & Smith, 1995; Sparks et al., 1992), the analysis method employed (Norman & Smith, 1995), or the focus on laboratory­ generated attitudes (Jonas et al., 1997). Our own research has attempted to address this issue through five cross-sectional studies on attitude-intention relationships (Conner & Flesch, 1998; Conner, Povey, & Sparks, 1995; Conner et al., 1 998b; Conner, Povey, Sparks, James, & Shepherd, in press; Sparks et

58

MARK CONNER AND PAUL SPARKS

measures of attitude (semantic differential measure) and ambivalence (split semantic differential measures) were taken in an initial questionnaire; while self-report measures of behavior (e.g., "I have eaten a low-fat diet in the past month", strongly agree-strongly disagree) were taken in a later question­ naire. The studies examined healthy eating (Conner et al., in press), eating a low-fat diet (Conner et a/., 199c; Conner et a/., 1995), or eating five portions of fruit and vegetables per day (Conner et al., 199c). The results from these studies are summarized in Table 2.4. For presentational purposes we again split the sample at the median on the ambivalence measure in order to create a higher and a lower ambivalence group. Correlations between attitude and behavior measures were computed separately for these two groups and then compared. As can be seen in Table 2.4 the results were fairly consistent in showing the predicted moderation effect. The one study failing to show a significant effect (Conner et al., in press) employed a single-item measure of ambivalence, which may have compromised its reliability. Across studies the frequency-weighted correlation shows this effect to be substantial, with more ambivalent attitudes being less strongly correlated with behavior assessed prospectively, r = 0.40 vs. r = 0.53 (Table 2.4). In addition, this effect was also apparent in within-subjects correlations for healthy eating in a prospective design (Conner et al., in press). The lower ambivalence group (M 0.57; SD = 0.37) and the higher ambivalence group (M = 0.47; SD = 0.35) differed signifi­ cantly on attitude-behavior correlations in the predicted direction (i.e., weaker correlations for the higher ambivalence group). One problem with the above studies is the failure to take account of the effect of ambivalence on attitude-intention and attitude-behavior relation­ ships simultaneously. Armitage and Conner (200, Study 1) attempted to rectify this problem by taking measures of attitude, intention, and behavior at three different time points. In addition, this study used structural equation modeling in order to remove any effects due to measurement error. Data were collected from a sample of 350 hospital workers in relation to eating a low-fat diet. The measure of ambivalence was a split semantic differential measure that was used to split the group into a higher and lower ambivalence group (based on a median split). We then fitted a model to the two groups with structural paths from attitude to intentions, from attitude to behavior, and from intentions to behavior. As the interest was in the relationship be­ tween the latent constructs (attitude, intentions, and behavior), we con­ strained the model such that the factor loadings for each construct were equal in the two groups (the factor loading constraint model). This allows us to make direct comparisons between the regression paths for the higher and lower ambivalence groups. As predicted, the path from attitude to intention was stronger in the lower compared to the higher ambivalence group. Also as predicted, the path from attitude to behavior was stronger in the lower com­ pared to the higher ambivalence group. Formal tests of these differences =

AMBIVALENCE AND ATTITUDES

59

revealed them both to be significant. These effects were present despite any evidence that ambivalence influenced attitude stability in the same data. Thus, there appears to be growing evidence that ambivalence moderates the attitude-intention relationship and, particularly, the attitude-behavior relation­ ship, such that higher levels of ambivalence are related to weaker relationships. It would also appear that such effects are not purely attributable to attitude stability effects. It has been suggested that this moderation effect may be due to ambivalence being confounded with attitude extremity. For example, Thompson et al. (1995) noted that attitudinal ambivalence is necessarily nega­ tively correlated with attitude extremity (p. 382), that is, univalent attitudes should be more extreme than ambivalent attitudes and therefore more predict­ ive of behavior. This hypothesis is directly supported by Lavine et al. (1998; but see Jonas et al., 1997) and indirectly supported by Kallgren and Wood (1986). The latter compared experimental participants, whose attitude extremity had been manipulated, with a no-treatment control group, and reported attitude­ behavior correlations of 0.85 and 0.19 for the experimental and control condi­ tions, respectively. One might therefore expect less ambivalent attitudes to be more predictive of behavior than ambivalent attitudes, just because of greater attitude extremity. Nevertheless, there is evidence that the moderation effects observed for ambivalence are not merely attributable to differences in attitude extremity. For example, Conner et al. (in press) while reporting the predicted attitude-intention and attitude-behavior moderation effects for ambivalence, found these relationships remained even when ambivalence was unconfounded from extremity. Most convincing, however, is the data reported by Priester and Petty (2000). In a study of safer sex behaviors, they compared a lower and a higher ambivalence group. Ambivalence was found to moderate the attitude­ behavior relationship. They then examined only those respondents checking the most positive (extreme) attitude category. In this case, lower-ambivalent respondents were significantly more likely to report performing the behavior than higher-ambivalence respondents, despite being identical on attitude ex­ tremity. In examining the moderation effects of ambivalence on attitude­ intention and attitude-behavior relationships, it is clearly important to ensure that these effects are not simply attributable to attitude extremity effects. In addition, these reported moderation effects need to be confirmed with objective measures of behavior and in experimental studies. Pliability of Attitudes

More ambivalent attitudes are expected to be more susceptible to the influ­ ence of a persuasive communication (i.e., more pliable) . For example, Eagly and Chaiken (1995) argue that, "attitudes are strong to the extent that they are well embedded in an existing attitudinal structure" (p. 414). Thus strong attitudes are held to be more securely "anchored" in knowledge structures.

60

MARK CONNER AND PAUL SPARKS

Given that ambivalent attitudes are based on conflicting evaluations and in­ consistent information, they should be more weakly anchored and hence more pliable than univalent attitudes. Only a few studies have investigated this prediction for ambivalence. Bassili (1996) tested the pliability of attitudes towards three attitude objects in the face of persuasive messages. Pliability was found to be significantly related to ambivalence for two attitude objects (r = 0.15 in both cases), but not for a third (two tests: r = 0.06 and r = 0.03) (see also Pomerantz, Chaiken, & Tordesillas, 1995). MacDonald and Zanna (1998) showed individuals high in (cross-dimensional) ambivalence about a particular social group to be more susceptible to a priming manipulation in terms of their judgment about that group. Finally, Maio, Esses, and Bell {200) found an interaction between strength of message and ambivalence on attitudes towards residents of Hong Kong, such that only in the high ambivalence group did message strength produce differential effects on attitude change. In our own research we have investigated the impact of ambivalence on the pliability of attitudes in one study {Armitage & Conner, 200, Study 2). The study was a simple pre-post quasi-experimental design. Attitudes and levels of ambivalence towards eating a low-fat diet were assessed at baseline in a sample of 344 hospital workers. Attitude measures were based on a multi­ item semantic differential measure, while ambivalence was assessed using a split semantic differential measure. Five months later, participants were ran­ domly assigned to either an attitude change or control condition and received experimental materials designed either to change attitudes or provide infor­ mation only. Both sets of intervention materials included UK Government recommendations on dietary fat intake, epidemiological data on population levels of fat intake, and a description of sources of fat in the diet. The experi­ mental intervention also included a section designed to change individual attitudes (see Armitage & Conner, in press b). Based on Fishbein and Ajzen's (1975) model of belief-attitude relations, we provided persuasive information that targeted five beliefs about eating a low-fat diet, which had previously shown to distinguish who did and did not intend to eat a low-fat diet (Armi­ tage & Conner, 1999). Analysis of variance revealed that the experimental and control groups did not differ at baseline on the attitude measures. In order to test the prediction that less ambivalent attitudes are more resistant to change, we split the group into higher and lower levels of ambivalence, based on a median split of this measure. Across both our intervention conditions, attitudes became more positive, although the change was greater in the experimental group. Of par­ ticular interest was the three-way interaction between condition, am­ bivalence, and time. This only approached significance. Nevertheless examination of the effects of the intervention demonstrated that it had dif­ ferential effects in the two ambivalence groups. For the lower ambivalence

AMBIVALENCE AND ATTITUDES

61

group, there was no difference between the control and experimental condi­ tions; while for the higher ambivalence group, attitudes becam� significantly more positive following the experimental intervention. Hence, the attitude intervention (experimental) group had significantly more impact on more ambivalent attitudes supporting the pliability prediction. Hence, the evidence to date generally supports the idea that ambivalent attitudes tend to be more pliable in the face of persuasive messages. Nev­ ertheless, further studies across a more diverse range of attitude objects are required. Information Processing Impacts of Ambivalence

A further predicted consequence of attitudinal ambivalence is an impact on the processing of attitude-relevant information. Attitudes characterized by higher levels of ambivalence are assumed to be less likely to guide process­ ing of information about the attitude object, because of their lower accessib­ ility. Several studies have found evidence to support the relationship of higher levels of ambivalence to lower attitude accessibility (Bargh et al. , 1 992; Bassili, 1 996; Bromer, 1 998; see Jonas et al., 2000, for a review). For example, Van der Pligt et al. (2000) hypothesized that those individuals with ambivalent attitudes would take longer to integrate information into an overall attitude judgment compared to non-ambivalent individuals, and hence ambivalent attitudes would show longer response latencies when compared to non-ambivalent attitudes. Across three different attitude ob­ jects support was found for this effect. In each case the response latency for an overall attitude judgment was longer for ambivalent compared to non­ ambivalent individuals. In our own research we have investigated this effect in one study (Conner, Bainbridge, & Halton, 1 999a). This study used a response latency paradigm to examine the time taken to respond to different attitude objects. Consistent with previous research, we expected bipolar attitude judgments to show longer latencies when the respondents reported higher degrees of am­ bivalence. Ambivalence was assessed using split semantic differential mea­ sures to separately assess positive and negative reactions to the attitude object. However, we further hypothesized that ambivalence would only influ­ ence the response latency to the bipolar attitude measure when the two unipolar judgments had similar short latencies. Consistent with Erber et al. 's (1995) attitudes-as-constructions model, this prediction was based on the view that, in arriving at an overall judgment about an attitude object (i.e., as tapped by the bipolar rating), respondents can take account of both negative and positive evaluations (i.e., as tapped by the unipolar ratings) , but will only do so if they are salient (i.e., have a shorter latency). Thus, only when both unipolar judgments are salient and in conflict (i.e., ambivalent) will response

62

MARK CONNER AND PAUL SPARKS

latencies for the overall attitude judgment be longer. Within-subject analyses supported this view, with latencies for bipolar attitude judgments being sig­ nificantly longer when the respondent was ambivalent about the same attitude object and both unipolar evaluations had similar and short response latencies (see McGregor, Newby-Clark, & Zanna, 1999, for a similar finding). Thus, there appears to be reasonably good evidence that ambivalence leads to lower accessibility. However, the evidence that this reduced accessibility has significant impacts on the processing of information about the attitude object has been addressed in fewer studies. Several studies appear to support the idea that higher levels of ambivalence lead to a more systematic process­ ing of attitude-relevant information. The assumption is that an ambivalent attitude is weaker than a non-ambivalent attitude and so is less likely to produce a biased processing of attitude-relevant information (e.g., informa­ tion consistent with one's initial attitude receives more favorable processing than information that is inconsistent). For example, Maio, Bell, and Esses (1996) showed that higher levels of ambivalence about a minority group resulted in more systematic processing of information about that group. Similarly, Bromer (1998) showed that only individuals lower in ambivalence showed a preference for attitude-congruent information in processing mes­ sages about European Monetary Union. Jonas et al. (1997) suggest that the mediating mechanism is confidence in one's attitude: high ambivalence pro­ duces a reduction in confidence in one's attitude and reduced confidence leads to more systematic processing of information. Summary of Research Findings on the Consequences of Ambivalence

As the above review demonstrates, there is a growing body of research on the differing consequences of ambivalence. The findings support the view that ambivalence is an important issue for those involved in attitude research (Breckler, 1994; Thompson et al., 1995; Priester & Petty, 1996), and that ambivalence produces many of the effects one might expect of a measure of attitude strength. Perhaps the strongest evidence concerns the moderating effects of ambivalence on attitude-intention and attitude-behavior relation­ ships. Those with ambivalent attitudes show significantly weaker relationships between their attitudes and intentions or attitudes and behavior. We would anticipate that a range of behaviors that show conflict between cognitive and affective influences (e.g., safer sex practices, exercise, food choices) might be characterized by high levels of ambivalence and so show attenuated attitude­ behavior relationships. However, the generality of the moderation effect across different behaviors still needs to be established and demonstrated experimentally. For example, Jonas et al. (1997) and Maio et al. (1996) suggest that ambivalence can, under certain circumstances, lead to stronger attitude­ behavior relationships.

AMBIVALENCE AND AITITUDES

63

The predicted effects of ambivalence on the stability and pliability of attitudes were less consistent. The prediction that less ambivalent attitudes will be more stable over time has only been confirmed with meta-attitudinal measures of stability and has been refuted in two studies using more appropriate measures of attitude stability (Bassili, 1996; Armitage & Conner, 200, Study 1; but see Bargh et al., 1992). Similarly, the prediction that ambivalent attitudes wil be more pliable has only received limited support to date (Armitage & Conner, 200, Study 2; Bassili, 1996). In relation to information-processing effects, several stud­ ies demonstrate that ambivalent attitudes are less accessible. However, only two studies (Bramer, 1998; Maio et al., 1996) have shown impacts of ambivalence on information processing; greater ambivalence appears to be associated with more systematic processing of attitude-relevant information. Given that ambivalence measures show some evidence of producing all the effects one might expect of a measure of attitude strength, one might ask what is the relationship of ambivalence to other measures of attitude strength. Although there have been several empirical studies of the relationships amongst attitude strength measures (e.g., Bassili, 1 996; Eagly & Chaiken, 1995; Jonas et al., 2000; Krosnick, Boninger, Chuang, Berent, & Carnot, 1993; Pomerantz et al., 1995) the relationship of ambivalence to the full range of attitude strength measures remains under-researched. Ambivalence does tend to be associated with low accessibility of the attitude (Bargh et al., 1992; Bramer, 1998), lower attitude extremity (Maio et al., 1996), and lower attitude certainty (Bassili, 1996; Jonas et al., 1997), and unrelated to evaluative­ cognitive or evaluative-affective consistency (Maio et al., 1996, 2000). However, the relationship of ambivalence to other attitude strength dimen­ sions, such as attitude importance, knowledge, interest, and intensity remains an issue for future research.

FUTURE DIRECTIONS FOR RESEARCH WITH THE AMBIVALENCE CONSTRUCT

Whilst our review has demonstrated that we now have a greater understand­ ing of the ambivalence construct, particularly in relation to some of the con­ sequences of ambivalence, we believe that a number of areas are worthy of further research attention. We would highlight three such key areas for re­ search. The first is the need to further explore the interrelationship of am­ bivalence measures. As the above review makes clear, there is evidence for the divergent validity of various different ambivalence measures. However, there is also a need for researchers working with the ambivalence concept to be clear about their interpretation of the concept, in order that the area is not dogged by differences of interpretation that are not made explicit. Further research is necessary to identify the relationships amongst ambivalence

64

MARK CONNER AND PAUL SPARKS

measures across different attitude objects. A multi-trait multi-method matrix study (see Campbell & Fiske, 1959) might be particularly informative in this regard in isolating the degree to which variations were due to differing types of ambivalence measurement method (e.g., open-ended vs. closed-ended measures; direct ambivalence vs. split semantic differential measures; split semantic differential vs. belief-based measures) or attributable to different attitude objects (e.g., events, objects or actions). A second key issue for future research is the factors that elicit ambivalence. We still know relatively little about which attitude objects produce consider­ able levels of ambivalence or the factors promoting the formation of am­ bivalence. Future studies might usefully employ experimental methods to study such effects (e.g., see Jonas et al., 1997), although it is not clear what might represent the best means for manipulating ambivalence (Wegener et al., 1995). A third key area is the consequence of holding an ambivalent attitude, and in particular the implications for information processing. Whilst a number of studies have examined other consequences of holding an ambivalent atti­ tude, only a few studies have considered the implications for information processing (but see Bromer, 1998). An investigation of these effects may throw light on a number of the other consequences of ambivalence. Other information-processing implications of holding an ambivalent attitude also deserve further attention.

CONCLUSIONS

In the present chapter, we have reviewed the various definitions of am­ bivalence that have been suggested in the literature and indicated a number of different forms of ambivalence. We have presented methods for measuring different forms of ambivalence and some preliminary findings about their degree of interrelationship. We have also described findings relating to the elicitors of ambivalence. In the major section of the chapter, we then reviewed findings for the ambivalence construct in relation to four major predictions based on considering ambivalence a measure of attitude strength: temporal stability of attitudes, impact of attitudes on intentions and behavior, pliability of attitudes, and impacts on information processing. There is evidence for each consequence, although the strongest evidence supports the moderating role of ambivalence on attitude-intention and attitude-behavior relation­ ships. Finally, we outlined a number of suggestions for further research: in particular, the need to further explore the interrelationship of ambivalence measures, the key determinants of ambivalence, and the information­ processing implications of holding ambivalent attitudes. Further exploration of the concept of ambivalence promises to provide us with further insights into broader questions about the structure of attitudes.

66

MARK CONNER AND PAUL SPARKS

Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant validity by the multitrait-multimethod matrix. Psychological Bulletin, 56, 81-105. Chaiken, S., Pomerantz, E. M., & Giner-Sorolla, R. ( 1995). Structural consistency and attitude strength. In R. E. Petty & J. A. Krosnick (Eds), Attitude Strength: Antece· dents and Consequences (pp. 387-412). Mahwah, NJ: Erlbaum. Conner, M., & Armitage, C. J. (1998). Extending the theory of planned behavior: A review and avenues for further research. Journal of Applied Social Psychology, 28, 1430-64. Conner, M., Bainbridge, J., & Halton, B. ( 199a). Within-subjects correlations be· tween bipolar and unipolar measures of attitude and intentions: Moderating effects of response latency (unpublished raw data). School of Psychology, University of Leeds, UK. Conner, M., & Flesch, D. (1998). Attitudes, intentions and ambivalence about having casual sex. Unpublished raw data. School of Psychology, University of Leeds, UK. Conner, M., & Flesch, D. (2001). Having casual sex: Additive and interactive effects of alcohol and condom availability on the determinants of intentions. Journal of Ap­ plied Social Psychology, 31, 89-1 12. Conner, M., Povey, R., & Sparks, P. ( 1 995). The role of attitudinal ambivalence in the theory of planned behaviour: A study of low fat diets (unpublished raw data). School of Psychology, University of Leeds, UK. Conner, M., Povey, R., Sparks, P., James, R., & Shepherd, R. (1998a). Understanding dietary choice and dietary change: Contributions from social psychology. In A. Murcott (Ed.), The Nation 's Diet: The Social Science of Food Choice (pp. 43-56). London: Longman. Conner, M., Povey, R., Sparks, P., James, R., & Shepherd, R. (in press). Moderating role of attitudinal ambivalence within the theory of planned behaviour. British

Journal of Social Psychology. Conner, M., Sheeran, P., Norman, P., & Armitage, C. J. (200). Temporal stability as a moderator of relationships in the theory of planned behaviour. British Journal of Social Psychology, 39, 469-93. Conner, M., & Sherlock, K. (1997). Attitudinal ambivalence and the use of ecstasy (unpublished raw data). School of Psychology, University of Leeds, UK. Conner, M., Sherlock, K., & Orbell, S. (1998b). Psychosocial determinants of ecstasy use in young people in the UK. British Journal of Health Psychology, 3, 295-3 18. Conner, M. T., & Sparks, P. (1996). The theory of planned behaviour and health behaviours. In M. Conner and P. Norman (Eds), Predicting Health Behaviour: Re­ search and Practice with Social Cognition Models (pp. 121-62). Milton Keynes: Open University Press. Conner, M., Sparks, P., & Armitage, C. (1 99b). The moderating role of attitudinal ambivalence on attitude-behaviour relationships. Proceedings of the British Psycho· logical Society, 7(1 ) , 58. Conner, M., Sparks, P., Povey, R., James, R., & Shepherd, R. (1996). Attitude­ intention-behaviour links: moderating role of attitudinal ambivalence. Paper pre­ sented at 1 1th General Meeting of the European Association of Experimental Social Psychology, Gmunden, Austria, 1 3-18 July 1996. Conner, M., Sparks, P., Povey, R., James, R., & Shepherd, R. ( 1 99c). Moderator effects of attitudinal ambivalence on attitude-behaviour relationships (manuscript submitted for publication ). Converse, P. E. (1995). Foreword. In R. E. Petty & J. A. Krosnick (Eds), Attitude Strength: Antecedents and Consequences (pp. xi-xvii). Hillsdale, NJ: Erlbaum. Doll, J., & Ajzen, I. (1992). Accessibility and stability of predictors in the theory of planned behavior. Journal of Personality and Social Psychology, 63, 754-65.

68

MARK CONNER AND PAUL SPARKS

Krosnick, J. A., Boninger, D. S., Chuang, Y. C., Berent, M. K., & Camot, C. G. ( 1993). Attitude strength: One construct or many related constructs? Journal of Personality and Social Psychology, 65, 1 132-5 1 . Krosnick, J . A . , & Petty, R . E. (1995). Attitude strength: An overview. In R. E. Petty & J. A. Krosnick (Eds), Attitude Strength: Antecedents and Consequences (pp. 1-24). Mahwah, NJ: Erlbaum. Kruglanski, A. W., Webster, D. M., & Klem, A. (1993). Motivated resistance and openness to persuasion in the presence or absence of prior information. Journal of Personality and Social Psychology, 65, 861-76. Lavine, H., Thomsen, C. J., Zanna, M. P., & Borgida, E. (1998). On the primacy of affect in the determination of attitudes and behavior: The moderating role of affective-cognitive ambivalence. Journal of Experimental Social Psychology, 34, 398-421 . Lipkus, I. M., Green, J . D., Feaganes, J . R., & Sedikides, C. (in press). The relationship between attitudinal ambivalence and desire to quit smoking among college smokers.

Journal of Applied Social Psychology. MacDonald, T. K., & Zanna, M. P. (1998). Cross-dimension ambivalence toward social groups: Can ambivalence affect intentions to hire feminists? Personality and Social Psychology Bulletin, 24, 427-41. Maio, G. R., Bell, D. W., & Esses, V. M. (1 996). Ambivalence and persuasion: The processing of messages about immigrant groups. Journal of Experimental Social Psychology, 32, 513-36. Maio, G. R., Esses, V. M., & Bell, D. W. (200). Examining the conflict between components of attitudes: Ambivalence and inconsistency are distinct concepts. Ca­ nadian Journal of Behavioral Science, 32, 71-83. McGregor, 1., Newby-Clark, I. R., & Zanna, M. P. ( 199). "Remembering" dissonance: Simultaneous accessibility of inconsistent cognitive elements moderates epistemic discomfort. In E. Harmon-Jones & J. Mills (Eds), Cognitive Dissonance: Progress on a Pivotal Theory in Social Psychology (pp. 325-53). Washington, DC: American Psychological Association. Merton, R. (1976). Sociological Ambivalence and Other Essays. New York: Free Press. Moore, M. (1973). Ambivalence in attitude measurement. Educational and Psycholog­ ical Measurement, 33, 481-3. Moore, M. ( 1 980). Validation of attitude toward any practice scale through use of am­ bivalence as a moderator. Educational and Psychological Measurement, 40, 205-8. Norman, P., & Smith, L. (1995). The theory of planned behaviour and exercise: An investigation into the role of prior behaviour, behavioural intentions and attitude variability. European Journal of Social Psychology, 25, 403-15. Ortony, A., Clore, G. L., & Collins, A. (1988). The Cognitive Structure of Emotions. Cambridge: Cambridge University Press. Otnes, C., Lowrey, T. M., & Shrum, L. J. (1997). Toward an understanding of attitudi­ nal ambivalence. Journal of Consumer Research, 24, 80-93. Petty, R. E., & Krosnick, J. A. (Eds) (1995). Attitude Strength: Antecedents and Con­ sequences. Mahwah, NJ: Erlbaum. Pomerantz, E. V., Chaiken, S., & Tordesillas, R. S. (1995). Attitude strength and resistance processes. Journal of Personality and Social Psychology, 69, 408-19. Pratkanis, A. R., Breckler, S. J., & Greenwald, A. G. (1989). Attitude Structure and Function. Hillsdale, NJ: Erlbaum. Priester, J. R., & Petty, R. E. (1996). The gradual threshold model of ambivalence: Relating the positive and negative bases of attitudes to subjective ambivalence. Journal of Personality and Social Psychology, 11, 431-49. Priester, J. R., & Petty, R. E. (200). The relationship of ambivalence to behavior: A reconceptualization of the antecedents and consequences of evaluative tension.

AMBIVALENCE AND AITITUDES

69

Paper presented at the European Association of Experimental Social Psychology Small Group Meeting on the Psychology of Attitudes, Newtown, Wales, 26-28 July 2000. Priester, J. R., & Petty, R. E. (2001). Extending the bases of subjective attitudinal ambivalence: Interpersonal and intrapersonal antecedents of evaluative tension. Journal of Personality and Social Psychology, 86, 19�34. Schwartz, S. H. (1978). Temporal instability as a moderator of the attitude-behavior relationship. Journal of Personality and Social Psychology, 36, 715-24. Scott, W. A. (1966). Measures of cognitive structure. Multivariate Behavioral Research, 1, 391-5. Scott, W. A. (1968). Attitude measurement. In G. Lindzey & E. Aronson (Eds), The Handbook of Social Psychology (Vol. 2, pp. 204-73). Reading, MA: Addison­ Wesley. Scott, W. A. (1969). Structure of natural cognitions. Journal of Personality and Social Psychology, 4, 261-78. Sheeran, P., Orbell, S., & Trafimow, D. (1999). Does the temporal stability of be­ havioral intentions moderate intention-behavior and past behavior-future behavior relations? Personality and Social Psychology Bulletin, 25, 721-30. Sheppard, B. H., Hartwick, J., & Warshaw, P. R. (1988). The theory of reasoned action: A meta-analysis of past research with recommendations for modifications and future research. Journal of Consumer Research, 15, 325-43. Snyder, M., & Gangestad, S. {1986). On the nature of self-monitoring: Matters of assessment, matters of validity. Journal of Personality and Social Psychology, 51, 125-39. Sparks, P. (1994). Attitudes towards food: Applying, assessing and extending the "theory of planned behaviour". In D. R. Rutter & L. Quine (Eds) Social Psychology and Health: European Perspectives (pp. 25-46). Aldershot: Avebury. Sparks, P., Conner, M., James, R., Povey, R., & Shepherd, R. (1997a). Taking am­ bivalence seriously in attitude research. Proceedings of the British Psychological Society, 5(1), 47. Sparks, P., Conner, M., James, R., Shepherd, R., & Povey, R. (1995). Ambivalence about health-related behaviours: Relationships to personality measures (un­ published raw data). Institute of Food Research, Reading, UK. Sparks, P., Conner, M., James, R., Shepherd, R., & Povey, R. (2001). Ambivalence about health-related behaviours: An exploration in the domain of food choice. British Journal of Health Psychology, 6, 53-68. Sparks, P., Hedderley, D., & Shepherd, R. (1992). An investigation into the relation­ ship between perceived control, attitude variability and the consumption of two common foods. European Journal of Social Psychology, 22, 55-71. Sparks, P., James, R., Conner, M. , Shepherd, R., & Povey, R. (1997b). Ambivalence about food: Implications for attitude theory and research. Appetite, 29, 259. Stocker, M. (1990). Plural and Conflicting Values. Oxford: Clarendon Press. · Tesser, A. (1978). Self-generated attitude change. In L. Berkowitz (Ed.), Advances in Experimental Social Psychology, 11, 289-338. Thompson, M. M., Naccarato, M. E., & Parker, K. (1989). Measuring cognitive needs: The development and validation of the personal need for structure (PNS) and personal fear of invalidity (PFI) measures. Paper presented at the annual meeting of the Canadian Psychological Association, Halifax, NS. Thompson, M. M., & Zanna, M. P. (1995). The conflicted individual: Personality-based and domain specific antecedents of ambivalent social attitudes. Journal of Person­ ality, 63, 259-88. Thompson, M. M., Zanna, M. P., & Griffin, D. W. (1995). Let's not be indifferent about (attitudinal) ambivalence. In R. E. Petty & J. A. Krosnick (Eds), Attitude Strength: Antecedents and Consequences (pp. 361 -86). Mahwah, NJ: Erlbaum.

70

MARK CONNER AND PAUL SPARKS

Tourangeau, R., Rasinski, K. A., Bradburn, N., & D'Andrade, R. ( 1 989). Belief access­ ibility and context effects in attitude measurement. Journal of Experimental Social Psychology, 25, 401-2 1 . Van den Putte, B. (1991). 2 0 years o f the theory o f reasoned action o f Fishbein and Ajzen: A meta-analysis (unpublished manuscript). University of Amsterdam. Van Harreveld, F., Van der Pligt, J., & De Vries, N. K. (1 99). Belief importance and ambivalence in expectancy value models (manuscript submitted for publication). Van der Pligt, J., De Vries, N. K., Manstead, A. S. R., & Van Harreveld, F. (200). The importance of being selective: Weighing the role of attribute importance in attitudi­ nal judgment. In M.P. Zanna (Ed.), Advances in Experimental Social Psychology, 32, 1 35-200. Verplanken, B., Hofstee, G., & Janssen, H. J. W. ( 1 998). Accessibility of affective versus cognitive components of attitudes. European Journal of Social Psychology, 28, 22-35. Waterman, M., & Conner, M. ( 199). An open-ended exploration of attitude objects which generate neutral, positive, negative, or both positive and negative evaluations (unpublished raw data). School of Psychology, University of Leeds, UK. Wegener, D. T., Downing, J., Krosnick, J. A., & Petty, R. E. ( 1 995). Measures and manipulations of strength-related properties of attitudes: Current practice and fu­ ture directions. In R. E. Petty, & J. A. Krosnick (Eds ), Attitude Strength: Antecedents and Consequences (pp. 456-87). Hillsdale, NJ: Erlbaum. Wilson, T. D., Lindsey, S., & Schooler, T. Y. (200). A model of dual attitudes. Psychological Review, 107, 1 01-26. Wilson, T. D., & Hodges, S. D. ( 1 992). Attitudes as temporary constructions. In A. Tesser & L. Martin (Eds), The Construction of Social Judgment (pp. 37-65). Hills­ dale, NJ: Erlbaum. Zaller, J., & Feldman, S. ( 1 992). A simple theory of the survey response: Answering questions versus revealing preferences. American Journal of Political Science, 36, 579-616.

Chapter 3

Expanding the Assessment of Attitude Components and Structure: The Benefits of Open-ended Measures Victoria M. Esses University of Western Ontario

Gregory

R.

Maio

Cardiff University, UK

ABSTRACT In this chapter, we describe open-ended measures for assessing the cognitive, affective, and behavioral components of attitudes, and the use of these measures to assess a variety of properties relating to attitude structure. In describing the utility of these measures to assess both intergroup and more general attitudes, we highlight situations in which open-ended measures may have advantages over closed-ended measures, as well as identifying situations in which closed-ended measures may be preferable. We conclude by suggesting that researchers inter­ ested in fully assessing attitudes would do well to include both open-ended and closed-ended measures in pursuit of this endeavor.

An advance in the study of attitudes in the last 10-15 years is the recognition that attitudes (i.e., overall evaluations of attitude objects) are based on several components, or sources of information (e.g., Eagly & Chaiken, 1 993, 1998; Zanna & Rempel, 1988). This has benefits over previous, less delineated Correspondence regarding this chapter should be sent to Victoria M. Esses, Department of Psychology, University of Western Ontario, London, Ontario, Canada N6A 5C2. E mail: [email protected]. European Review ofSocial Psychology, Volume

© 2002 John Wiley &

Sons Ltd.

12. Edited by Wolfgang Stroebe

and Miles Hewstone.

72

VICTORIA M. ESSES AND GREGORY R. MAIO

definitions of attitudes, in that it separates the general evaluative aspect of attitudes from the specific components on which attitudes are based. Two important components of attitudes are cognitions, specific beliefs about an attitude object, and affect, specific emotions and feelings about the attitude object (see, e.g., Eagly & Chaiken, 1993, 1 998; Zanna & Rempel, 1988). Abundant research has shown that it is important to assess these components in addition to the overall evaluation of an attitude object, because the compo­ nents may affect such processes as attitude-relevant message-processing, attitude-behavior consistency, and attitude change (e.g., Edwards, 1990; Edwards & von Hippel, 1995; Millar & Millar, 1990). Given the importance of these attitude components, measuring them accu­ rately is a valuable goal. The basic knowledge gained by assessing the compo­ nents is inextricably linked to the manner in which they are assessed. Consequently, researchers have been developing measures for the assessment of specific attitude components (e.g., Breckler & Wiggins, 1989; Eagly, Mladinic, & Otto, 1994; Esses, Haddock, & Zanna, 1993). In this context, we describe our open-ended measures for assessing the cognitive and affective components of attitudes, and the many benefits that recommend their use. We focus here predominantly on the affective and cognitive components of atti­ tudes because other research (e.g., Edwards, 1990; Millar & Millar, 1990) has highlighted the importance of these components, and they have been the primary focus of our research as well. Nevertheless, as we will discuss in a later section of this chapter, the open-ended procedure can easily be adapted to assess the behavioral component of attitudes as well. Our emphasis on open-ended measures is in contrast to the traditional predominant use of closed-ended measures in attitude research. Open-ended measures have generally been neglected in this domain (although some ex­ ceptions will be noted shortly), perhaps because of a perception that they are too complex or subjective. However, as in other domains (e.g., self­ discrepancy, Higgins, 1987; discourse analysis, Potter, 1998; Reicher & Hopkins, 1996), we believe that it is now time for attitude researchers to take advantage more frequently of the many benefits of open-ended measures. These benefits include not only their wide applicability, but also their poten­ tial for assessing a number of attitude properties related to attitude structure. On the basis of these benefits, and the complementary role that open-ended and closed-ended measures may play in assessing attitudes, we suggest that researchers consider utilizing open-ended measures of attitudes, in addition to the traditional use of closed-ended measures. We wil begin by providing an overview of our conceptual framework for measuring attitude components and by providing a description of the open-ended measures for assessing the cognitive and affective components of attitudes. We wil then discus some of the benefits of these measures as compared to closed­ ended measures. This wil be followed by a description of a more complex benefit

74

VIcrORIA M. ESSES AND GREGORY R. MAIO

because participants rather uniformly attach high subjective probabilities to the responses they spontaneously provide. Thus, our open-ended measures are related to the expectancy-value approach, but have some unique features, and the importance of these features will be discussed further in this chapter. The open-ended measures are also similar to more specialized thought-listing measures designed to assess cognitive responses to persuasion (e.g., Cacioppo & Petty, 1981; Greenwald, 1968; Petty & Cacioppo, 1984) and attitude-relevant information (e.g., Cronen & Conville, 1975a,b; Hackman & Anderson, 1968; Wood, 1982; see also Petty, Ostrom, & Brock, 1981, for a review of the histor­ ical foundations of cognitive response measures). Recent versions of these cognitive response measures include both a listing of thoughts, and a rating by participants of the valence of the thoughts they have listed (e.g., Breckler & Wiggins, 1991; Maio & Olson, 1998; Romero, Agnew, & Insko, 1996). In addition, the open-ended measures are related to open-ended techniques that have been used to examine attitude functions (e.g., Boninger, Krosnick, & Berent, 1995; Maio & Olson, 1994; Shavitt, 1990), the clustering of stereotypes (Niemann, Jennings, Rozelle, Baxter, & Sullivan, 1994), and the content of intergroup attitudes more generally (Herek, 1987; Monteith & Spicer, 200). They are also related to open-ended techniques for examining stereotyping and attributions. For example, Maass (1999) has conducted extensive research iden­ tifying variation in the categorization of ingroup and outgroup traits and be­ haviors. In one study (Maass, Montalcini, & Biciotti, 1998), participants were asked to provide open-ended descriptions of their ingroup and two outgroups. The results indicated that the descriptions of the ingroup included equally abstract positive and negative stereotypes, whereas the descriptions of the out­ groups included negative stereotypes that were more abstract than the positive stereotypes. This phenomenon is closely linked to the general tendency of people to make stable, dispositional attributions for negative behaviors from other people and outgroups, and such attributional tendencies have also been revealed using open-ended measures of causal attributions (e.g., Islam & Hewstone, 1993). In contrast to our approach, however, these studies have focused predominantly on coding non-evaluative aspects of participants' re­ sponses, rather than focusing on assessing properties of attitudes. DESCRIPTION OF THE OPEN-ENDED MEASURES FOR ASSESSING THE AFFECfiVE AND COGNITIVE COMPONENTS OF ATTITUDES

Based on the multicomponent perspective on attitudes (e.g., Eagly & Chaiken, 1993, 1998; Zanna & Rempel, 1988), our open-ended measures have focused on assessing two important components of attitudes: affect and cogni­ tion. To assess the affective component of attitudes, participants are asked to

76

VICfORIA M. ESSES AND GREGORY R. MAIO

with stereotypes having less of a unique role to play (e.g., Esses et a/., 1993; Haddock et a/., 1993). Similarly, symbolic beliefs have been found to predict discrimination toward homosexuals in simulated jury settings, whereas stereo­ types are not predictive of these effects (Akst & Esses, 1999). Psychometric examination of the open-ended measures indicates that they are reliable and valid. For example, Haddock and Zanna ( 1998b) found split-half reliabilities on the cognitive and affective open-ended measures averaging 0.75, and Eagly et al. (1994) reported coefficient alphas in a satis­ factory range, ranging from 0.61 for affect toward women to 0.94 for affect toward Republicans. In both cases, these internal consistency indices did not differ between the cognitive and affective measures, and the coefficient alphas reported for the open-ended measures by Eagly et a/. ( 1994) did not differ from those for semantic differential scales also utilized. In addition, using a multi-trait, multi-method analysis involving both the open-ended measures and some semantic differential scales developed on the basis of pretesting, Haddock and Zanna (1998b) found evidence of convergent and discriminant validity of the cognitive and affective open-ended measures. Scores on the open-ended and semantic differential measures of the same construct were highly correlated, demonstrating convergent validity (e.g., cognitions: 0.86). In addition, cognition-affect correlations for the open­ ended measures were significantly lower (0.58), demonstrating discriminant validity, and were significantly lower than those for the closed-ended mea­ sures (0.73).

BASIC BENEFITS OF THE OPEN-ENDED MEASURES

Data for Quantitative and Content Analyses

Unlike most measures of attitudes, the open-ended measures provide data that can be subjected to both direct quantitative analyses and content analyses. In many cases, both types of analyses are of potential interest. A direct quantitative analysis of the responses involves calculating a score for each participant for each attitude component-either the average valence (i.e. l:V/n, where V = valence of each belief or emotion and n = the number of beliefs or emotions listed) or the sum of valences (i.e., l:V), depending on the context. For example, to examine whether mood influences the valence of beliefs that people ascribe to groups, we utilized average valence as an indica­ tion of the typical response provided (Esses et a/., 1 994; Esses & Zanna, 1 995). In contrast, in using responses from the open-ended measures to calculate individuals' degree of ambivalence toward a group, we use the sum of positive valences and the sum of negative valences, as described below (Bell & Esses, 1 997; Bell, Esses, & Maio, 1996; Maio, Bell, & Esses, 1996). Thus, whether average valence or summed valences are utilized depends on the conceptual

EXPANDING ASSESSMENT OF ATTITUDE COMPONENTS AND STRUCTURE

77

variable of interest, although it should be noted that average valence and summed valences are very highly correlated (e.g., Esses et al., 1993, 1994; Esses & Zanna, 1995). In addition, a content analysis of the responses pro­ vided can be conducted, with the goal of examining whether similar or dif­ ferent content is provided for different targets under different conditions (e.g., Donakowski & Esses, 1 996; Esses & Beaufoy, 1994; Haddock et at., 1993, 1 994a). In a study examining whether the label used to describe Native Peoples in Canada influences attitudes toward the group, Donakowski and Esses ( 1 996) used the open-ended measures to elicit responses to different labels used to describe Native Peoples in Canada. They then conducted both a quantitative analysis of attitude component scores and a content analysis of the specific responses provided. The analysis of attitude component scores indicated that the label used to describe Native Peoples affected the positivity of symbolic beliefs expressed toward the group. The content anal­ ysis of symbolic beliefs then revealed that the specific symbolic beliefs listed differed as a function of the group label. Of particular interest were responses to the label "First Nations" . In addition to being the label that Native Peoples in Canada often use to describe themselves, this label has further connotations because of its association with a Native political organ­ ization in Canada. The Assembly of First Nations is a national organization made up of chiefs of First Nations, whose stated goal is the preservation of the right of self-determination for each First Nation. This organization is actively involved in promoting Native land claim settlements and Native access to resources, such as fishing rights. Results indicated that the First Nations label indeed elicited very different responses than the other labels (e.g., Native Peoples, Native Indians). In particular, when the label First Nations was presented, symbolic belief scores were less positive and the content of the symbolic beliefs was more likely to refer to threats to national well-being and unity than when other labels for the group were utilized. Thus, both analyses were of value in elucidating the specific mechanism through which labels influence attitudes toward the group. It should be noted that because the open-ended measures allow particip­ ants to provide their own content, these measures are not well suited for use in studies in which researchers wish to make direct comparisons between participants on the same content. In such cases, closed-ended measures would seem to be more appropriate for use. For example, in a study examining the effect of a manipulation on the extent of endorsement of specific stereotypes, direct assessment of these specific stereotypes through closed-ended measures would be of greatest value. Of course, if a researcher was interested in deter­ mining whether a manipulation influenced the likelihood of these stereotypes being spontaneously generated, open-ended measures would be of greater value .

78

VICfORIA M. ESSES AND GREGORY R. MAIO

Idiosyncratic Responses and Ease of Completion

Two related benefits of the open-ended measures are that they elicit idiosyncratic responses to an attitude object and that respondents find them relatively simple and easy to complete. Respondents are asked to list the beliefs and emotions that come to mind in response to an attitude object and to indicate the valence of each belief or emotion listed. Thus, individuals are asked to provide responses in their own words and, in comparison to closed-ended measures, are not constrained by predeter­ mined terms provided by the researcher. This procedure avoids the poten­ tial problem of participants being unsure of the meaning of a term or being asked to rate a term that is actually irrelevant to their attitude (see also Eagly et al., 1 994). Participants are also given the opportunity to indicate further their interpretation of each term through the use of valence ra­ tings. By assessing idiosyncratic beliefs and emotions, then, we are likely to gain a clearer understanding of the bases of attitudes. We should note, however, that the elicitation of idiosyncratic responses means that it is not simple to perform the item analyses that are often performed with closed­ ended measures. In terms of ease of responding, in general, individuals appear to have little difficulty in providing responses to the open-ended measures because of the clear-cut nature of the task. Indeed, Haddock and Zanna (1998b) reported that only 3-5o/o of university students fail to provide at least one response for a single attitude object, and Geer (1988) similarly indicated that only 5% of the public is unable to respond to open-ended questions. In our own research, participants readily provide lists of beliefs and emo­ tions, and when they do not, they often indicate that they have no basis on which to provide a response, which may, in fact, provide an accurate reflec­ tion of their attitude. For the most part, participants provide a number of responses to each measure. For example, in a study examining the effect of mood on the expression of ethnic stereotypes, we found that participants listed an average of five to eight stereotype-relevant characteristics for ethnic groups with which they were likely to be familiar (Esses et al., 1994; Esses & Zanna, 1 995). Similarly, in a study examining the bases of atti­ tudes toward people with disabilities (people who are amputees, have AIDS, are chronically depressed), participants listed an average of ap­ proximately four emotions, four symbolic beliefs, and five stereotype­ relevant characteristics per group (Esses & Beaufoy, 1994). Eagly et al. (1994) also noted that approximately seven characteristics and six emo­ tions were listed per group in a study examining attitudes toward men, women, Democrats, and Republicans. Together, these results suggest that participants find it relatively easy to provide responses to the open-ended measures.

82

VICTORIA M. ESSES AND GREGORY R. MAIO

which participants were most likely to base their attitudes toward Native Peoples. Specifically, prior to a crisis at Oka, Quebec, involving Native Indian People, emotions were the strongest unique predictor of attitudes toward Native Peoples, whereas during the Oka crisis, symbolic beliefs became the strongest unique predictor of attitudes, with emotions showing reduced pre­ dictive power. In sum, open-ended measures provide evidence that affect and cognition are distinguishable components of attitudes that are important to assess separately. Indeed, in support of this contention and of the ability of open-ended measures to assess stereotypes more distinctly from intergroup affect, Stangor et a/. (1991) found that open-ended stereotype measures pre­ dicted attitudes toward a number of social groups above and beyond the effects of affect, whereas closed-ended stereotype measures, for the most part, did not. Assessment of Salient Affect and Cognitions

In assessing the cognitive and affective components of attitudes, researchers attempt to assess beliefs and emotions that are most important and relevant to the attitude object at hand. Indeed, the usefulness of this exercise is supported by theory (Ajzen & Fishbein, 1980; Maio & Olson, 200) and research (van der Pligt & de Vries, 1998). In contrast, in utilizing closed-ended measures to assess beliefs and emotions, researchers must conduct pretests to determine beliefs and emotions relevant to the particular context and then attempt to provide those that are likely to be held in common by respondents in a study. In addition to the necessity of pretesting to build such a list of consensual items (which may change over time), basing measures on beliefs and emotions that are thought to be held in common by participants may result in the selection of items that are not the most relevant and important bases of particular individuals' attitudes (Eagly et a/., 1994; Esses et a/., 1993, 1994; Niemann et a/., 1994; Stangor & Lange, 1994; see also Devine & Elliot, 1995). As a result, when participants are asked to respond to closed-ended measures of attitude components, they may at times merely retrieve their overall atti­ tudes and then respond to items in an attitude-consistent fashion (Eagly et a/., 1 994). Alternatively, they may consider the items at hand and try to deter­ mine whether they fit with their thoughts and emotions toward the attitude object (Niemann et al., 1994). In either case, responses are less likely to reflect respondents' spontaneous reactions to an attitude object than to reflect whether, upon consideration, the reactions make sense to them. In contrast, open-ended measures that ask participants to list their own beliefs and emotions toward an attitude object are more likely to elicit respondents' personal reactions to the object. That is, participants respond to these measures by indicating what comes to mind-the reactions that are most accessible and highly linked to the attitude object in memory (Eagly et

EXPANDING ASSESSMENT OF ATTITUDE COMPONENTS AND STRUCTURE

83

al., 1994; Esses et al., 1993 Niemann et al., 1994; Stangor & Lange, 1994). For example, Stangor and Lange have proposed that, in assessing stereotypes of groups, a key advantage of open-ended measures is their better reflection of the associative strength between a group and a characteristic in memory. Stangor and Lange have suggested that traditional, closed-ended stereotype measures assess the availability of a trait within a representation of a group, whereas open-ended measures are likely to assess the accessibility of a trait upon exposure to a group. In support of these contentions, our early research found that participants' responses to open-ended measures primarily included beliefs and emotions that participants associated with a high percentage of group members (e.g., Esses et al., 1993, 1994; Esses & Beaufoy, 1994; Esses & Zanna, 1995; see also Eagly et al., 1994). In these early studies, participants were asked to indicate the percentage of a group to which each listed belief or emotion applied, as advocated by the Expectancy X Value approach (e.g., Fishbein, 1980). These percentage estimates were sufficiently and uniformly high (e.g., attribution of stereotypes on average to 74% of group members; Esses et al., 1994) that it never proved useful to utilize these estimates to weight the valences assigned to particular beliefs and emotions in computing attitude component scores (Esses et al., 1993, 1994; Esses & Zanna, 1995; Haddock et al., 1993, 1994a). In contrast, when using closed-ended measures to examine attitude components, it is necessary to weight the evaluation of each belief or emotion by a respondent's estimate of the probability that the belief or emo­ tion is associated with the attitude object. These multiplicative composite scores have been criticized on statistical grounds (e.g., Evans, 1991). However, as Eagly et al. (1994) have suggested, in using open-ended measures to elicit participants' idiosyncratic beliefs or emotions toward an attitude object, "it is unnecessary to allow them to deny these attributes or affects by assigning them low probabilities" (p. 1 1 9). Similarly, in examining the use of open­ ended measures to assess the basis of political attitudes, Geer (1991) found that participants were more likely to report important, salient issues relevant to their attitudes than unimportant or superficial issues, so that probability estimates were again unnecessary. Because probability estimates are not required in open-ended measures, multiplicative composites do not need to be utilized. A study reported in Stangor and Lange (1994) further supports the sugges­ tion that open-ended measures assess reactions that come to mind in response to an attitude object. This study demonstrated that responses to open-ended measures and to reaction-time measures of stereotypes were highly corre­ lated, forming a separate factor from responses to traditional, closed-ended measures of stereotypes. Thus, Stangor and Lange concluded that open­ ended measures are useful for assessing stereotypes that are highly associated with a category label in memory. Similarly, in a recent set of studies, Clow and

84

VICTORIA M. ESSES AND GREGORY R. MAIO

Esses (200) found that characteristics that were attributed to specific groups in the open-ended stereotype measure were judged to be more descriptive of the specific groups and were responded to faster in a separate feature verifica­ tion task. These associations with less controlled measures also suggest that responses to open-ended measures are not driven merely by social desirability considerations. Assessment of Respondents' Evaluative Interpretation of Their Afect and Cognitions

In assessing the cognitive and affective components of attitudes, a key issue is the valence of these thoughts and feelings. That is, researchers are almost always interested in determining whether thoughts and feelings about an attitude object are favorable or unfavorable. When researchers use closed-ended measures to assess attitude compo­ nents, they generally determine the evaluative implications of responses on the basis of intuition, or by comparing the terms they have used in their rating scales to normative data on the consensual meaning of the terms (e.g., Ander­ son's likeableness ratings of 555 personality-trait words; Anderson, 1968). For example, in assessing cognitive and affective responses to social groups, Stangor et a/. (1991; Study 1 ) determined the evaluative implications of traits used in closed-ended scales on the basis of pretesting with an independent sample and using the likeableness ratings of the terms on Anderson's (1968) list. The evaluative implications of emotions seemed to be based on the re­ searchers' own views of these emotions. Similarly, in assessing racial stereo­ types, Devine and Elliot (1995) determined the favorability ratings of stereotypes on the basis of pretesting and using Rothbart and Park's (1986) summary of the favorability of trait adjectives. In all of these cases, the as­ sumption is made that the evaluative implications of terms is invariant across individuals, situations, time, and attitude objects (see also Crites et a/., 1994). Logic and empirical data suggest otherwise. First, in the context of inter­ group attitudes, terms such as "religious", "conforming", and "sympathy" may well take on different meanings, depending on the target (e.g., "religious" Americans vs. "religious" immigrants), the year (e.g., "confor­ ming" in the 1960s vs. 1990s), and the perceiver (e.g., high prejudice vs. low prejudice individuals' "sympathy" toward Native Peoples). Indeed, earlier research suggested that stereotypes may assume a different meaning, or tone, depending on the context (Campbell, 1967; Peabody, 1968; Saenger & Flowerman, 1 954; Smedley & Bayton, 1978). Our recent research has also demonstrated that the perceived favorability of stereotypes may vary depending on factors such as the target group, the perceiver's level of prejudice, and the situation. That is, terms such as "femi­ nine" and "sexual" were considered to be more favorable when used to

EXPANDING ASSESSMENT OF A1TITUDE COMPONENTS AND STRUCIURE

85

describe women than when used to describe gay men (Esses, Jackson, & Nolan, 1 996). Low-prejudice individuals considered stereotypes of gay men, such as "sensitive" and "open", to be more favorable than did high-prejudice individuals (Esses et al., 1996). Also, stereotypes of a new immigrant group, such as "hard-working" and "ambitious", were considered to be less favor­ able when the group was perceived to be potentially competing for resources with non-immigrants than when it was not (Esses, Dovidio, Jackson, & Armstrong, 2001 ; Esses, Jackson, & Armstrong, 1998). Finally, even one's current mood state may influence the evaluative inter­ pretation of responses provided (Esses et al., 1994; Esses & Zanna, 1995; see also Haddock et al., 1994b; Schiff, Esses, & Lamon, 1 992). We have found that mood has a strong influence on the evaluative interpretation of stereotypes typically attributed to certain groups, with a weaker effect on the actual stereotypes utilized. In particular, negative mood caused individuals to inter­ pret characteristics they attributed to particular groups in a less positive light. Overall, then, these results demonstrate that it is essential to determine the valence of responses from the specific participants at hand for the specific attitude object under consideration and in the context in which they are assessed. Of course, closed-ended measures could include this assessment of valence, although they have not generally done so. ABILITY TO ASSESS A VARIETY OF ATTITUDE PROPERTIES RELATED TO ATTITUDE STRUCTURE

The above discussion has focused on the ability of the open-ended measures to assess the content and valence of participants' beliefs and feelings about an attitude object. In this manner, open-ended measures provide important information about the underlying basis of an attitude. This section describes how the measures can also be used to collect information about three import­ ant dimensions of attitude structure: ambivalence, inconsistency, and embeddedness. Attitude Ambivalence

People are said to have ambivalent attitudes when they are simultaneously favorable and unfavorable toward an attitude object. That is, people are ambivalent when their attitudes contain conflicting positive and negative elements (Eagly & Chaiken, 1993; Kaplan, 1 972; Katz & Hass, 1 988; Olson & Zanna, 1993). This ambivalence can exist within the components of an attitude or between them. For example, people might simultaneously possess positive and negative feelings about a particular ethnic group (intracompo ­ nent ambivalence). Alternatively, people might simultaneously possess

86

VIcrORIA M. ESSES AND GREGORY R. MAIO

positive beliefs about the group and negative feelings toward it (intercompo­ nent ambivalence). lntracomponent Ambivalence Intracomponent ambivalence (i.e., ambivalence within attitude components) has been examined extensively in research on prejudice (Dovidio & Mullen, 2001 ; Katz, 1981). Similarly, we have primarily measured intracomponent ambivalence using the open-ended measures of emotions, stereotypes, and symbolic beliefs (Bell & Esses, 1997; Bell et a/., 1996; Maio et a/., 1996). To assess ambivalence within a component (e.g., emotions), the first step involves obtaining positive dimension scores by summing the positive val­ ences across the items listed, and obtaining negative dimension scores by summing the negative valences across the items listed. We use summed dimension scores rather than averages because the use of averages to calcu­ late a dimension score would result in the dilution of the positivity or nega­ tivity of the score. For example, using averages, a person who listed one +3 characteristic would have the same score on the positive dimension as a person who listed 1 1 +3 characteristics. As discussed earlier, the open-ended procedure is likely to elicit accessible and salient information. Therefore, the person who provides 1 1 +3 characteristics likely has more positive informa­ tion accessible than the person who only provides one +3 characteristic. To take this into account, summed scores are used in calculating dimension scores for the purpose of assessing attitude properties. Participants' positive and negative dimension scores are then entered into a formula that is used to calculate intracomponent ambivalence. The formula for calculating ambivalence is based on two important characteristics of ambivalence. First, ambivalence is a positive function of the extent to which the positive and negative attitude dimensions are opposed (Bell et a/., 1996; Priester & Petty, 1996). As a hypothetical example, a positive dimension score of +2 and a negative dimension score of -8 would imply that there are two positive valences opposed by two negative valences; six negative valences would remain unopposed. These positive and negative dimension scores should reflect more ambivalence than, for example, scores of 0 and -8, which would have no opposing valences. Second, ambivalence is also a negative function of the amount of non-opposed valence (Bell et a/., 1996; Priester & Petty, 1 996). For example, although scores of +2 and -8 reflect identical amounts of opposing valence as do scores of +2 and -2 (four opposing val­ ences), the +2 and -8 scores have six unopposed valences, whereas the +2 and -2 scores have no unopposed valences. Therefore, the +2 and -8 scores reflect less ambivalence than the +2 and -2 scores. Thus, ambivalence is calculated as the number of opposed valences minus the number of non-opposed valences (plus a constant to preclude negative

EXPANDING ASSESSMENT OF ATTITUDE COMPONENTS AND STRUCTURE

87

ambivalence scores; see also Bell et al., 1996; Maio et al., 1996). Mathe­ matically, the formula becomes P + INI - 2 IP + Nl + 36, where P = net positive valence, N = net negative valence, and 36 is a constant that is added to preclude negative scores (when 12 responses, each ranging from -3 to +3, are permitted). The ambivalence scores calculated using this formula possess three desirable measurement properties of an ambivalence index: (a) am­ bivalence scores decrease when the smaller dimension is held constant and the larger dimension becomes increasingly polarized; (b) ambivalence scores increase when the value of the larger dimension is constant and the value of the smaller dimension increases; and (c) when dimension scores are equivalent, ambivalence increases as the dimension scores increase (see Breckler, 1994; Thompson, Zanna, & Griffin, 1995). After calculating an ambivalence score for each attitude component (e.g., stereotypes), the component ambivalence scores are then averaged to form scores that represent the average amount of intracomponent ambivalence. Supporting this aggregation of scores, mean intracomponent ambivalence strongly correlates with each subtype of intracomponent ambivalence (e.g., intra-emotion ambivalence), and the subtypes of intracomponent am­ bivalence significantly correlate with each other (see, e.g., Bell et al., 1996; Maio et al., 1996). Intercomponent Ambivalence

In past research, the calculation of ambivalence between attitude components appeared to be a complex challenge. Several studies have attempted to mea­ sure intercomponent ambivalence (Lavine, Thomson, Zanna, & Borgida, 1 998; MacDonald & Zanna, 1998; Thompson et al., 1995), but the measure­ ment approach used in those studies did not enable the calculation of degree of intercomponent ambivalence for all participants. However, we have recently developed a technique that uses the open-ended measures to assess intercomponent ambivalence for all respondents (see Maio, Esses, & Bell, 200), as in the assessment of intracomponent ambivalence (e.g., Bell et al., 1996; Katz & Hass, 1988). For example, the ambivalence between people's feelings and their stereotypes can be determined. High emotion-stereotype ambivalence would exist if the feelings are positive and the stereotypes are negative, or if the feelings are negative and the stereotypes are positive. To calculate intercomponent ambivalence, participants' valence ratings are first used to calculate the net favorability of each component of their attitudes. For each component, net favorability scores are obtained by summing the positive and negative valences (e.g., -2 + 3 + 3 = +4). Each type of intercom­ ponent ambivalence (e.g., emotions-stereotypes) is then calculated using a formula that is similar to the formula for calculating intracomponent am­ bivalence: IAI + IBI - 2 lA + Bl + 72, where A = the net favorability score for

88

VIcrORIA M. ESSES AND GREGORY R. MAIO

one component (e.g., emotions), B = the net favorability score for the other component (e.g., stereotypes), and 72 is a constant that is added to preclude negative ambivalence scores (when 12 responses, each ranging from -3 to +3, are permitted for each of the components). Importantly, once again, this calculation possesses the three desirable properties of an ambivalence for­ mula (see Breckler, 1994; Thompson et a/., 1995). After calculating an ambivalence score for each pair of components (e.g., emotions-stereotypes), the intercomponent ambivalence scores are then averaged to form an overall index of intercomponent ambivalence. Support­ ing this aggregation of scores, mean intercomponent ambivalence strongly correlates with each subtype of intercomponent ambivalence (e.g., emotion­ stereotype ambivalence), and the subtypes of intercomponent ambivalence strongly correlate with each other (Maio, Esses, & Bell, 200). Inconsistency

Attitude inconsistency exists when an attitude component is inconsistent with the overall attitude or with other attitude components. For example, one type of inconsistency is evaluative-cognitive inconsistency, which is oper­ ationalized as the magnitude (absolute value) of the difference between the favorability implied by one's overall attitude and the favorability implied by one's beliefs about the attitude object (Chaiken, Pomerantz, & Giner-Sorolla, 1995; Rosenberg, 1968). Another type of evaluative inconsistency is evaluative-affective inconsistency, which can be operationalized as the mag­ nitude (absolute value) of the difference between the favorability implied by one's overall attitude and the favorability implied by one's emotions about the attitude object (Chaiken et a/., 1995). Most past research has examined evaluative-cognitive consistency in atti­ tudes. This research has found that high evaluative-cognitive inconsistency is associated with: (a) low accessibility of attitudes from memory; (b) low atti­ tude stability; and (c) weak effects of attitudes on information processing (Chaiken et a/., 1995). Given these findings, it is provocative to regard evaluative-cognitive inconsistency as a feature of attitude structure that pre­ dicts attitude strength. Nevertheless, attitudes that possess low evaluative­ cognitive inconsistency may sometimes exhibit high evaluative-affective inconsistency (Chaiken et a/., 1995). In such cases, attitudes containing high evaluative-cognitive inconsistency may not be weak, because they are sup­ ported by consistent affect. Thus, it is important to simultaneously measure both evaluative-cognitive and evaluative-affective inconsistency. This goal is easily achieved using open-ended measures of attitude compo­ nents and a general measure of overall attitudes. Open-ended measures are used to calculate the net favorability of participants' cognitions and the net favorability of participants' emotions. The net favorability of the cognitions is

EXPANDING ASSESSMENT OF ATTITUDE COMPONENTS AND STRUCTURE

89

computed by summing the valence ratings across the stereotypes and sym­ bolic beliefs, and the net favorability of the emotions is computed by summing the valence ratings across the emotions. Next, the net cognition scores, net emotion scores, and attitude ratings are converted to Z-scores (Z-scores are used to re-scale the cognitions, emotions, and attitude scores to a common metric). Finally, the absolute values of the differences between participants' attitude Z-scores and (a) their cognition Z-scores (evaluative-cognitive inconsistency) and (b) their emotion Z-scores (evaluative-affective inconsis­ tency) are calculated. The inconsistency between components can also be calculated using open­ ended measures. To calculate intercomponent inconsistency, participants' val­ ence ratings are used to calculate the net favorability of each component of their attitudes, and each type of intercomponent inconsistency (e.g., emotions­ stereotypes) is determined as the absolute value of the difference in net favorability implied by the components. In the assessment of intergroup attitudes, the three intercomponent inconsistency scores (i.e., emotions-stereotypes, emotions-symbolic beliefs, stereotypes-symbolic beliefs) are then averaged to form an overall intercomponent inconsistency score for each participant. Attitude Embeddedness

Attitude embeddedness signifies the extent to which an attitude is embedded in or linked to a lot of attitude-relevant information (e.g., beliefs about the attitude object). This concept is similar to the concept of working knowledge, which refers to the attitude-relevant beliefs and experiences that spon­ taneously come to mind when one encounters an attitude object (Wood, 1982; Wood, Rhodes, & Biek, 1995). Using an open-ended measure, Wood and her colleagues have found that attitudes are more resistant to change and more predictive of future behavior when people possess many attitude-relevant beliefs and experiences than when people possess few attitude-relevant beliefs and experiences (e.g., Kallgren & Wood, 1986; Wood et al., 1 995). In short, attitudes that are embedded in many attitude-relevant beliefs and expe­ riences appear to be stronger than attitudes that are embedded in few attitude-relevant beliefs and experiences. Theoretically, however, one can also assess the extent to which an attitude is embedded in many attitude-relevant emotions, and it is possible that such embeddedness also confers attitude strength. In fact, it is easy to examine this issue using the open-ended measure of the affective component of attitudes because this measure can be used to calculate the number of emotions that participants associate with the attitude object. Thus, open-ended measures can be used to assess the extent to which an attitude is embedded in many cognitions (e.g., stereotypes, symbolic beliefs) and emotions by simply deter­ mining the number of responses indicated for each component.

90

VIcrORIA M. ESSES AND GREGORY R. MAIO

DOMAINS OF APPLICATIONS OF OPEN-ENDED MEASURES

In addition to the studies that we have briefly described earlier, our recent use of the open-ended measures within and beyond the intergroup context sug­ gests that they can easily be utilized in many contexts. Apptications within the Intergroup Context

The ability of open-ended measures to simultaneously assess ambivalence, inconsistency, and embeddedness is important because this simultaneous assessment enables an unconfounded examination of the correlations be­ tween these characteristics and other psychological variables. Unlike many closed-ended procedures, open-ended procedures do not require the use of different questionnaires and varying question phrasing. Indeed, the same data are used in the calculation of all three attitude characteristics. Although the use of the same data to calculate all three attitude characteris­ tics helps to remove potential confounds when comparing effects of these characteristics, a potential disadvantage is that any similar effects of all three characteristics might be driven partly by shared method variance. For this reason, evidence of discriminant validity among the characteristics may be more compelling than evidence of convergent validity. This observation is relevant to several recent studies that have used open­ ended measures to assess attitude ambivalence, attitude inconsistency, and attitude embeddedness in intergroup attitudes (e.g., Maio et al., 1996; Maio, Esses, & Bell, 200). As shown in Table 3. 1 , we have predicted and found moderate correlations between the attitude properties within a category (e.g., intracomponent ambivalence and intercomponent ambivalence), but mainly weak or non-significant correlations between the attitude properties in dif­ ferent categories (e.g., ambivalence and inconsistency). The one exception seems to be intercomponent inconsistency, which is not highly related to the other types of inconsistency. Nonetheless, this confirms our expectation that ambivalence and inconsistency are distinct attitude properties, because am­ bivalence is calculated using a formula that considers the amount of conflict and non-conflict between components, whereas inconsistency is simply a func­ tion of the absolute magnitude of the difference between evaluations (see also Maio, Esses, & Bell, 200). In general, then, our results demonstrate that the three categories of attitude properties appear to be distinct aspects of inter­ group attitudes. In addition, in the intergroup context, attitude ambivalence meaningfully predicts several variables that attitude inconsistency and attitude embedded­ ness do not predict (Bell & Esses, 1997; Maio et al., 1996; Maio, Esses, & Bell, 200). For example, we have predicted and found that high amounts of intra-

92

VICTORIA M. ESSES AND GREGORY R. MAIO

from England to Wales, we found that Welsh people who possessed positive attitudes toward English people were more opposed to devolution than were Welsh people who possessed negative attitudes toward English people. In contrast, English people who possessed positive attitudes toward Welsh people were more in favor of devolution than were English people who possessed negative attitudes toward Welsh people. Importantly, these effects occurred even when ambivalence, inconsistency, and embeddedness of parti­ cipants' intergroup attitudes were statistically controlled. Thus, the use of open-ended measures in this study enabled us to verify that the relations between intergroup attitudes and attitudes toward devolution were not medi­ ated by other attitude properties. Across a number of studies, then, open­ ended measures have facilitated our ability to investigate simultaneously the consequences of important attitude properties. Applications Outside of the Intergroup Context

Given the success of open-ended measures in the intergroup context, we more recently became interested in applying these measures in other contexts. By doing so, we were able to test the usefulness of these measures to assess more general attitudes. Indeed, several previous studies had utilized the open­ ended measures to assess the affective and cognitive components of attitudes toward social issues (e.g., Eagly et al., 1994; Haddock & Zanna, 1997, 1998a). Thus, in several recent studies, we have utilized open-ended measures to assess attitudes toward a variety of people and topics. For example, in two recent studies, we successfully used open-ended measures to study the rela­ tion between children's (aged 1 2-14 years) ambivalence toward their parents and their attachment styles in relationships (Maio, Fincham, & Lycett, 200). In a related study, we used open-ended measures to assess Italian adults' ambivalence and attitude valence toward their romantic partners, and the relations between these variables and the adults' attachment to their partners (Maio, Fincham, Regalia, & Paleari, 2001 ). In another study, we used open­ ended measures to examine the extent to which consensus information about a political debate influences people's evaluations of the issue at hand (Hod­ son, Maio, & Esses, 2001 ). In this case, open-ended measures were used to assess initial ambivalence toward the issue, which was social welfare (i.e., social assistance payments to the poor). In all cases, open-ended measures were easily utilized in a new context without extensive revision. Empirical Relations Among Attitude Properties In a particularly relevant set of studies (Maio, Esses, & Bell, 1997), we were interested in evaluating a new collection of open-ended measures. This included an open-ended assessment of past behaviors toward the attitude

94

VICfORIA M. ESSES AND GREGORY R. MAIO

that participants infer only their specific reactions (e.g., emotions, behaviors) to an attitude object. Using these reactions, the properties of ambivalence, inconsistency, and embeddedness are calculated without requiring that parti­ cipants infer these properties. Given the distinction between meta-judgmental versus operative indices, we expected responses to the meta-judgmental closed-ended measures of attitude properties to be distinct from responses to the operative closed­ ended measures of attitude properties and from responses to the operative open-ended measures of ambivalence, inconsistency, and embeddedness. Consistent with these predictions, factor analyses for each attitude object revealed a consistent five-factor solution. Three of these factors were the dimensions of attitude structure that were assessed using open-ended mea­ sures: intracomponent and intercomponent ambivalence loaded on a factor that we labeled ambivalence; evaluative-affective, evaluative-cognitive, evaluative-behavioral, and intercomponent inconsistency loaded on a factor that we labeled "inconsistency"; and the number of emotions, number of beliefs, and number of behaviors loaded on a factor that we labeled embed­ dedness. In addition, consistent with the previously mentioned distinction between meta-judgmental and operative indices (Bassili, 1996), the meta­ judgmental closed-ended measures possessed unique loadings on a fourth factor. Finally, the operative, closed-ended measures always loaded uniquely on their own factor. Thus, we consistently found that the attitude properties assessed using open-ended and closed-ended measures loaded on different factors. These results demonstrate: (a) the ability of open-ended measures to tap many attitude properties that are distinct from each other and from the properties assessed using closed-ended scales; (b) the ability of open-ended measures to incorporate measures of past behavior; and (c) the ease with which the mea­ sures can be applied to attitude objects other than social groups. CONCLUSIONS: OPEN-ENDED VERSUS CLOSED­ ENDED ATIITUDE MEASURES

As we have described, our open-ended measures of attitude components have a number of benefits that recommend their use. Thus, for certain purposes, open-ended measures may be preferable to the use of traditional closed­ ended measures. For example, open-ended measures are more likely to asss specific affect and cognitions strongly associated with an attitude object, rather than just picking up the general evaluation of the object. Open-ended measures may also be more easily utilized across numerous participant groups and targets, and for examining attitudes over time. However, the idiosyncratic nature of participants' responses to open-ended measures, which makes them

96

VICTORIA M. ESSES AND GREGORY R. MAIO

Bell, D. W., Esses, V. M., & Maio, G. R. ( 1996). The utility of open-ended measures to assess intergroup ambivalence. Canadian Journal of Behavioural Science, 28, 12-18. Bell, D. W., Ko Man Fung, J., & Esses, V. M. (1996). Ethnic attitudes and identity in Hong Kong Chinese and Canadian Chinese students. Paper presented at the Annual Meeting of the American Psychological Society, San Francisco, CA, June. Bern, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances in Experi­ mental Social Psychology (Vol. 6, pp. 1--62). San Diego, CA: Academic Press. Boninger, D. S., Krosnick, J. A., & Berent, M. K. ( 1 995). Origins of attitude import­ ance: Self-interest, social identification, and value relevance. Journal of Personality and Social Psychology, 68, 61-80. Breckler, S. J. (1984). Empirical validation of affect, behavior, and cognition as distinct components of attitude. Journal of Personality and Social Psychology, 47, 1 1 91-205. Breckler, S. J. ( 1 994). A comparison of numerical indexes for measuring attitude ambivalence. Educational and Psychological Measurement, 54, 350--65. Breckler, S. J., & Wiggins, E. C. (1989). Scales for the measurement of attitudes toward blood donation. Transfusion, 29, 401-4. Breckler, S. J., & Wiggins, E. C. (1991). Cognitive responses in persuasion: Affective and evaluative determinants. Journal of Experimental Social Psychology, 21, 180200. Cacioppo, J. T., & Petty, R. E. (1981). Social psychological procedures for cognitive response assessment: The thought-listing technique. In T. V. Merluzzi, C. R. Glass & M. Genest (Eds), Cognitive Assessment (pp. 309-43). New York: Guilford. Cacioppo, J. T., Petty, R. E., & Kao, C. F. (1984). The efficient assessment of need for cognition. Journal of Personality Assessment, 48, 306-7. Campbell, D. T. (1967). Stereotypes and the perception of group differences. American Psychologist, 22, 817-29. Chaiken, S., Pomerantz, E. M., & Giner-Sorolla, R. ( 1 995). Structural consistency and attitude strength. In R. E. Petty & J. A. Krosnick (Eds), Attitude Strength: Antece­ dents and Consequences (pp. 387-412). Mahwah, NJ: Erlbaum. Clow, K. A., & Esses, V. M. (200). A qualitative and quantitative approach to stereo­ type content. Paper presented at the Annual Meeting of the Canadian Psychological Association, Ottawa, Canada, June. Crites, S. L., Fabrigar, L. R., & Petty, R. E. ( 1 994). Measuring the affective and cognitive properties of attitudes: Conceptual and methodological issues. Personality and Social Psychology Bulletin, 20, 619-34. Cronen, V. E., & Conville, R. L. (1975a). Fishbein's conception of belief strength: A theoretical, methodological, and experimental critique. Speech Monographs, 42, 143-50. Cronen, V. M., & Conville, R. L. (1975b). Summation theory and the predictive power of subjects' own salient beliefs. Journal of Social Psychology, 97, 47-52. Devine, P. G., & Elliot, A. J. (1995). Are racial stereotypes really fading? The Prin­ ceton trilogy revisited. Personality and Social Psychology Bulletin, 21, 1 139-50. Donakowski, D. W., & Esses, V. M. (1996). Native Canadians, First Nations, or Abori­ ginals: The effect of labels on attitudes toward Native peoples. Canadian Journal of Behavioural Science, 28, 86-91. Dovidio, J. F., & Mullen, B. (2001 ). Are evaluations of minorities more extreme? A meta-analytic integration (unpublished manuscript). Eagly, A. H., & Chaiken, S. (1993). The Psychology of Attitudes. Mahwah, NJ: Erlbaum. Eagly, A. H., & Chaiken, S. (1998). Attitude structure and function. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds), Handbook of Social Psychology (Vol. 1. pp. 269322) . Boston: McGraw-Hill.

98

VICI'ORIA M. ESSES AND GREGORY R. MAIO

Greenwald, A. G. (1968). Cognitive learning, cognitive responses to persuasion, and attitude change. In A. G. Greenwald, T. C. Brock, & T. M. Ostrom (Eds), Psycho­ logical Foundations of Attitudes (pp. 147-70). New York: Academic Press. Hackman, J. R., & Anderson, L. R. ( 1 968). The strength, relevance, and source of beliefs about an object in Fishbein's attitude theory. Journal of Social Psychology,

76, 5�7. Haddock, G., & Zanna, M. P. ( 1 994). Preferring "housewives" to "feminists": Cate­ gorization and the favorability of attitudes toward women. Psychology of Women

Quanerly, 18, 25-52. Haddock, G., & Zanna, M. P. ( 1 997). Impact of negative advertising on evaluations of political candidates: The 1993 Canadian federal election. Basic and Applied Social

Psychology, 19, 205-23. Haddock, G., & Zanna, M. P. ( 1 998a). Assessing the impact of affective and cognitive information in predicting attitudes toward capital punishment. Law and Human

Behavior, 22, 325-39. Haddock, G., & Zanna, M. P. ( 1998b). On the use of open-ended measures to assess attitudinal components. British Journal ofSocial Psychology, 37, 1 29-49. Haddock, G., Zanna, M. P., & Esses, V. M. ( 1 993). Assessing the structure of prejudi­ cial attitudes: The case of attitudes toward homosexuals. Journal of Personality and Social Psychology, 65, 1 105-18. Haddock, G., Zanna, M. P., & Esses, V. M. ( 1 994a). The (limited) role of trait-laden stereotypes in predicting attitudes toward Native peoples. British Journal of Social Psychology, 33, 83-106. Haddock, G., Zanna, M. P., & Esses, V. M. ( 1 994b). Mood and the expression of intergroup attitudes: The moderating role of affect intensity. European Journal of Social Psychology, 24, 189-205. Herek, G. M. (1987). Can functions be measured? A new perspective on the functional approach to attitudes. Social Psychology Quanerly, SO, 285-303. Higgins, E. T. (1987). Self-discrepancy: A theory relating self and affect. Psychological Review, 94, 319-40. Hodson, G., Maio, G. R., & Esses, V. M. (2001). The role of attitudinal ambivalence in susceptibility to consensus information. Basic and Applied Social Psychology, 23,

1 97-205. Islam, M. R., & Hewstone, M. (1993). Intergroup attributions and affective con­ sequences in majority and minority groups. Journal of Personality and Social Psy­ chology, 64, 936-50. Jackson, L. M., & Esses, V. M. ( 1 997). Of scripture and ascription: The relation between religious fundamentalism and intergroup helping. Personality and Social Psychology Bulletin, 23, 893-906. Kallgren, C. A. III, & Wood, W. (1986). Access to attitude-relevant information in memory as a determinant of attitude-behavior consistency. Journal of Experimental

Social Psychology, 22, 328-38. Kaplan, K. J. (1972). On the ambivalence-indifference problem in attitude theory and measurement: A suggested modification of the semantic differential technique. Psy­ chological Bulletin, 77, 361-72. Katz, D., & Braly, K. (1933). Racial stereotypes of one hundred college students. Journal ofAbnormal and Social Psychology, 28, �90. Katz, I. (1981). Stigma: A Social Psychological Analysis, Hillsdale, NJ: Erlbaum. Katz, 1., & Hass, R. G. (1988). Racial ambivalence and American value conflict: Correlational and priming studies of dual cognitive structures. Journal of Personality

and Social Psychology, 55, 893-905. Kiger, G. ( 1 997). The structure of attitudes toward persons who are deaf: Emotions, values, and stereotypes. Journal of Psychology, 13, 554-60.

EXPANDING ASSESSMENT OF ATTITUDE COMPONENTS AND STRUCTURE

99

Lavine, H., Thomson, C. J., Zanna, M. P., & Borgida, E. (1998). On the primacy of affect in the determination of attitudes and behavior: The moderating role of affective-cognitive ambivalence. Journal of Experimental Social Psychology, 34, 398-421. Lord, C. G., & Lepper, M. R. (1999). Attitude representation theory. In M. P. Zanna (Ed.), Advances in Experimental Social Psychology (Vol. 31, pp. 265-343). San Diego, CA: Academic Press. Maass, A. (1999). Linguistic intergroup bias: Stereotype perpetuation through lan­ guage. In M. P. Zanna (Ed.), Advances in Experimental Social Psychology (Vol. 3 1 , pp. 79-121). San Diego, CA: Academic Press. Maass, A., Montalcini, F., & Biciotti, E. (1998). On the (dis)-confirmability of ster­ eotypic attributes. European Journal ofSocial Psychology, 28, 383-402. MacDonald, T. K., & Zanna, M. P. (1998). Cross-dimension ambivalence toward social groups: Can ambivalence affect intentions to hire feminists? Personality and Social Psychology Bulletin, 24, 427-41. Maio, G. R., Bell, D. W., & Esses, V. M. (1 996). Ambivalence and persuasion: The processing of messages about immigrant groups. Journal of Experimental Social Psychology, 32, 513-36. Maio, G. R., & Esses, V. M. (2001). The need for affect: Individual differences in the motivation to approach or avoid emotions. Journal of Personality, 69, 583-615. Maio, G. R., Esses, V. M., & Bell, D. W. (1997). Dimensions of attitude strength: The power and utility of open-ended measures of attitudes. Paper presented at the annual meeting of the Society of Experimental Social Psychology, Toronto, Canada, October. Maio, G. R., Esses, V. M., & Bell, D. W. (2000). Examining conflict between compo­ nents of attitudes: Ambivalence and inconsistency are distinct constructs. Canadian Journal ofBehavioural Science, 32, 58-70. Maio, G. R., Fincham, F. D., & Lycett, E. (2000). Attitudinal ambivalence toward parents and attachment style. Personality and Social Psychology Bulletin, 26, 145 164. Maio, G. R., Fincham, F. D., Regalia, C. & Paleari, G. (2001) Ambivalence and attachment in relationships. In K. Pillemer & K. Luscher (Eds), Intergenerational Ambivalences: New Perspectives on Parent-Child Relations in Later Life. Elsevier/ JAI. Maio, G. R., Hewstone, M., Esses, V. M., & Willis, H. (2001). Intergroup attitudes and attitudes toward splitting a nation: The devolution of power to Wales (manuscript submitted for publication). Maio, G. R., & Olson, J. M. (1994). Value-attitude-behaviour relations: The moderat­ ing role of attitude functions. British Journal of Social Psychology, 33, 301-12. Maio, G. R., & Olson, J. M. (1998). Attitude dissimulation and persuasion. Journal of Experimental Social Psychology, 34, 182-201. Maio, G. R., & Olson, J. M. (2000). Emergent themes and potential approaches to attitude function: The function-structure model of attitudes. In G. R. Maio & J. M. Olson (Eds), Why We Evaluate: Functions of A ttitudes (pp. 417-42). Mahwah, NJ: Erlbaum. McConahay, J. B. (1986). Modern racism, ambivalence, and the modern racism scale. In J. F. Dovidio & S. L. Gaertner (Eds), Prejudice, Discrimination. and Racism (pp. 91-125). Orlando, FL: Academic Press. Millar, M. G., & Millar, K. U. (1990). Attitude change as a function of attitude type and argument type. Journal of Personality and Social Psychology, 59, 217-28. Monteith, M. J., & Spicer, C. V. (2000). Contents and correlates of Whites' and Blacks' racial attitudes. Journal of Experimental Social Psychology, 36, 125-54.

100

VICTORIA M. ESSES AND GREGORY R. MAIO

Niemann, Y. F., Jennings, L., Rozelle, R. M., Baxter, J. C., & Sullivan, E. ( 1 994). Use of free responses and cluster analysis to determine stereotypes of eight groups. Personality and Social Psychology Bulletin, 20, 379-90. Olson, J. M. ( 1990). Self-inference processes in emotion. In J. M. Olson & M. P. Zanna (Eds), Self-inference Processes: The Ontario Symposium (Vol. 6, pp. 17-4 1 ). Hillsdale, NJ: Erlbaum. Olson, J. M. ( 1 992). Self-perception of humor. Evidence for discounting and augmen­ tation effects. Journal of Personality and Social Psychology, 62, 369-77. Olson, J. M., & Zanna, M. P. ( 1 993). Attitudes and attitude change. Annual Review of Psychology, 44, 1 17-54. Peabody, D. ( 1 968). Group judgments in the Philippines: Evaluative and descriptive aspects. Journal of Personality and Social Psychology, 10, 290-300. Petty, R. E., & Cacioppo, J. T. ( 1 984). The effects of involvement on responses to argument quantity and quality: Central and peripheral routes to persuasion. Journal of Personality and Social Psychology, 46, 69-81 . Petty, R . E . , & Krosnick, J. A . (Eds) ( 1 995). Attitude strength. Antecedents and con­ sequences. Mahwah, NJ: Erlbaum. Petty, R. E., Ostrom, T. M., & Brock, T. C. (1981 ). Historical foundations of the cognitive response approach to attitudes and persuasion. In R. E. Petty, T. M. Ostrom, & T. C. Brock (Eds), Cognitive Responses in Persuasion (pp. 5-29). Hillsdale, NJ: Erlbaum. Potter, J. ( 1 998). Discursive social psychology: From attitudes to evaluative practices. In W. Stroebe & M. Hewstone (Eds), European Review ofSocial Psychology (Vol. 9, pp. 233-66). Chichester: Wiley. Priester, J. R., & Petty, R. E. ( 1 996). The gradual threshold model of ambivalence: Relating the positive and negative bases of attitudes to subjective ambivalence. Journal of Personality and Social Psychology, 71, 431-49. Reicher, S., & Hopkins, N. ( 1 996). Seeking influence through categorizing self­ categories: An analysis of anti abortionist rhetoric. British Journal ofSocial Psychol­ ogy, 35, 297-31 1 . Romero, A . A., Agnew, C . R., & Insko, C . A . (1996). The cognitive mediation hypoth­ esis revisited: An empirical response to methodological and theoretical criticism. Personality and Social Psychology Bulletin, 22, 651-65. Rosenberg, M. ( 1 968). Hedonism, inauthenticity, and other goals toward expansion of a consistency theory. In R. P. Abelson, E. Aronson, W. J. McGuire, T. M. Newcomb, M. J. Rosenberg, & P. H. Tannenbaum (Eds), Theories of Cognitive Consistency: A Sourcebook (pp. 73-1 1 1 ). Chicago, IL: Rand McNally. Rothbart, M., & Park, B. ( 1 986). On the confirmability and disconfirmability of trait concepts. Journal of Personality and Social Psychology, 50, 131-42. Saenger, G., & Flowerman, S. (1954). Stereotypes and prejudicial attitudes. Human Relations, 7, 217-38. Schiff, B. B., Esses, V. M., & Lamon, M. ( 1 992). Unilateral facial contractions produce mood effects on social cognitive judgements. Cognition and Emotion, 6, 357-68. Shavitt, S. ( 1990). The role of attitude objects in attitude functions. Journal of Experi­ mental Social Psychology, 26, 124-48. Sherif, M., & Sherif, C. W. (1967). Attitudes as the individual's own categories: The social judgment-involvement approach to attitude and attitude change. In C. W. Sherif & M. Sherif (Eds), Attitude, Ego-involvement and Change (pp. 105-39). New York: Wiley. Smedley, J. W., & Bayton, J. A. (1978). Evaluative race-class stereotypes by race and perceived class of subjects. Journal of Personality and Social Psychology, 36, 530-35. Stangor, C., & Lange, J. E. (1994). Mental representations of social groups: Advances in understanding stereotypes and stereotyping. In M. P. Zanna (Ed.), Advances in Ex­ perimental Social Psychology (Vol. 26, pp. 357-416). San Diego, CA: Academic Pres.

EXPANDING ASSESSMENT OF ATIITUDE COMPONENTS AND STRUCTURE

101

Stangor, C., Sullivan, L. A., & Ford, T. E. (1991). Affective and cognitive determinants of prejudice. Social Cognition, 9, 359-80. Thompson, M. M., Zanna, M. P., & Griffin, D. W. (1995). Let's not be indifferent about (attitudinal) ambivalence. In R. E. Petty & J. A. Krosnick (Eds), Attitude Strength: Antecedents and Consequences (pp. 361-86). Mahwah, NJ: Erlbaum. van der Pligt, J., & de Vries, N. (1998). Belief importance in expectancy-value models of attitudes. Journal of Applied Social Psychology, 28, 1339-54. Wood, W. (1982) . The retrieval of attitude-relevant information from memory: Effects on susceptibility to persuasion and on intrinsic motivation. Journal of Personality and Social Psychology, 42, 798-810. Wood, W., Rhodes, N., & Biek, M. (1995). Working knowledge and attitude strength: An information processing analysis. In R. E. Petty and J. A. Krosnick (Eds), Attitude Strength: Antecedents and Consequences (pp. 455-87). Mahwah, NJ: Erlbaum. Zanna, M. P., & Rempel, J. K. (1988). Attitudes: a new look at an old concept. In D. Bar-Tal & A. W. Kruglanski (Eds), The Social Psychology of Knowledge (pp. 31534). Cambridge: Cambridge University Press.

This page intentionally left blank

Chapter 4

Shifting Standards and Contextual Variation in Stereotyping Moniea Biemat Elizabeth R. Thompson University of Kansas, USA

ABSTRACT This chapter provides an overview of some 10 years of research on the "Shifting Standards Model" (Biernat, Manis, & Nelson, 1991). This model suggests that stereotypes lead perceivers to evaluate individual group members relative to within·category standards. One result of this process is that subjective evalua­ tions (e.g., j udgments on semantic-differential scales) may "mask" the operation of stereotypes. For example, a man and a woman may both be described as equally "competent", even when more "objective" evaluations reveal substan­ tial stereotyping effects (e.g., the man is assumed to score higher on standardized tests of competency). This review focuses on the role of context in moderating the tendency to shift judgment standards, and provides a critical comparison of the shifting standards model with two other theories emphasizing contextual variation in stereotyping, Social Identity/Self-Categorization Theory (Tajfel, 1972; Turner, 1982; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987), and the Parallel Constraint Satisfaction Model of impression formation (Kunda & Thagard, 1996).

In contrast to early views of stereotypes as rigid cognitive structures, more recent research and theory suggests that stereotypes are subject to contextual Correspondence regarding this chapter should be addressed to Monica Biernat, Department of Psychology, 141 5 Jayhawk Blvd, University of Kansas, Lawrence, KS 66045 7556, USA. European Review of Social Psychology, Volume

© 2002 John

Wiley

&

Sons Ltd.

12.

Edited by Wolfgang Stroebe and Miles Hewstone.

104

MONICA BIERNAT AND ELIZABETH R. THOMPSON

variation and change (Coats & Smith, 1999; Doosje, Haslam, Spears, Oakes, & Koomen, 1 998; Ellemers & van Knippenberg, 1997; Oakes, Haslam, & Turner, 1 994; Spears & Manstead, 1989). Whether stereotypes even develop in the first place may depend on the extent to which a trait dimension dif­ ferentiates between groups (Ford & Stangor, 1992), and whether a group attribute is the subject of conversation (Stangor & Schaller, 1996; Schaller & Conway, 1999). There is also evidence that stereotypes incorporate not just person attributes, but rather contextualized descriptions of persons in situa­ tions (Cantor & Mischel, 1 977; Cantor, Mischel, & Schwartz, 1 982; Niemann, Pollak, Rogers, & O'Connor, 1998; Shoda & Mischel, 1993; Trope, 1998). Similarly, the application of stereotypes is subject to contextual variation. The strength of stereotyping effects may vary, depending on the availability of other information about the person (Locksley, Borgida, Brekke, & Hepburn, 1 980; Jussim, Coleman, & Lerch, 1987; Kunda & Thagard, 1996), the relative salience of the stereotyped category (Macrae, Bodenhausen, & Milne, 1995), and motivational concerns of the perceiver (Fein & Spencer, 1 997; Green­ berg, Pyszczynski, Solomon, Rosenblatt, Veeder, Kirkland, & Lyon, 1990; Sinclair & Kunda, 1999). Some situational contexts-for example, the pres­ ence of a solo or token member of a group--may also increase the extent to which stereotypes are applied (Biernat & Vescio, 1993; Taylor, Fiske, Etcoff, & Ruderman, 1 978). In other words, group stereotypes may differ, and the application of those stereotypes to individuals may differ, depending on context. We define "con­ text" very broadly, in terms of the setting, surrounding or environment in which a stereotype might be generated or applied. A definition from Webs­ ter's dictionary is appropriate to what we have in mind-context refers to "the interrelated conditions in which something exists or occurs". Our first goal in this chapter is to review a body of research from our laboratory that is consistent with this contextual view of stereotyping. This research is derived from the "shifting standards model" (SSM; Biernat & Manis, 1994; Biernat, Manis, & Nelson, 1991 ), which suggests that stereotypes activate differing standards against which individual group members are judged. Our second goal is to relate this work to the broader literature on "stereotypes in context". Although there are many findings in the field that suggest the importance of context, we will focus specifically on the two extant models that we see as offering the strongest theoretical statements about the importance of context. These two important perspectives are self-categorization theory (Turner, 1982; 1 984; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987), and the parallel-constraint satisfaction model of impression formation (PCSM; Kunda & Thagard, 1996). Both of these theories are broader in scope than the shifting standards model (and quite different from each other), but nonethe­ less we hope to highlight the similar themes on context that all three perspec-

SHIFTING STANDARDS IN STEREOTYPING

105

tives offer. After providing brief overviews of each of these perspectives and the manner in which they incorporate context, we will discuss connections between these models and our own. In short, we hope to highlight findings that have emerged in our laboratory, to integrate our perspective with other relevant models, and to discuss future directions in which research on stereo­ types in context may profitably move. THE SHIFTING STANDARDS MODEL

The basic premise of the SSM is that when we judge individual members of stereotyped groups on stereotype-relevant dimensions, we use within-category reference points or standards. For example, given stereotypes that men are better leaders than women, we are likely to judge the leadership competence of a particular woman relative to (lower) standards of competence for women, and the leadership competence of a particular man relative to (higher) stand­ ards of competence for men. The result is that evaluations of men and women on leadership competence may not be directly comparable, as their meaning is tied to different contexts: "good'' for a woman does not mean the same thing as "good" for a man. A standard incorporates the average and range that is expected from members of a group on a particular dimension, and aids the judge in anchoring the endpoints of a subjective rating scale (e.g., "high" to "low" competence). Rating points are defined to reflect the expected distribu­ tion of category members on the dimension, with high numbers reserved for targets with the highest expected level of the attribute among members of the category. When groups are expected to differ (i.e., when a stereotype is held), endpoints are anchored differently for the contrasting groups (for related themes in other judgment models, see Parducci, 1963; Postman & Miller, 1945, Upshaw, 1962; Volkmann, 195 1). Evidence supporting the operation of stereotype-based standard shifts can be gleaned from comparisons between judgments that are made on subjective rating scales to those made on objective rating scales. The terms "subjective" and "objective" require some clarification. A "subjective scale" refers to any rating system in which the units of judgment have no ties to external reality. For example, Likert-type scales, semantic differentials, and many instruments that display trait judgments as continua (e.g., "very unathletic" to "very ath­ letic") are subjective in the sense that their rating points can be differentially defined and adjusted (e.g., across time, across perceivers, and across targets). Subjective judgments are "slippery" in that they do not connote a fixed mean­ ing. On the other hand, an "objective scale" refers to an externally anchored, "common rule" scale, consisting of judgment units that maintain a constant meaning across contexts and/or types of targets. For example, judging the heaviness of an object in pounds or kilograms would be an objective

106

MONICA BIERNAT A N D ELIZABETH R. THOMPSON

judgment, whereas rating the object on a "light" to "heavy" scale would be a subjective judgment. In evaluating people, objective scales may include such judgments as estimated standardized test scores or grades (to assess compet­ ence stereotypes), monetary or time judgments (to assess salary, effort, or worth), and rank orderings (which fit our conception of objectivity in that the meaning of a ranking unit is clear-it places a target individual in a readily interpretable position relative to a given frame of reference; see Biernat & Manis, 1994; Biernat, Crandall, Young, Kobrynowicz, & Halpin, 1998). Objec­ tivity does not, then, refer to an accurate or unchanging judgment, but rather to one whose meaning can be readily interpreted because of the external frame of reference. The key prediction of the SSM is that objective or externally anchored judgments are more likely than subjective judgments to reveal the influence of stereotypes. That is, judgments of individual group members will generally be assimilated to group stereotypes when objective response scales are used, but will show less strong, null, or contrastive influences of stereotypes when sub­ jective scales are used. This occurs because objective scales call forth an external frame of reference, which then reveals or captures the available stereotype, whereas subjective scales can be differentially adjusted for dif­ ferent target categories (within-category frames of reference). In short, sub­ jective scales may mask the operation of stereotypes because the response language can be defined in category-specific terms.1 Evidence of Shifting Judgment Standards

We have now accumulated considerable evidence supporting the basic predic­ tion that stereotyping effects will be stronger in objective (externally

10ther researchers who have compared "objective" and "subjective" response scales have also noted that objective scales are less sensitive than subjective scales to context effects (e.g., judg mental contrast), precisely because they do not allow for semantic changes of meaning of the sort described here (e.g, Campbell, Hunt, & Lewis, 1 958; Krantz & Campbell, 1961; Manis, 1967, 1 97 1 ). However, recent research on social comparison and anchoring does suggest that both objective and subjective judgments are affected by changes in anchors or comparative standards (Mussweiler & Strack, 1 99, 200; Strack & Mussweiler, 1997). Nonetheless, a closer took at this latter work actually reveals results quite consistent with our own theorizing. For example, in studies on self-evaluative comparison, "objective" self-judgments revealed assimilation to com parative standards (akin to the stereotype-assimilative effects we anticipate on objective scales), whereas subjective judgments revealed contrast from those standards (akin to the use of within­ category or shifting frames of reference; Mussweiler & Strack, 200). In language similar to our own, these authors suggest that, "because [objective] judgments are externally anchored by consensual standards, the underlying scale cannot be shifted to include or exclude a specific reference point" (Mussweiler & Strack, 200, p. 25). Instead, these judgments simply reflect accessible knowledge. On the other hand, subjective scales require judges to "interpret the given scale labels" (p. 25) according to a given (but shifting) frame of reference, thereby leading to contrast from those (category specific) standards.

SHIFfiNG STANDARDS IN STEREOTYPING

109

their evaluations. If competence is assessed in a within-sex fashion, subjective judgments may mask the influence of gender stereotypes. In our study, parti­ cipants read an excerpt of a magazine article that was either attributed to Joan or John McKay, and that concerned either a masculine (e.g., bass fishing), feminine (e.g., trends in make-up), or gender-neutral (e.g., the mind-body problem) topic. Participants were asked to evaluate the quality, monetary worth, and interest level of the article, either using objective (e.g., dollar estimates of worth) or subjective (terrible-excellent) response scales. As the top panel of Figure 4.2 indicates, when judgments were made in objective units, evidence of gender stereotyping was clear: feminine articles were judged better when written by Joan than by John, masculine articles were better when written by John than by Joan, and neutral articles were judged equivalently regardless of the author sex. Subjective judgments, depicted in the bottom panel of Figure 4.2, produced results consistent with the meta­ analytic findings: author sex had no impact on evaluations. This suggests, then, that participants j udged article quality using an Author Sex x Context evaluative framework. Despite a greater view of worth when an author's sex "matched" the article topic (John deserves more than Joan for his article on bass fishing), this translated into equivalent subjective evaluations for the two authors. In an additional study on gender stereotypes and competence, participants evaluated the resume of "Katherine Anderson" or "Kenneth Anderson", who had applied for either a masculine (executive chief of staff) or feminine (executive secretary) position (Biernat & Kobrynowicz, 1 997). The job de­ scription in the two conditions was identical-only the label was changed, with the result that the "chief of staff" position was viewed as more mas­ culine, higher in status, and worthy of a higher salary than when the same position was labeled "executive secretary". Results in this study were compar­ able to those in the Joan-John study described above (Biernat & Manis, 1994, Study 1). Using externally-anchored rating scales (e.g., assigned letter grades, rank order relative to other candidates), Kenneth was judged better than Katherine for the chief of staff position, whereas Katherine was judged better than Kenneth for the secretarial position (see top panel of Figure 4.3). Again, participants rated candidates most favorably when their sex "matched" the job for which they applied. However, subjective judgments revealed reversals of these patterns (bottom panel of Figure 4.3). The SSM suggests that parti­ cipants may have subjectively perceived the male applicant for the feminine job as "very good, for a male", and the female applicant for the masculine job as "very good, for a female" .2 2The issue of when subjective judgments will produce null effects of stereotypes versus contrast effects is beyond the scope of this chapter. For a discussion, see Biernat, Vescio, and Manis (1998).

SHIFfiNG STANDARDS IN STEREOTYPING

113

gender stereotyping increased over time and, consistent with the SSM, rank­ ings were more likely than ratings to reveal evidence of gender­ stereotyping. Contextual Variations in the Tendency to Shift Standards

As noted above, some of our research has indicated that gender stereotypes do not operate in an aU-or-none fashion. Women are not uniformly judged as less competent writers or employees than men; rather, women are judged better than men when writing on "feminine" topics or applying for "femi­ nine" jobs, but worse than men in "masculine" domains (Biernat & Kobrynowicz, 1 997; Biernat & Manis, 1994). In this sense, context affects the application of stereotypes and, by extension, the tendency to use within­ category judgment standards. Norms

Other research from our laboratory also indicates that the tendency to shift judgment standards is moderated by contextual factors. For example, in the study of US Army captains described above, standards to judge leadership competence varied depending on officers' sex but not their race (Biernat, Crandall et al., 1 998). We attribute this to the Army's explicit norms against the use of within-race judgment standards (e.g., " . . . the Army does not patronize or infantilize Blacks by implying that they need special standards in order to succeed" [Moskos & Butler, 1 996, p. 72] ; " . . . the military has no hint of two promotion lists in which Whites are compared only with Whites, Blacks only with Blacks" [p. 70]). This stands in contrast with recommendations for military policy on the issue of sex set by the Presidential Commission on the Assignment of Women in the Armed Forces (1992): "Entry level training may be gender-specific as necessary" (p. 9), "Military pre-commissioning training may be gender-normed" (p. 1 1 ), and " . . . women should be excluded from direct land combat units and positions" (p. 24; see also Francke, 1 997). These norms provide strong contextual cues about the appropriateness of within-category comparisons-cues that our participants heeded. Interestingly, the lack of race-based judgment scale differences did not mean that all signs of bias based on race were absent. In two studies, officers who were racial minor­ ities were generally judged as less competent leaders than White officers, particularly by White judges, but this effect emerged regardless of the judg­ ment format (objective rankings vs. subjective ratings). This finding delimits the SSM by indicating that stereotyping need not always prompt standard shifts; contextual factors that discourage differential standard use may over­ ride this tendency.

1 14

MONICA BIERNAT AND ELIZABETH R. THOMPSON

Category Combinations The Army data set allowed for the examination of the joint influence of gender and racial category membership on judgments of leadership compet­ ence. In this context, an additional reason why judgment standards shifted based on sex but not race may have been that gender stereotypes regarding leadership competence were stronger than corresponding racial stereotypes. Indeed, when targets come from multiple categories (as all social targets do), we suggest that evidence of standard shifts will only occur along the category lines that are implicated in the stereotyped judgment dimension (i.e., the "focal dimension"; see Eiser & Stroebe, 1972; Tajfel, 1969; McGarty, 1999). For example, on the trait dimension of "intelligence", perceivers hold a race stereotype (Whites are more intelligent than Blacks) but not a gender stereotype (males and females are judged equally intelligent). When asked to judge how intelligent a series of male and female, Black and White targets may be, participants should shift standards based on target race (as evidenced in a Race x Response Scale interaction), but not based on sex (no Sex x Response Scale interaction). A recent study in which rankings were used to simulate "objective" judgments supported this pattern (Biernat & Thompson, 200). Similarly, we found that undergraduate judges viewed "caring" (or nurturance) as relevant to gender but not racial stereotypes, "athleticism" as relevant to both gender and racial stereotypes, and "outgoing" as an attribute irrelevant to both gender and race. In studies in which individual targets (represented in photographs) were judged along these trait dimensions, we generally supported predictions regarding specificity in shifting standards effects. For example, "caring" judgments produced a Sex x Response Scale interaction, "athleticism" judgments produced both Sex x Response Scale and Race x Response Scale interactions, and "outgoing" judgments were un­ affected by these factors (Biernat & Thompson, 200). In addition, however, there was a tendency for judges to evaluate members of certain category combinations most extremely on certain dimensions. For example, Black males were judged the most extremely athletic, and White females were judged the most extremely caring of all targets. These individ­ uals may represent the "prototypical" carriers of the stereotyped attributes we studied (indicating Race x Sex subtypes; see also Stangor, Lynch, Duan, & Glass, 1992). This finding also suggests the interesting possibility that not only are people likely to be described with traits that typify their groups, but that traits are interpreted with reference to particular types of people-athleticism calls to mind Black men; caring or nurturance calls to mind White women. When these traits are applied to other targets (e.g., athletic White females, caring Black men), their meaning may be substantially different. We return to the issue of context and trait construal later in this chapter.

1 16

MONICA BIERNAT AND ELIZABETH R. THOMPSON

conditions. It is also possible that perceived importance of the judgment di­ mensions moderated these effects, with standard shifting occurring on dimen­ sions less important to one's gender identity. Further research will be needed to address this possibility. Translating Subjective Language Implicit in research on the SSM is the notion that subjective evaluations are differentially translated or interpreted by perceivers, depending on a target's social category membership. That is, evaluative labels such as "tall" or "good" or "very smart" have no absolute meaning, but rather are interpreted in light of the group membership (and relevant stereotypes) of the person to whom the label is applied. For example, what does it mean to say that someone is a "good parent?" In one study, we asked participants to "decode" this evaluation into objective estimates of the frequency of parenting behavior (e.g., how many hours per week does the parent play interactively with her/his children?; How many days a week does the parent make lunch for the children?; How many school functions does the parent attend? etc). Of course, the gender of the parent described as "good" was manipulated. Consistent with the stereotype or ex­ pectation of women as more involved parents than men, mothers were judged to engage in parenting behaviors with greater frequency than fathers in each of the broad categories of activities, emotional needs, physical needs, and centrality of the children in the parent's life (Kobrynowicz & Biernat, 1997, Study 2). In another study, being "good at mathematics" was translated into lower objective grades and mathematics involvement for Black than for Asian targets (Kobrynowicz & Biernat, 1997, Study 3). These translation studies demonstrate a fairly direct contextual influence-language is understood dif­ ferentially, depending on the target to whom it is applied. Setting Judgment Standards The research described thus far has examined the judgment consequences of using differential group standards, but one can also examine a preceding step in this process, namely, the setting or instantiating of judgment standards. The SSM suggests that when a stereotype is called to mind, a lower standard is set for documenting the relevant trait in members of the group stereotyped as deficient on the attribute. For example, in two studies on judgments of work­ related competence, women were held to lower minimum standards than men, and Black men were held to lower minimum standards than White men. That is, the bar was set lower-women and Black men needed to provide less evidence of competence-to meet perceivers' low-competence expectations (Biernat & Kobrynowicz, 1 997).

1 18

MONICA BIERNAT AND ELIZABETH R. THOMPSON

confirms a trait, whereas an unexpected behavior's implications for trait diag­ nosis is uncertain. In keeping with the theme of contextual variation, these findings indicate that whether judges perceive that a target possesses an attribute will depend not only on the target's group membership, but also on the standard that is called to mind at the moment a judgment is rendered. A woman can be perceived as both aggressive (because her actions readily surpass low standards of aggression for women) and unaggressive (because a few aggressive behaviors cannot overcome stereotype-assimilative tendencies). Worth exploring is whether the attributions made for unexpected behaviors (relative to group-based standards) underlie this phenomenon. For ex­ ample, much research on reactions to counterstereotypical (or expectancy­ violating) information suggests that perceivers may deeply process this in­ formation in an attempt to reconstrue or assimilate it to expectations (e.g., Olson, Roese, & Zanna, 1 996; Stangor & McMillan, 1 992). To confirm ag­ gression in a woman or emotionality in a man, perceivers may feel that many examples are needed before overcoming the tendency to dismiss such be­ haviors as flukes.

Sumary A number of lines of research in our laboratory have demonstrated that the tendency to judge individual group members relative to within-group com­ parison standards is subject to contextual variation. Group stereotypes are themselves context-specific (e.g., gender stereotypes regarding competence vary depending on the masculinity or femininity of the judgment setting), norms may indicate when differential standard use is appropriate, ingroup/ outgroup status may determine which dimensions of judgment recruit within­ category comparison standards, and the translation and meaning of trait terms may shift depending not only on social category membership but the type of standard that is instantiated in a given situation (e.g., minimum vs. confirmatory). However, this work represents only an initial foray into the ways that context may affect judgments of individual members of stereotyped groups. Notably missing in our work thus far are considerations of how a variety of perceiver motivations may affect the tendency to shift judgment standards, and how group status relations and group-level identification may moderate shifting standards effects. As a model for the directions in which con­ textualized research on shifting standards should move, we tum next to a discussion of social identity and self-categorization theories, which have of­ fered much insight into the importance of context in intergroup judgment and relations.

SHIFfiNG STANDARDS IN STEREOTYPING

1 19

SOCIAL IDENTITY AND SELF-CATEGORIZATION THEORIES Overview

"Social identity" refers to a person's self-concept in terms of group member­ ships, and social identity theory (SIT) generally posits a need for positive social identity, "expressed through a desire to create, maintain, or enhance the positively valued distinctiveness of ingroups compared to outgroups on relevant dimensions" (Turner, 1999, p. 8). As Tajfel (1979) noted, categoriza­ tion as a member of a group arouses social identity concerns, which leads to social comparison between groups in the quest for positive ingroup dis­ tinctiveness (see also Tajfel, 1 972; Turner 1975, 1978). Self-categorization theory (SCT), an outgrowth of SIT, is also built on the distinction between social and personal identity, and suggests that when the self is categorized in social terms, this leads to "an enhanced perceptual identity between self and ingroup members and an enhanced perceptual contrast between ingroup and outgroup members" (Turner, 1 999, p. 1 1). Self-categorization leads to self­ stereotyping and the perception of relative interchangeability with other in­ group members, and this depersonalization "transforms individual into collec­ tive behavior" (Turner, 1999, p. 1 1 ; see also Turner, 1984; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). A consideration of context is central to both SIT and SCT. This is perhaps most clearly seen in the statement that stereotypes themselves are not fixed mental images, but rather vary across situations: A stereotype of the same people may vary in categorical level, kind, content, and prototypical meaning as a function of the relationship between self and others, the frame of reference, the dimensions of comparison, and the background knowledge, expectations, needs, values, and goals of the perceiver (Turner, 1999, p.

26).

In a recent chapter, Ellemers and van Knippenberg (1997) outlined four specific ways in which self-categorization theory suggests that stereotyping and other intergroup phenomena are affected by context. The first is that context determines the salience of particular social categorizations; that is, the same group might be perceived differently depending on which other groups are present in the judgment environment. For example, during the Persian Gulf war, Australians viewed Americans as aggressive when Soviets were included in the frame of reference, but as less so when Iraqis (a common enemy for Australians) were also included (Haslam, Turner, Oakes, McGarty, & Hayes, 1992; see also Diab, 1962). Second, the "degree of involvement or the perspective of the perceiver" may affect the way social groups are perceived (Ellemers & van Knippenberg, 1997, p. 211). For example, stereotypes are less likely to form via illusory

120

MONICA BIERNAT AND ELIZABETH R. THOMPSON

correlation principles when judges are members of one of the groups being considered (Maass & Schaller, 1991), and perceivers may "selectively" gener­ alize from sample information to judgments of their group as a whole, or self­ stereotype to maintain a positive group image (Biernat, Vescio, & Green, 1996; Doosje, Spears, & Koomen, 1995). A third manner in which context affects stereotyping is that status relations between groups may constrain group perceptions. Rather than uniformly striving for a positive identity, group level judgments may reflect or be con­ strained by social reality (e.g., individual members of low-status groups may negatively self-stereotype), so long as a distinctive identity can be established (see Mlicki & Ellemers, 1996). Finally, Ellemers and van Knippenberg (1997) suggest that group-level perceptions may differ depending on the kinds of between-group comparisons being made and the specificity or globality of the judgment dimensions. For example, members of a low-status group may ac­ cept the global superiority of a high-status group but view the ingroup as better on some more specific feature (Ellemers, van Rijswijk, Roefs, & Simons, in press), or members of different groups may mutually acknowledge each other's superiority on different dimensions (Spears & Manstead, 1989; van Knippenberg, 1984; van Knippenberg & Ellemers, 1990). Comparisons with the Shifting Standards Modei-Simlarities and Diferences

In the social identity and self-categorization theory traditions, each of these effects of context has most frequently been examined with respect to group­ level perceptions. That is, frames of reference, ingroup/outgroup status, re­ ality constraints, and judgment dimensions have all been shown to influence how groups as a whole are perceived. The SSM has focused instead on a different contextual theme-that judgments of individual members of stereo­ typed groups are made relative to expectations regarding the group as a whole. Furthermore, only the SSM has distinguished between subjective versus objective response language and its diferential susceptibility to stereo­ typing effects, and only self-categorization theory has emphasized the process of categorization itself. Despite these differences in level and type of the typical dependent variable, there are some similar themes in these approaches. Frames of Reference The key common theme in the shifting standards and social identity/self­ categorization theory perspectives is the importance of one's frame of refer­ ence for making social judgments. Both theoretical approaches build on clas­ sic judgment models in articulating this theme. In the shifting standards

122

MONICA BIERNAT AND ELIZABETH R. THOMPSON

theory's name implies, it is the selfs categorization as a member of one group versus another that sets in motion important judgmental and behavioral out­ comes. An interesting question is whether judgments of targets might be jointly influenced by that target's group membership as well as judges' self­ standing (e.g., in terms of own group membership or standing on the dimen­ sion of evaluation). For example, to what extent is the judgment of a woman's aggressiveness influenced not only by standards of aggression for women but by the judges' level of aggression and his/her self-categorization as a man/ woman? To date, we have not considered such issues in our own research. Distortions of Reality Both the shifting standards model and SCI' highlight the distinction between frames of reference that focus on within-group versus across- or between­ group comparisons. Within-group comparisons are prompted by the use of subjective judgment scales in the SSM and by other contextual cues (such as the presence of only ingroup members) in SCI'. Similarly, across-group com­ parisons are prompted by the use of objective judgment scales in the SSM and by the broader context (presence of both ingroup and outgroup members; between-group competition) in scr. This within- versus across-group distinction is important, as it produces dramatically different judgments of social stimuli. In the shifting standards model, for example, a woman can be simultaneously perceived as both an "excellent" softball player (relative to other women), and as someone who is not worth being selected for the team (when the field of choice includes men as well as women; Biernat & Vescio, in press). In self-categorization theory, an ingroup member may be perceived as both a distinct individual with idio­ syncratic traits (when within-group comparisons are made), or as relatively interchangeable with the self (when a salient outgroup is present; Hogg & Turner, 1987; Lorenzi-Cioldi, 1991). In work on the SSM, we have highlighted that this situation may produce a confusing state of affairs for the target of judgment, in that evaluations re­ ceived from others may be contradictory. Furthermore, we have typically treated the judgment effects that emerge on subjective responsive scales as distortions that mask underlying stereotyping effects. Thus, the fact that the female is not chosen for the team reflects the perceiver's "true" underlying (negative) representation, a representation that is masked when subjective, within-category evaluations are offered. However, from a self-categorization perspective, both judgments are equally true or real; the theory suggests that context-sensitive judgments reflect reality, or the "rational selectivity of per­ ception" (Oakes, Haslam, & Turner, 1994, p. 212). In short, scr and the SSM differ on whether contrast and assimilation effects represent response distortions or "true" reflections of the perceived

124

MONICA BIERNAT AND ELIZABETH R. THOMPSON

moderators of shifting standards effects. For example, members of lower­ status groups may be more likely to shift judgment standards (such that ingroup targets might be more favorably judged) than members of high-status groups, particularly when evaluations are being made on central, group­ defining dimensions. Sumary Social identity and self-categorization theories are broad, important, and far­ reaching theories that have implications for a wide range of psychological phenomena and processes (e.g., group structure and function, intergroup rela­ tions, prejudice, social influence, the self, etc.). We have focused solely on the theories' perspective on stereotyping in this chapter, highlighting the central role of context and contextual variation. Although more limited in scope, the shifting standards model shares with SCf an emphasis on the importance of frames of reference for social judgment, and a concern with the shifting nature of judgment and perception across different contexts. SCf's emphases on group-level judgments and the key role of self-categorization and motivation differentiate it from the SSM. Nonetheless, we think the perspectives can mutually inform each other, and jointly add to the growing body of literature on stereotyping in context.

THE PARALLEL-CONSTRAINT SATISFACfiON MODEL (PCSM) AND STEREOTYPE CONSTRUAL

Overview

Social identity and self-categorization theories offer a broad perspective on the nature of intergroup judgment and behavior. Closer in scope to the shift­ ing standards model is a recent theoretical account of impression formation based on parallel constraint satisfaction principles (Kunda & Thagard, 1996). The PCSM describes a network of constructs including cognitive, affective, and behavioral units (Kunda & Thagard, 1996). The model assumes that judgments of a target's traits arise from the network of connections or associ­ ates that are activated or inhibited at the time of use. This process of activa­ tion and inhibition occurs simultaneously or in parallel, such that pieces of information mutually constrain each others' meaning and the system ul­ timately "settles" on a judgment. Furthermore, the PCSM suggests that, while any given trait may have a broad and diverse network of associates, only a subset of these will be activated on any given occasion. These changing pat­ terns of activation for a particular trait result in changes of meaning for that trait from one occasion or context to another. For example, learning that James is aggressive may activate both the potential behaviors "punch" and

1 26

MONICA BIERNAT AND ELIZABETH R. THOMPSON

groups, the PCSM emphasizes qualitative differences. As Kunda et a/. (1997) note, "an aggressive man be not only more aggressive than an aggressive woman, but also aggressive in different ways" (p. 732). Still, these views are quite consistent with each other, and in a recent paper we considered the likely impact of perceived qualitative differences in judgment standards across groups (Biernat et a/., 1998). For example, on the dimension of athleticism, a woman might be judged to be "athletic" if she does aerobics several times a week (a stereotype­ consistent behavior for women), and a man may also receive the same "ath­ letic" label if he plays several competitive sports (stereotype-consistent be­ havior for men). In this case, the key difference between the male and female is not in quantity but in quality or kind of athleticism. Furthermore, we suspect that because of such qualitative differences in the nature of trait terms, perceivers would not spontaneously compare these two targets to one another. However, if required to rank order the "athletic" man and the "athletic" woman on the dimension of athleticism (thereby forcing a cross­ category comparison) we suspect that perceivers would judge the male target as "more" athletic than the female target. One reason for this is that the athleticism exhibited by males is more prototypical of the trait "athletic" (and perhaps more valued as an athletic behavior) than is the athleticism exhibited by females (Biernat et a/., 1998). In other words, "playing sports" is a better exemplar of the trait category than is "doing aerobics". In parallel-constraint satisfaction terms, the exemplars or nodes associated with "playing sports" should receive greater activation than those for "doing aerobics" for the trait "athletic", when both male and female categories are activated. In short, we agree with Kunda et al. (1997) that not only do stereotypes affect trait mean­ ing in terms of quantity or degree, but also in terms of quality or kind. In fact, it may be the case that standards of comparison shift precisely because the behaviors expected of one group differ in kind, or qualitatively, from the behaviors expected of the other group. Work by Dunning and his colleagues (Dunning & Sherman, 1997; Dunning, 1993; Dunning & Cohen, 1 992; Dunning, Meyerowitz, & Holzberg, 1989; Dunning, Perie, & Story, 1991) demonstrates that trait terms have adjustable meanings; they are "sponge" words that can take on the connotations of the context or the needs of the perceiver (Dunning, 1993). Furthermore, many traits have ambiguous behavioral referents, and therefore interpretations of and memory for these behaviors may be influenced by the social category membership of the actors (i.e., stereotypes). In a series of five studies, Dunning and Sherman (1997) found that stereotypes affected people's encoding and recall of ambiguous behavioral information. For example, participants were more likely to recall the sentence, "The nun was unhappy about the amount of liquor being served at the party", when cued with "didn't like all the alcohol" than when cued with "wanted more alcohol". Likewise, the sentence, "The rock musician was

SHIFTING STANDARDS IN STEREOTYPING

127

unhappy about the amount of liquor being served at the party", was better recalled when cued with "wanted more alcohol" than with "didn't like all the alcohol". Since the stereotype-consistent cues facilitated recall (rock musi­ cians like alcohol; nuns do not), it appears that participants made tacit in­ ferences at the time of encoding, based on their stereotypes, that were not explicit in the presented information (Dunning & Sherman, 1997; see also Slusher & Anderson, 1987). These findings are similar to the aforementioned work by Kunda and Sherman-Williams (1993) on the differential construal of "hitting someone who annoyed him/her" when applied to a housewife versus a construction worker. Participants in that study believed that the construction worker had "punched someone", while the housewife had "spanked a child". The SSM would predict that if the housewife and construction worker's aggression were assessed on subjective judgment scales (e.g., a 1-9 rating of aggressiveness), the two would be judged equivalently (based on the use of within-group standards). Spanking a child may be quite aggressive for a housewife, as is punching someone for a construction worker. However, if judges were asked to make a cross-category comparison (which is the more aggressive of the two?), we strongly suspect that the construction worker would be chosen. Not only does this judgment reflect the stereotype of construction workers as more aggressive than housewives, but also the perception that punching an adult is more aggressive than spanking a child (and perhaps more prototypical of the category "aggressive behavior"). We believe that these differences in meaning or construal contribute significantly to the signature shifting stand­ ards pattern-greater evidence of stereotyping on externally-anchored than on subjective judgment scales. Furthermore, our own research, described ear­ lier, on the "translation" of subjective evaluations into behavioral expecta­ tions is consistent with both the SSM and the PCSM in suggesting that the "same" traits may have very different meanings depending on the group stereotype that is activated (Kobrynowicz & Biernat, 1997; cf., Peabody, 1967). We also see one other important connection between the two models. In their review of the literature on how stereotypes combine with individuating information to influence judgments, Kunda and Thagard (1996) note that, in situations where diagnostic individuating information overrides the influence of stereotypes on trait ratings of targets, stereotypes may nonetheless con­ tinue to influence behavioral predictions relevant to the trait. This argument echoes back to the notion that a given trait or category has many associates. Not only do construction workers and accountants differ with respect to aggression, but they also differ with respect to social class; furthermore, there are different expectations or associates of aggression for different social classes. For example, although a construction worker and an accountant were judged equally unaggressive after "failing to react to an insult", judges

1 28

MONICA BIERNAT AND ELIZABETH R. THOMPSON

nonetheless assumed that the construction worker was more likely than the accountant to engage in aggressive behaviors such as "punching" and "cur­ sing" (Kunda et al., 1997). This finding parallels the subjective-objective dif­ ference that emerges in much of our shifting standards research, suggesting that behavioral predictions function in much the same manner as objective judgments. For Kunda and Thagard (1996), behavioral predictions continue to be influ­ enced by stereotypes because individuating information about a trait (e.g., knowledge that a construction worker has walked away from a fight) does not override any number of other associates to the category label. These other associates (such as "working class" or "unrefined" in the case of the con­ struction worker), can have connections to other stereotype-consistent be­ haviors (e.g., "punches people"). Thus, activation of an expected behavior can still occur, even in the face of behavioral evidence or individuating infor­ mation that is non-stereotypical. In the SSM, objective judgments are influ­ enced by stereotypes because they invite a cross-category perspective that better taps into available stereotyped mental representations of targets. The stereotype makes a given behavior seem more likely for a target from one group versus the other. These explanations obviously differ in a number of ways. However, both views suggest that a stereotype is not simply a static piece of information used by perceivers. The important point here is that our research on the greater susceptibility of objective judgment scales to ster­ eotyping effects can readily be incorporated into the PCSM of impression formation. Differences Alongside these similarities, the shifting standards model also differs from the PCSM in a number of ways. First, the models offer different accounts of how perceivers arrive at a judgment of a target person. In the PCSM, associ­ ates of observed information (including social category information) are processed simultaneously until the system settles on a judgment, and this processing occurs automatically. In the SSM, the assumption is that acti­ vated stereotypes instantiate judgment standards against which individuals are compared. Judgment is either assimilated to, or contrasted from, the expectation, depending on the stereotype-consistency of the target and the type of judgment called for (within- vs. across-category). Although we have also suggested that this occurs relatively automatically, models incorporat­ ing comparison typically assume a more controlled, sequential process (Petty & Wegener, 1993; Schwarz & Bless, 1 992). Furthermore, whereas Kunda and Thagard (1996) dismiss contrast effects as "too few" in number and too influenced by controlled processes to fall under the rubric of their model, contrast effects are a very likely outcome of the use of within-

SHIFTING STANDARDS IN STEREOTYPING

129

category judgment standards in the SSM. At this point, the process assump­ tions underlying the SSM have not been tested to the same degree as those of Kunda and Thagard's (1996) model, and it is therefore difficult to com­ ment on the differences between them. An intriguing possibility, however, is that judgment standards could be incorporated in an expanded version of a PCSM (see Read & Miller, 1998), thereby providing a further means of integrating these perspectives. Looked at in a slightly different manner, a connectionist model (such as the PCSM) could address the possibility that a judgment context asking for a subjective evaluation of a target will activate a different subset of associates than a judgment context asking for an objective evaluation. Some indirect evidence for this comes from research on the use of meta-informational cues in social judgments (Abele & Petzold, 1 998). In these studies, a series of targets belonging to one of two different occupational categories were either presented in blocks (e.g., a block of nurses was judged, followed by a block of stockbrokers) or in a mixed presentation mode with targets from the two occupations interspersed. When participants judged the helpfulness of nurses and stockbrokers presented in blocks (thereby receiving the meta­ informational cue that nurses were to be compared with nurses and stock­ brokers to stockbrokers), stockbrokers were judged overall to be more help­ ful than were nurses (a contrast effect). However, when judges evaluated these same targets in a mixed presentation format (thereby receiving the meta-informational cue that nurses and stockbrokers were to be compared to one another across categories) nurses were judged to be more helpful than were stockbrokers (an assimilation effect). These results parallel the subjective-objective judgment distinction that emerges in shifting standards research, and they also suggest that different associates may be triggered by different meta-informational cues. In this sense, judgment type (subjective vs. objective; within-category vs. cross-category) may also be part of the context that determines what judgments emerge. A second difference between the two models concerns the centrality of stereotypes. As indicated above, stereotypes have no privileged status in Kunda and Thagard's (1996) model-they are but one piece of information about a person that enters into and constrains judgment. However, stereo­ types play a central role in the shifting standards model-the array of pre­ dicted effects is based entirely on the assumption that stereotyped expectations are critically important. There is much to like about Kunda and Thagard's (1996) perspective on stereotype non-centrality. Perhaps most not­ ably, the demotion of stereotypes to a smaller role captures the spirit of the meta-analytic finding that stereotypes influence judgments of individuals to a lesser extent ( r = 0.25) than does individuating information ( r = 0.71; Kunda & Thagard, 1996, p. 292). Furthermore, it is often tricky to distinguish social category information from individuating information (is being a nurse an

SHIFTING STANDARDS IN STEREOTYPING

131

CONCLUSIONS

Over the past several decades, the social psychological literature has seen a considerable shift in emphasis from the view of stereotypes as rigid cognitive structures that are too readily applied, to a view of stereotypes as flexible constructs, subject to contextual variation and change. Our work on the shift­ ing standards model incorporates this contextual view by noting that ster­ eotyping effects are dependent on the judge's frame of reference (e.g., within­ category vs. across-category) and the type of judgment scale or evaluative language at hand. It has been our goal here to review this work and to relate it to other perspectives that address the broader issue of stereotypes in context. We have focused on comparisons with two important theoretical perspec­ tives that strongly reflect the contextual view of stereotypes and stereotyping. Self-categorization theory emphasizes context in its view that the very content of stereotypes varies across situations and according to perceiver goals, in the construct of the meta-contrast ratio, which describes when a social category will become salient in a given situation, and in the notion that the self can shift from personal to social levels of identification (see Turner, 1999, for a review). The PCSM of impression formation emphasizes context in its recognition that stereotypes serve to influence and constrain the meaning of other information one might learn about a target person (Kunda & Thagard, 1 996). There are many differences between our work on the SSM and these other approaches. Most notably, both SCT and the PCSM are broader in scope than the SSM; SCT in particular speaks to many domains beyond stereotyping, and also incorporates the self in a way the other perspectives do not. Other dif­ ferences have been reviewed earlier in this chapter, but we prefer to leave the reader focused on the similarities among these approaches. Each model recognizes that stereotypes are not uniformly applied to indi­ vidual group members, but rather are "part of the mix" of impression form­ ation, and that the judge's frame of reference is a critically important determinant of perception and judgment. We feel that the SSM can most profitably benefit from the other perspectives by better incorporating the self (in terms of group membership and identification, and standing on the judg­ ment dimension), and by focusing more specifically on the process by which perceivers judge members of stereotyped groups (e.g., does a parallel con­ straint process hold? Does judgment relative to a category standard involve controlled or automatic processes?) In turn, we think the SSM offers to the PCSM and reinforces for SCT the message that response language matters. Not only can people mean different things by trait terms they use, but they may have conflicting judgments and perceptions of the same target at the same time, depending on the type of judgment standard activated (e.g., min­ imum vs. confirmatory) and the form of evaluative language at hand (e.g., subjective vs. objective).

1 32

MONICA BIERNAT AND ELIZABETH R. THOMPSON

By highlighting several theoretical perspectives that strongly implicate the role of context in stereotyping, we have only sampled the large array of empirical findings that bear on the topic of contextual variation. As noted in the introductory paragraphs of this chapter, researchers have demonstrated in a number of ways that the nature of stereotypes and the manner in which they are applied may be substantially influenced by factors both internal and exter­ nal to the perceiver. We hope that our selective review demonstrates the value of the contextualized perspective in understanding stereotypes and the stereotyping process.

REFERENCES Abele, A. E., & Petzold, P. ( 1998) . Pragmatic use of categorical information in impres­ sion formation. Joum{ll of Personality and Social Psychology, 75, 347-58. Asch, S. E. ( 1946) . Forming impressions of personality. Journal of Abnormal and Social Psychology, 41, 258-90. Biernat, M., Crandall, C. S., Young, L. V., Kobrynowicz, D., & Halpin, S. M. ( 1998) . Al that you can be: Stereotyping of self and others in a military context. Journal of Personality and Social Psychology, 75, 301-17. Biernat, M., & Kobrynowicz, D. ( 1 997 ) . Gender- and race-based standards of compet­ ence: Lower minimum standards but higher ability standards for devalued groups. Journal of Personality and Social Psychology, 72, 544-57. Biernat, M., & Manis, M. ( 1994) . Shifting standards and stereotype-based judgments. Journal of Personality and Social Psychology, 66, 5-20. Biernat, M., Manis, M., & Nelson, T. F. ( 1991 ) . Comparison and expectancy processes in human judgment. Journal of Personality and Social Psychology, 61, 203-1 1 . Biernat, M . , Nario-Redmond, M . R . & Ma, J. E . (2001 ) . The role o f stereotypes and judgment standards in trait diagnosis (Manuscript under review). Biernat, M., & Thompson, E. R. ( 200) . Multiple category membership and shifting judgment standards (Unpublished manuscript). Biernat, M., & Vescio, T. K. ( 1993) . Categorization and stereotyping: Effects of group context on memory and social judgment. Journal of Experimental Social Psychology, 29, 166-202. Biernat M., & Vescio, T. K. (in press). She swings, she hits, she's great, she's benched: Implications of gender-based shifting standards for judgment and behavior. Person­

ality and Social Psychology Bulletin. & Green, M. L. ( 1 996) . Selective self-stereotyping. Journal of Personality and Social Psychology, 71, 1 1 94-209. Biernat, M., Vescio, T. K., & Manis, M. ( 1998) . Judging and behaving toward members Biernat, M., Vescio, T. K.,

of stereotyped groups: A shifting standards perspective. In C. Sedikides, J. Schopler & C. A. Insko (Eds), Intergroup Cognition and Intergroup Behavior (pp. 151-75 ) . Mahwah, NJ: Erlbaum. Brewer, M. B. ( 1988) . A dual process model of impression formation. In T. K. Srull & R. S. Wyer, Jr. (Eds), Advances in Social Cognition (Vol. 1, pp. 1-36) . Hillsdale, NJ: Erlbaum. Campbell, D. T., Hunt, W. A., & Lewis, N. A. ( 1958) . The relative susceptibility of two rating scales to disturbances resulting from shifts in stimulus contexts. Journal of Applied Psychology, 42, 213-17.

134

MONICA BIERNAT AND ELIZABETH R. THOMPSON

Francke, L. B. (1997). Ground Zero: The Gender Wars in the Military. New York: Simon & Schuster. Gidron, D., Koehler, D. J., & Tversky, A. (1993). Implicit quantification of personality traits. Personality and Social Psychology Bulletin, 19, 594-. Goldberg, P. (1968). Are women prejudiced against women? Transaction, S, 28-30. Greenberg, J., Pyszczynski, T., Solomon, S., Rosenblatt, A., Veeder, M., Kirkland, S., & Lyon, D. (1990). Evidence for terror management theory II: The effects of mor­ tality salience on reaction to those who threaten or bolster the cultural worldview.

Journal of Personality and Social Psychology, 58, 308-18. Hamilton, D. L., & Sherman, S. J. (1994). Stereotypes. In R. S. Wyer, Jr. & T. K. Srull (Eds), Handbook of Social Cognition (pp. 1--68). Hillsdale, NJ: Erlbaum. Haslam, S. A., & Turner, J. C. (1992). Context-dependent variation in social stereo­ typing 2: the relationship between frame of reference, self-categorization and accen­ tuation. European Journal of Social Psychology, 22, 251-77. Haslam, S. A., Turner, J. C., Oakes, P. J., McGarty, C., & Hayes, B. K. (1992). Context­ dependent variation in social stereotyping: I. The effects of intergroup relations as mediated by social change and frame of reference. European Journal of Social

Psychology, 22, 3-20. Hogg, M. A., & Turner, J. C. (1987). Intergroup behaviour, self-stereotyping and the salience of social categories. British Journal of Social Psychology, 26, 325-40. Jussim, L., Coleman, L. M., & Lerch, L. (1987). The nature of stereotypes: A com­ parison and integration of three theories. Journal of Personality and Social Psychol­

ogy, 52, 536--46. (1973) The processes of causal attribution. American Psychologist, 28, 107-28. Kobrynowicz, D., & Biernat, M. (1997). Decoding subjective evaluations: How stereo­ types provide shifting standards. Journal of Experimental Social Psychology, 33, 579-601. Krantz, D. L., & Campbell, D. T. (1961). Separating perceptual and linguistic effects of context shifts upon absolute judgments. Journal of Experimental Psychology, 62, 3542. Kunda, Z., & Thagard, P. (1996). Forming impressions from stereotypes, traits, and behaviors: A parallel-constraint satisfaction theory. Psychological Review, 103, 284308. Kunda, Z., & Sherman-Williams, B. (1993). Stereotypes and the construal of indi­ viduating information. Personality and Social Psychological Bulletin, 19, 90-99. Kunda, Z., Sinclair, L., & Griffin, D. (1997). Equal ratings but separate meanings: Stereotypes and the construal of traits. Journal of Personality and Social Psychology, 72, 720-34. Locksley, A., Borgida, E., Brekke, N., & Hepburn, C. (1980). Sex stereotypes and judgments of individuals: An instance of the base-rate fallacy. Journal of Experimen­ tal Social Psychology, 18, 23-42. Lorenzi-Cioldi, F. (1991 ) . Self-stereotyping and self-enhancement in gender groups. European Journal of Social Psychology, 21, 403-17. Lorenzi-Cioldi, F. (1998). Group status and perceptions of homogeneity. In W. Stroebe & M. Hewstone (Eds), European Review of Social Psychology (Vol. 9, pp. 31-75). Kelley, H. H.

Chichester: Wiley. Maass, A., Milesi, A., Zabbini, S., & Stahlberg, D. (1995). Linguistic intergroup bias: Differential expectancies or in-group protection? Journal of Personality and Social Psychology, 68, 1 16-26. Maass, A., & Schaller, M. (1991). Intergroup biases and the cognitive dynamics of stereotype formation. European Review of Social Psychology, 2, 189-209.

136

MONICA BIERNAT AND ELIZABETH R. THOMPSON

Reeder, G. D., & Brewer, M. B. (1979). A schematic model of dispositional attribution in interpersonal perception. Psychological Review, 86, 61-79. Reeder, G. D., Pryor, J. B., & Wojciszke, B. (1992). Trait-behavior relations in social information processing. In G. R. Semin & K. Fiedler (Eds), Language, Interaction and Social Cognition (pp. 37-57). London: Sage. Rothbart, M., & Park, B. (1986). On the confirmability and disconfirmability of trait concepts. Journal of Personality and Social Psychology, SO , 131-42. Schaller. M., & Conway, L. G., III (1999). Influence of impression management goals on the emerging contents of group stereotypes: Support for a social evolutionary process. Personality and Social Psychology Bulletin, 25, 819-33. Schwarz, N., & Bless, H. (1992). Constructing reality and its alternatives: An inclusion/ exclusion model of assimilation and contrast effects in social judgment. In L. L. Martin & A. Tesser (Eds), The Construction of Social Judgments (pp. 217-45). Hillsdale, NJ: Erlbaum. Sherif, M., & Hovland, C. I. (1961). Social Judgment: Assimilation and Contrast Effects in Communication and Anitude Change. New Haven, CT: Yale University Press. Shoda, Y., & Mischel, W. (1993). Cognitive social approach to dispositional inferences: What if the perceiver is a cognitive social theorist? Personality and Social Psychol­ ogy Bulletin, 19, 574-85. Sinclair, L., & Kunda, Z. (1999). Reactions to a Black professional: Motivated inhibi­ tion and activation of conflicting stereotypes. Journal of Personality and Social Psychology, 71, 885-904. Slusher, M. P., & Anderson, C. A. (1987). When reality monitoring fails: The role of imagination in stereotype maintenance. Journal of Personality and Social Psychol­ ogy, 52, 653-62. Spears, R., & Manstead, A. S. (1989). The social context of stereotyping and differen­ tiation. European Journal ofSocial Psychology, 19, 101-21. Stangor, C., Lynch, L., Duan, C., & Glass, B. (1992). Categorization of individuals on the basis of multiple social features. Journal of Personality and Social Psychology, 62, 207-18. Stangor, C., & McMillan, D. (1992). Memory for expectancy-congruent and expectancy-incongruent information: A review of the social and social developmen­ tal literatures. Psychological Bulletin, 111, 42-62. Stangor, C., & Schaller, M. (1996). Stereotypes as individual and collective represen­ tations. In C. N. Macrae, C. Stangor & M. Hewstone (Eds), Stereotypes and Stereo­ typing (pp. 3-37). New York: Guilford. Strack, F., & Mussweiler, T. (1997). Explaining the enigmatic anchoring effect: Mecha­ nisms of selective accessibility. Journal of Personality and Social Psychology, 73, 437-46. Swim, J., Borgida, E., Maruyama, G., & Myers, D. G. (1989). Joan McKay versus John McKay: Do gender stereotypes bias evaluations? Psychological Bulletin, 105, 40929. Tajfel, H. (1969). Cognitive aspects of prejudice. Journal of Social Issues, 25, 79-97. Tajfel, H. (1972). Some developments in European social psychology. European Jour­ nal of Social Psychology, 2, 307-21 . Tajfel, H . (1979). Individuals and groups i n social psychology. British Journal of Social and Clinical Psychology, 18, 183-90. Tajfel, H., & Wilkes, A. L. (1963). Classification and quantitative judgement. British Journal of Social and Clinical Psychology, 54, 101-14. Taylor, S. E., Fiske, S. T., Etcoff, N. L., & Ruderman, A. J. (1978). Categorical and contextual bases of person memory and stereotyping. Journal of Personality and Social Psychology, 36, 778-93.

SHIFfiNG STANDARDS IN STEREOTYPING

137

Trope, Y. (1998). Dispositional bias in person perception: A hypothesis-testing perspective. In J. M. Darley & J. Cooper (Eds), Attribution and Social Interaction: The Legacy of Edward E. Jones. Washington, DC: American Psychological Association. Turner, J. C. (1975). Social comparison and social identity: Some prospects for inter­ group behaviour. European Journal of Social Psychology, 5, 5-34. Turner, J. C. (1978). Social comparison, similarity and ingroup-favoritism. In H. Tajfel (Ed.), Differentiation between Social Groups: Studies in the Social Psychology of Intergroup Relations. London: Academic Press. Turner, J. C. (1982). Toward a cognitive redefinition of the group. In H. Tajfel (Ed.), Social Identity and Intergroup Relations. Cambridge: Cambridge University Press. Turner, J. C. (1984). Social identification and psychological group formation. In H. Tajfel (Ed.), The Social Dimension: European Developments in Social Psychology (Vol. 2). Cambridge: Cambridge University Press. Turner, J. C. (1985). Social categorization and the self-concept: A social cognitive theory of group behaviour. In E. J. Lawler (Ed.), Advances in Group Processes: Theory and Research (Vol. 2). Greenwich, CT: JAI. Turner, J. C. (1999). Some current issues in research on social identity and self­ categorization theories. In N. Ellemers, R. Spears & B. Doosje (Eds), Social Identity: Context, Commitment, Content (pp. 6-34). Oxford: Blackwell Scientific. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the Social Group: A Self-categorization Theory. New York: Basil Blackwell. Upshaw, H. S. (1962). Own attitudes as an anchor in equal appearing intervals. Journal of Abnormal and Social Psychology, 64, 85-96. Upshaw, H. S. (1965). The effect of variable perspectives on judgments of opinion statements for Thurstone scales: Equal-appearing intervals. Journal of Personality and Social Psychology, 2, 60-69. Upshaw, H. S. (1969). Stimulus range and the judgmental unit. Journal ofExperimental Social Psychology, 5, 1-1 1. van Knippenberg, A. (1984). Intergroup differences in group perceptions. In H. Tajfel (Ed.), The Social Dimension: European Developments in Social Psychology. Cambridge: Cambridge University Press. van Knippenberg, A., & Ellemers, N. ( 1 990). Social identity and intergroup differentia­ tion processes. In W. Stroebe & M. Hewstone (Eds), European Review of Social Psychology (Vol. 1, pp. 137-69) Chichester: Wiley. Volkmann, J. (1951). Scales of judgment and their implications for social psychology. In J. H. Rohrer & M. Sherif (Eds), Social Psychology at the Crossroads (pp. 273-94). New York: Harper. Wigboldus, D., Spears, R., & Semin, G. (1999). Categorization, content and the con­ text of communicative behaviour. In N. Ellemers, R. Spears & B. Doosje (Eds), Social Identity: Context, Commitment, Content. Oxford: Blackwell Scientific.

This page intentionally left blank

Chapter 5

A Group By Any Other N arne-The Role of Entitativity in Group Perception David L. Hamilton University of California at Santa Barbara, USA

Steven J. Sherman

Indiana University, USA

Luigi Castelli

University of Padova, Italy

ABSTRACT The social world consists of numerous and diverse groupings of people into meaningful and important collectives. As perceivers, we routinely encounter aggregates of people, some of which we endow with the property of groupness, and others we do not. Moreover, the variety of groups is enormous, yet per­ ceivers differentiate among them and understand their properties. This chapter discusses how and why perceivers "see" an aggregate of persons as a group, the distinctions among different types of groups that perceivers detect, the variation among groups in their perceived groupness or entitativity, and the consequences that follow from perceiving a group as an entitative unit. The results of our research program addressing these issues are summarized, and implications for remaining unanswered questions are discussed.

Correspondence regarding this chapter should be addressed to David L. Hamilton, Department of Psychology, University of California, Santa Barbara, CA 93106, USA.

European Review of Social Psychology, Volume

© 2002 John

Wiley

&

Sons Ltd.

12.

Edited by Wolfgang Stroebe and Miles Hewstone.

140

DAVID L. HAMILTON, STEVEN J. SHERMAN AND LUIGI CASTELLI

Committee-a group of men who individually can do nothing but as a group decide that nothing can be done (Fred Allen). Land of our birth, we pledge to thee Our love and toil in the years to be; When we are grown and take our place, As men and women with our race (Rudyard Kipling). To the family-that dear octopus from whose tentacles we never quite escape, nor, in our inmost hearts, ever quite wish to (Dodie Smith). Al of these quotes refer to groups with which we are very familiar. All of us

originate from a family context; we all have some national and ethnic identi­ fication, and (for better or worse) we have all served on committees. To a considerable extent our social life is organized according to group-like struc­ tures, and we all are members of numerous groups. Yet, as the quotations above illustrate, the range and variety of those groups is impressively broad. Indeed, the term "group" encompasses an enormous array of social struc­ tures. Not only are we members of, and participants in, groups of various kinds, but also we, individually and collectively, perceive and interact with groups continually in everyday life. As perceivers, we recognize that others belong to and function in groups, and our cognizance of that fact influences both our perceptions of and behavior toward those persons. This chapter is concerned with the perception of groups by perceivers. A group, at its minimum, is a collection of individuals. However, not all collec­ tions of individuals are groups. The fact that perceivers can somehow dif­ ferentiate between collections of persons that are and are not groups, and that they can further differentiate among groups that are families, committees, and nationalities, is perhaps trivially obvious. It is also fundamental to successful adaptation in social life. And, for the social psychologist interested in group perception, it also poses a number of intriguing and challenging questions, the answers to which are not so obvious. In this chapter we raise some of those questions and attempt to address them. Since its beginnings, social psychology has developed massive literatures concerned with both the nature of groups at various levels of analysis and how they function (Levine & Moreland, 1998; Pruitt, 1998) and the perception of groups by observers and how those perceptions guide behavior (Brewer & Brown, 1 998; Crocker, Major, & Steele, 1 998; Fiske, 1998). Despite the rich and extensive history of research on these topics, questions of how perceivers "see" a collection of persons as a group have received little systematic investigation. In our work, we have focused on one important aspect of group perception, the perceived "entitativity" or "groupness" of the group. Although the term "entitativity" was coined by Campbell (1958) over 40 years ago, it is only in recent years that researcheFS have attempted to understand how that percep-

ROLE OF ENTITATIVITY IN GROUP PERCEPTION

141

tion is achieved and what differences it makes. However, this recent work has already revealed the varieties and complexities inherent in our conception of groups. As we shall see, differences in perceived entitativity have important consequences for various perceptions, inferences, and evaluations of the group and its members, as well as for behaviors toward, and interactions with, group members. ENTITATIVITY: THE CONCEPT

To begin our analysis, it is useful to consider the meaning of the concept of entitativity. When he introduced the term, Campbell (1958) simply offered (in a footnote) the following: "Entitativity: the degree of being entitative. The degree of having the nature of an entity" (p. 17). Campbell's discussion made clear that the term referred to perceiving the "groupness" of groups, and his article was effective in delineating some of the cues that perceivers might use in detecting that an aggregate of individuals possesses that quality of group­ ness. Nevertheless, the definitional imprecision and the variety of cues on which such perceptions might be based has resulted in considerable latitude in the way the term has been defined, manipulated, and measured in research. Some authors (e.g., Brewer, Weber, & Carini, 1995; McGarty, Haslam, Hutchinson, & Grace, 1995; Yzerbyt, Rogier, & Fiske, 1998) have focused on perceived similarity or variability among group members as a cue to group­ ness. Others (e.g., Haslam, 1 998; Yzerbyt, Rocher, & Schadron, 1997) have emphasized a relation between group entitativity and the concept of essential­ ism in the perception of social categories. Another view (Brewer & Harasty, 1996) has related perceived entitativity to the extent to which the perceiver has a clear prototype representing the group. Finally, some writers (e.g., Gaertner & Schopler, 1998; Hamilton, Sherman, & Lickel, 1 998) have stressed the interdependence, interconnectedness, and organization among group members as an important basis for perceived entitativity. These various approaches often reflect different, and sometimes unspecified, assumptions about the nature of entitativity. To avoid such assumptions and any ideologi­ cal constraints they may entail, we choose to employ a neutral, generic defini­ tion of the term. We will use the term "entitativity" to refer to the perception that an aggregate of individuals is bonded together in some way to constitute a group. The variety inherent in current conceptions of entitativity has brought both advantages and disadvantages to progress in research inquiry. The disadvan­ tage, of course, is that there remains an unsatisfying lack of clarity in the meaning of this term and especially in the limits on its referents. On the other hand, and particularly in the early stages of a new line of inquiry, this concep­ tual liberalism is perhaps beneficial in that it can permit researchers to

142

DAVID L. HAMILTON, STEVEN J. SHERMAN AND LUIGI CASTELLI

consider and to explore numerous implications of the concept. We believe that this has, in fact, been the case in this new research on entitativity, and in the remainder of this chapter we hope to convey the richness and diversity of questions and issues that have been examined as a consequence. Neverthe­ less, it wil soon become important that researchers come to a common under­ standing of what is and is not meant by this term. As Campbell (1958) noted, an analysis of entitativity is not simply a matter that some collections of groups "have it" and others do not; groups vary considerably in the extent to which we perceive them as possessing this quality. Moreover, as suggested at the outset, the term "group" itself can have a variety of meanings; it can refer to entities that vary considerably in their properties. Do perceivers meaningfully diferentiate among different types of groups, and if so, on what basis? If perceivers do recognize groups as varying in entitativity, what difference does it make in their perceptions of those groups? Finally, do different types of groups differ in the extent to which they are perceived as entitative? In this chapter, we discuss the findings of recent research on perceiving group entitativity and attempt to use those findings to assess our understanding of these questions. ANTECEDENTS TO PERCEPTIONS OF ENTIT ATIVITY

Cues to Entitativity

Campbell's (1958) discussion of entitativity was focused on the question of how perceivers use available informational cues to infer that an aggregate of individuals has the quality of being a group. His analysis, based on a Gestalt approach, highlighted certain properties that, if present, might suggest that an aggregate is a group. For example, if the individual persons are in close proximity to each other, if they appear similar to each other, if they move together in coordinated action and appear to have common fate (e.g., shared goals and common outcomes), then it is more likely that they would be perceived as an entitative unit. Although Campbell reported no research evidence in support of his analysis, these ideas have provided useful guides to subsequent investigators. Following Campbell's conceptual analysis, our research has investigated the role of different types of stimulus information one might have about groups as bases for perceivers' inferences about the entitativity of groups. In a series of studies (Lickel, Hamilton, Wieczorkowska, Lewis, Sherman, & Uhles, 200), we sought to determine the extent to which several stimulus properties are predictive of perceptions of a group's entitativity. Parallel studies were con­ ducted at the University of California, Santa Barbara (UCSB) and the Uni­ versity of Warsaw, Poland. Participants rated a diverse array of 40 groups (e.g., a family, a committee, women, an orchestra, people at a bus stop) on a

ROLE OF ENTITA TIVITY IN GROUP PERCEPTION

143

variety of rating scales. Instructions indicated that some groups seem to be "groups" to a greater extent than others are (without suggesting why), and participants were asked to rate the extent to which each one of the 40 groups "qualified as a group", which we used as a measure of perceived group en­ titativity. Participants then rated each of those same groups on several group properties that previous authors had suggested as cues to entitativity. These included the extent to which members interact, the importance of the group to its members, the similarity among group members, the extent to which mem­ bers share common goals and common fate, the size of the group, the length of its history, and whether joining and leaving the group is easy or difficult. Among these group attributes, interaction, importance, common goals, com­ mon fate, and similarity were most strongly correlated with perceived group entitativity. In contrast, group size, length of group history, and the per­ meability of group boundaries were essentially uncorrelated with entitativity. These studies provided among the first empirical tests of the relative import­ ance of a variety of group properties as indicators of group entitativity across a broad spectrum of groups. Importantly, the results of these analyses, based on data from the UCSB and Warsaw samples, were remarkably similar (see Lickel et al., 2000, for details). Types of Groups

Our next step was to use these rated group properties to discern whether distinct types of groups could be identified. That is, the 40 stimulus groups rated by participants intentionally included a very diverse array of groups, which differed in many respects. Therefore, we considered the ratings on these eight group characteristics to comprise a profile for each group, and we then used these profiles to statistically identify several group types, each of which had a distinct pattern of features. For example, intimacy groups (e.g., family, close friends, local street gang) are small groups with high levels of interaction; membership is important to their members, the groups have long histories, and changing membership is not easy. Task-oriented groups (e.g., a committee, a jury, cast of a play, people working in a restaurant) are also relatively small and highly interactive and their mem­ bers have shared goals and common outcomes, but compared to intimacy groups, these groups are less centrally important to their members, are of shorter duration, and joining and leaving such groups is not difficult. In marked contrast to these first two types, social categories (e.g., women, Blacks, Jews, Americans) are very large groups with long histories and relatively impermeable boundaries, and lower levels of interaction among members. Loose associations (e.g., students at the same university, people living in the same neighborhood, people who like classical music) are also large groups but have very low levels of interaction, have shorter histories,

ROLE OF ENTITATIVITY IN GROUP PERCEPTION

145

groups to which they belonged, and subsequently were asked to rate each of these groups on a series of attributes. These rating scales assessed the same variables rated in the previous studies, as well as one additional attribute-the importance of membership in the group to the participants themselves. Cor­ relational analyses of these variables with ratings of groupness (entitativity) paralleled those relations found in the earlier studies: interaction, importance of membership, common goals, common fate, and similarity again were most strongly related to perceived group entitativity. In this case, however, ratings of the importance of one's own membership in the group were most strongly correlated with perceived entitativity. When the group profiles based on these ratings were analyzed to determine types of groups, the results were again similar to the earlier analyses, with intimacy groups, task groups, social cate­ gories, and loose associations evident in these results. In sum, in this series of studies we have investigated some of the antecedent cues upon which perceptions of group entitativity may be based. These studies have shown that certain group features or properties-high interaction among group members, importance of group membership, sharing of common goals and common outcomes, similarity among members-consistently were most highly correlated with ratings of group entitativity across a broad spectrum of groups. Moreover, structural analyses indicated that perceivers differentiate several distinct types of groups, including intimacy groups, task groups, social categories, and loose associations. Finally, these types of groups differed sig­ nificantly in the extent to which they were perceived as possessing entitativity. All of these findings consistently emerged in similar analyses: (a) in three separate studies (conducted in Santa Barbara, CA, USA; Warsaw, Poland; Bloomington, IN, USA); {b) in analyses based on ratings of groups on de­ scriptive scales and in structural analyses based on participants' sorting of the groups into categories; and (c) using groups provided by the researchers and also using groups generated by the participants to which they belonged (see Lickel et al., 2000, for details). Thus, our findings provide evidence of gener­ ality in the correlates of perceived entitativity, and also suggest that these types of groups reflect some fundamental distinctions in the ways people think about social groups. Spontaneous Dift'erentiation of Group Types

All of the analyses reported by Lickel et al. involved tasks in which the participants engaged in thoughtful, deliberative procedures. One might plau­ sibly wonder: do people spontaneously make these distinctions among types of groups in their everyday perceptions, thoughts, and categorizations, even at times and in contexts when they are not asked to rate or sort various groups on various attributes? The answer appears to be "Yes". We {Sherman, Castelli, & Hamilton, in press) have recently obtained evidence that people

146

DAVID L. HAMILTON, STEVEN J. SHERMAN AND LUIGI CASTELLI

spontaneously distinguish among these group types in their normal cognitive functioning. If the group types that emerged in the Lickel et a/. studies are indeed psychologically meaningful, then those types would represent cognitive struc­ tures that would influence the way information about social groups is spon­ taneously processed and stored in memory. We focused on how information about individuals who are presented as members of different social groups is organized in memory. If the defining characteristics of a group type are spon­ taneously extracted and used to categorize groups into different typologies, specific information about two groups that share similar defining characteris­ tics should be more likely to be confused, even when this information is unrelated to the distinct defining features. In the first experiment (Sherman et al., 200), we adopted a recognition memory paradigm that represents a useful, unobtrusive measure of how information is encoded and stored in memory (Taylor, Fiske, Etcoff, & Ruderman, 1 978). We presented participants with pictures of 60 male faces one time each for 3 seconds, along with a label indicating the group mem­ bership of the person portrayed in the picture. Six different group labels were used (10 members of each group), two for each of the three major types of groups identified in the Lickel et a/. (200) studies (i.e., family members and friends for intimacy groups, co-workers and jury members for task-oriented groups, and French and Presbyterians for social-categories). Participants simply looked at the faces and read aloud the group identifica­ tion labels. In a subsequent recognition test, only the pictures of the 60 faces were presented, and participants were asked to recall the original group membership of the person. Errors and response latencies constituted the main dependent variables. We predicted that the highest number of errors would occur from a given group to a group of the same type (e.g., misremembering a French person as a Presbyterian) rather than to a group of a different type (e.g., misremembering a French person as a jury member). In other words, "within-group-type er­ rors" should be more frequent than "between-group-type errors". In addi­ tion, response latencies should be faster for within-group-type errors than for between-group-type errors, because participants will be more certain of the (mis )identifications in the former case. The accessibility of the group-type label will make all groups of that type appear to be familiar. Results showed significantly more within-group-type errors than between­ group-type errors. Figure 5.1 shows the findings from Experiment 1. Thus, if a face had been labeled as a Frenchman (a social category), then participants were far more likely to misidentify the face as a Presbyterian (another social category) than as a family member (an intimacy group) or a jury member (a task-oriented group). In addition, the within-group-type errors were signifi­ cantly faster than the between-group-type-errors.

148

DAVID L. HAMILTON, STEVEN J. SHERMAN AND LUIGI CASTELLI

be more difficult to then conclude that he is not a member of another group of the same type (e.g., a Belgian, another social category). Knowledge that the target belongs to a group of a particular type (e.g., a social category) should lead one to infer (at least momentarily) that the target is a member of that type of group, and therefore groups of the same type will be confusable. In this study, participants learned, to a strict criterion, the group memberships of 24 social targets, by viewing their pictures and a category label. Six different category labels were used, two representing each of the three group types. Participants then performed a social category verification task. On the com­ puter screen, one of the six group labels appeared. One second later, a previously-seen picture appeared below the label, and participants had to respond as quickly and accurately as possible whether the target in the picture was or was not a member of the group indicated by the label. The data of interest are participants' responses, in terms of both accuracy and speed, for mismatching trials, when the label was not the correct membership category of the person in the picture. We expected that a higher number of incorrect "Yes" answers would occur in the case of members of a group of the same type. In addition, we predicted that response latencies to incorrectly say "Yes" would be faster if the incorrect group label were of the same group type as the actual membership. The results supported both predictions, and again indicated that groups of the same general type are highly confusable and thus that they are represented together in memory. These experiments clearly indicate that not all groups are alike and that the group typology identified by Lickel et a/. (200) is spontaneously used when information is encoded about various social groups. This spontaneous use of an implicit group typology has a powerful influence on the cognitive organiza­ tion of information about social groups and their members. These different group types are represented according to a typology that takes into account the phenomenal antecedents that serve as indicators of perceived "group­ ness" (Campbell, 1958; Hamilton et al., 1998). The ways in which these group types differ are systematic and involve some rather high-level, abstract char­ acteristics. The fact that people spontaneously differentiate among group types is especially interesting, given that there do not appear to be any simple words to describe the group types. There are important implications of this spontaneous use of a group typol­ ogy. For example, the impression one forms of any given social group should be more affected by any information about other groups of the same general type than by information about groups of different types. In addition, memory errors about which group might have been responsible for any given initiative or outcome may be greater when two groups are of the same type. Thus, source-monitoring processes (Johnson, Hashtroudi, & Lindsay, 1993) may be affected by our group typology, because membership in the same superordi­ nate category is known to affect source monitoring. Finally, impressions of

ROLE OF ENTIT A TIVITY IN GROUP PERCEPTION

149

individuals may be affected by group typology if information about individ­ uals is sought and organized according to similarities of the individual's group membership, as specified by this typology. The functionality of employing such a group typology in the cognitive organization of persons and groups will be discussed in a subsequent section. What We've Learned, and Some New Questions

These studies by Lickel et al. (200) and Sherman et al. (in press) provide evidence for several points that have important implications for social percep­ tion. First, they document that perceivers recognize variation among groups in the extent to which they constitute meaningful entities, in the sense of having the quality of "groupness" and bondedness among their members. Second, they provide evidence that perceivers distinguish among different types of groups and recognize the different types as having meaningfully different properties and characteristics. Third, the results show that these different types of groups are seen as differing in entitativity, as well as in other important characteristics. Fourth, evidence indicates that these distinctions are not the product merely of completing paper-and-pencil tasks involving group descriptions, which could reflect demand characteristics and/or seman­ tic and cultural distinctions regarding groups, but that people spontaneously utilize these distinctions in encoding and storing information about group members. The findings also raise a number of intriguing questions, the answers to which await further investigation. We briefly consider two such questions here. First, given that perceivers meaningfully and consistently differentiate among different types of groups, and that these different group types are all seen (to varying degrees) as possessing entitativity, one might pose the follow­ ing question: does entitativity mean the same thing for all group types? Is it meaningful to talk about entitativity in all of these cases? A complete answer to these questions is not currently available. We defined entitativity as the perception that a collection of persons is bonded together in some way and is therefore seen as a meaningful unit, and our results show that these very different types of groups are indeed seen (to varying degrees) as comprising groups. What our findings also show is that the different group types are defined by different features or properties. For example, intimacy groups are perceived as small, highly interactive, long-lasting, and unchanging groups that are highly important to their members. Task groups share some of these properties (small size, high interaction), but are also characterized by other features (lower importance to members, shorter history, ease in joining or leaving). Social categories are defined by quite different features, such as their long histories and their unchanging (impermeable) membership status. These differences in defining features (at least in how these groups are perceived)

150

DAVID L. HAMILTON, STEVEN J. SHERMAN AND LUIGI CASTELLI

suggest the possibility that the very perception of entitativity or groupness itself may be based on different group properties for different types of groups. Second, assuming that there is a psychological reality to these different group types, the question naturally arises as to what functions are served by having such a cognitive organization and categorization of groups. That is, it may be entirely plausible that perceivers make important distinctions be­ tween individuals and groups, that they think of others as individuals and also recognize that people join together to form groups. But why isn't that ade­ quate? Why does the perceiver go further, as our findings document, and perceptually differentiate among, and presumably form distinct cognitive structures for, different types of groups? What functions are served by making such distinctions? Again, at this point we can only speculate. As we saw, our perceivers differentiated among three major group types­ intimacy groups, task-oriented groups, and social categories. We propose that these three group types correspond to three of the most important social motivations that characterize humans. First, there is the need for belonging­ ness (Baumeister & Leary, 1995). This need involves aspects of affiliation (Schachter, 1959), emotional attachment and support (Bowlby, 1969), love, and acceptance. Membership in high-intimacy groups, such as friends or fam­ ily, serves to promote these goals, while at the same time preventing loneli­ ness, alienation, and rejection. A second major social human motivation is that of achievement (Atkinson & Feather, 1966; McClelland, Atkinson, Clark, & Lowell, 1953) and compet­ ence (White, 1959). Here the goals are toward mastery, cohesion, success, cooperation, and efficacy (Bandura, 1982). In many contexts, task completion and goal attainment are more easily achieved, and sometimes can only be achieved, as a group effort. Thus, participation in task-oriented groups can serve an important function in helping the individual to achieve important goals. At the same time, the possibilities of failure and conflict are reduced by membership in task-oriented groups. Third, humans also have an important need to establish a stable and secure sense of self and social identity (Tajfel & Turner, 1979, 1986; Turner, 1987). The specific goals may involve self-esteem maintenance and enhancement, self-knowledge, self-protection, self-validation, and reduction of uncertainty (Hogg & Mullin, 1999). As emphasized in social identity theory, these goals are often met by the broad social categories to which people belong. Individ­ uals take pride in and often find much of their personal identities in their group memberships, including their identification with large social categories, such as their religion, nationality, or ethnicity. Because it is of primary importance that these three kinds of social motives are satisfied, it should not be surprising that people think about groups in ways that are associated with the satisfaction of these different needs and motives. If one has an acute need for achievement or affiliation, it is important

ROLE OF ENTITA TIVITY IN GROUP PERCEPTION

151

to know exactly where to go in order to fulfill such a need. Thus, groups that fulfill the same motive may be represented together in memory, even if they share few lower level features in common. CONSEQUENCES OF PERCEIVED ENTITATIVITY

Given that the evidence reviewed above establishes that perceivers recognize variation among groups in their entitativity, the next question naturally con­ cerns the consequences of this variation. What difference does it make? Re­ search by several investigators has pursued that question in a number of different ways, resulting in documentation of several consequences of per­ ceived entitativity. Integrative Processing

Hamilton and Sherman (1996; Hamilton, Sherman, & Lickel, 1998) reasoned that entitativity reflects the perceived unity and coherence of a group target. They also recognized that different social targets vary considerably in the extent to which they are perceived as having these qualities. For example, there is a long and rich literature on forming impressions of individuals, with the inherent assumption that the target person has a unified and coherent personality and the perceiver's goal in forming an impression is to compre­ hend that personality. Individual targets, in other words, are assumed to be high in entitativity. If so, then differences in the perceptions of individual and group targets may parallel differences in the perceptions of groups that differ in perceived entitativity. Following Asch (1946), Hamilton and Sherman (1996) argued that the expectation of unity leads perceivers to use the information available to infer dispositional qualities in the target of impression formation, to assume consis­ tency across time and situation in the nature of that target's personality, to form an organized impression of the target, and to attempt to resolve any inconsistencies in the information acquired about the target. Hamilton and Sherman predicted that these processes would be more prevalent in forming impressions of individuals than of groups. More generally, they proposed that perceivers are more likely to engage in what they ( Hamilton, Sherman, & Maddox, 1999) called integrative processing in forming impressions of individ­ uals than of groups and, by extension, of groups high in entitativity than of groups low in entitativity. The strategy they adopted was to compare two experimental conditions, both of which involved the same information ( usually, behavior-descriptive sentences) and both of which employed the same task ( impression formation) . The only difference between conditions was in the target of the impression to

1 52

DAVID L. HAMILTON, STEVEN J. SHERMAN AND LUIGI CASTELLI

be formed. In one case, participants were told that the same person had performed all behaviors and their task was to form an impression of that person. In the other case, participants were told that a different person per­ formed each behavior, but that all of these persons belonged to the same group, and the task was to form an impression of the group. A number of studies have been conducted using this basic paradigm, and these studies have assessed a variety of dependent measures tapping different aspects of infor­ mation processing (Hilton & Von Hippel, 1990; McConnell, Sherman, & Hamilton, 1 994, 1997; Stem, Marrs, Millar, & Cole, 1984; Susskind et al., 199; Welboume, 1 999; Wyer, Bodenhausen, & Srull, 1 984). For example, Susskind et al. (1999) used this paradigm to investigate the inference processes engaged during the impression-formation process. They reasoned that the expectancy of unity and coherence in a social target would induce more on-line inference processes during information processing and, based on the assumptions outlined above, such integrative processing should be more evident for an individual target than for a group target. Participants were presented a series of 16 sentences describing behaviors performed: (a) by a single individual; (b) by individuals who all belong to the same group; or (c) by an aggregate of individual persons-and were told to form an impres­ sion of the target. After reading the stimulus sentences, they completed a number of dependent measures. The results, shown in Table 5.2, provided consistent support for the hypotheses. Participants rated the individual as more organized and coherent, and judged the stimulus behaviors to comprise a more coherent pattern, in the individual target condition than in the other conditions. Moreover, they made more extreme trait ratings of the individual target, made those judgments more quickly, and were more confident in those judgments for the individual target than for the group target. Table 5.2

Mean ratings on dependent measures Target type

Measure Individual

Group

Aggregate

Organization of target's behaviors

5.66

4.43

4.10

Perceived unity of the target

6.00

4.06

4.03

Trait ratings

5.98

5.29

4.92

Response latencies to trait ratings (in seconds)

4.74

6.16

6.16

Confidence ratings

5.54

5.23

4.21

Recall

5.72

5.09

4.26

From Susskind et al., 1 99, Experiment 1, with permission.

154

DAVID L. HAMILTON, STEVEN J. SHERMAN AND LUIGI CASTELLI

Results reported by McConnell et a/. (1997) substantiate these findings. Like Susskind et a/. (1999), these researchers compared impression processes for individual and group targets, but in addition orthogonally manipulated perceiver expectations. Specifically, they induced expectations of high or low entitativity by presenting brief paragraphs at the outset that informed the participant that the person or group was either highly consistent in behaviors and personality, or that dissimilarity and inconsistency characterized the per­ son or group. A third condition was given no prior expectancy. Participants then read 24 behavior-descriptive sentences (16 desirable, 8 undesirable be­ haviors), and subsequently were asked to recall freely as many of those be­ haviors as they could and to rate the target on a 1 0-point liking scale. Three measures were assessed as indicators of integrative processing: amount of free recall, primacy in recall (recalling more early than late stimulus behaviors), and the correlation between free recall and evaluative judgments. Thus, the experiment had a 2 (individual or group target) x 3 (high, low, or no expec­ tancy induction) design. The theoretical prediction was that three of the conditions-high-entitativity individual, high-entitativity group, and no­ expectancy individual-would evidence greater integrative processing than the other three conditions-low-entitativity individual, low-entitativity group, and no-expectancy group. As can be seen in Table 5.4, the results conformed to predictions for each of these measures. The middle column of Table 5.4 compares the individual versus group comparison made in several previous studies, as we have just discussed. Again, the results suggest greater integrative processing for the individual than for the group target (better recall of stimulus information, and that recall appears not to be memory-based, as reflected in the near-zero memory-judgment correlation). The unique contribution of McConnell et al. 's design was that the instructional manipulation of expected entitativity significantly altered this standard difference. Specifically, expectations of high Table 5.4 Effects of target type and entitativity manipulation on recall and memory­ judgment correlations

Expected entitativity

Target

Free recall Individual Group Memory-judgment correlation Individual Group Adapted from McConnell

et a/.

High

No information

Low

12.78 10.72

12.94 8.94

10.28 8.33

0. 1 1 0.33

-0.13 0.39

0.62 0.45

(1997), Experiment 1, Tables 2 and 3.

ROLE OF ENTITA TIVITY IN GROUP PERCEPTION

155

target (individual or group) entitativity produced results comparable to the "typical" individual target condition, whereas induction of low entitativity expectations (for either individual or group) led to results resembling those for the "typical" group target condition. These findings are useful in docu­ menting the importance of entitativity expectations in driving these differing outcomes for individual and group targets. Welboume (1999), using a different paradigm, has also found differences in processing information about individual and group targets, and, like McCon­ nell et al. (1997), also manipulated expectations of entitativity as well. Her results were quite consistent with those summarized above, and also provided evidence suggesting that the effects of expected entitativity reflect assump­ tions about the coherence and organization of the target, rather than expecta­ tions of consistency (Hamilton et al., 1998). The results of several studies, then, tell a consistent story. Despite the fact that participants in the different conditions were given the same task and identical information, differences between conditions have consistently been obtained, and across studies, on a number of dependent measures. Compared to people forming impressions of a group target, people forming impressions of an individual make more extreme trait ratings; they make those judgments more quickly; they are more confident in those judgments; they recall more of the behavioral information, especially behaviors that are incongruent with the emerging impression; those incongruencies trigger causal attributional think­ ing to explain those inconsistencies; and memory-judgment correlations are typically low (for reviews of these studies, see Hamilton & Sherman, 1 996; Hamilton, Sherman, & Maddox, 1999). Hamilton et al. (1999) have referred to this syndrome of processing characteristics as integrative processing. For all of these measures, differences have been obtained in comparisons of individual and group impression formation. Several of these studies have also compared conditions in which groups are described as being high or low in entitativity­ related qualities, and parallel differences have been obtained. All of these findings suggest that perceivers engage in more extensive and more integrative processing of information when the target is: (a) an individ­ ual person, rather than a group; and (b) a high-entitativity, rather than a low­ entitativity, group. Higher levels of perceived target unity and coherence, or entitativity, induce more integrative processing of information. The per­ ceiver's assumption is that high-entitativity targets have unified, organized "personalities" and hence use the available information to uncover the im­ portant themes and characteristics of that target. Judgment Polarization

If expectations or perceptions of entitativity increase integrative processing as information about a group is acquired, then one would expect differences in

156

DAVID L. HAMILTON, STEVEN J. SHERMAN AND LUIGI CASTELLI

resulting judgments made of high- and low-entitativity targets. In some of the studies summarized earlier (e.g., Susskind et al., 1999; Welbourne, 199) we have already seen evidence of such an effect. However, the specific nature of those effects of entitativity on judgments can take several different forms, and these possibilities have been explored in recent research. Dasgupta, Banaji, and Abelson (1999; see also Abelson, Dasgupta, Park, & Banaji, 1998) have argued that group entitativity increases the perceived threat value of a stimulus group. Therefore, they proposed that greater en­ titativity of groups is associated with more negative, hostile perceptions of the group. For example, in one of their studies (Dasgupta et a/., 1999, Experiment 2) participants viewed computer-generated groups of five human-like crea­ tures, presumably created as a species for a new science fiction story. The five figures shown in the display were either highly similar (all of the same color) or dissimilar (of differing colors). The participants' task was to rate their impressions of the creatures on a variety of attributes and behaviors. Highly similar (entitative) groups of creatures were rated more negatively than were the less entitative groups. Thus, Dasgupta et al. concluded that perceived entitativity leads to more negative impressions of groups. The results reported by Susskind et al. (1999) seem to reflect a different pattern. These authors compared impressions of individual and group targets. As described earlier, participants read 16 behavior-descriptive sentences, de­ scribing either a person or members of a group, and were asked to form an impression of the target. All behaviors were desirable, and Susskind et a/. found significantly more favorable ratings of the individual than of the group target. Presumably, individuals are perceived as possessing more coherence and unity than groups (Hamilton & Sherman, 1996), suggesting that highly entitative targets induce stronger trait ratings, in this case more positive judgments. These findings are only seemingly incompatible, however. As noted, Sus­ skind et al. presented only desirable behavior descriptions. In contrast, in the Dasgupta et al. research the human-like creatures were described as members of some alien species and participants were explicitly told that the science fiction story would include not only this species but another as well, clearly suggesting an intergroup context that might imply hostile relations. Thus, in both cases participants made more extreme ratings of the high-entitativity target, but in opposite directions. Taken together, these findings suggest that entitativity may produce a polarization of judgments, the direction depending on the informational and situational context in which the groups are judged. Thakkar (2001 ) recently tested this hypothesis. Participants learned about a group and were asked to form an impression of it. Thakkar (2001 ) manip­ ulated expectations of group entitativity through instructions given at the outset. For example, one group was told that the group pursues a common set of goals, they depend on each other a lot, meet frequently to update each

ROLE OF ENTITATIVITY IN GROUP PERCEPTION

157

other on their progress toward completing assignments, etc., and ended with the statement that it is a "tightly-structured" group. In contrast, in another condition the instructions described the group as pursuing a variety of goals relevant to the group's objectives, that group members focused on completing their individual task assignments and hence did not depend on each other, that they met only once in a while, and the group was described as a "loosely­ structured" group. Following these instructions, participants read a series of 24 behavior-descriptive sentences, each of which described a different mem­ ber of the group, and the participants' task was to form an impression of the group. Eighteen of the behaviors were selected to reflect three trait domains (adventurous-unadventurous, friendly-unfriendly, intelligent-unintelligent; six behaviors representing each dimension), with the remaining six behaviors being neutral on all three of these dimensions. For half of the participants, the trait-relevant behaviors were all positively-valenced behaviors; for the other half, they were negatively-valenced behaviors. After reading these materials, participants then rated their impression of the group on a series of attribute scales, including scales representing the three traits incorporated in the be­ havior descriptions. Thakkar found that participants made more extreme ratings of the groups in the high-entitativity than in the low-entitativity condition. That is, the positively-described group was rated more favorably, and the negatively­ described group was rated less favorably, when it was perceived as being high, rather than low, in entitativity. Moreover, the evidence indicated that percep­ tions of group entitativity (as assessed on scales measuring several entitativity-related properties) mediated the effect of the entitativity manip­ ulation on the extremity of judgments. Thus, the perceived entitativity of a group leads to more polarized judgments, not simply to more negative views of the group. Results of other studies (Welboume, 1999; Yzerbyt et al., 1998), although not designed to test a polarization hypothesis, obtained results that can be interpreted as supporting that hypothesis. Perceived Entitativity and the Perception of Group Members

Recently, we have begun to investigate another way in which processing dif­ ferences about groups and their members are affected by perceived entitativity (Crawford, Sherman, & Hamilton, 2001 ). This work is concerned not with the impact of perceived entitativity on the perceptions of groups, but rather on the perceptions of individual group members. As we have shown previously (Mc­ Connell et al., 1997), in the case of highly entitative groups, there is an on-line abstraction of stereotypes and a search for the ''essence'' of such groups. Conse­ quently, inferences about individual group members should be guided by this abstract essence, and group level characteristics should be applied rather indis­ criminately across all members of highly entitative groups.

158

DAVID L. HAMILTON, STEVEN J. SHERMAN AND LUIGI CASTELLI

Based on this reasoning, we hypothesized that, for highly entitative groups, abstract group-level information will easily transfer from one group member to another. If so, then unique behavioral information about individual group members might well be lost in this case. For low-entitativity groups, on the other hand, individual members should be associated with their own unique behaviors, and the transference of attributes across different group members will be less likely. In other words, members of high-entitativity groups should be treated categorically, whereas members of low-entitativity groups should be treated individualistically. This prediction is compatible with Brewer and Harasty's (1996) proposal that perceivers have prototypic representations of high-entitativity groups but are more likely to have exemplar representations of low-entitativity groups. In order to examine this issue, Crawford et al. (2001 ) employed the savings­ in-relearning paradigm (Carlston & Skowronski, 1 994). In the first phase of the experiment, behavioral information was presented about individual mem­ bers of two different groups, Group A and Group B. Each group contained 16 members. On each trial, a photograph was presented along with a first-person description of a behavior (e.g., "Where I work, I share a desk with another person who puts his stuff all over the place. I finally got fed up with it and ran all of his papers through the paper shredder. That should teach him"). For Group A, eight members performed a behavior that was aggressive in nature (as in the example), and eight performed intelligent behaviors. For Group B, eight members performed a behavior that was honest, and eight performed lazy behaviors. Prior to presentation of the stimulus information, participants read, in the instructions, descriptions creating expectations of either high or low entitativity for the two groups, similar to the strategy used in previous studies (McConnell et al., 1997; Thakkar, 2001 ). The next phase involved a learning task. Each of the previously seen 32 photographs was paired with a single trait term. Half of the time, a photo­ graph was paired with the trait term implied by the behavior that was pre­ viously associated with the person in the photograph. Thus, an individual who had performed an aggressive behavior was paired with the trait "ag­ gressive". These are called inference trials. In the other half of the cases, the photograph was paired with the trait term implied by the behavior of other members of the group, but not by the behavior of the specific person in the photograph. For example, a person who performed an aggressive behavior earlier was now paired with the trait "intelligent", a trait implied by the behaviors of other members of his group (Group A). These are called trans­ ference trials. Inference trials should be easier in the case of low entitativity groups, where individuals are perceived in terms of their unique behaviors and characteris­ tics. Thus, an individual from a low-entitativity group who has engaged in a behavior indicative of a trait should quickly and easily be paired with that

160

DAVID L . HAMILTON, STEVEN J . SHERMAN AND LUIGI CASTELLI

Collective responsibility is hypothesized to be more likely when the group of which the wrongdoer is a member is a highly interdependent group. Inter­ dependence reflects the extent to which the group has high interaction among its members, is important to its members, and is an organized group. These variables are among the most important predictors of group entitativity (Lic­ kel et a/., 200). Therefore, judgments of collective responsibility are most likely to occur under conditions of high perceived group entitativity. Why would other group members be held accountable for the actions of another group member, when they in fact had no direct involvement in the wrongdoing? Lickel (200) has differentiated two causal explanations upon which perceivers' judgments might be based. One possibility is that perceivers may believe that the other group members may benefit from the wrongdoer's action, that they may ultimately experience some indirect benefit, or even that they simply were sympathetic to or supportive of the wrongdoer's actions. For example, consider a case in which a member of a protest group or a terrorist group is arrested for some illegal activity, performed individually. Even though they may have had no direct role in the wrongdoer's illegal action, other group members may be judged critically, simply because they are viewed as being sympathetic to those actions, as sharing the actor's views or motives, and as viewing the action as useful in advancing their overall cause. This is referred to as responsibility by commission. On the other hand, percep­ tions of collective responsibility may be based on perceptions of responsibility by omission, in which group members are judged accountable because it is believed that they should have been able to prevent the wrongdoer from performing this act. This is referred to as responsibility by omission. In this case, responsibility does not reflect perceptions of possible support or benefit; in fact, the other group members may not have been at all sympathetic to the wrongdoer's behavior. However, it is perceived that they had the ability to prevent the action, and should have done so, and therefore they share in the responsibility for the action. The earlier example of the fraternity brother who punched a customer fits this model; the other brothers should have recognized that their friend was getting out of control and should have prevented the incident from escalating. Research conducted by Lickel (Lickel, 200; Lickel, Schmader, & Hamilton, 2001) provides support for this model. In several studies, measures of perceived interdependence among group members have been significantly predictive of judgments of collective responsibility. Furthermore, mediational analyses have shown that both responsibility by commission and by omission are significant mediators of this relationship. Moreover, Lickel, Schmader, and Hamilton (2001 ) tested the role of collec­ tive responsibility in people's perceptions of responsibility for the 199 shoot­ ings at Columbine High School, in which two teenaged boys shot and killed 12 schoolmates and one teacher before committing suicide. In the aftermath of

ROLE OF ENTITA TIVITY IN GROUP PERCEPTION

161

this tragedy, there was considerable discussion throughout the country, as well as in the media, about the role of two important groups in these boys' lives, namely, their parents and a social group to which they belonged, called the Trenchcoat Mafia. In two questionnaire studies, one conducted 9 days after the incident and the second conducted a month later, participants (UCSB students) rated a number of groups with whom the boys had contact, including both the parents and the Trenchcoat Mafia, on the extent to which they viewed these groups as responsible and accountable for the boys' actions. They also completed several scales assessing the perceived interdependence of the boys with the members of these groups, as well as their perceptions of responsibility by commission and by omission. In general, perceptions of interdependence of a group's members with the shooters significantly pre­ dicted judgments of collective responsibility. Moreover, this relationship was mediated by the two causal explanations. Of particular interest was the fact that responsibility by commission was the more important mediator for the Trenchcoat Mafia. Also, the parents received high ratings of responsibility by both commission and omission. Thus, entitativity-related properties of groups can influence the extent to which group members are viewed as responsible and accountable for the actions of another group member, even when those other members are not directly involved in the action in question. SUMMARY AND CONCLUSIONS

We began this chapter with quotations reflecting thoughts and perspectives on three different groups-a committee, a country, and a family. There is no question that each of these three entities constitutes, in the minds of most observers, groups with which we are quite familiar. Yet the very nature of these entities, as well as our perceptions of them, differ in many important respects. Our goal in this chapter has been to consider several questions concerning the perception of groups as entities. These questions have be­ come, in recent years, the focus of a considerable amount of research, our own as well as that of others. The accumulated findings document the importance of one particular factor, perceived group entitativity, in social perception. Our social world is rich in the variety of groups with which we are engaged, both as members of groups in which we participate and with which we iden­ tify, and as observers of and interactants with groups that we encounter in our everyday lives. Given that we routinely deal with a vast variety of aggregates of people, why is it that we perceive some of those aggregates to be groups and others not? Why, how, and on what bases do we endow some of those aggregates with the quality of groupness to a considerable degree, while oth­ ers are perceived as possessing that quality to a much lesser extent? Given the diversity of the social groups that we encounter, what differences are there in

1 62

DAVID L. HAMILTON, STEVEN J. SHERMAN AND LUIGI CASTELLI

our perceptions of the nature of their groupness, and in the bases on which we endow that property on them? And given that we perceive varying degrees of entitativity in the groups we encounter, what differences follow as con­ sequences of perceiving that groupness? Our research program has provided at least partial answers to these ques­ tions. We have documented that people do recognize considerable variety among groups in the extent to which they possess the quality of groupness. We have discovered several variables that are meaningfully correlated with the perception of entitativity in a broad array of groups, properties such as interaction among group members, importance of membership in the group to its members, the sharing of common goals and common fate, and similarity among group members. These group characteristics provide information about the group that can serve as functional cues, which the perceiver can use to determine the relative entitativity of groups (Lickel et al., 200). Our re­ search has also established an empirically-based typology of groups, as per­ ceived by our research participants. These group types-intimacy groups, task groups, social categories, loose associations-were derived from data pro­ vided by participants in two countries (USA and Poland) and on two different tasks (ratings of groups on entitativity-related group properties, free sortings of groups), with great consistency in the results obtained across these tasks in the data obtained in both countries (see Lickel et a/., 200, for details of these comparisons). Moreover, our research has also employed a categorization and recognition task to verify that our participants spontaneously employ these group-type distinctions in their processing of information about individual members of such groups (Sherman et a/., in press). Our research (along with that of others) has also shown that, once a group is perceived as possessing entitativity to a certain degree, that perception can itself influence subsequent perceptions of and processing of information about that group and its members. Variations in degree of perceived target entitativity influence the extent to which perceivers engage in integrative processing of information about group members (Hamilton et al., 1999), with consequent effects on the resulting impressions and perceptions of the group. These effects have been documented in more extreme (polarized) judgments of the target person or group (Susskind et a/., 1 999; Thakkar, 2001 ); faster judgments about the group, based on inferences formed on-line (McConnell et al., 1 994, 1 997; Susskind et a/., 1 999); greater confidence in one's impres­ sions of the group (Susskind et al., 1 999); greater spontaneous transfer of attributes from one group member to another (Crawford et al., 2001 ); better recall of information acquired about the group and its members, especially information that contradicts an existing impression (McConnell et a/., 1 997; Susskind et a/., 1 999); and greater use of attributional thinking to understand such expectancy-violating information (Susskind et a/., 1999). In addition, the perception of group entitativity can influence the perception of moral

164

DAVID L . HAMILTON, STEVEN J . SHERMAN AND LUIGI CASTELLI

Asch, S. E. (1946). Forming impressions of personality. Journal of Abnormal and Social Psychology, 41, 258-90. Atkinson, J. W., & Feather, N. (196). A Theory of Achievement Motivation. New York: Wiley. Bandura, A. (1982). Self-efficacy mechanism in human agency. American Psychologist, 37, 122-47. Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497529. Bowlby, J. (1969). Attachment and Loss. Volume 1: Attachment. New York: Basic Books. Brewer, M. B., & Brown, R. J. (1998). Intergroup relations. In D. Gilbert, S. T. Fiske & G. Lindzey (Eds), Handbook of Social Psychology (4th Edn, Vol. 2, pp. 554-94). Boston: McGraw-Hill. Brewer, M. B., & Harasty, A. S. ( 1996). Seeing groups as entities: The role of perceiver motivation. In R. Sorrentino & E. T. Higgins (Eds), Handbook of Motivation and Cognition (Vol. 3, pp. 347-70). New York: Guilford. Brewer, M. B., Weber, J. G., & Carini, B. (1995). Person memory in intergroup contexts: Categorization versus individuation. Journal of Personality and Social Psy­ chology, 69, 29-40. Campbell, D. T. (1958). Common fate, similarity, and other indices of status of aggre­ gates of persons as social entities. Behavioral Science, 3, 14-25. Carlston, D. E., & Skowronski, J. J. (1994). Savings in the relearning of trait informa­ tion as evidence for spontaneous inference generation. Journal of Personality and Social Psychology, 66, 840-56. Crawford, M. T., Sherman, S. J. & Hamilton, D. L. (2001 ). Perceived entitativity, stereotype formation, and the interchangeability of group members (unpublished manuscript). Indiana University, Bloomington, IN. Crocker, J., Major, B., & Steele, C. (1998). Social stigma. In D. Gilbert, S. T. Fiske, & G. Lindzey (Eds), Handbook of Social Psychology (4th Edn, Vol. 2, pp. 504-53). Boston: McGraw-Hill. Dasgupta, N., Banaji, M. R., & Abelson, R. P. (199). Group entitativity and group perception: Associations between physical features and psychological judgment. Journal of Personality and Social Psychology, 77, 991-1003. Deaux, K., Reid, A., Mizrahi, K., & Ethier, K. A. (1995). Parameters of social identity. Journal of Personality and Social Psychology, 68, 280-91. Fiske, S. T. (1998). Stereotyping, prejudice, and discrimination. In D. Gilbert, S. T. Fiske & G. Lindzey (Eds), Handbook of Social Psychology (4th Edn, Vol. 2, pp. 357-411). Boston: McGraw-Hill. Gaertner, L., & Schopler, J. (1998). Perceived ingroup entitativity and intergroup bias: An interconnection of self and others. European Journal of Social Psychology, 28, 963-0. Hamilton, D. L., & Sherman, S. J. (1996). Perceiving persons and groups. Psychological Review, 103, 336-55. Hamilton, D. L., Sherman, S. J., & Lickel, B. (1998). Perceiving social groups: The importance of the Entitativity Continuum. In C. Sedikides, J. Schopler, & C. A. Insko (Eds), Intergroup Cognition and Intergroup Behavior (pp. 47 74). Mahwah, NJ: Erlbaum. Hamilton, D. L., Sherman, S. J., & Maddox, K. B.. (199). Dualities and continua: Implications for understanding perceptions of persons and groups. In S. Chaiken & Y. Trope (Eds), Dual-process Theories in Social Psychology (pp. 606-26). New York: Guilford.

166

DAVID L . HAMILTON, STEVEN J . SHERMAN AND LUIGI CASTELLI

Susskind, J., Maurer, K. L., Thakkar, V., Hamilton, D. L., & Sherman, J. ( 199). Perceiving individuals and groups: Expectancies, inferences, and causal attributions. Journal of Personality and Social Psychology, 76, 181-91. Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In W. G. Austin & S. Worchel (Eds), The Social Psychology of Intergroup Relations (pp. 3347). Monterey, CA: Brooks-Cole. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S. Worchel & W. G. Austin (Eds), Psychology ofIntergroup Relations (2nd Edn, pp. 7-24). Chicago: Nelson-Hall. Taylor, S. E., Fiske, S. T., Etcoff, N. L., & Ruderman, A. J. (1978). Categorical and contextual bases of person memory and stereotyping. Journal of Personality and Social Psychology, 36, 778-93. Thakkar, V. (200). The role of entitativity in judgments about groups (unpublished doctoral dissertation). University of California, Santa Barbara, CA. Turner, J. C. (1987). Rediscovering the Social Group: Self-Categorization Theory. Ox­ ford: Blackwell. Welbourne, J. L. ( 199). The impact of perceived entitativity on inconsistency resolu­ tion for groups and individuals. Journal of Experimental Social Psychology, 35. 481508. White, R. W. (1959). Motivation reconsidered: The concept of competence. Psycholog­ ical Review, 66, 297-334. Wilder, D., & Simon, A. F. (1998). Categorical and dynamic groups: Implications for social perception and intergroup behavior. In C. Sedikides, J. Schopler, & C. A. Insko (Eds), Intergroup Cognition and Intergroup Behavior (pp. 27-44). Mahwah, NJ: Erlbaum. Wyer, R. S. Jr., Bodenhausen, G. V., & Srull, T. K. (1984). The cognitive represen­ tation of persons and groups and its effect on recall and recognition memory. Jour­ nal of Experimental Social Psychology, 20, 445-69. Yzerbyt, V. Y., Rocher, S., & Schadron, G. (1997). Stereotypes as explanations: A subjective essentialistic view of group perception. In R. Spears, P. J. Oakes, N. Ellemers & S. A. Haslam (Eds), The Social Psychology of Stereotyping and Group Life (pp. 20-50). Oxford: Blackwell. Yzerbyt, V. Y, Rogier, A, & Fiske, S. T. (1998). Group entitativity and social attribu­ tion: On translating situational constraints into stereotypes. Personality and Social Psychology Bulletin, 24, 1089-103.

Chapter 6

The Meaning and Consequences of Perceived Discrimination in Disadvantaged and Privileged Social Groups Michael T. Schmitt Nyla R. Branscombe University of Kansas, USA

ABSTRACT The subjective meaning and consequences of perceived discrimination depend on the position of one's group in the social structure. For members of disadvantaged groups, attributions to prejudice are likely to be internal, stable, uncontrollable, and convey widespread exclusion and devaluation of one's group. For members of privileged groups, the meaning of attributions to prejudice is more localized. Be­ cause of such meaning differences, attributions to prejudice are considerably more harmful for the psychological well-being of members of disadvantaged groups than they are for members of privileged groups. According to the Rejection­ Identification Model (Branscombe, Schmitt, & Harvey, 1999), members of disad­ vantaged groups cope with the pain of attributions to prejudice by increasing identification with their disadvantaged group. We conclude with an exploration of the social contextual factors that can affect how the disadvantaged cope, and a discussion of the challenges facing future research on attributions to prejudice.

Correspondence regarding this chapter should be addressed to Nyla R. Branscombe, Department of Psychology, 1415 Jayhawk Blvd, University of Kansas, Lawrence, KS 66045, USA. E mail: nyla@ukans. edu European Review of Social Psychology, Volume

© 2002 John

Wiley

&

Sons Ltd.

12.

Edited by Wolfgang Stroebe and Miles Hewstone.

168

MICHAEL T. SCHMIT AND NYLA R. BRANSCOMBE

One of the most enduring ideas in social psychology is that the subjective experience of social life is moderated by attributions (Weiner, 1985). We con­ sider the psychological consequences of attributing one's negative outcomes to prejudice against one's group membership. We review and integrate previous theoretical and empirical work on such attributions to prejudice, and consider how attributing negative events to prejudice influences psychological well­ being. We first outline our theoretical reasoning for how and why attributions to prejudice have very different meanings for disadvantaged and privileged groups. We conclude that these attributions are more psychologically harmful for members of disadvantaged groups than for members of privileged groups and review evidence supporting this conclusion. We then discuss how the psy­ chological benefits of identification with one's disadvantaged group can counter some of the painful consequences of recognizing prejudice. Finally, we consider factors that limit disadvantaged group members' ability to cope via group iden­ tification, and make suggestions for future research on attributions to prejudice.

THE TOTAL RELEVANCE OF ATTRIBUTIONS TO PREJUDICE What an attribution to prejudice means to the person making it is central to understanding its emotional consequences. In its most basic sense, an attribu­ tion to prejudice explains a negative event as the result of someone 's bias against one's category membership. While this description of an attribution to prejudice describes what the attribution implies for the immediate situation­ its "local relevance"-it does not necessarily capture all of its larger potential meanings and implications-its "total relevance". Heider ( 1 958, p. 253) sug­ gested that it is the total relevance of an attribution, not its local relevance, that guides our interpretation of and reaction to events. Thus, understanding the meaning and consequences of an attribution to prejudice requires a con­ sideration of its total relevance for the person making it. Because prejudice and discrimination are intergroup phenomena, an anal­ ysis of attributions to prejudice calls for an examination of how the state of intergroup relations determines the meaning of those attributions. Our cen­ tral argument is that the total relevance of an attribution to prejudice is a function of the ingroup's and the outgroup's

the social structure.

relative positions ofpower within

The meaning and consequences of an attribution to preju­

dice will depend on the nature of the existing intergroup relations at a particu­ lar historical juncture and, more importantly, the position of one's own group within the social structure. The psychological experience of perceiving preju­ dice will be very different depending on whether one belongs to a group that is disadvantaged relative to other groups, or to a group that is relatively privileged. For the reasons we outline below, we argue that because of their

MEANING AND CONSEQUENCES OF PERCEIVED DISCRIMINATION

169

differing social realities, members of disadvantaged groups experience more harmful psychological consequences when making attributions to prejudice than do members of privileged groups. Severity of Discrimination Experiences

Clearly, disadvantaged groups suffer because the outcomes they receive are poorer relative to higher-status groups. We consider the psychological con­ sequences of how those negative events are interpreted, rather than the con­ sequences of the events themselves. However, the differential severity of events potentially labeled as "discrimination" is also likely to moderate the subjective experience of interpreting those events as due to prejudice. Be­ cause the magnitude of an event is likely to influence the magnitude of its effects, minor instances of discrimination will have minimal implications for well-being, while more severe discriminatory experiences will have a greater impact. Thus, one of the most important ways in which attributions to preju­ dice differ for disadvantaged and privileged groups is in the severity of the type of the events typically attributed to prejudice. For instance, Branscombe (1998) found that when men and women were asked to describe the disadvan­ tages that they had experienced based upon their gender group membership, women described relatively severe events, such as a lack of freedom, fear of sexual assault, and job discrimination. In contrast, men described less severe and more circumscribed events, such as having to pay for dates, or being more likely to get a speeding ticket. In a related study (Kappen, Branscombe, Kobrynowicz, & Schmitt, 2001 ) women and men listed discriminatory events they had personally experienced, and coders who were blind to the gender of the participants rated the discriminatory events for their potential life con­ sequences. The events generated by women were coded as having a substan­ tially greater impact on the recipient's life, compared to the events generated by men. Likewise, when White Americans were asked about their experiences with racial discrimination, they reported relatively minor types of discrimina­ tion, such as not having exclusively White organizations and being seen as racially biased (Branscombe, Schiffbauer, Schmitt, & Valencia, 2001), while Black Americans reported discrimination in a wide range of important life domains, including housing, education, and employment (Sigelman & Welch, 1 991 ) Thus, because of real differences in the severity of the discriminatory events experienced, attributions to prejudice are likely to have greater emo­ tional consequences for disadvantaged than for privileged groups. ,

.

Attributional Dimensions

An analysis of attributions to prejudice requires consideration of where these attributions fall on the three main dimensions of attribution: locus of the cause

170

MICHAEL T. SCHMIT

AND

NYLA R. BRANSCOMBE

(Heider, 1958), stability (Weiner et al., 1971), and controllability (Weiner, 1979). 1ndeed, "knowing the dimensional locations of a given attribution allows accurate prediction of the consequences of that attribution" (Anderson, Krul, & Weiner, 1996, p. 280). In this section, we explore the total relevance of an attribution to prejudice by considering its locus (the extent to which it reflects something about the self), stability (the implications the attribution has for future situations), and controllability (the extent to which the outcome is per­ ceived as controllable). We limit our analysis to these three aspects of attribu­ tion because they alone can account for a wide variety of effects, and the evidence for the existence or importance of other attributional dimensions (e.g., intentionality, globality) is inconclusive (Anderson et al., 1996). Attributions to Prejudice as both Internal and External Social psychologists often consider the causal locus of attributions as reflect­ ing a single dimension, ranging from internal to external, or even as a categor­ ical distinction where attributions are either internal or external. However, assuming that external and internal attributions are inversely related may be an oversimplification (Kelley, 1983), and numerous empirical studies have found that, in fact, internal and external attributional ratings are often inde­ pendent (McClure, 1998). Thus, the internal and external dimensions of at­ tribution are best conceptualized and measured separately. Most of the existing theoretical work on attributions to prejudice has as­ sumed that the locus of such attributions is purely external (e.g., Crocker & Major, 1989; Crocker, Major, & Steele, 1998). This assumption led to the "discounting hypothesis", the idea that attributions to prejudice discount the self as a cause of negative events (Crocker & Major, 1989; Crocker et al., 1998; Crocker & Quinn, 1 998). Accordingly, it is argued that "because prejudice against one's group is an external attribution, making this attribution should protect the self-esteem of stigmatized people" (Crocker, Voelkl, Testa, & Major, 1991, p. 219). Attributions to prejudice certainly do have an external component, with the prejudice of the source being outside of the self. On the other hand, we argue that they also have a substantial internal component-the target's own group membership. Although attributions to prejudice are often described as at­ tributions to "group membership" (e.g. Crocker & Major, 1989, pp. 613-14), the fact that group membership is an internal aspect of the self has been largely ignored. Because the prejudice of the source (external) and the tar­ get's group membership (internal) are both necessary but not sufficient causes of prejudicial treatment, neither can be discounted when making an attribu­ tion to prejudice (McClure, 1998). The characterization of attributions to prejudice as partially internal is important, because it leads to very different predictions about the

172

MICHAEL T. SCHMITT AND NYLA R. BRANSCOMBE

qualities implausible. In the "Others accepted" condition, the professor granted the requests of about 10 other students, making an attribution to the self more plausible. In the "Other gender accepted" condition, the professor only let in students who were not of the participant's gender, making an attribution to gender prejudice highly plausible. Indeed, as revealed by a manipulation check, attributions to prejudice were much higher in this condi­ tion than in the other two. As predicted, the locus of the attributions made differed by condition. Attributions were significantly more internal in the Other gender accepted condition than in either of the other conditions, while externality of attributions did not differ by condition. Taken together, these two studies suggest that, like attributions to other aspects of the self, attribu­ tions to prejudice against one's group membership do have a substantial internal component. Our findings are inconsistent with the fundamental assumption of the dis­ counting hypothesis-that attributions to prejudice are exclusively external. Because these data illustrate that attributions to prejudice are in part internal, they undermine the theoretical basis for predicting that attributions to preju­ dice are self-protective in disadvantaged groups. By treating attributions to prejudice as purely external, the discounting hypothesis may greatly overesti­ mate their self-protective properties. Consistent with this argument, in the two studies just described, women reported less positive affect when rejection was attributable to sexism compared to when the rejection was attributable to an exclusively external cause, the professor's disposition. This suggests that, for members of disadvantaged groups, attributions to prejudice do not offer the level of self-protection provided by purely external attributions. Stability Although the internal aspects of attributions to prejudice call into question their potential for protecting the self, it is important to note that not all internal attributions for negative events are equally harmful (Rhodewalt, Morf, Hazlett, & Fairfield, 1 991; Snyder & Higgins, 1985). Attributional sta­ bility is likely to moderate their effects. Attributions to prejudice are likely to be more stable for members of disadvantaged groups than for members of privileged groups. Disadvantaged groups, by definition, experience discrimi­ nation more frequently and across a wider variety of contexts than do privil­ eged groups (Sigelman & Welch, 1 991; Swim, Cohen, & Hyers, 1998). Women report experiencing more prejudice against their group and perceive preju­ dice against them as more pervasive across a wider variety of social contexts than do men (Kobrynowicz & Branscombe, 1997; Schmitt, Branscombe, Kobrynowicz, & Owen, in press). Indeed, in the USA there is considerable consensus among men and women and among different ethnic groups regard­ ing the relative status differences that exist between them (Sidanius & Pratto.

MEANING AND CONSEQUENCES OF PERCEIVED DISCRIMINATION

173

1999; Stewart, Vassar, Sanchez, & David, 200). Precisely because members of disadvantaged groups are aware that they may face negative treatment on the basis of their stigma in many situations, attributions for negative events that implicate their group membership are likely to be stable and painful (Branscombe, Schmitt, & Harvey, 1 999; Galin, Sweeney, & Shaeffer, 1 98 1 ; Robins, 1 988; Weiner, 1 985). I n contrast, because privileged groups d o not see prejudice against them as pervasive, they will see negative treatment based on their group membership as anomalies or localized occurrences. Thus, for privileged groups, attributions to prejudice are more likely to be unstable and lack meaning beyond the immediate situation. In summary, the differential pervasiveness of discrimination faced by privileged and disadvantaged groups makes attributions to prejudice much more stable, and as a result, more psychologically costly for the disadvantaged. Because the level of stability of attributions to prejudice reflects the degree to which prejudice will affect one's outcomes in the future, the stability di­ mension is of critical importance for understanding the total relevance of an attribution to prejudice. Because an attribution to prejudice among disadvan­ taged groups reflects more stable treatment, the total relevance of the attribu­ tion will be much greater that its local relevance. However, among privileged groups, for whom an attribution to prejudice is more unstable, the total relev­ ance of an attribution to prejudice is likely to be equivalent to its local relevance. Controllability The controllability dimension of attribution refers to the extent to which the cause of an outcome is under the volitional control of the self or others (Anderson, et al., 1 996). The subjective consequences of the perceived con­ trollability of attributions to prejudice can best be captured by considering who, in the context of prejudicial treatment, has the power to inflict their prejudices on others. Because disadvantaged groups are afforded less power within the social structure than privileged groups, encounters with prejudicial treatment are likely to be seen by disadvantaged groups as relatively uncon­ trollable. Conversely, the outcomes of privileged group members are unlikely to be seen as generally under the control of members of disadvantaged groups, because their privileged status affords them a position of greater power. Consequently, attributions to prejudice made by privileged groups are less likely to harm feelings of control. Empirical research supports the idea that attributions to prejudice have negative consequences for perceived control among disadvantaged groups. Ruggiero and Taylor ( 1 997} found that disadvantaged groups who at­ tributed negative feedback to discrimination suffered a reduction in per­ ceived control. Efficacy-based approaches to self-esteem (Bandura, 1 997)

1 74

MICHAEL T. SCHMITT AND NYLA R. BRANSCOMBE

suggest that the perception of control over one's outcomes is an important component of well-being. According to Weiner (1985), negative, internal, uncontrollable events (as attributions to prejudice among disadvantaged groups are likely to be) result in depression (Brown & Siegel, 1 988), and can even undermine cognitive ability (von Hecker & Sedek, 1999). Further­ more, when prejudice is recognized by the disadvantaged, making discrimi­ nation claims may be perceived as a futile effort because they could result in even more negative treatment from members of the privileged group (Kai­ ser & Miller, 2001 ). While members of disadvantaged groups may engage in a number of strat­ egies to avoid encounters with prejudicial treatment (see Miller & Major, 200), the use of such strategies may not leave one feeling very empowered. For instance, groups with "invisible" stigmas have the option of attempting to pass as members of the privileged group. Attempts to pass in order to achieve more positive outcomes clearly places the control over one's life in the hands of the privileged group. For example, members of the US military who are gay, lesbian, or bisexual are required to hide their sexual orientation or face exclusion. In such contexts, passing may give stigmatized group members some control in terms of whether they will avoid negative outcomes, but being forced to hide an important aspect of one's identity severely limits one's freedom and is likely to be experienced as a vital lack of control. Similarly, having to prove oneself as worthy of equal treatment is a severe disadvantage in itself, and implies that one has less control over one's life than members of privileged groups, who are more likely to be assumed to be worthy of positive treatment even before they have the opportunity to demonstrate (or discon­ firm) the validity of that assumption. In summary, attributions to prejudice are not simply external attributions that permit a discounting of the selfs causal role in the production of negative outcomes. For disadvantaged groups, attributions to prejudice will be harm­ ful, because they implicate an aspect of the self that can result in uncontroll­ able negative treatment across a wide variety of situations. In privileged groups, prejudicial treatment is attributed to an aspect of the self that only infrequently results in negative outcomes. Because of their instability, attribu­ tions to prejudice should be less harmful for privileged groups than for disad­ vantaged groups, and could even be beneficial if they serve to remind the privileged of the positive treatment that they normally receive. Despite these attributional reasons for expecting attributions to prejudice to have very dif­ ferent consequences for disadvantaged and privileged groups, the total relev­ ance of an attribution to prejudice cannot be fully appreciated without looking beyond locus, stability, and controllability to other factors that are intimately intertwined with the power of one's group in the social structure. Two other important ways in which attributions to prejudice might differ for disadvantaged and privileged groups are the degree of ambiguity about the

1 76

MICHAEL T. SCHMITT AND NYLA R. BRANSCOMBE

prejudice, and suggest that attributional ambiguity and attributions to preju­ dice may have some costs (p. 522). However, they also suggest that the clear evidence of minimization in Ruggiero's studies is not generalizeable to other disadvantaged groups (p. 522). To reconcile the evidence of minimization of attributions to prejudice with the attributional ambiguity perspective, Crocker et al. ( 1998) offer a revised perspective on the potential costs and benefits of attributional ambiguity. As Crocker et al. (1991 ) found, attributional ambigu­ ity is costly when it leads to the discounting of one's personal role in bringing about positive outcomes. For negative events, they suggest that the self­ protective properties of attributional ambiguity "may be limited to circum­ stances in which negative outcomes are unrelated to one's negative social identity" (Crocker et al., 1 998, p. 520). In other words, in their revised view, attributional ambiguity is threatening when "negative events that are caused by prejudice cannot be easily attributed to prejudice" (p. 521 ). This statement suggests that, while attributional ambiguity for stigmatized group members "affords them a degree of latitude" to make attributions to prejudice (p. 521 ), such ambiguity is threatening when it interferes with reaping the benefits of making an accurate attribution to prejudice. Furthermore, they suggest that, while attributions to prejudice for personal outcomes might be minimized because of their costs, attributions to prejudice for group outcomes are not as likely to be minimized. However, the presumed difference between personal and group discrimination perceptions has been shown to be due to the two ratings involving different comparisons and does not reflect distancing one's own experience from that of the ingroup (see Kessler, Mummendey, & Leisse, 200; Postmes, Branscombe, Spears, & Young, 1999). Indeed, as Crocker et al. (1998) note, its not clear theoretically why the potential costs and benefits of making attributions to prejudice would differ for judgments made at the personal and group levels (p. 524). Our perspective on attributional ambiguity is much simpler. We argue that chronic attributional ambiguity-having to regularly consider prejudice as an explanation for one's negative outcomes-will be detrimental to the well­ being of disadvantaged groups. Such ambiguity about the role of prejudice in one's outcomes is stressful because it is an ongoing reminder of the rejection and devaluation in the broader culture. Indeed, the fear of confirming other people's negative views of the ingroup undermines the performance of Black Americans (Steele & Aronson, 1 995), women (Spencer, Steele, & Quinn, 1 999), and members of the working class ( Croizet & Claire, 1998). The mental work required to navigate through social contexts in which one may be a target of prejudice is likely to be extremely taxing (Frable, Blackstone, & Scherbaum, 1990; Goffman, 1963; Kramer & Wei, 1999). Furthermore, not having to consider the role of group membership as a cause of one's outcomes is one of the major advantages of belonging to a privileged social group (Mcintosh, 1 998). From our perspective, attributional ambiguity is generally

MEANING AND CONSEQUENCES OF PERCEIVED DISCRIMINATION

177

costly to the psychological well-being of disadvantaged groups and, in contrast to the original attributional ambiguity perspective, it is only self-protective in that it provides opportunities to deny that prejudice might actually be the true cause of negative treatment. Implications for Acceptance into Society

Among disadvantaged groups, a pattern of stable attributions to prejudice reflects perceived systematic exclusion by the privileged group. Because privileged groups have the power to define who is and who is not fully accepted, rejection by privileged groups implies that one's group is excluded from the most valued positions in society. Many theoretical approaches predict that feeling excluded in this way will harm self-esteem (Baumeister & Leary, 1995; Cooley, 1956; Mead, 1934; Rosenberg, 1979; Tajfel & Turner, 1 986), and empirical research has supported the contention that such exclu­ sion is painful, resulting in anxiety, depression, feeling a lack of control, and lowered self-esteem (Baumeister & Tice, 1 990; B owlby, 1973; Cozzarelli & Karafa, 1998; Frable, 1 993; Leary, Tambor, T,erdal, & Downs, 1 995; Sol­ omon, Greenberg, & Pyszczynski, 1 99 1 ; Williams, Shore, & Grabe, 1 998) . In contrast, when privileged group members are rejected by the disadvantaged, it carries no implications for exclusion from the wider culture. Thus, because attributions to prejudice among privileged groups do not represent exclu­ sion of one's social identity from the dominant society as a whole, they are less painful than when the same attributions are made by disadvantaged group members.

RECOGNIZING DISADVANTAGE IS PAINFUL

As we have argued, the attributions to prejudice made by disadvantaged groups differ on a number of dimensions from those made by privileged groups. Attributions to prejudice made by privileged groups are less stable, and are less likely to reflect a general lack of control or rejection by the dominant culture. In contrast, attributions to prejudice among members of disadvantaged groups will be more stable, more likely to reflect a lack of control over one's life, and more likely to represent exclusion from the cul­ ture. Stated another way, in disadvantaged groups, the total relevance of an attribution to prejudice extends far beyond the immediate situation; however, for members of privileged groups, the total relevance of an attribution to prejudice is virtually identical to its local relevance. Thus, attributions to prejudice should be more harmful to the well-being of disadvantaged groups than for privileged groups.

1 78

MICHAEL T. SCHMITT AND NYLA R. BRANSCOMBE

Empirical Support

Research on a variety of disadvantaged groups has demonstrated that recog­ nizing prejudice against one's social group is negatively related to psychologi­ cal well-being. For instance, women (Kobrynowicz & Branscombe, 1 997; Landrine, Klonoff, Gibbs, Manning, & Lund, 1 995) and African-Americans (Branscombe et a/., 1 999; Cross & Strauss, 1 998; Klonoff & Landrine, 1 99; Williams, Yu, Jackson, & Anderson, 1 997) who perceive pervasive discrimi­ nation against their group are more likely to exhibit debilitating psychiatric and physical health symptoms, compared to those who perceive less discrimi­ nation. In a recent review, Clark, Anderson, Clark, and Williams ( 199) con­ cluded that perceived racism among African-Americans results in psychological and physiological stress responses that have a number of long­ term negative health effects. Such costs of making attributions to prejudice are not limited to women or ethnic minority groups. In a study of lesbian and gay crime victims, those who attributed the crime to prejudice against their sexual orientation experienced more symptoms of depression, anxiety, vul­ nerability, and post-traumatic stress than did those who attributed the crime to other causes (Herek, Gillis, & Cogan, 1 999). Consistent with our perspec­ tive, Herek et a/. concluded that the greater costs associated with attributing victimization to prejudice occur because they link the negative outcome with one's stable social identity. Although the repeatedly observed negative relationship between perceived discrimination and well-being is compelling, one might argue that this finding results from the reverse causal direction. This alternative explanation would suggest that those who generally see the world as "out to get them" are consequently more likely to see themselves as the target of prejudice (e.g., Crocker, Luhtanen, Broadnax, & Blaine, 1 999; Kramer, 1 998). However, this "paranoia" explanation is less plausible as an account of the relationship between physical health outcomes and perceived discrimination (Clark et a/., 1 999). Further, experimental studies have found that, in fact, attributions to prejudice do harm well-being among members of disadvantaged groups. In one of the earliest experimental investigations of this issue, Dion and Earn (1975) found that Jewish participants who could attribute their failure to a Gentile's anti-Semitism felt more stress and negative affect than those in experimental conditions where an attribution to prejudice was implausible. In a recent study from our own laboratory (Schmitt, Branscombe, & Postmes, 2001 ), women read an article that argued that sexism and discrimi­ nation against women was either pervasive or rare. The manipulation was very successful. In the rare condition, participants reported that 27o/o of men were biased against women, while women in the pervasive condition believed that 5 1 % of men were sexist. Participants also reported expecting to encoun­ ter gender discrimination more often in the pervasive condition than in the

180

MICHAEL T. SCHMITT AND NYLA R. BRANSCOMBE

self-protective (Crocker & Major, 1989; Crocker et a/., 1998). However, stud­ ies of disadvantaged groups have not provided strong support for the self­ protection hypothesis. Dion's (1975) early study of attributions to prejudice manipulated the perceived likelihood that sexism could explain the negative treatment that female participants received by varying the gender of the source of the treatment. Contrary to the self-protection hypothesis, particip­ ants felt worse about themselves when an attribution to prejudice was most plausible (i.e., when the source of the negative treatment was a man), com­ pared to when it was less plausible (the negative treatment came from a woman). In favor of the discounting perspective, Crocker et al. ( 1991 , Study 1 ) found that women who received negative evaluations from a sexist male evaluator reported less depressed affect compared to those who received negative feedback from a non-sexist evaluator, although no support for the discounting hypothesis was obtained on a measure of self-esteem. In Study 2, the discounting hypothesis was not supported. Black participants who re­ ceived negative feedback did not significantly differ in terms of mood or self­ esteem as a function of whether the feedback was attributable to prejudice or not. If attributions to prejudice do have self-protective consequences for mem­ bers of disadvantaged groups, it is logical to suspect that members of disad­ vantaged groups might be motivated to make attributions to prejudice when those attributions are at least somewhat plausible. Any evidence that stig­ matized individuals "overuse" attributions to prejudice and discrimination could be therefore interpreted as support for the discounting hypothesis. However, the evidence of vigilance on the part of disadvantaged groups is especially dubious. Not only is this hypothesis inconsistent with a large body of work finding that the disadvantaged groups minimize the extent to which they perceive the painful reality of their own disadvantage (Crosby, Pufall, Snyder, O'Connell, & Whalen, 1989), when it has been carefully tested ex­ perimentally it has not been supported. Ruggiero and Taylor (1995, 1997) found that women, Asian Americans, and Black Americans who had received negative feedback were quite reluctant to attribute their failure to an evalua­ tor's prejudice, and did so only when prejudice was presented as a virtual certainty. Kleck and Strenta (1980) are often cited as demonstrating that the self­ protective properties of an attribution to prejudice lead the disadvantaged to make such attributions, even when they are a logical impossibility. In their study, the experimenter applied a fake facial scar on participants, but then secretly removed it while ostensibly applying moisturizer to the make-up. After interacting with another person, participants who believed that they had been given the facial disfigurement perceived a strong bias against them. However, attributions to prejudice against a fake, temporary scar would nei­ ther be internal nor stable, and would not raise concerns of potential

MEANING AND CONSEQUENCES OF PERCEIVED DISCRIMINATION

181

devaluation or rejection in the future. For that reason, the meaning of the attributions to prejudice made in the context of Kleck and Strenta's experi­ ment are very different from those made by members of historically disadvan­ taged groups. The aversive qualities of attributions to prejudice that reflect the experience of real disadvantaged groups is absent. In a study comparing the attributions of Black and White participants, Crocker et al. (1991) found a racial group main effect where, "Black subjects were more likely than White subjects to attribute the feedback they received to prejudice when they received negative rather than positive feedback and when the evaluator could see them, hence was aware of their race, than when they were not seen" (p. 225), and interpreted this as evidence that disadvan­ taged groups might be overly vigilant of prejudice against them. However, because of a confound of the design, there is a more plausible explanation for these data. In their research, both Black and White participants received negative feedback from a White evaluator-an outgroup member in the for­ mer case, but an ingroup member in the latter. While the Black participants' racial prejudice attribution ratings were indeed higher than those of the White participants, as the authors themselves note, this is likely to have resulted from a confound between participant race and whether the evaluator was an ingroup or outgroup member. In order to compare Whites and Blacks in terms of the degree of attributional ambiguity and willingness to make at­ tributions to prejudice, attributions to prejudice must be at least somewhat plausible for both groups. For Whites, being evaluated by a White evaluator clearly creates a context in which attributions to prejudice are implausible, and for that reason White participants were less likely to make attributions to prejudice than Blacks, for whom an attribution to prejudice was plausible. Despite this alternative explanation for the observed racial difference, social psychologists continue to cite this study as evidence that disadvantaged groups are more motivated to make attributions to prejudice than are privil­ eged groups. However, this racial difference in attributions really indicates that participants realistically assume that Whites are more likely to be preju­ diced against Blacks than Whites are likely to be prejudiced against Whites. Recent empirical work that does not contain the confound present in Crocker et al. (1991), has found that the differences between privileged and disadvantaged groups in willingness to make attributions to prejudice show a quite different pattern than Crocker et al. found. When Ruggiero and Major ( 1998) controlled for the plausibility of discrimination for a particular event by manipulating the probability of discrimination from an outgroup member orthogonally to group membership, disadvantaged groups (Blacks and women) minimized perceptions of discrimination against them, but privileged groups (Whites and men) appeared to be vigilant in detecting it. Thus, the motivation to exaggerate the possibility of discrimination (given the oppor­ tunity) appears to be present in privileged groups, while the opposite-the

182

MICHAEL T. SCHMITI AND NYLA R. BRANSCOMBE

motivation to minimize the possibility of discrimination-appears to be pres­ ent in disadvantaged groups. In another study with interpretational problems, Crocker et al. ( 199) con­ cluded that, because Black Americans rated anti-Black conspiracies as more plausible than did Whites, Blacks are vigilant in detecting systematic discrimi­ nation against them. However, this racial difference could also be explained by Whites' lack of awareness or denial of the existence of real anti-Black conspiracies. Recently, we conducted a study (Nelson, Branscombe & Schmitt, 2001 ) in which Black and White participants evaluated the plau­ sibility of a bogus newspaper article describing an alleged White conspiracy against Blacks. As in the Crocker et al. (1999) study, Blacks did rate the anti­ Black conspiracy as more plausible than did Whites. However, we found that this racial difference was eliminated when we controlled for Blacks' greater knowledge of real past conspiracies against Blacks (e.g., the Tuskegee experi­ ment). These results suggest that Crocker et al.'s finding that Blacks see anti­ Black conspiracies as more plausible than do Whites is not due to Blacks' motivation to protect personal or group-based self-esteem, but rather that Whites are unaware of, or deny the reality of, past conspiracies by Whites against Blacks. Are Attributions to Prejudice Ever Self-protective?

As Crocker and Major (1989) suggested, attributions to prejudice can, under some circumstances, protect perceptions of one's performance or ability by discounting them as causes of the negative outcome (Britt & Crandall, 200; Ruggiero & Taylor, 1997). However, for disadvantaged groups, attributions to prejudice also imply stable exclusion, making them mostly harmful, as the research we reviewed indicates. In contrast, when members of privileged groups make attributions to prejudice they endure less harmful consequences and reap more psychological benefits than do members of disadvantaged groups. Privileged groups can discount their own personal qualities as causes of the specific negative outcome without simultaneously feeling systematically devalued and disadvantaged. Research on disadvantaged groups provides strong evidence that attribu­ tions to prejudice are even more costly than attributions to one's performance (Ruggiero & Taylor, 1997). Perhaps this is because one's performance, qualifications, and even ability in a domain are perceived as less stable than prejudice against one's group membership. We can, however, imagine a poss­ ible situation where an attribution to personal qualities is more stable, and therefore more threatening, than an attribution to prejudice. Attributing a lack of long-term success in a highly valued domain to a stable personal characteristic might well be a worse alternative to an attribution to prejudice. In such situations, discrimination may be seen as a less stable impediment to

MEANING AND CONSEQUENCES OF PERCEIVED DISCRIMINATION

183

future success than one's personal traits. In that context, an attribution to prejudice, while still being an unpleasant attribution to make, might be some­ what less painful than the alternative. That said, we suggest that such situations are relatively rare, in part because in most situations many other more psychologically comforting explanations for negative outcomes exist. The discounting perspective considers only two attributional alternatives for negative events: that they "could be due to one's lack of merit, inferior qualifications, poor performance, or other shortcom­ ings", or "due to prejudice and discrimination against one's social identity" (Crocker et al., 1998, p. 520). However, in many situations where people face a negative outcome, these are not the only two types of attributions that might be plausible. For instance, one could attribute a negative outcome to truly external factors (e.g., another's negative disposition toward everyone, aspects of the situation that could interfere with performance, task difficulty, or chance) or transient internal factors (e.g., fatigue or mood). Furthermore, in many situations in which members of disadvantaged groups regularly face negative treatment, an attribution to personal qualities, ability, or perfor­ mance are highly implausible, because the negative treatment occurs before there is even an opportunity to demonstrate one's deservingness or lack thereof. In future research, to fully gauge the costs or benefits of attributions to prejudice, we must consider such attributions within a wider range of plausible attributions. The preponderance of the data suggests that attributions to prejudice among disadvantaged groups are painful. Thus, our review is inconsistent with Crocker et al.'s (1991 ) speculation that "members of stigmatized groups who believe that they are discriminated against or that others are racist should be more likely to attribute negative feedback to prejudice and therefore may be higher in self-esteem" (p. 226). In contrast, our perspective suggests that the relative frequency of disadvantaged groups' encounters with prejudice make such attributions especially harmful, while the self-protective properties of such attributions are mainly limited to members of privileged groups, who are likely to encounter prejudice relatively infrequently. COPING WITH PREJUDICE IN DISADVANTAGED GROUPS

Both correlational and experimental evidence attests to the psychological costs of attributions to prejudice among members of disadvantaged groups. Given these costs, we now consider how members of disadvantaged groups cope with the pain of recognizing their disadvantaged status. Crocker and Major (1989) argued that members of disadvantaged groups do not simply passively accept the dominant culture's devaluation of their social identity,

MEANING AND CONSEQUENCES OF PERCEIVED DISCRIMINATION

185

Threats to the ingroup generally encourage group identification (Turner, Hogg, Turner, & Smith, 1 984). As the Rejection-Identification Model pre­ dicts, recognizing prejudice and minority group identification are correlated among Jews (Rollins, 1 973), women (Gurin & Townsend, 1 986), African­ Americans (Gurin, Gurin, Lao, & Beattie, 1 969), lesbians (Crosby et al., 1 989), and non-mainstream college groups (e.g., punks, hippies, nerds; Cozzarelli & Karafa, 1 998). In order to experimentally test the causal dir­ ection of this relationship, we conducted a study in which we manipulated, rather than measured, perceived discrimination against one's group (Jetten, Branscombe, Schmitt, & Spears, 2001). In a sample of people with body piercings, we provided participants with bogus information about the per­ vasiveness of anti-piercing prejudice among members of the mainstream. In one condition, participants were told that the majority of the mainstream disliked them and would discriminate against them. In another condition, participants were led to believe that treatment by the mainstream was more positive. Consistent with predictions, participants in the pervasive discrimina­ tion condition reported significantly more identification with other group members than did participants in the positive treatment condition. Although the reverse effect of group identification on perceptions of dis­ crimination might also be possible (as hypothesized by Crocker & Major, 1 989; Crocker et al., 1 998), we are aware of no experimental evidence support­ ing this causal direction. Furthermore, a number of other researchers have found experimental evidence of the Rejection-Identification Model's predic­ tion that attributions to prejudice encourage minority group identification. Dion (1975) found that women who had attributed their failure to sexism subsequently described themselves more favorably on positive aspects of the stereotype of women. Similarly, Dion and Earn (1975) found that Jewish participants who were led to believe that their failure at a task was attribut­ able to a Gentile's anti-Semitism described themselves more in terms of the positive aspects of the Jewish stereotype compared to participants in the non­ prejudice condition. Dion, Earn, and Yee (1978) argued that these two studies suggest that attributions to prejudice elicit a heightened identification with the positive aspects of one's minority group. Similarly, when women are made aware of their stigmatized status, they are more likely to self-stereotype in terms of their gender (Foster & Matheson, 1999; Hogg & Turner, 1 987). Reminding gay men of anti-gay prejudice increases their identification with the gay movement (Simon et al., 1998) . More generally, when power relation­ ships among experimentally created groups were manipulated, Depret and Fiske (1993) found that the lower-power group showed increased allegiance to their ingroup compared to the more powerful group. All of these studies provide support for the hypothesis that perceiving prejudice leads members of disadvantaged groups to see themselves more in group terms, and to feel greater emotional attachment to that group.

188

MICHAEL T. SCHMITT AND NYLA R. BRANSCOMBE

FUTURE DIRECfiONS FOR THE REJECfiON­ IDENTIFICATION MODEL

The Rejection-Identification Model provides a basis for studying the subjec­ tive experience of disadvantaged groups from a social identity perspective. Social identity theory guided the creation of the model, and offers a number of ways in which the model can be expanded in the future. Below we describe how the Rejection-Identification Model can be used as a theoretical vehicle for examining the processes by which group identification protects well-being, and the factors that might moderate how well the disadvantaged cope.

The Psychological Benefits of Group Identification

Although a number of empirical studies have documented that identification with one's disadvantaged group is associated with psychological well-being, more work needs to be done to examine the processes by which group identi­ fication protects the well-being of members of disadvantaged groups. At first it may seem paradoxical to suggest, as the Rejection-Identification Model does, that people can make a threatening intergroup comparison in which they recog­ nize their group's disadvantage relative to other more privileged groups, and simultaneously benefit psychologically by identifying with that group. Social identity theory suggests that group identification can lead to two general strat­ egies for constructing a positive social identity, despite threatening comparisons with more privileged groups-viz., social creativity and social competition (Taj­ fel, 1978). When adopting the social creativity strategy, the disadvantaged shift their intergroup comparisons to dimensions on which their group fares more favorably, and increase the relevance of those dimensions to the group's iden­ tity. Crocker and Major (1989) also suggest that such "selective devaluing" is one way that disadvantaged groups cope with a negative social identity. Re­ search supports the idea that disadvantaged groups acknowledge reality and admit that the outgroup is superior on status-defining dimensions (e.g. wealth), but see the ingroup as superior on non-status-defining dimensions (Ellemers, van Rijswijk, Roefs, & Simons, 1997; Lemaine, 1974). In political terms, social creativity amounts to rejecting the norms and standards of the dominant culture that disadvantage and devalue the ingroup. In fact, rejection of mainstream norms-which serve the interests of privileged groups-is probably a necessary condition for building a positive minority identity. In Tajfel's (1978) words, "differentiation [from the mainstream] often represents, socially and politically, a rejection of the status quo by groups which perceive themselves as separate and socially disadvantaged" (p. 7). Tajfel (1978) identified social competition as another outcome of group identification that will be adopted by disadvantaged groups who perceive

190

MICHAEL T . SCHMITT A ND NYL A R. BRANSCOMBE

Klink, & Mielke, 1 999), when alternatives to the current social structure are difficult to imagine (Ellemers et al., 1990; Martin, 1986), and when the status quo is legitimized (Ellemers, Wilke, & van Knippenberg, 1993; Commins & Lockwood, 1979). Crocker and Major (1994) have also considered the importance of the perceived legitimacy of prejudice and discrimination as a moderator of the costs of making attributions to prejudice. They argued that, while an attribu­ tion to illegitimately held prejudice can be self-protective, making an attribu­ tion to prejudice that is seen as justifiable is costly. Research on "overweight" women supports the idea that attributions to legitimately held prejudice are harmful to well-being (Crocker, Cornwell, & Major, 1993), and that believing in or being reminded of ideologies that legitimize prejudice against the over­ weight harms psychological well-being (Quinn & Crocker, 1999). While we do agree that attributions to prejudice are more harmful when prejudice is seen as legitimate, compared to when it is seen as illegitimate, we reach this conclu­ sion for a different reason. Social identity theory suggests that the perceived legitimacy of prejudice blocks the possibility of coping with disadvantage through group identification, making perceptions of prejudice and discrimina­ tion especially harmful. We (Garza, Branscombe, Schmitt, & Zarate, 2001 ) found support for this idea i n a study o f Mexican-Americans. Participants completed measures of the perceived legitimacy of discrimination, perceived pervasiveness of prejudice, group identification, and psychological well-being. As expected, we replicated our findings with African-Americans and women among those who saw prejudice against Mexican-Americans as illegitimate. Among those participants, perceived prejudice was positively related to group identification, and group identification partially attenuated the negative con­ sequences of perceiving discrimination on well-being. However, among Mexican-Americans who saw prejudice against them as legitimate, the rela­ tionship between perceived discrimination and minority group identification was non-significant. As a result, the overall effect of perceived discrimination was more harmful among those who saw it as legitimate, compared to those who saw it as illegitimate, even though the direct negative effect was equivalent in both cases. Like legitimacy, other factors, such as the perceived mutability of the social structure and perceived opportunities for individual upward social mobility, are likely to moderate when perceived rejection from the mainstream will encourage coping via group identification. Factors that block coping via group identification can arise from a number of sources, but of particular interest are aspects of culture that are primarily controlled by the powerful. For example, disadvantage can be legitimized by beliefs in the inferiority of disadvantaged groups (e.g., "old-fashioned" racism), or the controllability of their stigma (e.g. weight, sexual orientation). Similarly, group identification might be dis­ couraged by ideologies that make individual upward mobility appear possible

192

MICHAEL T. SCHMITT AND NYLA R. BRANSCOMBE

disadvantaged groups are, in part, internal. Second, because stability and perceived legitimacy are important moderators of the effects of attributions to prejudice, researchers should be cognizant of what the laboratory context suggests to participants about the pervasiveness and legitimacy of prejudice. For example, discrimination coming from a single individual might imply that discrimination is less pervasive than if the discrimination comes from the outgroup as a whole (see Abelson, Dasgupta, Park, & Banaji, 1998). In addi­ tion, when groups are created ostensibly on the basis of performance, coping via group identification is less likely because discrimination against the low­ performance group is likely to be perceived as legitimate. In sum, all forms of discrimination are not experienced identically. Thus, researchers must create in their laboratories or examine in the field the kinds of psychological contexts that actually reflect the psychological dilemmas faced by the disadvantaged groups of interest. In addition, researchers should consider the important role of identification with one's social group as a means of coping with perceived discrimination. As we have shown, group identification suppresses some of the harmful effects of attributions to prejudice; consequently, the harmful effects of such attributions are more evident when identification is included in the analysis. Thus, researchers who do not include group identification in their analyses are likely to underestimate the direct negative effects of attributions to prejudice.

CONCLUSIONS

Crocker and Major (1989) brought the issue of attributions to prejudice in disadvantaged groups to the forefront of mainstream social psychology. That being said, there are important differences between our theoretical perspec­ tive and that of our colleagues. Crocker and Major conceptualized attribu­ tions to prejudice in comparison to attributions to personal aspects of the self, such as ability, performance, and qualifications. Accordingly, they hypoth­ esized that attributions to prejudice are self-protective because they avoid the alternative attributional possibility-that negative outcomes are the result of a personal lack of deservingness. Approaching this topic from a social identity perspective, we conceptualized attributions to prejudice in disadvantaged groups using a different point of comparison-attributions to prejudice made by members of privileged groups. Attributions to prejudice do not just dis­ count personal aspects of the self, but actually implicate another important aspect of the self-one's group membership. Because the consequences of attributions to prejudice are likely to depend on the meaning of its internal component, we considered how attributions to prejudice have different mean­ ings, depending on the position of the ingroup within the social structure. For disadvantaged groups, attributions to prejudice are likely to be relatively

MEANING AND CONSEQUENCES OF PERCEIVED DISCRIMINATION

193

stable, uncontrollable, and reflective of widespread rejection and devaluation. For privileged groups, however, attributions to prejudice are likely to be far more unstable and controllable, with very localized implications. For this reason, we hypothesized that, while attributions to prejudice are less harmful and potentially self-protective for members of privileged groups, they are detrimental to the psychological well-being of the disadvantaged. We perceive social psychologists as having generally accepted the idea that disadvantaged groups protect their psychological well-being by making at­ tributions to prejudice for negative events, with very little empirical evidence to support this conclusion. Our review of the empirical research revealed strong support for the hypothesis that for disadvantaged groups, recognizing prejudice is not self-protective and is, in fact, painful. In a variety of disadvan­ taged social groups, we found that the disadvantaged cope with the psycholog­ ical pain of perceiving prejudice by increasing identification with their ingroup. However, this coping strategy only partially alleviates the harm to psychological well-being caused by attributions to prejudice. We suggest that future work on attributions to prejudice examine how aspects of the culture discourage group identification as a means of coping with the recognition of devaluation. ACKNOWLEDGMENTS

We thank Monica Biernat, Jack Brehm, Chris Burris, Jean-Claude Croizet, Jake Harwood, Alex Haslam, Miles Hewstone, Jolanda J etten, Nicolas N'gbala, Tom Postmes, Paul Silvia, Heather Smith, Joey Sprague and Wolf­ gang Stroebe for helpful comments on earlier versions of this manuscript. REFERENCES Abelson, R. P., Dasgupta, N., Park, J., & Banaji, M. R. (1998). Perceptions of the collective other. Personality and Social Psychology Review, 2, 243-50. Anderson, C. A., Krull, D. S., & Weiner, B. (1996). Explanations: Processes and consequences. In E. T. Higgins & A. W. Kruglanski (Eds), Social Psychology: Hand­ book of Basic Principles (pp. 271-96). New York: Guilford. Bandura, A. (1997). Self-efficacy: The Exercise of Control. New York: W. H. Freeman. Bat-Chava, Y. (1994). Group identification and self-esteem of deaf adults. Personality and Social Psychology Bulletin, 20, 494-502. Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497529. Baumeister, R. F., & Tice, D. M. (1990). Anxiety and social exclusion. Journal of Social and Clinical Psychology, 9, 165-95. Biernat, M., Vescio, T. K., & Theno, S. A. (1996). Violating American values: A "value congruence" approach to understanding outgroup attitudes. Journal of Experimen­ tal Social Psychology, 32, 387-410.

1 94

MICHAEL T. SCHMITT AND NYLA R. BRANSCOMBE

Bowlby, J. (1973). Attachment and Loss, Vol. 2: Separation Anxiety and Anger. New York: Basic Books. Branscombe, N. R. (1998). Thinking about one's gender group's privileges or disad­ vantages: Consequences for well-being in women and men. British Journal ofSocial Psychology, 37, 167--84. Branscombe, N. R., Schiffbauer, K., Schmitt, M. T., & Valencia, L. (2001). Effects of thinking about White privilege or disadvantage and degree of White racial identi­ fication for feelings about the ingroup and outgroup (unpublished manuscript). University of Kansas. Branscombe, N. R., Schmitt, M. T., & Harvey, R. D. (199). Perceiving pervasive discrimination among African-Americans: Implications for group identification and well-being. Journal of Personality and Social Psychology, 77, 135-49. Britt T. W., & Crandall, C. S. (200). Acceptance of feedback by the stigmatized and the non-stigmatized: The mediating role of the motive of the evaluator. Group Processes and Intergroup Relations, 3, 79-95. Brown, J. D., & Siegel, J. M. (1988). Attributions for negative life events and depres­ sion: The role of perceived control. Journal of Personality and Social Psychology, 54, 316-22. Clark, R., Anderson, N. B., Clark, V. R., & Williams, D. R. (199). Racism as a stressor for African-Americans: A biopsychosocial model. American Psychologist, 54, 80516. Commins, B., & Lockwood, J. (1979). The effects of status differences, favoured treatment and equity on intergroup comparisons. European Journal of Social Psy­ chology, 9, 281-9. Cooley, C. H. (1956). Human Nature and the Social Order. New York: Free Press. Cozzarelli, C., & Karafa, J. A. (1998). Cultural estrangement and terror management theory. Personality and Social Psychology Bulletin, 24, 253-67. Crocker, J., Cornwell, B., & Major, B. (1993). The stigma of overweight: Affective consequences of attributional ambiguity. Journal of Personality and Social Psychol­ ogy, 64, 60-70. Crocker, J., Luhtanen, R., Broadnex, S., & Blaine, B. E. (199). Belief in U. S. govern­ ment conspiracies against Blacks among Black and White college students: Powerlessness or system blame? Personality and Social Psychology Bulletin, 25, 94153. Crocker, J., & Major, B. (1989). Social stigma and self-esteem: The self-protective properties of stigma. Psychological Review, 96, 608-30. Crocker, J., & Major, B. (1994). Reactions to stigma: The moderating role of justifica­ tions. In M. P. Zanna & J. M. Olson (Eds), The Psychology of Prejudice: The Ontario Symposium, Vol. 7. (pp. 289-314). Hillsdale, NJ: Erlbaum. Crocker, J., Major, B., & Steele, C. M. (1998). Social stigma. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds), The Handbook of Social Psychology 4th Edn, (Vol. 2, pp. 504-53). Boston, MA: McGraw-Hill. Crocker, J., & Quinn, D. (1998). Racism and self-esteem. In J. L. Eberhardt & S. T. Fiske, (Eds), Confronting Racism: The Problem and the Response (pp. 169-87). Thousand Oaks, CA: Sage. Crocker, J., Voelkl, K., Testa, M., & Major, B. (1991). Social stigma: The affective consequences of attributional ambiguity. Journal of Personality and Social Psychol­ ogy, 60, 218-28. Croizet, J. C., & Claire, T. (1998). Extending the concept of stereotype threat to social class: The intellectual underperformance of students from low socioeconomic back­ grounds. Personality and Social Psychology Bulletin, 24, 588-94. Crosby, F. J., Pufall, A., Snyder, R. C., O'Connell, M., & Whalen, P. (1989). The denial of personal disadvantage among you, me, and all the other ostriches. In M. Crawford & M. Gentry (Eds), Gender and Thought (pp. 79-99). New York: Springer-Verlag.

1 98

MICHAEL T. SCHMITT AND NYLA R. B RANSCOMBE

Robins, C. J. (1988). Attributions and depression: Why is the literature so inconsistent? Journal of Personality and Social Psychology, 54, 880-89. Rollins, J. H. (1973). Reference identification of youth of differing ethnicity. Journal of Personality and Social Psychology, 26, 222-31 . Rosenberg, M. (1979). Conceiving the Self. New York: Basic Books. Rowley, S. J., Sellers, R. M., Chavous, T. M., & Smith, M. A. (1998). The relationship between racial identity and self-esteem in African-American college and high school students. Journal of Personality and Social Psychology, 74, 715-24. Ruggiero, K. M., & Major, B. (1998). Group status and attributions to discrimination: Are low or high status group members more likely to blame their failure on discrimi­ nation? Personality and Social Psychology Bulletin, 24, 821-37. Ruggiero, K. M., & Taylor, D. M. (1995). Coping with discrimination: How minority group members perceive the discrimination that confronts them. Journal of Person­ ality and Social Psychology, 68, 826-38. Ruggiero, K. M., & Taylor, D. M. (1997). Why minority group members perceive or do not perceive the discrimination that confronts them: The role of self-esteem and perceived control. Journal of Personality and Social Psychology, 72, 373-89. Schmitt, M. T. & Branscombe, N. R. (in press). The causal loci of attributions to prejudice. Personality and Social Psychology Bulletin. Schmitt, M. T. & Branscombe, N. R. (2001 ) The psychological consequences of gender-based exclusion (manuscript in preparation). University of Kansas. Schmitt, M. T., Branscombe, N. R., Kobrynowicz, D., & Owen, S. (in press). Perceiving discrimination against one's gender group has different implications for well-being in women and men. Personality and Social Psychology Bulletin. Schmitt, M. T., Branscombe, N. R., & Postmes, T. (2001 ). Women's emotional re­ sponses to the perception of pervasive gender discrimination (manuscript submitted for publication). Sidanius, J., & Pratto, F. (1999). Social Dominance: An Intergroup Theory of Social Hierarchy and Oppression. New York: Cambridge University Press. Sigelman, L., & Welch, S. (1991). Black Americans ' Views of Racial Inequality: The Dream Deferred. New York: Cambridge University Press. Simon, B., Loewy, M., Sturmer, S., Weber, U., Freytag, P., Habig, C., Kampmeier, C., & Spahlinger, P. (1998). Collective identification and social movement participation. Journal of Personality and Social Psychology, 74, 646-58. Snyder, C. R., & Higgins, R. L. (1985). Excuses: Their effective role in the negotiation of reality. Psychological Bulletin, 104, 23-35. Solomon, S., Greenberg, J. & Pyszczynski, T. (1991 ). Terror management theory of self-esteem. In C. R. Snyder & D. R. Forsyth (Eds), Handbook ofSocial and Clinical Psychology: The Health Perspective (pp. 21-40). New York: Pergamon. Spencer, S. J., Steele, C. M., & Quinn, D. M. (1999). Stereotype threat and women's math performance. Journal of Experimental Social Psychology, 35, 4-28. Steele, C. M., & Aronson, J. (1995). Stereotype threat and the intellectual test perfor­ mance of African-Americans. Journal of Personality and Social Psychology, 69. 797Sll.

Stewart, T. L., Vassar, P. M., Sanchez, D. T., & David, S. E. (200). Attitude toward women's societal roles moderates the effect of gender cues on target individuation. Journal of Personality and Social Psychology, 79, 143-57. Stipek, D. J. (1983). A developmental analysis of pride and shame. Human Develop­ ment, 26, 42-54. Swim, J. K., Cohen, L. L., & Hyers, L. L. (1998). Experiencing everyday prejudice and discrimination. In J. K. Swim & C. Stangor (Eds), Prejudice: The Target's Perspective (pp. 37-60). New York: Academic Press.

This page intentionally left blank

Chapter 7

Goal-based Explanations of Actions and Outcomes John McClure Victoria University of Wellington, New Zealand

ABSTRACT Most attribution theories focus on inductive inferences and abstract causal cate­ gories. By contrast, goal-based and knowledge structure theories focus on people's perceptions of intentional actions, their deductive inferences, and their concrete explanations of actions. Goal-based theories have demonstrated the importance of goals and intentions as explanations. However, research shows that explanations of goal-based actions refiect the extremity and controllability of the actions, the presence of obstructions, and the type of causal question eliciting the explanation. These factors determine whether people prefer to invoke goals or enabling conditions (or conjunctions) as explanations. Judg­ ments of explanations also refiect communicative principles or informativeness and relevance, more than logical judgments of probability and necessity. The goal-based approach is often seen as an alternative to the inductive co­ variational approach, but the two lines of research can be integrated.

How do people explain other people's actions? This is an issue of interest in social psychology, law, and philosophy, to name just a few of the relevant areas. This chapter focuses on social psychological theories of explanation, but theories in law and philosophy are often relevant and have stimulated Correspondence regarding this chapter should be addressed to John McClure, Department of Psychology, Victoria University of Wellington, PO Box 60 Wellington, New Zealand. E mail: John.Mcclure@vuw .ac.nz European Review ofSocial Psychology, Volume

© 2002 John Wiley &

Sons Ltd.

12.

Edited by Wolfgang Stroebe and Miles Hewstone.

202

JOHN MCCLURE

social psychological theories. In several instances, ideas from other disciplines have been translated by social psychologists into testable concepts and propo­ sitions (e.g., Hilton, 1990; Malle, 1 999). In social psychology, the dominant model of explanations has been attribution theory, an orientation that focuses on logical and inductive judgments about the causes of actions (e.g., Forster­ ling, 1 989; Kelley, 1 967). However, a second line of research that has flour­ ished in the 1980s and 1990s is the goal-based approach, which focuses on the intentional nature of human action and real-world knowledge. The flowering of goal-based theories of explanations has been strengthened by the emer­ gence of research on knowledge structures (e.g., Schank & Abelson, 1977), folk psychology (e.g, Fletcher, 1 995), and theory of mind (e.g., Wellman, 1990). This emergence of interest in goal-based approaches is illustrated by the fact that the preconference for the Society for Personality and Social Psychology in 1998 focused on Heider's (1958) theory of a naive psychology. This theory typifies a goal-based approach. The present chapter focuses on goal-based approaches to explanations, but it also examines potential connec­ tions between goal-based theories and orthodox attribution theories. To place goal-based theories in context, it is useful to provide a brief overview of theories of explanations in social psychology. The first systematic theory of explanations was Heider's (1958) interpersonal theory, which has two aspects. First, it focused on judgments of intentional actions and the ways that those judgments differed from judgments about non-intentional actions and physical events. This aspect of Heider's model revolved around lay people's concepts, such as "can" (as in, "I can do this task") and "want" (as in, "I want to do it"). A second branch of Heider's work set out to define logical processes underlying causal inferences about behavior. In this analysis, Heider drew on the co-variational reasoning defined by J. S. Mill (187211973), which proposes that people attribute an event to the cause that is present when the effect is present and that is absent when the effect is absent. Follow­ ing Heider, attribution theories focused primarily on this logical inductive strand of Heider's theory, and research examined co-variational reasoning, the distinction between the person and the situation, and the processes that underlie dispositional inferences (e.g., Jones, 1979; Kelley, 1967). Researchers continue to examine co-variational reasoning and dispositional inferences (e.g., Fiedler, Walther & Nickel, 1999; Fosterling, 1989; Morris & Larrick, 1995). However, recently theorists have returned to the other main issue examined by Heider (1958): the nature of judgments about intentional actions. For example, the knowledge structure framework focuses on goal­ based explanations, and uses categories, such as goals and preconditions, rather than the person and the situation (e.g., Abelson & Lalljee, 1988; Lalljee & Abelson, 1983; Leddo, Abelson, & Gross, 1984; Read, 1987, 1988; Read & Marcus-Newhall, 1 993; Schank & Abelson, 1977). With Schank and Abelson's (1977) classic example of going to a restaurant, for example, one goal is

GOAL-B ASED EXPLANATIONS

203

wanting a meal (or wanting to meet someone at the restaurant), and precondi­ tions include having money to buy a meal and the restaurant being open. Knowledge structure theorists have framed questions about explanations in the following terms. How do judgments about actions tie in with people's scripts and knowledge structures about actions? Do people judge goals to be better explanations of actions than the preconditions that are needed for an action to occur? Recent research also links these questions to folk psychology (e.g., Malle, 1999). This chapter outlines the leading theories of goal-based explanations and the research on those theories. Three key theories are Heider's (1958) theory of personal causality, Hart and Honore's (1985) theory of causation in the law, and the knowledge structure approach (e.g., Lalljee & Abelson, 1983). After reviewing these theories, the chapter outlines research on goal-based explanations, focusing on knowledge structures, the extremity of actions, the controllability of actions, and the causal question soliciting the explanations. This research also examines the role of pragmatic factors in explanations, and points out the importance of distinguishing between logical measures of ex­ planations, such as the probability and necessity of causes, and pragmatic criteria, such as the informativeness and relevance of explanations (Hilton, 1990; Schwarz, 1999). The chapter then examines whether goal-based theories of explanation can be integrated with other approaches: folk psychology on the one hand, and inductive attribution theories on the other. A final section discusses the broader relevance of goal-based explanations. THEORIES OF GOAL-BASED EXPLANATIONS

A number of theories have been applied to goal-based explanations, but the most important are Heider's (1958) theory of personal causality and equifinality, Hart and Honore's (1985) theory of causation in the law, and the knowledge structure approach (e.g., Lalljee & Abelson, 1983). These theories are described in turn here, before turning to the relevant research. Intentional Action and Equifinality

Heider focused on intentional action, and the defining features of intentional action, which he sometimes called "personal causality", as opposed to non­ intentional action and physical causes. He claimed that a key characteristic of intentional action is that it demonstrates equifinality, which is the condition where the means to achieve a goal may vary, but the goal itself stays constant (i.e., it is equi-final). When one route (or precondition) toward a goal is blocked, people with goals pursue other routes to achieve their goals. Take the case of a man who wishes to visit a friend. If he tries to start his car and

204

JOHN MCCLURE

finds it does not start, he is likely to seek an alternative means of fulfilg his goal of visiting his friend (e.g., catch a taxi). Heider contrasted this equifinal quality of intentional action with physical causality: to adapt Hume 's ( 1739/1888) classic example, when a billiard ball encounters an obstacle on the path to the hole, it doesn't keep trying out different routes to get into the hole. The ball merely ricochets away from the obstacle. Heider's theory recognized that preconditions such as abilities, and en­ vironmental forces such as task difficulty, may serve as obstacles that impede actions. But Heider considered that human action is typically equifinal in nature, and that by pursuing different routes to a goal, people can circumvent obstacles. He claimed that within a wide range of environmental conditions, the person's intention or goal is the sufficient cause to effect an action: "If there are obstacles in his way, the goal-directed person will circumvent them" (Heider, 1958, p. 102). Heider discussed environmental conditions that im­ pose limits to a person's goals as sufficient causes, but he did not fully define those limits. In general, Heider's view would suggest that goals are likely to be better explanations than the particular means used to achieve an action, be­ cause goals are normally sufficient to produce the action (see also Kruglanski, 1 996). Causation in the Law

The perceived role of intentionality in action is important in the legal con­ text. Hart and Honore (1985) examined the issue of how people impute causality for actions in the context of legal judgments. Like Heider, Hart and Honore proposed that the voluntary nature of intentional actions influ­ ences causal judgments and that people see voluntary causes as better ex­ planations of actions than non-voluntary causes. To illustrate this claim, they cited the case of a forest fire that results when a person discards a cigarette butt when driving through a forest and a breeze fans the flames, leading to the forest fire. They claimed that people are likely to attribute the forest fire to the voluntary action, which in this case is the person throwing the cigarette butt, because this cause is more abnormal than the breeze. According to Hart and Honore, one reason why this happens is that people see voluntary causes as deviations from the natural order, and see them as more abnormal than natural causes. In some cases both of two causes preceding an event may be voluntary. Hart and Honore claimed that in this case people attribute the event to the most recent voluntary cause. For example, with the forest fire scenario, the first cause may again be a person throwing a cigarette butt, but the second cause, rather than being a breeze, may be a person fanning the flames ignited by the cigarette butt. The two actions result in a forest fire. In this case, Hart and Honore claimed, people are likely to attribute the fire to the most recent

206

JOHN MCCLURE

claim, that explanations in terms of a person with an intention are seen as good explanations. The Knowledge Structure Approach

Several authors have applied the knowledge structure approach to goal-based explanations (e.g., Abelson & Lalljee, 1988; Abraham, 1988; Lalljee & Abelson, 1983; Leddo et a/., 1984; McClure & Hilton, 1997, 1998; Read, 1987; Read & Marcus-Newhall, 1993; Schank & Abelson, 1977). Knowledge struc­ tures refer to the schemas and scripts that people use to comprehend every­ day reality and understand stories, and that influence people's explanations for actions. In place of the abstract categories preferred in covariational ap­ proaches, such as, "something about the person" and "something about the situation", the knowledge structure approach examines specific concrete ex­ planations, such as, "He did it because he wanted to get revenge". Categories like "the person" and "the situation" are broad and heterogeneous, and fail to distinguish important differences in explanations. For example, when a listener complements a speaker, the "person" category fails to distinguish between the goal of flattering the speaker and the goal of expressing genuine admiration for the speaker. In contrast to the inductive inferences emphasized in covariational attribu­ tion models, the knowledge structure approach claims that people's explana­ tions often reflect their prior knowledge, and that people use real-world knowledge to determine which explanations are more plausible. The know­ ledge structure approach claims that explanations are derived deductively from people's knowledge of the world, rather than an abstract inductive pro­ cess. This knowledge can be represented in the form of a script, a stereotyped sequence of actions that serve the goals of the actor( s) in the script (Lalljee & Abelson, 1983; Abelson & Lalljee, 1988). With regard to explanations of actions, the knowledge structure approach focuses on explanations that account for intentional actions. The most import­ ant of these are goals, but they also include preconditions and other causes. Goals typically refer to the intention behind an action, and preconditions refer to causal factors that enable actions to occur. Goals and preconditions do not map directly onto the person-situation distinction, because only some preconditions are situational (e.g., when a cafe is closed), and others are located in the person (e.g., when a person lacks the ability to perform an action). Of course, goals may sometimes serve as a precondition to an action, but the knowledge structure approach treats goals and preconditions as dis­ tinct categories. In support of this goal-based taxonomy of explanations, Lall­ jee and Abelson (1983) noted that, when research has examined the categories that people use to distinguish different actions, the dimensions that people use most comprise the goals and motives of actors, rather than the

208

JOHN MCCLURE

money in his pocket was judged more probable than the explanation that merely stated that he was hungry. This preference for conjunctions over indi­ vidual causes is consistent with research in other domains demonstrating conjunction effects (e.g., Tversky & Kahneman, 1983). The finding that people rate goals and preconditions together as more probable than goals on their own suggests that preconditions are an important component of good explanations, even though preconditions on their own are seen as less import­ ant than goals. Preferences for conjunctions differed, however, for completed and failed actions (Leddo et al., 1984). For completed actions like going to the restaur­ ant, participants judged that both goals and preconditions were probable, but for uncompleted or failed actions, like not going to the restaurant or not passing an examination, people judged that the absence of one cause, either the goal or the precondition, was likely. This finding suggests that conjunction effects are more likely for completed actions than uncompleted actions. For someone to go to the restaurant, he/she needs to want a meal and have the money to buy it. But for someone to not go to the restaurant, only one relevant cause needs to be absent: for example, the person isn't hungry or the restaurant is closed. This pattern is analogous to many physical causes. To start a car, petrol in the tank and a charged battery are both necessary, but the absence of either cause, such as running out of petrol or a fiat battery, will prevent a car from moving. To explain this difference in conjunctions for completed and failed actions, Leddo et al. (1984) drew on people's knowledge structures about causal rela­ tions. They proposed that when people explain an action, they combine their knowledge about the particular situation with their abstract knowledge about causal relations in an explanatory frame. These frames represent people's beliefs about the structure of a completed explanation for the action. With completed actions, the frames contain slots for typical goals for the action and enabling conditions that make the action possible. With uncompleted actions, however, the frame contains a single slot for any cause that could prevent the action. So with uncompleted actions, people seek a single suitable cause that fills the available slot. This cause may be a goal (or the absence of a goal) or it may be an absent precondition. This interpretation accounts for Leddo et a/. 's findings, but it fails to account for the importance of competing goals and competing causes. Wilensky (1983) proposed that a common reason why people do not complete actions is that they have competing goals. For example, a person's failure to go to a restaurant may ocur not because he/she lacks the goal of going to the restaurant or lacks a precondition such as money that would enable him/her to go, but because he/ she has the confficting goal to go to a movie. In support of this view, McClure et al. (1989) showed that when people evaluate confficting goals as explanations, they judge that a conjunction of a confticting goal and a relevant precondition is

GOAL-BASED EXPLANATIONS

209

a more probable explanation for the failure to complete an action than the conflicting goal or the precondition in isolation. When explaining Fred's failure to go to the restaurant, for example, the presence of the desire to go to the football match and his possession of a ticket for the match is judged a better explanation than either the competing goal or the precondition on their own. So the pattern of conjunctions and single causes found with goals that motivate the action is reversed with goals that conflict with the action. This finding shows that people's explanations take account of goal conflict and the oppositions between causes that arise in cases of goal conflict (Wilensky, 1 983). In sum, the initial research applying knowledge structures to explanations suggested that goals are better explanations of scripted actions than precondi­ tions. The presence or absence of conflicting goals plays a key role in whether people prefer a mere goal as an explanation rather than a conjunction of a goal and precondition. The Effect of Extremity and Difculty

The research described up to this point is consistent with the view that goals are good and sufficient explanations of actions. Recent research, however, has shown a number of boundaries to the idea that people prefer goals as explana­ tions for actions. This research has demonstrated that preconditions some­ times play an important role in explanations and shows that judgments about goals as explanations are influenced by a number of factors. These factors include the extremity of actions, the controllability of actions, and the causal question that prompts the explanations. This research also shows that judg­ ments differ, depending on whether the measures assess the probability, necessity, or informativeness of explanations. Research examining boundaries to preferences for goals first showed that judgments of explanations are influenced by the extremity or difficulty of the relevant action. Previous research on attributions had pointed in this direction, although earlier goal-based theories did not explore this point. Kelley's (1972) theory of discounting and causal schemata proposed that, when people are presented with two or more alternative causes for a common action, they tend to discount one of the alternative causes. In contrast, with extreme and difficult actions, rather than discounting one of the causes, people tend to see several causes as contributing to the action. Kelley linked discounting to the multiple sufficient schema, a causal schema where people see either of two alternative causes as sufficient to explain the effect. Kelley linked the lower levels of discounting with extreme actions to the multiple necessary schema, a schema where multiple causes are necessary to explain an action. So, whereas people are likely to attribute a moderate achievement to the presence of either ability or effort, they are likely to attribute an outstanding (extreme) success to the combination of ability and effort.

210

J O HN MCCLURE

Kelley's proposition applies equally to all causes and makes no distinctions between different explanations, such as goals and preconditions. However, the theory implies that when people are explaining extreme actions or achieve­ ments, goals in isolation are likely to

be seen as insufficient explanations.

Research on Kelley's theory has shown that, for some outcomes, people judge actions in the way that the theory implies, and give more conjunctive explanations for extreme actions than moderate actions (Kun 1 973; Morris

& Larrick,

&

Weiner,

1995). However, with many extreme actions, people

invoke a single extreme cause as an explanation, and this cause is often a precondition (McClure, 1998; McClure, Lalljee, Brewer, 1979). For example, McClure

et al.

&

Jaspars, 1991 ; Reeder

&

( 1 99 1 ) showed that when people

explained Einstein's success as a scientist, they did not invoke a conjunction of working hard and intelligence, as Kelley's theory might suggest, but they said that Einstein's success occurred because he was brilliant. In other words, they invoked an extreme level of intelligence to explain the success, and made no reference to effort or hard work. In contrast with effort or the goal to succeed, brilliance is a precondition, in the form of extreme ability. This finding sug­ gests that for some extreme events, preconditions such as ability are seen as better explanations than a person's effort or goals. This finding supports Reeder and Brewer's ( 1 979) claim, that actions in general are diagnostic in terms of preferences (goals), but only extreme actions are diagnostic in terms of abilities, which are a subset of preconditions. McClure and Hilton ( 1 997) applied this idea to goal-based explanations, a domain where most previous studies found that goals are better explanations than preconditions (e.g., Leddo

et al.,

1 984). In studies comparing goals and

preconditions as explanations, the events being explained have typically described relatively common scripted actions, at least in Western or indus­ trialized cultures, such as going to a restaurant or having a barbecue (e.g., Leddo

et al.,

1 984) . For these relatively common actions, goals are typically

achieved by moderate effort and goals are preferred as explanations. McClure and Hilton ( 1 997) suggested that this finding would not generalize to difficult and extreme actions, and that for these actions preconditions would be pre­ ferred as explanations. To test this view, they manipulated the extremity of different actions and the level of obstruction facing the actor who was attempting to carry out the action. Extremity was defined in terms of the resources required to perform the action. With a purchasing scenario, for example, the extreme version of one scenario referred to Mary buying a new expensive car, an action that requires substantial financial resources, and the moderate version of the scenario referred to Mary buying a new dress, an action that requires rela­ tively modest resources. "Obstruction" was defined in terms of the level of resources that the given target person possessed. The obstructed version of the purchasing scenario described Mary as poor, whereas the unobstructed

212

J O HN MCCLURE

sufficient explanation of the action. People make this assumption even when common actions have been obstructed, although in this case they give more weight to preconditions (see also Johnson, Boyd, & Magnani, 1994; Lalljee & Abelson, 1983). In contrast, when an actor performs an extreme action that was previously obstructed by a precondition deficit people attribute the action to the precondition becoming available. With extreme actions, such as buying a very expensive car, people judge goals to be better explanations only when the action is unobstructed, such as when the person buying the car is very rich. In the case of a poor person, the availability of a precondition, such as winning a lottery or acquiring an inheritance, is a better explanation of the action. In sum, these findings suggest the use of goals and preconditions reflects the difficulty of the action being performed and the presence of obstructions hindering the actions (Kelley, 1972; McClure, 1991, 1992; McClure, Lalljee, & Jaspars, 1991 ; Read, 1987; Wilensky, 1983). Pragmatic Influences on Explanations

Recent research on goal-based explanations for actions shows that the mea­ sure used to judge causes is important. In the initial research on knowledge structures described earlier, the judgments assessed the probability of the causes (e.g., Leddo et a/., 1984; McClure et a/., 1989). This research found that people preferred conjunctions over mere goals as explanations of scripted actions. However, subsequent research has shown that this preference for conjunctions on probability ratings disappears on measures of pragmatic as­ pects of explanations. The pragmatic approach claims that people evaluate statements in terms of communicative principles (Grice, 1975; Hilton, 1990, 1 991, 1995; Schwarz, 1999; Strack, Schwarz, & Waenke, 1991). For example, people evaluate explanations according to whether the explanations are infor­ mative and relevant, rather than in terms of their probability and necessity. To apply this perspective to goal-based explanations, research has asked par­ ticipants to judge goals and preconditions in terms of their informativeness and relevance, as well as their probability and necessity (Hilton & Knott, 1 996; McClure & Hilton, 1 997; McGill, 1990). McClure and Hilton (1997) also examined which of these different specific measures of explanations (e.g., probability) was the best predictor of judgments of the overall quality of explanations, as assessed by questions such as, "How good overall is the statement 'Fred was hungry' as an explanation of Fred stopping in to a res­ taurant to eat?" McClure and Hilton (1997) replicated previous findings that, when common actions occur, people judge that both goals and preconditions are probable and necessary for the action to occur. However, on measures of the overall quality of explanations, participants judged goals on their own to be better explanations of common actions than conjunctions of goals and preconditions,

214

JOHN MCCLURE

overall quality. In addition, people judge that the informativeness of an ex­ planation is a better predictor of the overall quality of explanations than its probability. The Effect of ControUabilty

The research described in previous sections shows that the extremity of actions influences judgments of goals and preconditions as explanations for those actions. This finding suggests that explanations may be influenced by the controllability of actions. With extreme actions and outcomes that require substantial preconditions, such as buying an expensive house or completing a highly skilled task, many people find it difficult or impossible to obtain the preconditions required for the action. An event that requires substantial pre­ conditions is thus likely to be harder to control, and this inability to control the action explains why a goal is often insufficient to explain the action. In contrast, with common scenarios that require modest preconditions, people are likely to view the actions and outcomes as controllable, and to judge goals to be a good explanation. This reasoning suggests that goal-based theories of explanation need to integrate a control dimension (Malle & Knobe, 1997). Research on action theories has shown that control is an important variable in relation to actual actions, as distinct from lay people's explanations of actions (Gollwitzer, 1 993; Skinner, Chapman, & Baltes, 1988). In the area of lay explanations, the con­ trollability of causes is also recognized as an important variable. It is recog­ nized in achievement theories of attribution, along with other dimensions, such as locus and stability (e.g., Weiner, 1 986), and in research on motivated social cognition (Kruglanski, 1996), and counterfactual reasoning (Girotto, Legrenzi, & Rizzo, 1 991; Mandel & Lehman, 1996). Girotto et a/. ( 1 991 ), for example, showed that, for negative events, people judge controllable factors, such as an actor's decision, to be more preventable and mutable, than uncon­ trollable causes. This was the case even when other causes, such as abnormal circumstances and the actions of other people, were seen as being more important in causal explanations of the event. When an actor suffers a road accident, for example, observers judge that the accident could have been prevented by changes to the actor's choice, such as the choice of route, rather than non-voluntary causes, such as road conditions and the actions of other drivers. These findings indicate the importance of control in relation to actions and events, but research on goal-based explanations has yet to fully integrate this variable (Malle & Knobe, 1997). To address this issue, McClure, Densley, Liu, and Allen (2001 ) examined the relation between the controllability of events and judgments about goals and preconditions as explanations. They aimed to show that preferences for goals apply only to controllable events. As with previous studies, the research

216

JOHN MCCLURE

This finding clarifies the boundaries to the idea that goals are better explan­ ations than preconditions, and shows that the lower preference for goals for extreme events reflects the low controllability of those events. Even in the context of explanations of goal-based actions, participants judge that some outcomes are outside most people's control and that the goal is not the best explanation of the action. The finding that controllability is the best predictor of goals has important implications. It suggests that judgments about goals as explanations for extreme events are determined not by the improbability per se of improbable events, but by the fact that improbable desirable events are typically uncontrollable (Mandel & Lehman, 1996; Roese & Olson, 1995). These findings suggest that goals are good explanations only for controllable actions and outcomes. It is interesting to consider how these findings may relate to the concept of equifinality (Heider, 1 958; Kruglanski, 1996). The findings on the effects of controllability suggest that observers judge everyday scripted actions to be relatively controllable for the actor. This finding suggests that observers judge that, with these scripted actions, the actor can circumvent obstacles and ob­ tain the means (or preconditions) necessary to fulfill their goals. This finding thus seems to be consistent with the concept of equifinality, which proposes that when one means to achieve a goal is blocked, people circumvent the obstacle and obtain other means to achieve their goals (e.g., Heider, 1958; K.ruglanski, 1996). However, the findings on controllability also suggest that equifinality may apply only with outcomes that the actor can control, and may not extend to events that fall beyond people's control (Malle & Knobe, 1997). The findings show that observers prefer goals as explanations with a target person who has more resources (e.g., money, skills), and preconditions with a target person who has fewer resources. This suggests that what is controllable for one person or group may be uncontrollable for other persons or groups (McClure, 1 985; McClure & Hilton, 1997). What is controllable for a rich person or a skilled person may be uncontrollable for a poor person or an unskilled person. For people with more resources, equifinality may apply in many domains (although not necessarily all domains: "money can't buy me love"). For people with access to only limited resources, equifinality may apply only within a narrow domain of their lives (Wilensky, 1983). As the concept of equifinality suggests, a wealthy woman wanting a good meal has an almost infinite number of ways of achieving her goal, and can easily circum­ vent an obstacle that prevents her from eating at the first restaurant she comes to. In contrast, a starving woman in a famine-struck country who wants to eat a good meal may have no ways of achieving her goal, and equifinality is close to non-existent. She has no ways of circumventing the obstacles that prevent her from achieving her goal. This suggests that equifinality may be con­ strained within the bounds of what is possible for a given person or group (Cross & Markus, 1 991 ; Oyserman, Ager, & Gant, 1995). This reasoning

218

JOHN MCCLURE

The finding that "why" questions about actions tend to elicit goal explana­ tions may account for the high use of goal explanations for common actions in McClure and Hilton's (1997) studies on the effects of extremity. These studies showed that people preferred goals as explanations of common actions, even when those actions previously were obstructed by a precondition default. The measures of the informativeness and the overall quality of explanations in these studies were framed as "why" questions. The measure of informative­ ness, for example, asked, "How informative is the explanation 'Fred was hungry' in regard to the question 'Why did Fred go the restaurant'?". The measure of the overall goodness of explanations was framed in similar terms. Graesser et al.'s (1980) analysis suggests that the "why" framing of these questions may have implied that the best explanation was a goal. Of course, this interpretation does not explain the finding that preconditions were pre­ ferred for difficult events, whereas goals were preferred for common events, because both extreme and moderate scenarios used the same "why" ques­ tions. But this interpretation does suggest that the preference for goals with common actions was enhanced by the "why" wording of the questions (see also Graesser, Langston, & Baggett, 1993; Graesser & Person, 1994; McClure, 1 998; Schwarz, 1999). In related research on causal questions, McGill and Klein (1993) found that the framing of causal questions influenced the focus of information that people sought about explanations. Participants were more likely to seek infor­ mation verifying the necessity of causes when a question was framed as, "What do you think caused [the target event] to happen?", than when a question identified the candidate factor and asked, "Did something about X cause the target event to happen?". Given that the "What do you think caused X?" question enhanced a focus on necessity, this type of question should elicit preconditions as much as goal explanations, because precondi­ tions are regarded as necessary causes, even where goals are preferred as explanations (Hilton & Knott, 1996; McClure & Hilton, 1 997). This reasoning suggests that framing a causal question as a request for an explanation should elicit preconditions more than "why" prompts. To test this proposition, McClure and Hilton (1998) examined the effect of causal questions on goals and preconditions as explanations for actions. They examined whether people explain common actions more in terms of precondi­ tions than goals when the question is phrased as requesting a causal explana­ tion, rather than when posed as a "why" question. As in previous studies, McClure and Hilton (1998) obtained explanations for obstructed and unob­ structed actions; the actions being explained were all common actions adapted from Leddo et a/.'s (1984) scenarios, such as the "going to the restaurant" scenario. The studies also took account of the fact that previous findings on the issue may reflect the use of structured measures that might suggest an explana­ tion that participants would not otherwise have thought of. To counter this

220

JOHN MCCLURE

Ratings of causes when causal question is phrased as a "why" question and a "causal explanation" question

Table 7.2

Question

Scenario action version Why

Explanation

Normal

Obstructed

Normal

Obstructed

Goodness, goal Goodness, precondition

5.97 3.38

5.63 4.96

4.75 4.03

4.08 4.72

Informativeness, goal Informativeness, precondition

6.03 3.78

5.68 5.10

4.84 4.23

4.33 4.89

Probability, goal Probability, precondition

5.67 4.28

5.80 5.43

5.35 4. 19

5.22 5.20

Necessity, goal Necessity, precondition

5.31 3.92

5.66 5.56

5.28 4.28

5.16 4.92

Inhibitory

2.86

4.89

3.15

4.79

Ratings

From McClure & Hilton, 1998. © John Wiley & Sons Limited. Reproduced with permission.

claim that pragmatic principles of communication, such as informativeness, play a stronger role in judgments of the quality of explanations than probabil­ ity judgments (Grice, 1 975; Hilton, 1990, 1991; Schwarz, 1999). This research by McClure and Hilton (1998) manipulated only two catego­ ries of causal question: "why" questions, and requests for a "causal explana­ tion". These questions reflected the key categories of causal question proposed by Graesser et al. (1980), but they represent only two of a number of possible causal questions that can be asked in naturalistic settings (Graesser & Person, 1994; Turnbull, 1986; White, 1992). Using the same scenarios de­ veloped by Leddo et al. (1984), McClure, Hilton, Cowan, Ishida, and Wilson (2001) examined the effects of a wider range of causal questions, including not only "why" and "give a causal explanation", but also two questions that are often posed in naturalistic settings: "suggest why" and "how come?''. The "why" and "suggest why" questions elicited more goals, but the "give a causal explanation" and "how come" questions elicited more preconditions. These findings also suggest a relationship between the availability of pre­ conditions and what is abnormal or surprising about an action. When people ask causal questions in naturalistic situations, these questions take account of people's existing knowledge about the relevant events ( Graesser & Person, 1994). So questions about actions reflect the availability of the preconditions for the action. This point can be illustrated with the scenarios used by Mc­ Clure and Hilton (1997). When explaining rich people making an expensive

GOAL-BASED EXPLANATIONS

221

purchase, the preconditions necessary for the actions are likely to be assumed, and people will be interested in the actors' goals. They are thus more likely to ask questions that presuppose the relevance of goals, such as "why" or "what was their reason?" to obtain an explanation of the action. In contrast, when explaining poor people making an expensive purchase, observers are more likely to be interested in explaining the means with which they completed the actions (i.e., preconditions), and may ask "how" in order to explain their action. These results, of course, are obtained in controlled experimental contexts and may not generalize to other more naturalistic contexts. However, both linguistic and knowledge structure analyses suggest that "why" questions about actions are usually seen as goal-orientated questions (e.g., Graesser et al., 1 980; Schank & Abelson, 1977), so these findings are likely to generalize to naturalistic contexts. The results are also consistent with the claim that, in the context of establishing responsibility for crimes, "why" questions are linked to judgments of intention and goal-based action more than questions about causal factors (Hart & Honore, 1985). In sum, these findings reinforce the view that neither goals nor preconditions are inherently better explanations for actions. Causal selections are based not only on the difficulty and con­ trollability of the action being explained, but the way the causal question is asked. These findings suggest that the wording of causal questions plays a signifi­ cant role in determining explanations of goal-based actions. Previous research on language and attribution has shown that the linguistic category used in sentences influences attributions (e.g., Semin & Fiedler, 1 991). The research described here shows that the wording of causal questions also influences explanations. In sum, people's explanations of goal-based actions are shaped by causal questions. Causal Chains: Sequences of Goals and Preconditions

In the research up to this point, observers judge goals and preconditions as explanations as concurrent influences on the outcome, and evaluate their goodness as explanations at a single point in time. These studies clarify people's judgments of goals and preconditions as concurrent influences on actions, but they show little about judgments regarding the consecutive influ­ ence of the causes on the outcome and on each other. For example, these studies give no idea of whether these causes are judged to have a temporal dependence on each other. Yet it is likely that goals and preconditions both influence each other sequentially (Hart & Honore, 1985; Hilton, 1 988; Lalljee & Abelson, 1983; Mandel & Lehman, 1 996; Read & Marcus-Newhall, 1 993; Spellman, 1997).

222

JOHN MCCLURE

One reason why preconditions are judged good explanations for extreme events may be that preconditions affect the likelihood of goals with these events. With difficult actions that require substantial preconditions, the pres­ ence of preconditions is likely to influence the emergence of goals. People who lack the money to buy a new Mercedes Benz are less likely to form the goal to buy one than those who have the means. By the same token, when a poor person who has never seriously thought of buying a new Mercedes wins a lottery, the acquisition of the precondition of money may enhance the goal to buy a Mercedes, or something else requiring a lot of money. The same principle applies with other preconditions, such as abilities. In contrast, pre­ conditions may have weaker effects on those cognitions that hold a less direct link to actions, such as fantasies. Goals and fantasies both represent a desire or aspiration, but fantasies have a weaker connection to the performance of actions (Bybee, Luthar, Zigler, & Merisca, 1997; Gollwitzer, 1993; Oettingen, 1 996). People may have a fantasy about a difficult action, such as buying a new Mercedes or winning an international award, even if they lack the precondi­ tions to complete the action. This idea can be related to the concept of equifinality, which proposes that with intentional actions, people circumvent obstacles to obtain the means necessary to achieve their goals (e.g., Heider, 1958; Kruglanski, 1996; Lalljee & Abelson, 1983). The present reasoning does not directly contradict this claim, but it does suggest that the prior availability of the preconditions necessary for an action affects the likelihood that the relevant goal will emerge. Most people don't generate the goal to walk across water, because people cannot walk across water. This reasoning implies that the relation between goals and preconditions (means) is more bidirectional than the rela­ tion covered by the concept of equifinality. Whereas the concept of equifinality proposes that the person with a goal will attempt to circumvent obstacles and obtain the means required to achieve his/her goal, the present reasoning implies that when people have no chance of obtaining the means to achieve a goal, they will not form the goal in the first place. However, there has been little research on this issue of the sequential effects of preconditions on goals and actions. Clearly, this issue needs more research. Diferent Criteria of Causation and Explanation

The research described in previous sections shows that different explanations reflect different causal questions, and that there are different criteria of causal influence and causal explanation. People's judgments of what makes a good explanation can be defined and measured in a number of ways. Many studies use unitary measures of causal judgments and treat these measures as ade­ quate reflections of people's judgment about the relation between causes and actions (e.g., Einhorn & Hogarth, 1986). The tendency to use a single measure

GOAL-BASED EXPLANATIONS

223

of causal influence is apparent in research and theory on causal discounting, where many analyses focus on single measures, such as the probability or sufficiency of contending causes (e.g., Morris & Larrick, 1995). However, there are different criteria for relations between cause and effects and there are different criteria for a good explanation, and it is important to distinguish between these criteria (McClure, 1 998) . For example, researchers should distinguish between perceived causes and communicated causes, or explanations. There is a difference between the issue of whether a cause is seen as contributing to an effect and the issue of whether a cause is included in an explanation. Communicated explanations may refer only to one cause, but it is possible that the person perceives that several causes contribute to the action. Where people select a single cause and apparently discount others, they may do so for reasons other than thinking that the other causes are weak or uncertain. Causes may be omitted from explanations because people see them as irrelevant or uninformative, and because people take the causes for granted, even though those causes contrib­ uted to the action. For example, when athletes who win a contest are asked to explain their win, they usually do not mention that they wanted to win, even though this goal is likely to have contributed to the outcome. The goal is taken for granted (Cheng & Novick, 1 991; Grice, 1 975; Hilton, 1 990; McClure & Hilton, 1 998; Schwarz, 1994, 1999). The cause that is presented in an explanation may be selected for a range of reasons. It may be the cause that distinguishes the event being explained from other similar situations, rather than being the primary determinant of the effect (Cheng & Novick, 1991; Hilton, 1 990). In addition, an explanation may include only the most recent or proximate cause that preceded an action, and omit more distal causes that are important links in a causal chain (Hart & Honore, 1985). Research on explanations needs to clarify the status of different causes, including causes that people discount on structured measures or omit from their explanations on unstructured measures. Research can then distinguish between causes that are seen as exerting little influence on an action, and causes that are seen as important, but omitted for some pragmatic reason. Research can best clarify the status of causal candidates by distinguishing judgments of probability, necessity, sufficiency, informativeness, and ex­ planatory value (Hilton & Erb, 1996; Hilton & Knott, 1996; Johnson et al., 1 994; McClure & Hilton, 1997, 1 998; McGill, 1990, 1 99 1 ; N'gbala & Branscombe, 1 995) . Necessity can be assessed by judgments of whether an action occurs when the cause is absent, whereas sufficiency can be assessed by judgments that the action always occurs when the cause is present. For example, one cause may be judged to be a sufficient explanation of an action, even when other causes are seen as important on other measures, such as probability and necessity. Informativeness can be assessed by

224

JOHN MCCLURE

judgments that the explanation is informative or relevant to explaining the action. McClure ( 1 998) proposed that if researchers examining explanations take account of these different measures, they are likely to produce co­ herent findings that are consistent with other findings on causal judgments and communication (Cheng & Novick, 1 991 ; Hart & Honore, 1985; Hilton, 1 990; Read & Marcus-Newhall, 1 993).

EXPLANATIONS AND FOLK PSYCHOLOGY

Theorists have recently taken up the idea of a folk psychology, which involves the systematic analysis of natural categories in people's perceptions of inten­ tions and intentional actions (e.g., Fletcher, 1995; Kashima et al., 1998; Malle, 1 999; Malle & Knobe, 1 997; Smedslund, 1997). This new interest in folk psychology echoes Heider's (1958) interest in intentional action and naive psychology, but it examines these domains with more systematic research. Malle and Knobe (1997) examined the features that people perceived as characterizing intentional actions, and showed that people view these actions as comprising two core elements: the desire to perform the action and the belief that a particular action would fulfill that desire. People judged that when both desire and belief are present the actor will have the intention to perform the action, but when either desire or belief is absent, the intention is unlikely to emerge. Further support for this model was obtained by Kashima et al. (1998), who gave participants scenarios in which either a desire or a belief was reversed (said to be absent) or substituted by an inappropriate option. As Malle and Knobe 's model predicts, only scenarios that included appropriate desires and beliefs made sense to participants. This finding also seems to be consistent with the notion of scripts proposed by the knowledge structure approach. Although beliefs and desires are central to folk psychology, Malle and Knobe (1997) claimed that lay people's idea of intentional actions is charac­ terized by two further elements: the actor's awareness of the desire-action relation, and the actor's possession of the skill required to perform the action. Malle (1999) more recently extended the model to include preconditions (or enabling factors), to take account of research showing that people cite precon­ ditions in explanations of intentional actions (e.g., McClure & Hilton, 1997). Malle claimed that the folk psychology of intention takes account of the preconditions that enable an action and that are necessary for the action. Malle claimed that, even though extreme achievements are intentionally brought about by the agent, people may invoke enabling factors rather than reasons to explain them. Enabling factors explain how, rather than why, inten­ tional actions or accomplishments occur (see also Ahn, Kalish, Medin, & Gelman, 1 995).

226

JOHN MCCLURE

People's judgments about goals and preconditions are also consistent with Hilton and Slugoski's (1986) abnormal conditions focus theory, which claims that explanations reflect whatever cause is most abnormal in the circum­ stances, and which has been applied to covariation-based judgments. For example, McClure and Hilton's (1997) finding that participants elevate pre­ conditions with obstructed actions is consistent with the abnormal conditions model. When actions previously have been obstructed by some precondition default, the occurrence of the action is abnormal by comparison with the previous occasions where it was obstructed. The abnormal conditions model has mostly been applied to the person-situation categories of explanation. However, the model also applies to goal-based explanations. There are also potential links between goal-based explanations and theories about correspondent inferences (Jones, 1979). Correspondent in­ ferences involve the judgment that an actor has a disposition that corresponds to a given action. As Hilton (1990) noted, the concept of category-based expectancies in the theory of correspondent inferences deals with similar contrasts to the abnormal conditions focus model. A category-based expec­ tancy is an expectancy that derives from the perceiver's knowledge that the target person belongs to a particular class or group (Jones & McGillis, 1976). According to correspondent inference theory, where an action deviates from that expectancy (i.e. it is abnormal), the perceiver is likely to make a corres­ pondent inference, and infer that the person has a disposition that corres­ ponds to the action. Applying this model to goal-based explanations, if goals are seen as similar to dispositional inferences, the model would suggest that, when an action deviates from category-based expectancies, people will attribute the action to the actor's goals. It is interesting that this prediction contrasts with McClure and Hilton's (1997) finding with actions that would seem to deviate from category-based expectancies, such as where a poor person buys a new expensive car. Participants explain these actions more in terms of preconditions becoming available than goals. Clearly there is a need for work on comparing and integrating these models. Despite these links between goal-based models and the more inferential attribution models, research suggests that knowledge structures and co­ variation information can have competing effects. Hilton and Knibbs (1988) examined the competing effects of knowledge structures and covariation in­ formation. They presented an action and two types of information: goal­ relevant information and covariation information (e.g., consensus information indicating how many other people performed an action). For example with the scenario, "Laura acts coolly on the date with Justin", the goal-relevant information was, "Laura knows she likes acting hard-to-get", and the con­ sensus information was, "almost everyone acts coolly on a date with Justin". The two classes of information either converged or conflicted in the causes that they implicated. For example, one condition presented goal-relevant

GOAL-BASED EXPLANATIONS

227

information implicating the person as a cause and covariational information implicating the situation as a cause. Participants then judged different causes of the effect. The results showed that goal-based information was not swamped by covariation information, and that people made slower and less confident judgments when there was a conflict between goal-based and covariational information. These findings suggest that goal-based and covariation information have different and sometimes conflicting effects. However, Sutton and McClure (2001) proposed that goal-based and co-variation approaches can be integ­ rated in a single model that overcomes the weaknesses of these two approaches. As noted earlier, covariational models deal with people's in­ ferences in terms of broad categories of explanation, such as a person explanation, but they fail to predict important distinctions between sub­ categories of explanation. For example, with the "person going to the restaur­ ant" scenario, the covariation model fails to distinguish between two different types of "person" explanation for the action: "she went to the restaurant because she is a foodie" and "she went to the restaurant because she was hungry". In addition, few researchers have applied covariational models to intentional actions and goals. Goal-based approaches, on the other hand, distinguish specific categories of explanations within a given causal locus and apply to intentional actions and goals. However, they fail to specify clearly when a given explanation is likely to be preferred over others. The integrative model proposes that goal-based explanations can be linked to the reasoning rules specified by the covariational approach. The model proposes that when people observe intentional actions, they infer antecedent events from their real-world beliefs. These potential explanations include motivating causes, such as goals, and enabling causes, such as preconditions. People then select the causes that co-vary with the action on dimensions such as consensus and distinctiveness, and offer these causes as explanations of the action. The model predicts not only when people invoke causes within the person, the situation, and the occasion, but also when people will invoke a stable cause ("he is always violent") rather than an unstable cause ("he wanted to defend himself"), and when people will use a motivating cause (a goal) rather than an enabling cause (a precondition) . Sutton and McClure (2001 ) demonstrated this model with scenarios that manipulated the consensus, consistency and distinctiveness of motivating and enabling causes. For example, one version of the "buying the car" scenario said that many were willing to buy a car (motivating cause, high consensus), but only the actor was able to buy the car (enabling cause, low consensus). The scenarios also gave information about the actor's and other people's purchases on other occasions (i.e. combinations of consensus and distinctive­ ness information). Participants then judged motivating and enabling causes as explanations of the action, and gave the explanations predicted by the

228

JOHN MCCLURE

integrative model. With this version of the "buying car" scenario, for example, participants attributed the action to an enabling cause. The integrative model also accounts for the finding that goals are used more for common events and preconditions more for extreme events (McClure, 1998; McClure & Hilton, 1997; McClure et al., in press). According to the model, the reason why people often use enabling causes, such as having an abnormal amount of money, to explain extreme actions is that there tends to be low consensus in the possession of those enabling causes. Sutton and McClure's research showed that with both extreme and moderate events, participants used enabling causes (preconditions) rather than motivating causes whenever the covariational information suggested that enabling causes were likely candidates. This finding contrasts with earlier goal-based models that predict the predominant use of goals as explanations, particularly for common actions (e.g., Leddo et a/., 1984; McClure & Hilton, 1997). The integrative model provides a conceptual bridge between goal-based accounts and covariation models of explanations, while retaining the strengths of goal-based models. This synthesis of goal-based and co-variational models shows how covariational information influences a range of goal-based explan­ ations, and when such information will favour the selection of goals and preconditions that are either stable or unstable. In sum, although the goal-based and inductive attributional approaches to explanations have a number of differences, there are potential links between them.

IMPLICATIONS AND APPLICATIONS

The present chapter has attempted to clarify the current state of theory and research on goal-based explanations of actions and outcomes. There is an interesting relation between research on lay people's explanations of actions and research on motivation and actions. The increase in research on goal­ based explanations parallels the expansion in research on actual actions, in terms of intentions, willpower, and goals (e.g., Gollwitzer, 1993; Gollwitzer & Bargh, 1996). The analysis of goal-based explanations examines lay persons' explanations of actions, and fits within the realm of "folk psychology". By comparison, the analysis of action examines factors that determine whether people's actual intentions lead to actions. Researchers need to take care not to confuse the two domains (Fletcher, 1995), but there is potential for researchers to exchange concepts between the two domains. For example, research on intentions and actions tends to focus on cognitive processes and mechanisms such as attention, the ability to delay gratification, and the importance of subgoals in implementing superordinate goals. These processes often determine whether a given intention is implemented in an

230

JOHN MCCLURE

The research described here also shows that the way people ask causal questions has a significant effect on the explanations that people give for actions. This research complements research showing the importance of ques­ tion framing in relation to attitudes (e.g., Schwarz, 1 99), and shows the importance of question framing in regard to explanations. This finding has implications for the researchers' design of questions intended to elicit explan­ ations, in that the type of causal questions used may favor different types of explanation. Researchers should be aware that their choice of causal ques­ tions is likely to shape the explanations that they obtain. The findings with causal questions also have implications for legal contexts and procedures. The framing of causal questions about intentions and actions may influence witnesses' and defendants' statements. For example, when a lawyer asks a defendant why he/she was in the place of a crime at a given time, the question implies that the defendant had a particular reason or motive for being in the vicinity of the crime (Graesser & Person, 1994). If the defendant gives an answer that implies that he/she had no particular motive for his/her presence, the answer may seem evasive, because the answer would violate the assumption implicit in the lawyer's question. So the effect of causal questions is not only a methodological issue; it also has applied implications. The present research also shows that people's explanations of goal-based actions depend on how their explanatory task is defined. For example, people may offer a different explanation if they are being asked what causes are necessary for an action to the explanation they would give if they are asked to explain what purpose an action has to a child or to an overseas tourist who doesn't understand the action. As the research here shows, the fact that someone thinks a cause is highly probable does not mean that the person thinks that the cause provides a good explanation of an action. So researchers should be aware that there are very different meanings to explanations and to different measures such as probability and informativeness that tap these different meanings. CONCLUSIONS

Most attribution theories have focused on the distinction between the person and the situation and inductive models of inference. However, recent years have seen the emergence of models that focus on people's perceptions of intentional actions: knowledge structure theories, folk psychology, and theory of mind. These theories have prompted research on goal-based explanations that focus on people's concrete concepts about actions and that link explana­ tions to real-world knowledge, rather than inductive or statistical reasoning. Initial research on goal-based theories focused on common actions and sug­ gested that people predominantly use goals as explanations, but recent

GOAL-BASED EXPLANATIONS

231

research clarifies where people use other explanations, such as preconditions, to explain actions. Explanations reflect the extremity of actions, the degree of obstruction facing actions, and the type of causal question that solicits the explanation. People's judgments of explanations are also influenced by prag­ matic factors, such as the relevance and informativeness of the explanations. Goal-based theories can be integrated with covariational theories, but this synthesis requires covariational theories to use goal-based categories of action, rather than abstract categories such as "the person" and "the situa­ tion''. There are useful potential links between the psychology of explanations and the psychology of action, but many of these links remain to be explored. ACKNOWLEDGMENTS

I thank Denis Hilton, Yoshihisa Kashima, Jim Liu, Bertram Malle, David Mandel, Michael Morris, Sik Hung Ng, Giin Semin, Ben Slugoski, and Robbie Sutton for valuable comments on the ideas in this manuscript. REFERENCES Abelson, R. P., & Lalljee, M. (1988). Knowledge structures and causal explanations. In D. J. Hilton (Ed.) Contemporary Science and Natural Explanation: Commonsense Conceptions of Causality. Brighton: Harvester. Abraham, C. (1988). Seeing the connections in lay causal comprehension: A return to Heider. In D. J. Hilton (Ed.), Contemporary Science and Natural Explanation: Com­ monsense Conceptions of Causality. Brighton: Harvester. Ahn, W. K., Kalish, C. W., Medin, D. L., & Gelman, S. A. (1995). The role of covaria­ tion versus mechanism information in causal attribution. Cognition, 54, 299-352. Antaki, C., & Leudar, I. (1992). Explanation in conversation: Towards an argument model. European Journal of Social Psychology, 22, 181-94. Buss, A. R. (1978). Causes and reasons in attribution theory: a conceptual critique. Journal of Personality and Social Psychology, 36, 131 1-21. Bybee, J., Luthar, S., Zigler, E., & Merisca, R. (1997). The fantasy, ideal, and ought selves: Content, relationships to mental health, and functions. Social Cognition, 15, 37-53. Cheng, P. W., & Novick, L. R. (1991). Causes versus enabling conditions. Cognition, 40, 83-120. Cross, S., & Markus, H. (1991). Possible selves across the life-span. Human Develop­ ment, 54, 230-55. Einhorn, H., & Hogarth, R. (1986). Judging probable cause. Psychological Bulletin, 99, 3-19. Fiedler, K., Walther, E., & Nickel, S. (1999). Covariation-based attribution: On the ability to assess multiple covariates of an effect. Personality and Social Psychology Bulletin, 25, 607-22. Fiske, S. T. (1998). Heider's pragmatism today: Perceiving is for doing. Paper pre­ sented at the Society of Personality and Social Psychology, Washington, DC. Fletcher, G. (1995). The Scientific Credibility of Folk Psychology. Mahwah, NJ: Erlbaum.

232

JOHN MCCLURE

Forsterling, F. (1989). Models of covariation and attribution: How do they relate to the analogy of analysis of variance. Journal of Personality and Social Psychology, 51, 615-25. Girotto, V., Legrenzi, P., & Riz, A. (1991 ). Event controllability in counterfactual thinking. Acta Psychologia, 18, 1 1 1-33. GoUwitzer, P. M. (1993). Goal achievements: The role of intentions. In W. Stroebe & M. Hewstone (Eds), European Review of Social Psychology (Vol. 4, pp. 141-85). Chichester: Wiley. GoUwitzer, P. M., & Bargh, J. H. (1996). The Psychology of Action: Linking Cognition and Motivation to Behavior. New York: Guilford. Graesser, A. C., Robertson, S. P., & Anderson, P. A. (1981 ). Incorporating inferences in narrative representations: A study of how and why. Cognitive Psychology, 13, 1-26. Graesser, A. C., Robertson, S. P., Lovelace, E. R., & Swinehart, D. M. (1980). Answers to why-questions expose the organization of story plot and predict recall of actions. Journal of Verbal Learning and Verbal Behavior, 19, 1 10--19. Graesser, A. C., Langston, M. C., & Baggett, W. B. (1993). Exploring information about concepts by asking questions. The Psychology of Learning and Motivation, 29, 41 1-36. Graesser, A. C., & Person, N. K. (1994). Question asking during tutoring. American Educational Research Journal, 31, 104-37. Grice, H. P. (1975). Logic and conversation. In P. Cole & J. L. Morgan (Eds), Syntax and Semantics 3: Speech Acts (pp. 41-58). New York: Academic Press. Hart, H. L. A., & Honore, T. (1985). Causation in the Law (2nd Edn). Oxford: Claren­ don Press. Heider, F. (1958). The Psychology of Interpersonal Relations. New York: Wiley. Hilton, D. J. (1988). Logic and causal attribution. In D. J. Hilton (Ed.), Contemporary Science and Natural Explanation: Commonsense Conceptions of Causality. Brighton: Harvester. Hilton, D. J. (1990). Conversational processes and causal explanation. Psychological Bulletin, 107, 65-81. Hilton, D. J. (1991). A conversational model of causal explanation. In W. Stroebe & M. Hewstone (Eds), European Review of Social Psychology (Vol. 2, pp. 31-50). Chichester: Wiley. Hilton, D. J. (1995). The social context of reasoning: Conversational inference and rational judgment. Psychological Bulletin, 118, 248-7 1 . Hilton, D. J . , & Erb, H.-P. (1996). Mental models and causal explanation: Judgments of probable cause and explanatory relevance. Thinking and Reasoning, 2, 273-308. Hilton D. J., & Knibbs, C. S. (1988). The knowledge-structure and inductivist strategies in causal attribution: A direct comparison. European Journal of Social Psychology, 18, 79-92.

Hilton, D. J., & Knott, I. C. (1996). Explanatory relevance: Pragmatic constraints on the selection of causes from conditions (unpublished manuscript). Hilton, D. J., & Slugoski, B. R. (1986). Knowledge-based causal attributions: The abnormal conditions focus model. Psychological Review, 93, 75-8. Hume, D. (1888). A Treatise on Human Nature (L. A. Selby-Bigge, Ed.). London: Oxford University Press (original published 1739). Jones, E. E. (1979). The rocky road from acts to attributions. American Psychologist, 34, 107-17. Jones, E. E., & McGillis, D. (1976). Correspondent inferences and the attribution cube: A comparative appraisal. In J. H. Harvey, W. J. Ickes & R. F. Kidd (Eds), New Directions in Attribution Research (Vol. 1 ) Hillsdale, NJ: Erlbaum. .

GOAL-BASED EXPLANATIONS

233

Johnson. J., Boyd, K., & Magnani, P. (1994). Causal reasoning about rare and common events. Journal of Personality and Social Psychology, 66, 229-42. Kashima, Y., McKintyre, C., & Clifford, P. (1998). The category of mind: Folk psychol­ ogy of belief, desire, and intention. Asian Journal of Social Psychology, 1, 289-313. Kelley, H. H. ( 1 967). Attribution theory in social psychology. Nebraska Symposium on Motivation, 15, 192-238. Kelley, H. H. (1972) . Causal schemata and the attribution process. In E. E. Jones, D. E. Kanouse, H. H. Kelley, R. E. Nisbett, S. Valins & B. Weiner (Eds), Attribution: Perceiving the Causes of Behavior (pp. 151-74). Morristown, NJ: General Learning Press. Kruglanski, A. W. (1996). Goals as knowledge structures. In J. A. Bargh & P. Goll­ witzer (Eds) Motivation and Action. New York: Guilford. Kun, A., & Weiner, B. (1973). Necessary versus sufficient causal schemata for success and failure. Journal ofResearch in Personality, 7, 197-207. Lalljee, M., & Abelson, R. P. (1983). The organization of explanations. In M. Hewstone (Ed.), Attribution Theory: Social and Functional Extensions. Oxford: Basil Blackwell. Leddo, J., & Abelson, R. P. (1986) . The nature of explanations. In J. Galambos, R. P. Abelson & J. B. Black, (Eds), Knowledge Structures (pp. 103-22). Hillsdale, NJ: Erlbaum. Leddo, J., Abelson, R. P., & Gross, P. H. (1984). Conjunctive explanations: When two reasons are better than one. Journal of Personality and Social Psychology, 47, 93343. Locke, D., & Pennington, D. (1982). Reasons and causes in attribution processes. Journal of Personality and Social Psychology, 42, 212-33. Malle, B. F. ( 1999). How people explain behavior: A new theoretical framework. Personality and Social Psychology Review, 3, 23-48. Malle, B. F., & Knabe, J. (1997). The folk concept of intentionality. Journal of Experi­ mental Social Psychology, 33, 171-89. Malle, B. F., & O'Laughlin, M. (1998). The study of free-response behavior explana­ tions: Theory and method (paper in preparation). University of Oregon. Mandel, D. R., & Lehman, D. R. (1996). Counterfactual thinking and ascriptions of cause and preventability. Journal of Personality and Social Psychology, 71, 450-63. McClure, J. L. ( 1983). Telling more than they can know: The positivist account of verbal reports and mental processes. Journal for the Theory of Social Behaviour, 13, 1 1 1-27. McClure, J. L. (1984). On necessity and commonsense: A discussion of central axioms in new approaches to lay explanation. European Journal of Social Psychology, 14, 123-50. McClure, J. L. (1985). The social parameter of "learned" helplessness: Its recognition and implications. Journal of Personality and Social Psychology, 48, 1534-9. McClure, J. L. (1991 ). Explanations, Accounts and Illusions: A Critical Analysis. Cambridge: Cambridge University Press. McClure, J. L. (1992). An economy of explanations. In M. L. McLaughlin, M. J. Cody & S. J. Read (Eds), Explaining One's Self to Others: Reason-giving in a Social Context (pp. 61-82). Hillsdale, NJ: Erlbaum. McClure, J. L. (1998). Discounting causes of behavior: Are two reasons better than one? Journal of Personality and Social Psychology, 74, 7-20. McClure, J. L., Densley, L., Liu, J. H., & Allen, M. (2001). Constraints on equifinality: Goals are good explanations only for controllable outcomes. British Journal of Social Psychology, 40, 99-115. McClure, J. L., & Hilton, D. J. ( 1 997). For you can't always get what you want: When preconditions are better explanations than goals. British Journal of Social Psychol­ ogy, 36, 223-40.

234

JOHN MCCLURE

McClure, J. L., & Hilton, D. J. (1998). Are goals or preconditions better explanations? It depends on the question. European Journal of Social Psychology, 28, 897-91 1 . McClure, J . L., Hilton, D. J., Cowan, J., Ishida, L . & Wilson, M . (2001 ). When rich or poor people buy expensive objects: Is the question how or why? Journal of Lan­ guage and Social Psychology, 20, 339-57. McClure, J. L., Lalljee, M., & Jaspars, J. (1991). Explanations of moderate and extreme events. Journal of Research in Personality, 25, 146-6. McClure, J. L., Lalljee, M., Jaspars, J., & Abelson, R. P. ( 1989). Conjunctive explana­ tions of success and failure: The effect of different types of causes. Journal of Personality and Social Psychology, 56, 19-26. McGill, A. L. (1990). Conjunctive explanations: The effect of comparison of the target episode to a contrasting background instance. Social Cognition, 8, 362�. McGill, A. L. (1991). The influence of the causal background on the selection of causal explanations. British Journal of Social Psychology, 30, 7�7. McGill, A. L., & Klein, J. L. (1993). Contrastive and counterfactual reasoning in causal judgment. Journal of Personality and Social Psychology, 64, 897-905. Mill, J. S. (1973). "System of logic", 8th Edn. In J. M. Robson (Ed.), Collected Works of John Stuart Mill, Vols 7 and 8. Toronto: University of Toronto Press (Original work published 1872). Morris, M. W., & Larrick, R. P. (1995). When one cause casts doubt on another: A normative analysis of discounting in causal attribution. Psychological Review, 102, 331-5. N'gbala, A., & Branscombe, N. R. (1995). Mental simulation and causal attribution: When simulating an event does not affect fault assignment. Journal of Experimental Social Psychology, 31, 139-62. Oettingen, G. (1996). Positive fantasy and motivation. In P. W. Gollwitzer & J. A. Bargh (Eds), The Psychology ofAction: Linking Cognition and Motivation to Action (pp. 236-59). New York: Guilford. Oyserman, D., Ager, J., & Gant, L. (1995). A socially contextualised model of African American identity: Possible selves and school persistence. Journal ofPersonality and Social Psychology, 69, 1216-32. Passer, M. W., Kelley, H. H., & Michela, J. L. (1978). Multidimensional scaling of the causes for negative interpersonal behavior. Journal of Personality and Social Psy­ chology, 36, 951-62. Read, S. J. (1987). Constructing causal scenarios: A knowledge structure approach to causal reasoning. Journal of Personality and Social Psychology, 52, 28302. Read, S. J. (1988). Conjunctive explanations: The effect of a comparison between a chosen and a non-chosen alternative. Journal of Experimental Social Psychology, 24, 146-62. Read, S. J., & Marcus-Newhall, A. (1993). Explanatory coherence in social explana­ tions: A parallel distributed processing account. Journal of Personality and Social Psychology, 65, 429-47. Reeder, G. D., & Brewer, M. B. (1979). A schematic model of dispositional attribution in interpersonal perception. Psychological Review, 86, 61-79. Rips, L. J. (1998). Reasoning and conversation. Psychological Review, lOS, 41 1-41. Roese, N. J., & Olson, J. M. (1995). Outcome controllability and counterfactual think­ ing. Personality and Social Psychology Bulletin, 21, 62�28. Schank, R. C., & Abelson, R. (1977). Scripts, Plans, Goals, and Understanding. Hillsdale, NJ: Erlbaum. Schwarz, N. (1994). Judgment in a social context: biases, shortcomings and the logic of conversation. Advances in Experimental Social Psychology, 26, 123-62. Schwarz, N. (1999). Self-reports: How the questions shape the answers. American Psychologist, 54, 93-105.

GOAL-BASED EXPLANATIONS

235

Semin, G. R., & Fiedler, K. ( 1991). The linguistic category model, its bases, applica­ tions and range. European Review of Social Psychology, 2, 1-30. Skinner, E. A., Chapman, M., & Baltes, P. B. (1988). Control, means-ends, and agency beliefs: A new conceptualization and its measurement during childhood. Journal of Personality and Social Psychology, 54, 1 17-33. Smedslund, J. ( 1997). The Structure of Psychological Common Sense. Mahwah, NJ: Erlbaum. Spellman, B. A. (1997). Crediting causality. Journal of Experimental Psychology: Gen­ eral, 126, 323-48. Sutton, R., & McClure, J. L. (2001). Covariational influences on goal-based explana­ tions: An integrative model. Journal of Personality and Social Psychology, 80, 22236. Strack, F., Schwarz, N., & Waenke, M. (1991). Semantic and pragmatic aspects of context effects in social and psychological research. Social Cognition, 9, 1 1 1-25. Turnbull, W. ( 1 986). Everyday explanation: The pragmatics of puzzle resolution. Jour­ nal for the Theory of Social Behaviour, 16, 141-60. Tversky, A., & Kahneman, D. ( 1983). Extensional versus intuitive reasoning: The conjunction fallacy in probability judgment. Psychological Review, 90, 293-315. Weiner, B. (1986). An Attributional Theory of Motivation and Emotion. New York: Springer-Verlag. Wellman, H. M. (1990). The Child's Theory of Mind. Cambridge, MA: MIT Press. White, P. A. (1992). Causal powers, causal questions, and the place of regularity information in causal attribution. British Journal of Psychology, 83, 161-88. Wilensky, R. W., (1983). Planning and Understanding: A Computational Approach to Human Reasoning. Reading, MA: Addison-Wesley.

This page intentionally left blank

238

CONSTANTIN E SEDIKIDES ET A L.

favorable impression of the partner; and (b) expecting attributional generosity from the partner. In fact, when the partner violates this expectancy (i.e., when he/she displays the SSB ), members of close dyads respond by manifesting the SSB in tum. We discuss these and several other contingencies that are likely to keep an individual's self-enhancement tendencies in check.

Individuals enhance the self in diverse and remarkable ways. They consider themselves more moral, trustworthy, kind, and physically attractive than oth­ ers. They rate themselves as above-average teachers, managers, and leaders. They also believe that they are happier than others, that they are likely to be healthier and live longer than others, and that they are more likely than others to experience positive life events but less likely than others to experience negative life events. They even believe they are better drivers! Such overblown self-evaluations are well documented (Sedikides & Strube, 1 997). These beliefs are maintained through several mechanisms. One mechanism is biased memorial processes, such as better memory for positive than negative self-attributes (Skowronski, Betz, Thompson, & Shannon, 1991 ) and for feedback pertaining to one's strengths rather than one's weaknesses (Sedikides & Green, 200). Another mechanism is the selective reconstruction, generation, and evaluation of confirming causal theories (Kunda, 1990; Ross, 1989). A third mechanism is the idiosyncratic (i.e., favorable to the self) definition of traits and abilities (Dunning, 1993). Other mechanisms include denial (Janoff-Bulman & Timko, 1987), psycho­ logical distancing from others (Schimel, Pyszczynski, Greenberg, O'Mahen, & Arndt, 200), affirmation of a self-domain that is unrelated to the self­ domain under threat (Steele, 1988), downward social comparison (Wills, 1 981), favorable self-presentation (Schlenker & Pontari, 200), and self­ favoring causal attributions for success and failure - an attributional pattern known as the self-serving bias (SSB). The SSB is the individual's propensity to make internal attributions for success, but external attributions for failure. Stated otherwise, the SSB refers to the individual's taking responsibility for successful task outcomes, but denying responsibility, or blaming other persons or circumstances, for failed task outcomes. For example, students will take credit for passing a difficult examination, but will attribute failing the same examination to its difficulty or the instructor's tough grading policy; business partners working on an account will claim disproportionate credit if the account is gained, but will avoid personal responsibility if the account is lost; and group members will over­ emphasize their individual contribution on a group project, but will blame other members for the group's inferior output. The SSB is pervasive, and is one of the most robust and easily replicable phenomena in social and person­ ality psychology (for both narrative and quantitative reviews, see Arkin, Cooper, & Kolditz, 1980; Campbell & Sedikides, 1999; Mullen & Riordan, 1988; Weary""Bradley, 1978; Zuckerman, 1979).

HOW RELATIONSHIPS CHANGE THE SELF

239

Imagine that George and Alexi, two perfect strangers, find themselves in an interesting predicament as rookie university students. In their first intro­ ductory psychology class session, the instructor invites them to complete a short assignment. She asks one of them to generate a list of as many "why" questions as possible, and asks the other student to generate a list of as many "because" answers as possible. The instructor explains that the two contes­ tants will work independently of one another. However, the outcome of the assignment is interdependent, as the rest of the class will be given the two lists and will judge how imaginative and humorous each why-because com­ bination is. George and Alexi work fervently for a few minutes. Then, the moment of truth arrives: the two helpless contestants learn that their joint output is poor and disappointing. How will they respond? Based on the robustness of the SSB, one would expect for them to blame (at least pri­ vately) each other for the dyadic failure. Now, consider the scenario in which the contestants are told that their collaborative exercise produced highly imaginative and humorous question-and-answer items. How will they respond in this case? Again, based on extant literature, one would expect them to claim privately disproportionate personal credit for the success of the dyad. The situation, though, becomes less predictable if we assume that George and Alexi are high-school friends. Would the outcome of the assignment (i.e., success vs. failure feedback) determine whether each person displaced responsibility for the failure of the dyad or claimed credit for its success? Would their friendship stop George and Alexi from being selfish, how so, and when? These are the questions that sparked the present investigation. More formally, the scope of our research is the role of close relationships in self-enhancement. The central issue with which we are concerned is whether close relationships attenuate self-enhancement tendencies. Do close relation­ ships put the self "in its place"? How so? When does this relational function (i.e., curtailing individual self-enhancement) break down? We will focus on the SSB as a traditional and well-validated signature of self-enhancement. Rephrasing our research questions, we want to know: (a) whether relation­ ship closeness reduces the SSB; (b) what are the mechanisms by which this reduction occurs; and (c) when relational closeness attenuates or accentuates the SSB. THE EFFECT OF RELATIONSHIP CLOSENESS ON THE SSB : HYPOTHESES

We propose two competing hypotheses that address the role of relationship closeness in the SSB: the relationships-as-bound hypothesis and the relationships-as-enabler hypothesis.

240

CONSTANTINE SEDIKIDES ET A L.

The Relationships-as-bound Hypothesis Several broad theoretical perspectives offer a sound rationale for the hypoth­ esis that close relationships will reduce self-enhancement, or that closely related partners will not manifest the SSB, We labeled this hypothesis the

relationships-as-bound hypothesis. According to balance theory (Heider,

1958),

attitudes toward the self extend

to close others. Self-expansion theory (Aron, Aron, Tudor,

&

Nelson,

1991 )

posits that the self-concept expands to incorporate a close partner ("the you in me", according to Hornstein,

1976). &

natural category (Sedikides, Olsen,

Oose others are indeed perceived as a Reis,

1993).

Interdependence theory and

the communal-exchange relationships literature maintain that an important goal in close relationships is the maximization of outcomes for both partners

1997), with partners being genuinely concerned for each & Mil, 1979), and being trustful (Holmes & Rempel, 1989; Wieselquist, Rusbult, Foster, & Agnew, 199), committed (Rusbult, 1983), and even sacrificial (Van Lange, Rusbult, Drigotas, Arriaga, Witcher, & Cox, 1997) (Rusbult

&

Arriaga,

other (Oark

to each other. In a similar vein, the extended self-evaluation maintenance model (Beach

&

Tesser,

1995) proffers that

close individuals are motivated to

protect not only their own but also their partner's self-concept. Indeed, as suggested by Sedikides and Strube's

(1997)

self-concept enhancing tactician

(SCENT) model, one important function of close relationships is to keep an individual's self-enhancing tendencies in check. Studies of self-presentational

& Stillwell, 1995) are consistent with this suggestion. If the need to belong is a fundamental human motivation (Baumeister & Leary, 1995), individuals will be willing to

modesty among closely related partners (Tice, Butler, Muraven,

relegate self-interest in exchange for (perceived or real) relational benefits. All of these theoretical statements converge in advocating the notion that, in close dyadic relationships, one will be as likely to protect or enhance the partner as one will be to protect or enhance the self. Close individuals will share the responsibility for a task outcome with their partner, regardless of whether the dyadic outcome is successful or unsuccessful. The SSB will be absent in close relationships.

The Relationships-as-enabler Hypothesis As a counterpoint, an equally broad and diverse body of literature emphasizes the view that close relationships serve a self-protective or self-enhancing role. Close relationships will augment self-enhancement, or enable closely related partners to display the SSB. We labeled this hypothesis the

relationships-as­

enabler hypothesis. Individuals are rather inaccurate in how they think they are viewed by specific related others (Kenny

&

DePaulo,

1993),

perhaps because others do

242

CONSTANTINE SEDIKIDES ET A L .

As stated previously, the SSB is an empirically robust phenomenon. Parti­ cipants manifest the SSB in individual (Miller, 1976) and group (Schlenker & Miller, 1 977) settings. Also, participants manifest the SSB in both interperso­ nal influence and skills-orientated tasks (Weary-Bradley, 1978). Interpersonal influence tasks include two major paradigms. One is the teacher-student para­ digm, in which participants assigned the role of the teacher report their influ­ ence on ostensibly successful or unsuccessful (fictitious) students (Beckman, 1970). That is, participants are asked to take responsibility for their "stu­ dent's" progress, or lack of it, on an instructional assignment. The second is the therapist-patient paradigm, in which participants assigned the role of the therapist report their level of contribution to the successful or unsuccessful therapeutic course of a patient (actually, a confederate) (Arkin, Gleason, & Johnston, 1976). Skills-orientated tasks are classified under two categories. The first category includes independent outcomes tasks. In these settings, participants work on a task independently (Luginbuhl, Crowe, & Kahan, 1 975). The second category includes dyadic interdependent outcomes tasks. In our research, we were concerned exclusively with dyadic interdependent outcomes (or "pooled interdependence") tasks. Such tasks involve collabora­ tion between participants. Success or failure of the dyad depends on the joint rather than unique contribution of the members. Given the dyad's common fate, several factors-such as interaction, effort coordination, and a good working relationship between partners-constitute important prerequisites for an optimal task outcome. Explanations for the SSB have been the topic of considerable debate in social and personality psychology. One early review proposed a cognitive explanation-more specifically, differential access to information due to selective attention and information accessibility in memory (Miller & Ross, 1 975). For example, individuals display the SSB because they restrict their attention to information readily available to them, failing to realize that this information is incomplete. Another explanation advocated the role of mo­ tives. Individuals manifest the SSB because they are motivated to think positively or avoid thinking negatively of the self (Weary-Bradley, 1978). Despite early pessimism about the conclusiveness of the cognition versus motivation debate (Tetlock & Levi, 1 982), there is now consensus that cogni­ tive explanations alone cannot account fully for the SSB. Instead, motiva­ tional reasons have emerged as a sufficient explanation for the SSB (Campbell & Sedikides, 1 999; Zuckerman, 1979). In our research, we are interested in the motivational underpinning of the SSB. Indeed, as we stated previously, we regard the SSB as a valid signature of the motivation to enhance the self. For this reason, we sought to outline clearly the features of an adequate test for the motivationally-based emergence of the SSB in dyadic interdependent tasks. This test would minimize the contributions of cognitive factors and

HOW RELATIONSHIPS CHANGE THE SELF

243

maximize the contributions of motivational factors. Such a test would need to satisfy the following four procedural criteria: (a) members of the dyad ought to be separated (i.e. seated in different rooms), so that they are unable to interact during the completion of the experimental task; (1?) the dyad ought to complete a relatively unfamiliar task; (c) each dyad member ought to receive false and randomly determined success or failure feedback at the dyadic rather than individual level; and (d) each dyad member ought to attribute privately the task outcome to the self or the partner. These procedural criteria arguably limit cognitive explanations for the ob­ tained SSB. One reason for this is the timing of the attributions: the attribu­ tions of the dyad members follow the temporally immediate rather than distant completion of a task, thus effectively neutralizing memorial biases (i.e., distortions or intrusions). Another reason is that the unambiguous ap­ praisal of the magnitude of own contribution to the task is virtually imposs­ ible, given that the task is novel and the feedback is at the dyadic level. A third and final reason is that the members of the dyad are unaware of the quality of each other's task performance. In our literature review, then, we searched for experiments whose pro­ cedures satisfied the four criteria named above. Both in the review and our research we were concerned with relativistic attributions, that is, attributing the task outcome, on a continuum, either to the self or the partner. Our choice of relativistic attributions was theory-guided: The relationships-as-bound and relationships-as-enabler hypotheses make relativistic predictions, which would be tested best by relativistic attributions. We focused on two types of dyadic relationships: close (i.e., friends) and distant (i.e., strangers). We pres­ ent the relevant research below. The SSB in Interdependent Outcomes Tasks: The Case of Close Dyads

Several studies have examined the presence of the SSB in close dyads (Chris­ tensen, Sullaway, & King, 1983; Fincham & Bradbury, 1989; Maass & Vol­ pato, 1988; Ross & Sicoly, 1979; Thompson & Kelley, 1 981). Some of these studies have obtained evidence for the SSB, others have found a reversal of the SSB (what we term the other-serving bias or OSB), and others have reported null findings. We will not discuss the results of these studies in detail, because two of their procedural features fall short of satisfying our review criteria. These studies allowed for memorial distortions or intrusions because they tested particip­ ants' recall of past interdependent activities, rather than participants' im­ mediate appraisal of their recently completed task performance. More importantly, these studies did not include a manipulation of success or failure performance feedback. Hence, prior to conducting our own research, we found no published studies that tested conclusively whether close participants

244

CONSTANTINE SEDIKIDES ET A L .

working on dyadic interdependent tasks display a motivationally-based SSB. We found no evidence in favor of or against the relationships-as-bound and the relationships-as-enabler hypotheses.

The SSB in Interdependent Outcomes Tasks: The Case of Distant Dyads Three experiments have examined the presence of the SSB in distant dyads that are involved in interdependent outcomes tasks. These experiments were reported by Johnston ( 1 967) and Wolosin

et a/.

( 1973, Experiments 1 and

2).

Participants in Johnston's ( 1 967) experiment believed (falsely) that they worked on a tracking task with a partner who was in a neighboring room. The task required holding a moving cursor steady at zero by manipulating a con­ trol knob. Participants repeated this task in three experimental sessions, each of which consisted of 20 trials, each of 60 seconds duration. Then, participants received randomly determined and bogus success or failure feedback. The results revealed an OSB: participants who received success feedback tended to give less credit to the self than the partner, whereas those who received failure feedback accepted more responsibility for the task outcome. Participants in Wolosin

et al. 's

( 1 973) research either cooperated (Experi­

ment 1 ) or competed (Experiment from each of

20

2) on a task. The task consisted of choosing

geographic location pairs the location in which participants

were more likely to reunite with a friend. At the conclusion of the task, participants received randomly determined success or failure feedback. Co­ operative participants displayed the SSB: they accepted individual respon­ sibility for the dyadic success, but blamed the partner for the dyadic failure. Likewise, competitive participants accepted individual responsibility for dyadic success, but blamed the self and partner to an equal extent for dyadic failure. In all, the evidence for the presence of the SSB in dyadic interdependent

et a/., et a/., Experiment 2, success condition), the OSB

outcomes tasks is weak. The SSB was present in two cases (Wolosin 1973, Experiment 1 ; Wolosin

was obtained in another case (Johnston, 1967), and null findings were re­ ported in still another case (Wolosin ment

2).

et al.,

1 973, failure condition of Experi­

How can this inconsistency be accounted for? We maintain that a

reason for the weak evidence for the SSB is task importance. The SSB inten­ sifies as task importance and a looming threat to the self increase (Campbell

& Sedikides,

1999). We contend that Johnston's ( 1 967) tracking task-a task

that produced an OSB-was probably perceived by participants as a low­ importance task. Our research controlled for differential task importance by standardizing it. We used a task that was a face valid measure of creativity, an attribute that university students, our targeted population, consider important (Amabile, 1983). We believed that the introduction of an important task, coupled with

HOW RELATIONSHIPS CHANGE THE SELF

245

the adoption of the four critical procedural features outlined above, would allow us to demonstrate a motivationally-based SSB in dyads consisting of distant partners. HYPOTHESIS TESTING

We conducted several experiments in an effort to test the viability of the relationships-as-bound and relationships-as-enabler hypotheses. We began by asking the question of whether relationship closeness reduces the SSB. We proceeded by addressing the issue of how relationship closeness might reduce the SSB. We concluded by examining when relationship closeness reduces or augments the SSB. Does Relationship Closeness Reduce the SSB?

We attempted to address the question of whether relationship closeness re­ duces the SSB in an experiment conducted by Campbell, Sedikides, Reeder, and Elliot (200). One objective of this experiment was to test whether the SSB emerges in distant dyads. In the absence of a relational bond and in light of the robustness of the SSB, we expected such dyads to display the SSB. We predicted that participants whose dyad received success feedback would regard the self as more responsible for the task outcome than the partner, whereas participants whose dyad received failure feedback would blame the partner more than the self for the task outcome. The primary objective of this experiment, though, was to find out whether the SSB emerges in close dyads. We were interested in comparing the attribu­ tional pattern of close dyads with that of distant dyads. Suppose that close participants were equally likely to assume personal responsibility for the dyadic success and its failure (compared to distant dyads). This pattern would be consistent with the relationships-as-bound hypothesis. Such a finding would demonstrate that relationship closeness poses limits on self­ enhancement. On the other hand, suppose that close participants outdid each other in taking responsibility for the dyadic success rather than its failure (compared to distant dyads). This pattern would be in line with the relationships-as-enabler hypothesis. Such a finding would show that relation­ ship closeness creates self-enhancement tendencies. We began by examining the presence of the SSB among strangers and among friends. Half of the participants reported to the laboratory with a friend, whereas the other half reported with a stranger. The experimenter placed each dyad member in a separate room, in which he/she stayed for the rest of the experiment. Not surprisingly, a manipulation check indicated that friends reported a higher degree of closeness than strangers.

246

CONSTANTINE SEDIKIDES ET A L.

For exploratory purposes, we asked participants to record their mood state. Next, participants learned that they would take the "Lange-Elliot Creativity Test", which was described as a well-validated test. Participants were in­ structed that the purpose of the experiment was to study how brainstorming affects dyadic creativity, Brainstorming was defined ("coming up with cre­ ative ideas in a group setting") and operationalized for the purposes of the experiment ("coming up with as many uses for an object as you can"). Parti­ cipants were informed further that they had been assigned to the control group and, as such, they would not interact with their partner. Specific instructions regarding the completion of the Lange-Elliot Creativity Test followed. Participants were told that they and their partner would be presented with the names of two objects, one at a time. The objects were "brick" and "candle". Participants would have 5 minutes to generate individually as many functionally distinct uses as possible for each object. They were to write each object use on a slip of paper, fold the paper in half, and drop it in the box next to them. In the end, the experimenter would place the slips in another box. Then, the experimenter would sum up the total number of functionally distinct and non-common uses that the two particip­ ants had generated. This sum would constitute the overall creativity score for the dyad. The experimenter could only know the dyadic, not the individual, creativity score. After rating the importance of creativity to them (creativity was equally important to friends and strangers), participants completed the test. Subsequently, participants received bogus and randomly determined suc­ cess or failure feedback at the dyadic level. The feedback was administered both graphically and verbally. The graphic form included a bell-shaped histo­ gram, in which the vertical axis was labeled "count" (representing the com­ bined total number of object uses) and the horizontal axis was labeled "Z­ score". Each participant in the success feedback condition was shown a mark at the 93rd percentile, was told "You did well", and was further informed, "Your and the other participant's combined score on the Lange-Elliot Creativity Test was at the 93rd percentile. You scored better than 93% of the individuals used in our normative reference sample". In contrast, each parti­ cipant in the failure feedback condition was shown a mark at the 31st percen­ tile, was told "You did poorly", and was further informed that, "Your and the other participant's combined score on the Lange-Elliot Creativity Test was at the 31st percentile. You scored worse than 69o/o of the individuals used in our normative reference sample". Participants were also asked to initial the feed­ back as evidence that they understood it. The completion of the dependent measures followed. Participants were assured that their responses to the scales would be anonymous and con­ fidential, and that they would not interact with their partner at the end of the experiment. The two key dependent measures were preceded by a statement

247

HOW RELATIONSHIPS CHANGE THE SELF

that the experimenters were "unable to determine which of you was more responsible for the overall positive or negative results obtained by the pair". The two measures were as follows: "Who was more responsible for the out­ come of this test?" and "Who made the greater positive contribution to this test?". Finally, participants filled out a success and failure feedback manip­ ulation check, composed of two questions: "How well do you think that both you and the other participant did on this test?", and 'How important was the outcome of this test to you?". The feedback manipulation was effective. Dyads who received success feedback reported having performed better than dyads who received failure feedback. Additionally, participants considered the creativity test more im­ portant after receiving success than failure feedback; that is, participants com­ pensated for their failure by downgrading the importance of the test (Wyer & Frey, 1983) . We display the results i n Table 8.1 . Participants in distant relationships displayed the SSB: they assumed greater personal responsibility for the suc­ cess of the dyad than its failure. This pattern is consistent with our prediction and demonstrates the motivational significance of the SSB. Interestingly, however, participants in close relationships did not manifest the SSB: they assumed equivalent levels of personal responsibility for the dyad's success and failure. As a conceptual replication of the above findings, close participants claimed a less positive contribution to the test outcome than distant particip­ ants. Thus, close participants refrained from the SSB. These findings render support to the relationships-as-bound hypothesis at the expense of the relationships-as-enabler hypothesis. As a reminder, we had assessed participants' mood states upon arrival at the laboratory. Friends were in a happier mood than strangers. However, Attributions as a function of feedback and relationship closeness in Campbell, Sedikides, Reeder, and Elliot (2000)

Table 8.1

I.

Task outcome question Distant dyads Close dyads

II.

Failure feedback 5.0la 5.45a

Positive contribution question Distant dyads 6.16a

Success feedback 6.22b 5.73a Close dyads 5 .52b

Note 1: Responses to the task outcome question were made on a 1 (the other participant was more responsible) to 10 (I was more responsible) scale; re sponses to the positive contribution question were made on a 1 (the other participant made the greater positive contribution) to 10 (I made the greater positive contribution) scale. Note 2: Within a row, means with different superscripts are significantly different.

248

CONSTANTINE SEDIKIDES ET A L.

these mood differences did not qualify the findings. Additionally, the findings generalized over participant gender.

How Does Relationship Closenes Reduce the SSB? In Search of Mechanisms What are the mechanisms by which relationship closeness reduces the SSB? We addressed this issue in two experiments reported by Sedikides, Campbell, Reeder, and Elliot (1998).

Participant Selection and Anticipated Interaction as Explanations for the Relationship-as-bound Hypothesis The Campbell et a/. (200) experiment showed that close partners keep self­ enhancement beliefs (i.e., the SSB) in check. However, this study can be criticized on the grounds of a critical confounding, participant selection. It is likely that participants who agreed to come to the laboratory with a friend were particularly gregarious and sharing. The results, then, may be due to personality differences rather than relational closeness. An important objec­ tive of our next experiment (Sedikides, Campbell, Reeder, & Elliot, 1998, Experiment 1 ) was to control for this potential confounding. We also intended to address another, more critical issue: what is the mecha­ nism that is responsible for the elimination of the SSB in close dyads? One such mechanism may be the anticipation of future interactions. Friends antici­ pate seeing each other outside the context of the experiment and sharing their experiences. Thus, friends may have acted generously in the Campbell et a/. (200) experiment in order to avoid the possibility of having to provide their friend with an account for selfish attributions. In this way, close partners might have refrained from the SSB for the sake of relationship maintenance. In fact, an influential literature on communal-exchange relationships (Clark, 1 984; Mills & Clark, 1 982) suggests that anticipation of future and rewarding interaction with a partner can lead to a less selfish (i.e., more communal) orientation in outcome distribution. We addressed this possibility in our next experiment (Sedikides et al., 1998, Experiment 1 ) by using unacquainted par­ ticipants exclusively. Given that they were unacquainted, participants were unlikely to expect future interactions, let alone rewarding interactions. In fact, care was taken so that no participant (a) anticipated interacting with his/her partner following the experiment, or (b) intended to discuss the experiment with the partner in incidental encounters outside the laboratory. The crucial manipulation in this experiment (Sedikides et a/., 1998, Experi­ ment 1 ) was aimed at inducing closeness in half of the participants. Reciprocal and escalating self-disclosure is a vital feature in the development of a close relationship (Derlega, Metts, Petronio, & Margulis, 1993). We devised a task

HOW RELATIONSHIPS CHANGE THE SELF

249

to simulate such self-disclosure between unacquainted participants, which we named the Relationship Closeness Induction Task (RCIT; Sedikides, Camp­ bell, Reeder, & Elliot, 1 999). Will members of close dyads, who do not anticipate interacting with one another in the future, still refrain from the SSB? Stated otherwise, is antici­ pated interaction a necessary condition for the elimination of the SSB in close dyads? If anticipated interaction is not a necessary condition, the results of the Campbell et al. (200) experiment will be replicated (i.e., the SSB will be absent in close dyads). On the other hand, if anticipated interaction is indeed a necessary condition, close partners will display the SSB. Four unacquainted participants arrived at the laboratory for each experi­ mental session and were split into two separate dyads by the experimenter. Participants in each dyad were seated across from each other, were informed that they would not interact with each other at the end of the experiment, and were asked not to talk about the experiment if they encountered each other on campus. Participants also learned that they would engage in two short and ostensibly unrelated studies. The first study would involve a communication task. This study actually was the RCIT, a structured self-disclosure task. The RCIT consists of three lists of questions and instructs participants to self­ disclose mutually for 9 minutes while engaging in as natural a conversation as possible. Participants spend 1 minute on List I (7 questions, e.g., "Where are you from?", "What year are you at this university?"), 3 minutes on List II (12 questions, e.g., "What are your hobbies?", "What would you like to do after graduating from this university?"), and 5 minutes on List III (10 questions, e.g., "Describe the last time you felt lonely", "Tell me one thing about your­ self that most individuals who already know you don't know"). The three lists of questions become progressively personal. While participants take turns answering the questions, the experimenter waits outside the room and acts as a time keeper, intervening only for the purpose of instructing participants to proceed to the next list. The majority of dyads managed to respond to almost all questions within each list. They also reported that they had adequate privacy, felt comfortable, considered conversation a valid way to become familiar with a stranger, and had experienced conversations of this sort in the past with strangers. Most importantly, participants in the close relationship condition reported feeling closer to their partner than participants in the distant relationship condition. The results of this manipulation check attest to the effectiveness of the RCIT in inducing relationship closeness. Then, participants were told that it was time to proceed with Study 2, in which they would take a creativity test. Participants in the close condition stayed with the same partner, but participants in the distant condition were switched to a new partner, who had just completed the RCIT with another participant. This practice ensured that participants in the close and distant conditions went through an

250

CONSTANTINE SEDIKIDES ET AL.

identical relationship induction procedure. From this point on, the experiment followed the same procedure and used the same manipulation checks and dependent measures as the Campbell et al. (20) experiment. Close and distant participants reported equivalent mood levels. Also, creativity was equally important to the two types of participants. Finally, the feedback manipulation was effective: participants in the success feedback condition reported that both they and their partner performed better com­ pared to reports of participants in the failure feedback condition. Also, parti­ cipants who received success feedback ended up regarding the creativity test as more important than participants who received failure feedback. As shown in Table 8.2, this experiment replicated the results of the Camp­ bell et a/. (200) experiment. Distant participants assumed greater personal responsibility for the dyad's success than its failure. However, close particip­ ants assumed equivalent levels of responsibility for the dyad's success and failure. Furthermore, close participants reported making a less positive contri­ bution to the test outcome relative to distant participants. In all, the SSB emerged in distant but not close dyads. The absence of the SSB in relationally close dyads is consistent with the relationships-as-bound hypothesis. (Note that, in contrast to the Campbell et al. experiment, men displayed the SSB to a greater degree than women.) Table 8.2

Attributions as a function of feedback and relationship closeness in Sedikides, Campbell, Reeder, and Elliot ( 1998, Experi­ ment 1 )

I.

Task outcome question Distant dyads Close dyads

II.

Failure feedback 5.233 5.6()a

Positive contribution question Distant dyads 5.9(}a

Success feedback 6.2()b 5.3()a Close dyads 5.27b

Note 1: Responses to the task outcome question were made on a 1 (the other participant was more responsible) to 10 (I was more responsible) scale; re sponses to the positive contribution question were made on a 1 (the other participant made the greater positive contribution) to 10 (I made the greater positive contribution) scale. Note 2: Within a row, means with different superscripts are significantly different.

This experiment (Sedikides et al., 1998, Experiment 1 ) ruled out participant selection as an explanation for the relationships-as-bound hypothesis and increases our confidence in the relationships-as-bound hypothesis. The ex­ periment also demonstrated that the anticipation of future rewarding interac­ tions is not a necessary condition for the absence of the SSB in close dyads.

252

CONSTANTINE SEDIKIDES ET A L .

Table 8.3

Attributions as a function of feedback and relationship closeness in Sedikides, Campbell, Reeder, and Elliot (1998, Experi­ ment 2) I.

Task outcome question Distant dyads Close dyads

II.

Failure feedback 4. 778 5.928

Positive contribution question Distant dyads 6.038

Success feedback 5.88b 5.758 Close dyads 5.31b

Note 1: Responses to the task outcome question were made on a 1 (the other participant was more responsible) to 10 (I was more responsible) scale; re sponses to the positive contribution question were made on a 1 (the other participant made the greater positive contribution) to 10 (I made the greater positive contribution) scale. Note 2: Within a row, means with different superscripts are significantly different.

Our next objective was to test whether partner impressions mediated the SSB. Before we proceed with a description of the relevant analyses, however, we wish to clarify how we intended our experimental results to map onto the actual conceptual explanation that we evoke. Conceptually, we propose that a reason for the non-manifestation of the SSB in closely-related participants is that they view their partner in relatively favorable terms. Empirically, we manipulated closeness of relationship and, therefore, labeled the mediator as the formation of a favorable impression. That is, we hypothesized and demon­ strated empirically that relational closeness leads to the formation of a favor­ able impression. We now want to know whether these impressions are partially responsible for keeping the SSB in check. We proceeded with a two-step, interactional mediation analysis (see Sed­ ikides et a/., 1 998, p. 384, for more details). In the first step, we established that partner impression predicts attributions of responsibility for the task outcome and for positive contribution to the task. In the second and critical step, we established that the relatively favorable partner impressions among close par­ ticipants (relative to distant participants) accounted for close partners refrain­ ing from the SSB. In summary, the interactional mediation analyses revealed that partner impression mediated the association between relationship type and the SSB. The favorable impressions that close dyad members formed of each other accounted for the non-manifestation of the SSB. Stated otherwise, a reason that close participants (a) were equally likely to take responsibility for the success and failure of the dyad and (b) claimed a lesser positive contribution to the test outcome was that they formed a more favorable impression of each other.

253

HOW RELATIONSHIPS CHANGE THE SELF

When Does Relational Closeness Reduce or Augment the SSB?

Our research so far succeeded in establishing the viability of the relationships­ as-bound hypothesis: relational closeness curtails self-enhancement. Our re­ search also identified a critical psychological mechanism for this effect. Closely related partners are attributionally generous to each other, because they form a positive impression of each other. However, are close partners always generous to each other? (Alternatively, are distant partners always selfish in their attributional patterns?) In our final round of work (Sedikides, Campbell, Reeder, & Elliot, 2001), we considered some of the circumstances under which relationship closeness is likely to reduce or augment the SSB. Attributional Expectancies

We reasoned that distant and close partners probably have different expec­ tancies about each other's behavior in interdependent outcomes tasks. Perhaps because they have a neutral impression of each other (as shown in Sedikides et al., 1998, Experiment 2), distant partners have competition ex­ pectancies. On the other hand, perhaps because they have a positive impres­ sion of each other (also as shown in Sedikides et al., 1998, Experiment 2), close partners have cooperation expectancies (Argyle & Henderson, 1984; Clark, 1984; Rusbult & Van Lange, 1996). We conducted an experiment (Sedikides et al., 2001, Experiment 1) to test directly the above derivations. Do distant and close partners have different expectancies about each other's behavior in an interdependent outcomes task, such as the one used in the present program of research? More specifically, do close partners have mutual expectations for refraining from the SSB? Do distant partners hold mutual expectations for displaying the SSB? Participants imagined a scenario. They and a same-gender other (friend or stranger) reported to the laboratory for a psychology experiment. Each mem­ ber of the dyad was seated in a separate room. They imagined completing the Lange-Elliot Creativity Test jointly with their partner and receiving either success or failure feedback. In order to increase the realism of the procedure, the instructions that accompanied the visualizations pertaining to the creativity test and the performance feedback were taken verbatim from the previously described experiments. In the end, participants indicated how they expected their partner to fill out the dependent measures. The wording of the dependent measures was identical to that used in our previous experiments. Close and distant participants indeed harbored different expectancies about their partner's attributions for task performance (Table 8.4 ) Distant participants expected a selfish response (i.e., the display of the SSB) from their partner. That is, they expected their partner to assume more personal responsibility for, and claim a greater positive contribution to, the dyadic .

HOW RELATIONSHIPS CHANGE THE SELF

255

participants would have their expectancies disconfirmed. Hence, they would reciprocate in a gracious attributional manner-by refraining from the SSB. We predicted a different pattern for close participants. Upon learning that their partner behaved selfishly, close participants would have their expectan­ cies disconfirmed. Hence, they would behave in an attributionally retaliatory manner, i.e., by assuming credit and displacing blame. However, upon learn­ ing that their partner behaved graciously, close participants would have their expectancies confirmed and would behave as they would have behaved any­ way, that is by refraining from the SSB. We tested these predictions empirically (Sedikides et al., 2001, Experiment 2). As in previous experiments, we induced relationship closeness in the laboratory via the RCIT. After completing the Lange-Elliot Creativity Test, participants received either success or failure performance feedback. However, prior to the completion of the dependent measures, we introduced another manipulation. Two-thirds of the participants were told that they were selected randomly to preview their partner's responses, although their partner would not be granted the same privilege. In actuality, the partner's responses were fabricated by the experimenter. The partner responses fell into two categories: constructive and destructive. The remaining one-third of particip­ ants were unaware of their partner's response. Specifically, some participants (constructive partner response condition) , previewed a generous partner response (i.e., an OSB). They learned that their partner attributed the success of the dyad to them, whereas the partner took personal responsibility for the failure of the dyad. However, other particip­ ants (destructive partner response condition) previewed a selfish partner response (i.e., an SSB). They learned that their partner blamed them for the dyadic failure, but took credit for the dyadic success. As stated above, the other one-third of participants comprised the no partner response (or control) condition. The RCIT was successful in inducing relationship closeness. Additionally, close and distant participants did not differ in their mood states at the start of the testing session. Moreover, close and distant participants regarded creativity as equally important. Finally, the success and failure feedback manipulations were effective. We will examine distant and close dyads separately (Tables 8.5 and 8.6). Let us first consider the attributions of distant dyad members. In the no partner response condition, distant participants manifested the SSB: they took dis­ proportionate responsibility for the success than the failure of the dyad, and they also claimed a more positive contribution for the success of the dyad. This finding replicates our past research. Interestingly, the SSB of distant dyad members was magnified in the destructive partner response condition. On being informed of their partner's selfish claim, distant members reacted in kind by taking the lion's share of the responsibility for the dyadic success and

258

CONSTANTINE SEDIKIDES ET A L.

failure feedback that was bogus and randomly determined, and attributed the dyad's performance relativistically to the self or the partner. According to the relationships-as-bound hypothesis, relational closeness inhibits self­ enhancement. Oosely related individuals refrain from the SSB. However, according to the relationships-as-enabler hypothesis, relational closeness pro­ motes self-enhancement. Oosely related individuals display the SSB. The results of several experiments provided unequivocal support for the relationships-as-bound hypothesis. In contrast to distant dyad members, close members refrained from the SSB. Relational closeness practically eliminates perhaps the most robust of self-enhancement tendencies, the SSB. Relational closeness puts the self "in its place". How do relationships keep self-enhancement tendencies in check? What is the mechanism through which the SSB is eliminated in close relationships? Furthermore, what are the circumstances under which relational closeness augments the SSB? We tested several explanations for the gracious attribu­ tional pattern of close others. One explanation was based on participant selection-that the disappearance of the SSB in close relationships was due to close partners having different personality characteristics than distant partners. Another explanation was based on the anticipation of future interactions-that close participants expect to have future truthful and rewarding interactions, and they cannot manifest the SSB in the laboratory without taking into account the dilemma of possibly having to lie to their partner when the topic is discussed after the experiment. The third explana­ tion was based on mood-that close participants refrain from the SSB because they are in a happier mood than distant participants. The fourth explanation was based on the differential importance of the creativity test for close and distant participants-that close partners regard the test as more important than distant partners. None of these explanations received empiri­ cal support. However, two other explanations based on favorable partner impressions (Sedikides et al., 1 998, Experiment 2) and expectancies of a cooperative part­ ner response (Sedikides et al., 200, Experiments 1 and 2) did receive empiri­ cal support. The absence of the SSB in close (compared to distant) relationships is due, in part, to close participants forming a favorable impres­ sion of each other and expecting each other to perform well on the interde­ pendent outcomes task. Furthermore, the absence of the SSB is probably due, in part, to the expectancies that close participants have of each other. That is, close participants expect their partner to refrain from the SSB. In fact, when they are informed that the partner's attributions for the task outcome have been self-serving, close participants retaliate with an equally self-serving attributional response. An obvious implication of the attributional reciprocity findings is that rela­ tionship closeness per se is not sufficient for the attenuation of self-

260

CONSTANTINE SEDIK.IDES ET A L.

pendent outcomes tasks: social value orientation and communal orientation. Research by Van Lange and his colleagues (Van Lange, Agnew, Harinck, & Steemers, 1 997; Van Lange & Liebrand, 1 991 ; Van Lange, Otten, De Bruin, & Joireman, 1997; Van Lange & Semin-Goossens, 1998; see also Kuhlman, Camac, & Cunha, 1 986) has compared the responses of three classes of indi­ viduals: prosocials (those who are predisposed to distribute outcomes so as to maximize joint gains or minimize reciprocal losses), individualists (those who are predisposed to maximize own gains in an absolute sense), and competitors (those who are predisposed to maximize own outcomes relative to others' outcomes). In comparison to individualists and competitors, prosocials are more likely to respond cooperatively, regardless of their partner's response. In the context of our research, prosocials in closely related dyads may be least likely to display the SSB, regardless of whether their partner initiates a selfish or gracious attributional pattern. The role of communal orientation is also relevant. Individuals with a com­ munal orientation are concerned with other persons' welfare, feel responsible for them, are more likely to help them (Clark & Mills, 1979; Clark, Quellette, Powell, & Milberg, 1987), and are more likely to allocate resources equally in a dyadic setting (Thompson & DeHarpport, 1998). Like prosocials, individ­ uals high in communal orientation may be less likely to manifest the SSB. More interestingly, both prosocials and communally-orientated persons may be even less likely to display the SSB in the context of a close, dyadic relation­ ship. Indeed, recent research by McCall, Reno, Jalbert, and West (200) has documented that communal orientation, along with the perception of the partner as community-orientated, reduces the SSB. Although our research has made substantive forays into the ways in which relational closeness affects self-enhancement, additional important issues remain. What are some other mechanisms, besides impression favorability and differen­ tial expectancies of partner responses, that reduce self-enhancement? What is the role of normative concerns (e.g., the norm to be loyal to the partner) in the attenuation of self-enhancement? How far beyond the immediate relational set­ ting does a gracious attributional pattern last? At what point is this unselfish response internalized to become a permanent fixture of the relationship? Does relational closeness afect other forms of self-enhancement, besides the SSB? Can high levels of self-threat lead the individual to behave selfishly rather than gra­ ciously toward a close partner? We hope that our research wil stimulate progres toward empirical resolution of these and other issues. ACKNOWLEDGMENTS

We are grateful to Wolfgang Stroebe, Miles Hewstone, and two anonymous reviewers for their perceptive and constructive feedback on an earlier draft.

262

CONSTANTINE SEDIKIDES ET A L .

Cohen, S., & Hoberman, H. (1983). Positive events and social supports as buffers of life change stress. Journal of Applied Social Psychology, 13, 99-125. Cohen, S., & Wills, T. A. (1985). Stress, social support, and the enabling hypothesis. Psychological Bulletin, 98, 31�57. Coyne, J. C., Burchill, S. A. L., & Stiles, W. B. (1991). An interactional perspective on depression. In C. R. Snyder & D. R. Forsyth (Eds), Handbook of Social and Clinical Psychology: The Health Perspective (pp. 327-49). New York: Pergamon. Deaux, K., & Farris, E. (1977). Attributing causes for one's own performance: The effects of sex, norms, and outcome. Journal of Research in Personality, 11, 59-72. Derlega, V. J., Metts, S., Petronio, S., & Margulis, S. T. (1993). Self-disclosure. Newbury Park, CA: Sage. Drigotas, S. M., Rusbult, C. E., Wieselquist, J., & Whitton, S. W. ( 199). Close partner as sculptor of the ideal self: Behavioral affirmation and the Michelangelo phenom­ enon. Journal of Personality and Social Psychology, 77, 293-323. Dunning, D. (1993). Words to live by: The self and definitions of social concepts and categories. In J. Suls (Ed.), Psychological Perspectives on the Self (Vol. 4, pp. 99126). Hillsdale, NJ: Erlbaum. Felson, R. B. (1993). The (somewhat) social self: How others affect self-appraisals. In J. Suls (Ed.), Psychological Perspectives on the Self (Vol. 4, pp. 1-26). Hillsdale, NJ: Erlbaum. Fincham, F. D., & Bradbury, T. N. (1989). Perceived responsibility for marital events: Egocentric or partner-centric bias? Journal of Marriage and the Family, 51, 27-35. Fletcher, G. J. 0., Fincham, F. D., Cramer, L., & Heron, N. (1987). The role of attributions in the development of dating relationships. Journal of Personality and Social Psychology, 53, 481-9. Goffman, E. (1959). The Presentation of Self in Everyday Life. New York: Doubleday. Gouldner, A. W. (1960). The norm of reciprocity: A preliminary statement. American Sociological Review, 25, 161-78. Hall, J. A., & Taylor, S. E. (1976). When love is blind: Maintaining idealized images of one's spouse. Human Relations, 29, 751--61 . Heider, F. (1958). The Psychology of Interpersonal Relations. New York: Wiley. Holmes, J. G., & Rempel, J. K. (1989). Trust in close relationships. In C. Hendrick (Ed.), Review of Personality and Social Psychology (Vol. 10, pp. 187-220). London: Sage. Hornstein, H. A. (1976). Cruelty and Kindness. Englewood Cliffs, NJ: Prentice Hall. Janoff-Bulman, R., & Timko, C. (1987). Coping with traumatic events: The role of denial in light of individuals' assumptive worlds. In C. R. Snyder & C. E. Ford (Eds), Coping with Negative Life Events: Clinical and Social Psychological Perspectives (pp. 135-59). New York: Plenum. Johnston, W. A. ( 1%7). Individual performance and self-evaluation in a simulated team. Organizational Behavior and Human Performance, 2, 309-28. Kenny, D. A., & DePaulo, B. M. (1993). Do individuals know how others view them? An empirical and theoretical account. Psychological Bulletin, 114, 145--61. Kling, K. C., Hyde, J. S., Showers, C. J., & Buswell, B. N. (199). Gender differences in self-esteem: A meta-analysis. Psychological Bulletin, 12S, 47�500. Kuhlman, D. M., Camac, C., & Cunha, D. A. (1986). Individual differences in social orientation. In H. Wilke, D. Messic & C. Rutte (Eds), Experimental Social Dilemmas (pp. 151-76). New York: Verlag Peter Lang. Kunda, Z. (1990). The case for motivated reasoning. Psychological Bulletin, 108, 48098. Long, B. H. (1983). Heterosexual involvement of unmarried undergraduate females in relation to self-evaluations. Journal of Youth and Adolescence. 18, 489-500.

264

CONSTANTINE SEDIKIDES ET A L .

J. M. Suls (Eds), Psychological Perspectives on Self and Identity (pp. 1 99-232). Washington, DC: APA Books. Schlenker, B. R., & Britt, T. W. ( 199). Beneficial impression management: Strate­ gically controlling information to help friends. Journal of Personality and Social Psychology, 76, 559-73. Schlenker, B. R., & Miller, R. S. (1977). Egocentrism in groups: Self-serving bias or logical information processing? Journal of Personality and Social Psychology, 35, 755-64. Sedikides, C., & Brewer. M. B. (2001 ). Individual Self, Relational Self, Collective Self. Philadelphia, PA: Psychology Press. Sedikides, C., Campbell, W. K., Reeder, G., & Elliot, A. J. (1998). The self-serving bias in relational context. Journal of Personality and Social Psychology, 74, 37S-86. Sedikides, C., Campbell, W. K. Reeder, G., & Elliot, A. J. (199). The relationship closeness induction task. Representative Research in Social Psychology, 23, 1-4. Sedikides, C., Campbell, W. K., Reeder, G., & Elliot, A. J. (2001 ). The self-serving bil in relational context: Attributional expectancies, expectancy violation, and attribu­ tional reciprocity (unpublished manuscript). University of Southampton. Sedikides, C., & Green, J. D. (200). On the self-protective nature of inconsistency/ negativity management: Using the person memory paradigm to examine self­ referent memory. Journal of Personality and Social Psychology, 79, 906-22. Sedikides, C., Olsen, N., & Reis, H. T. (1993). Relationships as natural categories. Journal of Personality and Social Psychology, 64, 71-82. Sedikides, C., Oliver, M. B., & Campbell, W. K. (1994). Perceived benefits and costs of romantic relationships for women and men: Implications for exchange theory. Per­ sonal Relationships, 1, 5-21. Sedikides, C., & Strube, M. J. (1997). Self-evaluation: To thine own self be good, to thine own self be sure, to thine own self be true, and to thine own self be better. In M. P. Zanna (Ed.), Advances in Experimental Social Psychology (Vol. 29, pp. 20969). New York: Academic Press. Skowronski, J. J., Betz, A. L., Thompson, C. P., & Shannon, L. (1991). Social memory in everyday life: The recall of self-events and other-events. Journal of Personality and Social Psychology, 60, 831-43. Steele, C. M. (1988). The psychology of self-affirmation: Sustaining the integrity of the self. In L. Berkowitz (Ed.), Advances in Experimental Social Psychology (Vol. 21, pp. 261-302). New York: Academic Press. Tesser, A., & Rosen, S. (1975). The reluctance to transmit bad news. In L. Berkowitz (Ed.), Ad�ances in Experimental Social Psychology (Vol. 8, pp. 193-232). New York: Academic Press. Tetlock, P. E., & Levi, A. (1982). Attribution bias: On the inconclusiveness of the cognition-motivation debate. Journal of Experimental Social Psychology, 18, 68-8. Thompson, L., & De Harpport, T. (1998). Relationships, goal incompatibility, and communal orientation in negotiations. Basic and Applied Social Psychology, 20, 3344. Thompson, S. C., & Kelley, H. H. ( 1981 ). Judgements of responsibility for activities in close relationships. Journal of Personality and Social Psychology, 41, 469-77. Tice, D. M., Butler, J. L., Muraven, M. B., & Stillwell, A. M. (1995). When modesty prevails: Differential favorability of self-presentation to friends and strangers. Jour­ nal of Personality and Social Psychology, 69, 1 120-38. Van Lange, P. A. M., & Kuhlman, D. M. (1994). Social value orientations and impres­ sions of partner's honesty and intelligence: A test of the might versus morality effect. Journal of Personality and Social Psychology, 67, 126-41. Van Lange, P. A., Agnew, C. R., Harinck, F., & Steemers, G. E. M. (1997). From game theory to real life: How social value orientation affects willingness to sacrifice in

HOW RELATIONSHIPS CHANGE THE SELF

265

ongoing close relationships. Journal of Personality and Social Psychology, 73, 133044.

Van Lange, P. A. M., & Liebrand, W. B. G. (1991). Social value orientation and intelligence: A test of the goal-prescribes-rationality principle. European Journal of Social Psychology, 21, 273-92. Van Lange, P. A. M., Otten, W., De Bruin, E. M. N., & Joireman, J. A. (1997). Development of prosocial, individualistic, and competitive orientations: Theory and preliminary evidence. Journal of Personality and Social Psychology, 73, 733-46. Van Lange, P. A. M., Rusbult, C. E., Drigotas, S. M., Arriaga, X. B., Witcher, B. S., & Cox, C. L. (1997). Willingness to sacrifice in close relationships. Journal of Person­ ality and Social Psychology, 72, 1373-95. Van Lange, P. A. M., & Semin-Goossens, A. (1998). The boundaries of reciprocal cooperation. European Journal of Social Psychology, 28, 847-54. Weary-Bradley, G. (1978). Self-serving biases in the attribution process: A re­ examination of the fact or fiction question. Journal of Personality and Social Psy­ chology, 36, 56-71. Wills, T. A. ( 1981 ). Downward comparison principles in social psychology. Psychologi­ cal Bulletin, 90, 245-71 . Wieselquist, J ., Rusbult, C . E., Foster, C . A . , & Agnew, C . R . (1999). Commitment, pro-relationship behavior, and trust in close relationships. Journal of Personality and Social Psychology, 77, 942-66. Wolosin, R. J., Sherman, S. J., & Till, A. (1973). Effects of cooperation and competi­ tion on responsibility attribution after success and failure. Journal of Experimental Social Psychology, 9, 220-35. Wyer, R. S., & Frey, D. (1983). The effects of feedback about self and others on the recall and judgments of feedback-relevant information. Journal of Experimental Social Psychology, 19, 540-59. Zuckerman, M. (1979). Attribution of success and failure revisited, or: The motiva­ tional bias is alive and well in attribution theory. Journal of Personality, 47, 245-87.

This page intentionally left blank

Chapter 9

Prospects, Pitfalls, and Plans: A Proactive Perspective on Social Comparison Activity Lisa G. Aspinwall University of Utah, USA

Douglas L. Hill Samantha L. Leaf

University of Maryland, USA

ABSTRACT Most studies of social comparison activity have focused on reactive comparisons (those directed toward understanding outcomes that are already known), with little attention to proactive comparisons (those directed toward understanding future possibilities). We examine the usefulness of this distinction for under­ standing the motives underlying comparison activity and responses to upward and downward comparisons. When an outcome is known, comparison motives are often circumscribed, such that upward comparisons are deflating and frus­ trating and downward comparisons are self-enhancing and reassuring. In con­ trast, prior to an event, when one's own outcomes are unknown and potentially malleable, social comparisons may provide useful information about likely out­ comes (prospects), what to avoid (pitfalls), and different ways to achieve or

Correspondence regarding this chapter should be addressed to Lisa G. Aspinwall, Department of Psychology, University of Utah, 380 South 1 530 East, Room 502, Salt Lake City, UT 8411 2 0251 . E mail: [email protected]. Samantha L . Leaf is now a t the Department of Psychol­ ogy, University of Utah. European Review of Social Psychology, Volume

© 2002 John Wiley &

Sons Ltd.

12.

Edited by Wolfgang Stroebe and Miles Hewstone.

268

LISA G. ASPINWALL, DOUGLAS L. HILL AND SAMANTHA L. LEAF

avoid particular outcomes (plans). Importantly, because people may have time and other resources to respond constructively, they may benefit from both up­ ward and downward comparisons. We review recent evidence consistent with this framework and discuss the implications of proactive comparisons for self­ regulation in such domains as achievement, aging, and health. Social comparison activity has been extensively studied as a central way in which people evaluate their current abilities, opinions, and outcomes (Buunk bons, 1997; Festinger, 1954; Suls

& Wil,

&

Gib­

1991; for recent reviews, se Biernat

&

Billings, in press; Blanton, 201 ). Comparison to others' outcomes has also been shown to be an important aspect of coping with adversity among people experi­ encing chronic

iles,

physical disability, burnout, marital discord, unfavorable

medical test results, por academic performance, and other stressors (Buunk Gibbons, 1997; Taylor, Buunk,

&

&

Aspinwall, 1990). Much of this evidence has

necesy focused on the impact of social comparisons on people's feelings and beliefs about outcomes that have already taken place. In this article, we wil consider ways in which social comparison information may be used proactively, that is, not only to evaluate current outcomes, but also to make important deci­ sions about future courses of action, such as which goals to select, how to pursue them, and what to expect along the way. In doing so, we hope to illustrate the substantial, but largely unexplored, role of social comparisons in such future­ oriented processes as proactive coping, self-regulation, and planning.

PROACfiVE COPING AS A BLEND OF SELF­ REGULATION AND COPING

Proactive coping refers

to a set of processes through which people anticipate

or detect potential stressors and act in advance to prevent them or to mute their impact (Aspinwall

& Taylor,

1997). As such, proactive coping is a blend

of two extensively studied processes, coping and self-regulation. Coping con­ sists of activities undertaken to master, reduce, or tolerate environmental or intrapsychic demands perceived as representing threat, harm, or loss (Lazarus

&

Folkman, 1984).

Self-regulation

is the process through which people con­

trol, direct, and correct their own actions as they move toward or away from various goals (Carver

&

Scheier, 1990). The study of proactive coping com­

bines these two processes by examining people's emotions, thoughts, and behaviors as they anticipate and address potential sources of adversity that might interfere with the pursuit of their goals (see Aspinwall, 2001 , in press; Aspinwall

&

Taylor, 1 997; McGrath

&

Beehr, 1990; Sansone

&

Berg, 1 993).

Recently, Schwarzer (in press) has suggested that the study of proactivity may be expanded to include thoughts and actions, not only in the service of pre­ venting adverse outcomes, but also in creating future opportunities. The purpose of this chapter is to examine ways in which social comparison activity may be used to accomplish both kinds of proactivity-troubleshooting

PROACTIVE SOCIAL COMPARISONS

269

potential problems and identifying potential opportunities. We wil present a framework for understanding different kinds of proactive comparison situations, present a description of the differences between proactive and reactive com­ parison situations, and discuss the implications of these differences for under­ standing comparison motives and outcomes. We wil then review recent evidence consistent with the proactive social comparison framework. We conclude with a discussion of the ways in which proactive comparisons may be used to guide future-oriented behavior in such domains as achievement, aging, and health. A FRAMEWORK FOR UNDERSTANDING PROACTIVE COMPARISON ACTIVITY Distinguishing Proactive from Reactive Comparison Situations

Our analysis suggests that the location of the comparer and target with re­ spect to the temporal development of some event or task is essential to understanding both the likely motives of the comparer and the likely out­ comes of comparisons to a particular target. If the comparer is either con­ sidering some new course of action or is already involved in it, and is making comparisons to a target who is further along in the course of events than the comparer, the comparison is proactive, rather than reactive. The hallmark of a proactive comparison situation is that some focal event (a new task, a stressor, or an opportunity) has yet to occur. Therefore, the event itself and/or the impact of the event on the self are potentially malleable.l Of central import­ ance are the ideas that one's own outcome is not yet known or circumscribed (many potential outcomes are possible), and that one has the potential to alter one's thinking or behavior in order to approach or avoid the outcome experi­ enced by a particular comparison target. In such situations, information about other people who have adopted a new goal (e.g., a career change or parent­ hood), who are more advanced in their standing on the comparison dimension (e.g., the development of a skill, such as golfing or cooking), and/or who are further along on the developmental continuum of the event (e.g., adjustment to college; parents of children older than one's own; recovery from major surgery; retirement; aging) has the potential to be highly informative with respect to one's own actions. In such situations, both upward comparisons to those who did well and downward comparisons to those who experienced difficulty may be informative. These ideas are shown in Figure 9. 1 . 1Even if a n event o r outcome i s not malleable, people may use social comparisons to prepare for managing such eventualities (see Aspinwall & Taylor, 1997, for discus­ sion) . For example, learning how others handled an unalterable future event, such as forced relocation following highway construction or transition to a nursing home, may be very useful in altering one's own preparations for and reactions to that event.

272

LISA G. ASPINWALL, DOUGLAS L. HILL AND SAMANTHA L. LEAF

interview. Now, responses to Mr Clean might differ dramatically. The person so exposed to Mr Clean might purchase a suit and prepare for the interview. Alternatively, he might be intimidated by Mr Clean's level of preparation and give up on the interview. In either case, exposure to Mr Clean provides important information about the level of preparation and professionalism required in the interview setting. To recap these examples in terms of our figure: in the original Mr Clean/Mr Dirty study, both the comparer (C2) and target (T2) await the interview. In this situation, the target is optimally situated to provide self-evaluative infor­ mation, as both comparer and target are at the same point with respect to the upcoming interview. However, because the interview is imminent, the infor­ mation provided by Mr Clean's superior preparation cannot be used by the comparer to alter his outcomes in this particular situation; thus, the com­ parison is experienced as aversive ("I'm no good"). In contrast, when the participant encounters Mr Clean 2 days before his own interview, the com­ parer is now at Cl, well before the interview, and the superior target at T2, just before. The comparison is now proactive and can serve a wide range of motives and self-regulatory functions, as exposure to Mr Clean provides use­ ful information about standards (the level of preparation required by the task), possible courses of action (dress up, cancel the interview), and likely outcomes for the comparer (if he does nothing to improve, he won't compete successfully for the job). Under such circumstances, participants might well be glad they had run into Mr Clean, and it is unlikely that his polished self­ presentation would be as deflating as it was in the original study.

Diferent Kinds of Proactive Comparison Our proactive version of the Mr Clean/Mr Dirty study provided an example in which both the target and the comparer were in the left part of the figure, as they encountered each other prior to the focal event (the job interview). A second important kind of proactive comparison may occur when the target has already experienced an event that awaits the comparer. The best illustra­ tion of such a situation, and its dramatic influence on self-regulation, is the series of field experiments on hospital roommate assignments conducted by Kulik and Mahler (1997; see also Kulik, Mahler, & Moore, 1996). People awaiting major surgery were randomly assigned to either preoperative or postoperative roommates. With respect to our Figure 9.1 , in the preoperative­ roommate condition, both the patients and their roommates are at time 2 with respect to the surgery, whereas in the postoperative-roommate condition, the patients are at C2 and their roommates at T4, presumably on the upward curve indicating post-surgical improvement. As we know from research on fear and affiliation, exposure to another person awaiting the same stressful event (the C2-T2 configuration) may be

276

LISA G. ASPINWALL, DOUGLAS L. HILL AND SAMANTHA L. LEAF

second domain, whereas people who failed an initial task were more interested in learning about their strengths in the second domain. Explana­ tions of such findings have focused on the potential of positive mood to bolster people's sense of the resources they have available to withstand ex­ posure to negative information (Aspinwall, 1998; Trope & Pomerantz, 1998; see Steele, Spencer, & Lynch, 1993, and Tesser, Crepaz, Beach, Cornell & Collins, 200, for discussion of alternative explanations involving self-esteem). To the extent that an upward social comparison can be thought of as useful, but negative, information relevant to the self (useful in the sense that it contains information relevant to one's own prospects and future actions, but negative in the sense that it contains information that one has been outper­ formed), the presence of positive mood may help people manage the typical affective and esteem costs of such comparisons. Consistent with this pos­ sibility, Tesser and his colleagues have shown that the typical pattern of unfavorable reactions to upward comparisons (i.e., derogating upward targets or decreasing closeness to them) can be attenuated completely by some prior opportunity to feel good about the self. Specifically, participants given a chance to affirm an important personal value were less likely to take actions to harm the performance of another participant who had outperformed them than those not given the opportunity to self-affirm (Tesser & Cornell, 1991 , Study 1 ). It is unknown how negative moods affect the motives and outcomes of proactive comparisons, but evidence from studies of reactive comparisons suggests that such effects will not be favorable. Negative moods, whether they are transitory or chronic, seem to predict that people will focus on the adverse or unfavorable aspects of upward comparisons. For example, relative to non­ depressed people, dysphoric and depressed people seem to engage in more such comparisons, to request more detailed information about them, and to be particularly threatened by them (Swallow & Kuiper, 1988, 1992, 1993; see Ahrens & Alloy, 1 997, for review). For example, work by Swallow and Kuiper (1988) suggests that depressed people may adopt unrealistic comparison standards and fail to adjust them for any advantages an upward target may have had, thus perpetuating feelings of inferiority. Other reactions to upward targets may minimize the usefulness of such comparisons. For example, Wenzlaff and Beevers (1998) found that depressed individuals, as well as those whose mood was manipulated with sad music, were more likely to choose negative questions for a happy target, in an apparent attempt to find out negative information about the target. Such a strategy might reduce the unpleasant feeling generated when individuals are confronted by an upward comparison target whose outcome is seen as unattainable, but would also seem to preclude finding benefits in the target's superior outcome. The literature on individual differences related to chronic mood yields similar findings, suggesting that positive moods facilitate, and negative moods

278

LISA G. ASPINWALL, DOUGLAS L. HILL AND SAMANTHA L. LEAF

comparisons, especially their implications for the self. Specifically, people have been shown to downgrade the importance of the task (Tesser, 1988), to shift the basis of their self-evaluations to other dimensions (Crocker & Major, 1 989; Gibbons, Benbow, & Gerrard, 1994), and to distance themselves from superior others (Tesser, 1 988), even to the point of elevating the upward target to a superior class (such as genius) in order to make the comparison irrelevant to the self (Alicke, LoSchiavo, Zerbst, & Zhang, 1997). Thus, as a whole, the contemporary research base of social comparison theory, especially among people under stress, provides strong evidence for predomi­ nantly negative reactions to upward comparisons and a preference for, as well as favorable reactions to, downward comparisons (Aspinwall, 1997; Gibbons & Gerrard, 1991; Taylor et a/., 1990; Taylor, Wood & Lichtman, 1983; Wills, 1 981). Although efforts to avoid or defuse upward comparisons and to create novel downward comparisons provide fascinating examples of creativity in social behavior and self-definition, researchers have long been concerned by the possibility that avoiding upward comparisons and courting downward ones may have detrimental effects. In particular, as we and others have sug­ gested, upward comparisons have the potential to provide useful information about task performance, ways of coping with problems, standards and upper boundaries of performance, and motivation or inspiration. How can people learn from others' good outcomes if they actively avoid such information? Worse yet, how will people improve if they focus on people who are perform­ ing worse than they are (see, e.g., Gibbons et a/., 1991 ; Gibbons, Blanton, Gerrard, Buunk, & Eggleston, 200; Wills, 1997)? In light of these concerns, the identification of the conditions that allow people to profit from upward comparisons and avoid the risks of downward comparisons has been a central research topic. To date, there are several compelling accounts of moderators of responses to upward and downward comparisons. We will review these briefly and then examine what the proac­ tive perspective may add to them. Moderators of Responses to Upward and Downward Comparisons

One class of explanations centers on the comparer's perceived degree ofper­ sonal control over future outcomes (Buunk, Collins, Taylor, VanYperen, & Dakof, 1 990; Major, Testa, & Bylsma, 1991 ; Ybema & Buunk, 1993). Specifi­ cally, if one's future outcomes are thought to be improvable through one's own efforts, then upward comparisons are not as aversive, because one can match the upward target's superior standing through effort or practice. In this vein, several studies have demonstrated that people's beliefs about the mal­ leability of their performance on a particular attribute sharply moderate their interest in, and reactions to, upward comparison information (Butler, 1992,

PROACTIVE SOCIAL COMPARISONS

279

1993, 2000; Dunning, 1995; Lockwood & Kunda, 1997). For example, when mathematics performance is perceived to be an incremental skill (one that is improvable through practice), as opposed to a fixed entity (an ability that cannot be increased), students do not find another student's superior perfor­ mance on a mathematics test to be aversive (Butler, 1993; see Butler, 2000, and Dweck & Leggett, 1988, for discussion). With respect to downward com­ parisons, exposure to another person's poor outcomes may not be upsetting if one believes one can work to avoid such outcomes. A second, related factor is the anticipated trajectory of one's future standing on the dimension, that is, whether one expects to improve, decline, or stay the same (Aspinwall, 1997; Ruble & Frey, 1991). Such expectations are not syn­ onymous with personal control beliefs, as factors generally outside of one's control, such as maturation and aging, may influence people's beliefs about their likely future standing on some dimension (Aspinwall, 1997). For dimen­ sions on which improvement may be expected (such as adjustment to college, or maturation), upward comparisons may not carry their usual sting, in that one may reasonably anticipate matching the target's good outcome in due time. If this is the case, the upward target may provide an inspiring example of the good outcomes one may soon experience. However, in cases where one expects stability or decline in one's own outcomes (such as progressive illness, or age-related declines), someone else's good outcome may be especially aversive, as the gap between one's own standing and the target's will only increase. In a similar fashion, beliefs about one's likely future standing are likely to moderate the effects of exposure to downward comparisons. If one believes one will decline, then downward comparisons may provide frighten­ ing visions of what one's future holds. In contrast, these risks are minimal if one believes that one's future standing is fixed or subject to improvement (see Aspinwall, 1997, for review and discussion). A third set of accounts has focused on whether the comparison activity is primarily affiliative, rather than explicitly self-evaluative, in nature (Taylor & Lobel, 1989). Drawing on their work with cancer patients, Taylor and Lobel noted that people may affiliate with upward targets to derive motivational benefits, such as inspiration, from them (see also Helgeson & Mickelson, 1995; Taylor, Aspinwall, Giuliano, Dakof, & Reardon, 1 993), but they may avoid explicit comparisons to such targets to preserve self-esteem. Our anal­ ysis suggests that these benefits may be more easily obtained in the proactive situation, because comparisons to someone at a different stage are less direct than those to someone at the same stage. In the case of downward targets, Taylor and Lobel note that people appear to avoid affiliation with people who are worse off (such as other cancer patients in waiting rooms), but report frequent, explicit self-evaluative comparisons to them. In this way, people can simultaneously maintain hope and appreciate the relative advantages of their own situation.

280

LISA G. ASPINWALL, DOUGLAS L. HILL AND SAMANTHA L. LEAF

A fourth set of conditions in which upward comparisons may not be experi­ enced as aversive is when there is shared group membership or some other sort of bond to the upward target. In such cases, the target's success reflects well on one's group and/or oneself (Blanton, Crocker, & Miller, 200; Brewer & Weber, 1 994; Brown, Novick, Lord, & Richards, 1992; Cialdini et al., 1976; Helgeson & Mickelson, 1995; Tesser, 1988). In the collectivist context, people may also be motivated to make upward comparisons to improve the outcomes of their group (Chung & Mallery, 1 999). With respect to downward com­ parisons, the poor performance of an ingroup member is, through this lens of reflected self-evaluation, unlikely to be seen as reassuring or self-enhancing (Snyder, Lassegard, & Ford, 1 986). A Proactive Perspective on Upward and Downward Comparisons

We believe that the proactive versus reactive distinction is useful in understand­ ing reactions to both upward and downward comparisons, both in its own right and in conjunction with these other processes. As we have described, if the comparer and the target are at the same point with respect to the development of an event (e.g., they have both just had cancer surgery) or some developmen­ tal continuum (e.g., they are the same age), a comparison to the target provides information about one's relative standing at the moment, but relatively little information about future actions and outcomes. To be sure, one could infer from such a comparison that if the target is relatively advantaged at one point, he/she may also experience better outcomes at some later point, but the com­ parison itself provides little information about the demands of some future task, its likely trajectory, or its range of future outcomes. However, the moment the target is more advanced along some temporal or developmental continuum, the meaningfulness of the comparison, and its implications for one's own future actions, increase dramatically. In particular, information about targets who are further along in the development of some event can provide information useful in prediction (what to expect), prospects (what outcomes are possible, for better or for worse), and plans (what to do to approach or avoid certain out­ comes). In this way, proactive comparisons may be an important means through which people acquire information about the amenability of their standing to personal control efforts, the likely future course of events, and the outcomes experienced by ingroup members.

NEW EVIDENCE CONCERNING PROACfiVE USES OF SOCIAL COMPARISON INFORMATION

A final task of this chapter is to use the proactive versus reactive comparison framework to review several recent studies that have examined the usefulness

PROACTIVE SOCIAL COMPARISONS

281

of different kinds of social comparison information to people considering different courses of action or making predictions about their own future outcomes. These studies have in common the idea that people can learn valuable information from targets who are further along in some course of events than they are. We also present two studies designed specifically to test our prediction that the typical pattern of responses to upward and downward comparisons obtained in reactive comparison situations may be reversed in a proactive situation. Using Social Comparison Information to Evaluate Future Outcomes

An interesting illustration of the psychological differences between adjusting to events (reactive) and considering future possibilities (proactive), and the implications of such differences for social comparison processes, comes from a recent experiment on pay satisfaction by Blanton, George, and Crocker (2001 ) In an extension of Major's (1989) work suggesting that women remain content with inferior compensation because they compare themselves to other women instead of to more highly paid men, Blanton et al. had college men and women complete a laboratory task for which they were told that the pay range was $5 .25 to $7.35/hour. All participants then received a pay rate of $ 6.25/hour. This pay rate was framed either as compensation for work just completed or as part of an offer for future employment. Participants were then told that they could see the pay rate of past participants and were asked to rank order their preferences for viewing the average male pay rate, the average female pay rate, or the average combined pay rate. As predicted, women who were given a pay rate for work already com­ pleted had lower pay expectations and sought intragroup comparisons to other women, whereas women who were given the same pay rate framed as a job offer had higher pay expectations and sought intergroup comparisons to men. According to Blanton et al., the women who have already received their pay rate are focusing on what Jost and Banaji (1994) term system justification (feeling better with what they have), whereas those who are not yet invested (those considering the pay rate as an offer of future employment) are focused on system evaluation (Is what I have good enough? Could I do better?). These conclusions are highly consistent with the predictions of our proactive versus reactive comparison framework, in that our analysis suggests that the women who know their outcomes (those who have been given a pay rate) focus on feeling better about them, but those who are considering a course of action (in this case, the job offer) are willing to consider a wider range of comparison information, including the potentially upsetting comparison to what men are being paid for the same work. In the proactive situation, exposure to the upward comparison is worthwhile, because it will help one make a good decision about whether to accept the job at the offered rate of compensation. .

282

LISA G. ASPINWALL, DOUGLAS L. HILL AND SAMANTHA L. LEAF

Using Social Comparisons to Predict and Improve Performance

Smith and Sachs (1997) examined two different proactive uses of social com­ parison information: predicting and improving future outcomes. In their terms, a proxy comparison involves using someone else's performance to gauge how a similar other (in this case, the self) will do on a task (see also Wheeler, Martin, & Suls, 1997), whereas comparison to a model involves considering what someone else did that worked, that could be used to im­ prove one's own performance. Smith and Sachs. were able to distinguish the implications of each of these functions for social comparison activity in an experiment. The experiment involved two tests (Tests A and B) that assessed similar, but not identical, abilities. All participants received a score on Test A. In the proxy condition, participants were told that the purpose of the study was to see how well participants could predict their performance on Test B, whereas in the model condition, participants were told that the purpose of the study was to see how much people could improve on their Test A perfor­ mance with practice on Test B. Participants were then allowed to choose a comparison target. Of the six choices, three had outperformed the participant on Test A and three had performed worse. Three findings were of interest. First, in the proxy condition, where accu­ rate prediction of performance on Test B was given as the explicit goal, comparison choices were similar (lateral) or slightly upward, with 90% of the participants choosing a target just above them in performance on Test A. Second, in the model condition, in which self-improvement was given as the explicit goal, choices were predominantly upward, with 90% of participants choosing one of the top two scorers for comparison. Third, no participant in either condition chose a downward target. Thus, when the comparer's goals are either prediction or improvement of task performance, downward com­ parisons appear to have little use. Instead, in these two proactive situations, lateral and slightly upward (prediction) or strongly upward (improvement), comparisons are selected. A second study examined the proxy function of social comparisons in greater detail by examining how participants used a comparison target to predict their own performance. Participants learned how another participant had done on a first task (which the participants had already completed) and then learned the target's score on a second, related task. This information influenced participants' expectations of their own performance on the second task, such that those who had performed similarly to the target on Task 1 expected to perform similarly on Task 2, those who had outperformed the target on Task 1 expected to do so on Task 2, and those who had underper­ formed the target on Task 1 tended to expect to perform worse than the target on Task 2. Interestingly, confidence in one's predictions for Task 2 were highest in the similar target condition, and higher in the superior or inferior

284

LISA G. ASPINWALL, DOUGLAS L. HILL AND SAMANTHA L. LEAF

article, but fourth-year students did not. In the terms of Figure 9. 1 , if being successful in college is taken as the focal event, the first-year students are making a C1-T4 comparison to a target who has already achieved great success in college, whereas the fourth-year students are making a C4-T4 comparison to the same target. In this latter case, because the comparer and target are at the same stage with respect to the comparison dimension (being fourth-year students), any discrepancy between their standing on academics or other dimensions provides meaningful (and unflattering) information about differences in ability, skill, or effort devoted to one's college career. In this case, the fourth-year students have no possibility of "catching up" with the upward target by matching his/her achievements. Thus, the upward target is likely to be experienced as a painful and permanent instance of someone outperforming the self, rather than as a source of inspiration. Consistent with this interpretation, Lockwood and Kunda found substantial evidence among the fourth-year students of efforts to denigrate the comparison process and to see the target as dissimilar to the self. In contrast, for first-year students, the information about the successful target served as a model of what might be achieved, and 82% of the first-year students reported feeling inspired by the news account of the superstar's success. It is interesting to note how delicate such effects are. In a second series of studies, Lockwood and Kunda (1999) found that priming people to think of their best possible achievements in college before exposure to the information about the superstar wiped out the motivational benefits of such exposure. Specifically, among students asked to think of the academic and career achievements they hoped to accomplish in the next 10 years, exposure to the superstar resulted in decreased self-evaluation relative to no-comparison con­ trols, whereas benefits of exposure to the superstar were obtained among college students asked to think about their current academic success. Similarly, in another study in the series, Lockwood and Kunda showed that being exposed to the outstanding target after one had already described one's aspirations for the next 10 years had a deflating effect on self-evaluations, whereas exposure to an outstanding target prior to generating one's own expectations resulted in higher aspirations. These latter studies highlight the importance of considering where the com­ parer is, relative to the target, in terms of some developmental sequence. As noted earlier, in terms of our Figure 9.1, in the current-status and pre­ comparison conditions, the comparer is at C1 and the superstar target at T4. The instructions to think about one's future accomplishments may have had the psychological effect of moving the comparer from Cl to C4, thus making the comparison reactive, rather than proactive. As we have seen in other studies, when both superior target and comparer are at the same stage, direct com­ parisons are likely, and in this particular case, decreased self-evaluation is the result (for related evidence and discussion, see Blanton, 2001 ; Reis-Bergan,

286

LISA G. ASPINWALL, DOUGLAS L. HILL AND SAMANTHA L. LEAF

The results of this first study revealed that participants exposed to the superior tape 2 days before their own performance outperformed no­ comparison baseline participants and all other participants, as rated by objec­ tive observers. They also rated success at the speech task as more personally important to them than did other participants. Thus, when participants had time to alter their behavior to meet the high standard set by an upward comparison target, exposure to upward comparisons appeared to be benefi­ cial, and no adverse effects were obtained. Our second study expanded this design to include exposure to a tape of either a superior performance (upward comparison) or an inferior perfor­ mance (downward comparison) at one of the three exposure times (2 days before, just before, or just after participants' own speeches). As in Study 1 , another group o f participants completed the speech task without seeing another student's performance. When participants saw another student's speech 2 days before their own, upward comparisons facilitated perfor­ mance on the speech task, whereas downward comparisons compromised it, relative to a no-comparison baseline. Thus, as predicted by the proactive framework, exposure to upward comparisons prior to performance in an achievement context can improve objective performance. Interestingly, in this situation, exposure to downward comparisons sharply decreased performance. The timing of exposure to the comparison information had strong effects on other aspects of self-regulation of performance, particularly appraisals of task difficulty, preparation in the 2 days between sessions of the study, and self­ evaluations. Specifically, participants exposed to the superior tape 2 days prior to their own speech rated the task as easier and less challengng than did baseline participants, whereas participants exposed to the superior tape just before or right after their own speech rated the task as more difcult and more challenging. Consistent with our proactive versus reactive framework, these results suggest that the meaning of the upward comparison may change, depending on where the comparer is with respect to some action involving the comparison dimension. For the person who has yet to give the speech, the upward tape sets the standard and may make the task seem easier by provid­ ing a model of good performance. However, to the person who is just about to give the speech or who has just given the speech, the upward tape may seem like a noxious benchmark that makes one's own efforts look bad by comparison. As was the case for objective performance, these results were reversed in the downward comparison condition. Perhaps the most dramatic effects were found among the participants who were exposed to the inferior model 2 days prior to their own performance. In addition to performing worse than baseline participants, these participants appraised the task as much more difficult than did other participants and tended to report lower intended preparation times.

288

LISA G. ASPINWALL, DOUGLAS L. HILL AND SAMANTHA L. LEAF

was beneficial and exposure to downward comparisons was detrimental. These latter findings are also highly consistent with Gibbons et al. 's (200) findings indicating that people who reported making downward comparisons experienced worse academic performance over time and that this perfor­ mance decrement was mediated by decreased expectations.

DISCUSSION AND DIRECfiONS FOR FUTURE RESEARCH

Expanding the Comparison Space to Include Proactive Comparisons

Our goal in this chapter was to examine ways in which the understanding of social comparison activity and its consequences could be expanded by con­ sidering proactive comparisons. We suggested that most contemporary stud­ ies of social comparison processes have focused on reactive comparisons. As a result, they have examined ways in which people use downward comparisons to feel better after adversity, and ways in which upward comparisons, al­ though useful in many ways, are often rejected. We presented several reasons why comparisons undertaken in advance of some focal stressor, event, ac­ tivity, or life task might serve a wider range of motives and functions. Specifi­ cally, in proactive situations, when one's future is not written in stone, all kinds of comparisons-upward, lateral, and downward-may provide useful information about one's own likely performance, what to expect, and dif­ ferent ways to achieve or avoid particular outcomes. In addition, because one has not yet experienced the full impact of a negative event, comparison pref­ erences are less likely to be driven downward in the service of the regulation of negative affect. Finally, because many proactive comparison situations involve comparisons to targets who are further along in some course of events, the risks of aversive direct comparisons to superior others may be more easily navigated, and the informational and motivational benefits of upward comparisons appreciated. We reviewed many potential benefits of proactive comparisons. In particu­ lar, upward comparisons appear to serve a wide range of self-regulatory func­ tions, including the evaluation of future outcomes, prediction of performance, inspiration for performance, and the acquisition of high performance stan­ dards. Ultimately, these benefits may facilitate better outcomes, as they did in our two public speaking studies (Aspinwall et al., 200). These findings under­ score the same point-in situations in which one believes one has time and other resources to achieve what the upward target has achieved, upward comparisons may be both useful and inspiring. Upward comparisons seem to be harmful only when one has been baldly outperformed by someone at the same stage and when improvement is either impossible or seems that way (Aspinwall et a/., 200; Lockwood & Kunda, 1997, 199; Major et a/. 1991).

290

LISA G. ASPINWALL, DOUGLAS L. HILL AND SAMANfHA L. LEAF

all points in the self-regulatory process, starting with decisions about whether to adopt a specific goal (Blanton et al., 2001; Ruble & Frey, 1991 ); appraisals of tasks and their demands (Aspinwall et a/., 20); predictions about one's own performance (Smith & Sachs, 1997; Wheeler et al., 1997); the acquisition of specific coping or problem-solving methods (Butler, 1992, 1993, 20); and the allocation of effort toward task preparation {Aspinwall et a/., 200). Social comparisons have also been found to influence perceptions of self-efficacy dur­ ing the course of task engagement (Bandura & Jourden, 1991) and thus may play an important role in sustaining goal-directed behavior. Proactive comparisons may also play an important affective function in generating such sustaining emotions as inspiration and hope (Lockwood & Kunda, 1 997, 1999; Taylor & Lobel, 1989). Of course, proactive comparisons to unattainable good outcomes or to frightening future outcomes have the potential to generate potent negative affects, such as resentment, resignation, fear, and dread, but it is important to note that our review suggested several reasons why such effects are less likely in the proactive setting. Specifically, when the information in the comparison may help the comparer to achieve a positive outcome or to avoid or prepare for the negative outcome, the com­ parison is less likely to be experienced as aversive. Finally, we suggested that proactive comparisons may be an important means through which people learn about the properties of comparison dimen­ sions, such as their range of potential outcomes, their amenability to personal control, and their prospects for improvement or decline. Such comparisons may also provide valuable information about group- or identity-linked out­ comes. In these ways, information gleaned from such comparisons may play a critical role in the adoption of goals (if he can do it, I can do it), the exercise of personal control in pursuit of one's goals (he tried hard, and it worked), and in disengagement from unworkable goals (if he couldn't do it, I can't do it). Taken together, these findings suggest that proactive social comparisons may play an important role in planning, enacting, sustaining, and modifying behavior. Continuing to Expand Social Comparison Research Beyond the Achievement Context

Most of the examples we reviewed were concentrated in the achievement context. Both task acquisition and longer-term academic goals, such as com­ pleting college, provide important examples of the costs and benefits of social comparison activity {Aspinwall & Taylor, 1993; Blanton, Buunk, Gibbons, & Kuyper, 1 999; Gibbons et al., 200). However, these comparison dimensions may be limited in that both discrete task performance and longer-term achievement goals are goals for which one can reasonably expect improve­ ment or at least stability (in the form of a performance plateau), but for which

292

LISA G. ASPINWALL, DOUGLAS L. HILL AND SAMANTHA L. LEAF

program. In both cases, comparers may wish to know what an older target did when she was young that led to her current outcomes. Exposure to information about other people's favorable and unfavorable outcomes may also have strong effects on beliefs about preventive health behavior and its likely consequences. Ample research evidence suggests that comparison instances may be more memorable and more influential than abstract statistics about health risks (see, e.g., Taylor & Thompson, 1982; Nisbett, Borgida, Crandall, & Reed, 1976; see Gibbons & Gerrard, 1997, for discussion of social comparison prototypes and their influence on health-risk behavior). Learning the (unfortunately true) story of a friend's mother who skipped her annual gynecological examination for 3 years and was then diag­ nosed with terminal ovarian cancer that could have been treated earlier may serve as a powerful incentive to have regular cancer screenings. So, too, may the positive instance of someone who had the annual examination and whose cancer was detected early and treated successfully. Research on positive and negative message framing may be useful in delineating the different condi­ tions under which each kind of comparison (the cautionary tale or the success story) is more motivating (Rothman & Salovey, 1997). As people live longer and as more information about the long-term effects of health behavior (e.g., exercise, diet, smoking, drinking) in young adulthood be­ comes available, we can expect people to be paying more attention to diferent kinds of outcomes in aging in an effort to understand which are amenable to personal control earlier in life and how to prepare for those that are not. To date, studies of social comparison and aging have focused on how older people main­ tain a sense of competence in the face of actual or expected declines in health and other areas of life (see, e.g., Brandtstiidter & Greve, 1994; Heckhausen & Brim, 1997; Heidrich & Ryff, 1993; Ruble & Frey, 1991). A more proactive approach to this domain would involve asssing younger people's reactions to older targets who illustrate a variety of potential outcomes in aging, and any changes in current behavior or expectations about aging that result. If proactive upward and downward comparisons to older people prove to be influential in this way, then health-promotion, illness-prevention, and illness-detection campaigns may profitably use such comparisons. One inter­ esting practical and conceptual question that arises from this discussion is how and when such messages should be presented. For example, at what ages do messages about health and age-related topics, such as retirement savings, become relevant and impactful? Is there a latitude of acceptance (Sherif & Hovland, 1961) for such information,2 such that people will be more or less responsive at different ages to pleas for cancer screening or retirement plan­ ning? An intriguing possibility is that these latitudes of acceptance may de­ pend on the social comparison models to which one has been exposed, their zwe are grateful to Miles Hewstone for suggesting this idea.

PROACTIVE SOCIAL COMPARISONS

293

similarity to the self, and other factors, such as cultural views of aging. Com­ bining models of social comparison prototypes and similarity (Gibbons & Gerrard, 1 997) with developmental lifespan approaches to comparison pro­ cesses (Aspinwall, 1997; Heckhausen & Schulz, 1 995; Reis-Bergan et al., 200; Ruble & Frey, 1 991) may provide answers to such questions. CONCLUSION

Examining how people look to others' outcomes to see what the future holds for them, and what they might do about it, may reveal a multiplicity of uses of proactive comparisons for self-regulatory purposes. Proactive comparison ac­ tivity may serve a wide range of goals, including prediction, preparation, motivation, or inspiration, and efforts to understand relevant standards, as people strive to achieve or avoid particular outcomes. We have focused on the potential of upward comparisons to meet these needs, but proactive downward comparisons may prove to be important as well, as they may illustrate what not to do to achieve a particular goal or what may happen if one does not take action. Future work on these topics may increase our understanding of how people use information about other people to shape their life choices and outcomes. ACKNOWLEDGMENTS

The authors wish to thank Hart Blanton, Rick Gibbons, two anonymous reviewers, and the editors for helpful comments on an earlier version of this manuscript. Preparation of this work was supported in part by a grant from the National Science Foundation (SBR-9709677) to the first author. REFERENCES Ahrens, A. H., & Alloy, L. B. (1997). Social comparison processes in depression. In B.P. Buunk & F. X. Gibbons (Eds), Health, Coping, and Well-Being: Perspectives from Social Comparison Theory (pp. 39-410). Mahwah, NJ: Erlbaum. Alicke, M. D., LoSchiavo, F. M., Zerbst, J., & Zhang, S. (1997). The person who outperforms me is a genius: Maintaining perceived competence in upward social comparison. Journal of Personality and Social Psychology, 73, 781-9. Arrowood, A. J., & Friend, R. (1969). Other factors determining the choice of a comparison other. Journal of Experimental Social Psychology, 5, 233-9. Aspinwall, L. G. (1997). Future-oriented aspects of social comparisons: A framework for studying health-related comparison activity. In B. P. Buunk & F. X. Gibbons (Eds), Health, Coping, and Well-being: Perspectives from Social Comparison Theory (pp. 125-65). Mahwah, NJ: Erlbaum.

294

LISA G. ASPINWALL, DOUGLAS L. HILL AND SAMANTHA L. LEAF

Aspinwall, L. G. (1998). Rethinking the role of positive affect in self-regulation. Motivation and Emotion, 22, 1-32. Aspinwall, L. G. (2001). Dealing with adversity: Self-regulation, coping, adaptation, and health. In A. Tesser & N. Schwarz (Eds), Blackwell Handbook of Social Psy­ chology: Vol. 1. lntraindividual Processes (pp. 591-614). Malden, MA: Blackwell. Aspinwall, L. G. (in press, b). Proactive coping, well-being, and health. In N. J. Smelser & P. B. Baltes (Eds), The International Encyclopedia of the Social and Behavioral Sciences. Oxford: Elsevier. Aspinwall, L. G., Frazier, L. E., & Cooper, D. A. (200). Social comparison, self­ regulation, and task performance: Timing is everything (manuscript in preparation). Aspinwall, L. G., & Taylor, S. E. (1993). The effects of social comparison direction, threat, and self-esteem on affect, self-evaluation, and expected success. Journal of Personality and Social Psychology, 64, 708-22. Aspinwall, L. G., & Taylor, S. E. (1997). A stitch in time: Self-regulation and proactive coping. Psychological Bulletin, 121, 417-36. Bandura, A. (1986). Social Foundations of Thought and Action: A Social Cognitive Theory. Englewood Cliffs, NJ: Prentice-Hall. Bandura, A., & Jourden, F. J. (1991). Self-regulatory mechanisms governing the im­ pact of social comparison on complex decision-making. Journal of Personality and Social Psychology, 60, 941-51. Biernat, M., & Billings, L. S. (2001 ). Standards, expectancies, and social comparison. In N. Schwarz & A. Tesser (Eds), The Blackwell Handbook of Social Psychology: Vol. 1, lntraindividual Processes (pp. 257-83). Malden, MA: Blackwell. Blanton, H. (2001 ). Evaluating the self in the context of another: The three-selves model of social comparison assimilation and contrast. In G. B. Moskowitz (Ed.), Cognitive Social Psychology. The Princeton Symposium on the Legacy and Future of Social Cognition (pp. 75-87) Mahwah, NJ: Erlbaum. Blanton, H., Buunk, B. P, Gibbons, F. X., & Kuyper, H. (199). When better-than­ others compare upward: Choice of comparison and comparative evaluation as inde­ pendent predictors of academic performance. Journal of Personality and Social Psychology, 76, 420-30. Blanton, H., Crocker, J., & Miller, D. T. (200) The effects of in-group versus out­ group social comparison on self-esteem in the context of a negative stereotype. Journal of Experimental Social Psychology, 36, 519-30. Blanton, H., George, G., & Crocker, J. (2001 ). Contexts of system justification and system evaluation: Exploring the social comparison strategies of the (not yet) con­ tented female worker, Group Processes and Intergroup Relations, 4, 126-37. BrandsUidter, J., & Greve, W. (1994). The aging self: Stabilizing and protective pro­ cesses. Developmental Review, 14, 52-80. Brewer, M. B. & Weber, J. G. (1994). Self-evaluation effects of interpersonal versus intergroup social comparisons. Journal of Personality and Social Psychology, 66, 268-75. Brown, J. D., Novick, N. J., Lord, K. A., & Richards, J. M. ( 1 992). When Gulliver travels: Social context, psychological closeness, and self-appraisals. Journal of Per­ sonality and Social Psychology, 62, 717-27. Butler, R. (1992). What young people want to know when: Effects of mastery and ability goals on interest in different kinds of social comparisons. Journal of Person­ ality and Social Psychology, 62, 934-43. Butler, R. (1993). Effects of task- and ego-achievement goals on information seeking during task engagement. Journal of Personality and Social Psychology, 65, 18-31 . Butler, R. (200). What learners want to know: The role of achievement goals in shaping information seeking, learning, and interest. In C. Sansone & J. M.

296

LISA G. ASPINWALL, DOUGLAS L. HILL AND SAMANTHA L. LEAF

Heckhausen, J., & Schulz, R. (1995). A life-span theory of control. Psychological Review, 102, 284-304. Heidrich, S. M., & Ryf, C. D. (1993). The role of social comparisons processes in the psychological adaptation of elderly adults. Journal of Gerontology: Psychological Sciences, 48, 1 27-36. Helgeson, V. S., & Mickelson, K. D. (1995). Motives for social comparison. Personality and Social Psychology Bulletin, 21, 1200-209. Jost, J. T., & Banaji, M. R. (1994). The role of stereotyping in system-justification and the production of false consciousness. British Journal of Social Psychology, 33, 1-27 Kahana, E., & Kahana, B. (1996). Conceptual and empirical advances in understand­ ing aging well through proactive adaptation. In V. L. Bengston (Ed.), Adulthood and Aging: Research on Continuities and Discontinuities (pp. 18-40). New York: Springer. Kulik, J. A., & Mahler, H. I. M. (1997). Social comparison, affiliation, and coping with acute medical threats. In B. P. Buunk & F. X. Gibbons (Eds), Health, Coping, and Well-Being: Perspectives from Social Comparison Theory (pp. 227-61). Mahwah, NJ: Erlbaum. Kulik, J. A., Mahler, H. I. M., & Moore, P. J. ( 1996). Social comparison and affiliation under threat: Effects on recovery from major surgery. Journal of Personality and Social Psychology, 71, 967-79. Lazarus, R. S., & Folkman, S. (1984). Stress, Appraisal, and Coping. New York: Springer. Lockwood, P., & Kunda, Z. (1997). Superstars and me: Predicting the impact of role models on the self. Journal of Personality and Social Psychology, 13, 91-103. Lockwood, P., & Kunda, Z. ( 199). Increasing the salience of one's best selves can undermine inspiration by outstanding role models. Journal of Personality and Social Psychology, 76, 214-28. Lyubomirsky, S. & Ross, L. (1997). Hedonic consequences of social comparison: A contrast of happy and unhappy people. Journal of Personality and Social Psychol­ ogy, 73, 1 141-57. Markus, H., & Nurius, P. (1986). Possible selves. American Psychologist, 41, 954-69. Major, B. (1989). Gender differences in comparisons and entitlement: Implications for comparable worth. Journal of Social Issues, 45(4), 99-1 15. Major, B., Testa, M., & Bylsma, W. H. ( 1991). Responses to upward and downward social comparisons: The impact of esteem-relevance and perceived control. In J. Suls & T. A. Wills (Eds), Social Comparison: Contemporary Theory and Research (pp. 237-59). Hillsdale, NJ: Erlbaum. Marsh, H. W. & Parker, J. W. (1984). Determinants of student self-concept: Is it better to be a relatively large fish in a small pond even if you don't learn to swim as well? Journal of Personality & Social Psychology, 47, 213-31. McGrath, J. E., & Beehr, T. A. (1990). Time and the stress process: Some temporal issues in the conceptualization and measurement of stress. Stress Medicine, 6, 93104. Morse, S., & Gergen, K. J. (1970). Social comparison, self-consistency, and the concept of the self. Journal of Personality and Social Psychology, 36, 148-56. Nisbett, R. E., Borgida, E., Crandall, R., & Reed, H. (1976). Popular induction: Infor­ mation is not necessarily informative. In J. S. Carroll & J. W. Payne (Eds), Cognition and Social Behavior (pp. 1 13-33). Hillsdale, NJ: Erlbaum. Reed, M. B., & Aspinwall, L. G. (1998). Self-affirmation reduces biased processing of health-risk information. Motivation and Emotion, 22, 99-132. Reis-Bergan, M., Gibbons, F. X., & Gerrard, M. (200). The impact of reminiscence on socially active elderly women's reactions to social comparisons. Basic and Applied Social Psychology, 22, 225-36.

PROACTIVE SOCIAL COMPARISONS

297

Rothman, A. J., & Salovey, P. ( 1 997). Shaping perceptions to motivate healthy be­ havior: The role of message framing. Psychological Bulletin, 121, 3-19. Ruble, D. N., & Frey, K. S. (1991). Changing patterns of comparative behavior as skills are acquired: A functional model of self-evaluation. In J. Suls, & T. A. Wills (Eds), Social Comparison: Contemporary Theory and Research (pp. 79-1 13). Hillsdale, NJ: Erlbaum. Salovey, P., & Rodin, J. (1984). Some antecedents and consequences of social­ comparison jealousy. Journal of Personality and Social Psychology, 41, 780-92. Sansone, C., & Berg, C. A. (1993). Adapting to the environment across the life span: Different process or different inputs? International Journal of Behavioral Develop­ ment, 16, 215-41. Schachter, S. (1959). The Psychology of Affiliation. Stanford, CA: Stanford University Press. Schwarzer, R. (2000). Manage stress at work through preventive and proactive coping. In E. A. Locke (Ed.) , A Handbook of Basic Principles of Organizational Behavior. Oxford: Blackwell. Sherif, M., & Hovland, C. I. (1961). Social Judgment: Assimilation and Contrast Efects in Communication and Attitude Change. New Haven, CT: Yale University Press. Smith, W. P., & Sachs, P. R. ( 1 997). Social comparison and task prediction: Ability similarity and the use of a proxy. British Journal of Social Psychology, 36, 587-602. Snyder, C. R., Lassegard, M., & Ford, C. E. (1986). Distancing after group success and failure: Basking in reflected glory and cutting off reflected failure. Journal of Person­ ality & Social Psychology, 51, 382-8. Steele, C. M. (1988). The psychology of self-affirmation: Sustaining the integrity of the self. In L. Berkowitz (Ed.), Advances in Experimental Social Psychology (Vol. 21, pp. 261-302). New York: Academic Press. Steele, C. M., Spencer, S. J., & Lynch, M. (1993). Self-image resilience and dissonance: The role of affirmational resources. Journal ofPersonality and Social Psychology, 64, 885-96. Suls, J., & Wills, T. A. (1991). Social Comparison: Contemporary Theory and Research. Hillsdale, NJ: Erlbaum. Swallow, S. R., & Kuiper, N. A. (1988). Social comparison and negative self­ evaluations: An application to depression. Clinical Psychology Review, 8, 55-76. Swallow, S. R., & Kuiper, N. A. (1992). Mild depression and frequency of social comparison. Journal ofSocial and Clinical Psychology, 11, 167-80. Swallow, S. R., & Kuiper, N. A. (1993). Social comparison in dysphoria and non­ dysphoria: Differences in target similarity and specificity. Cognitive Therapy and Research, 17, 103-22. Taylor, S. E., Aspinwall, L. G., Giuliano, T. A., Dakof, G. A., & Reardon, K. (1993). Storytelling and coping with stressful events. Journal of Applied Social Psychology, 23, 703-33. Taylor, S. E., Buunk, B. P., & Aspinwall, L. G. (1990). Social comparison, stress and coping. Personality and Social Psychology Bulletin, 16, 74-89. Taylor, S. E., & Lobel, M. (1989). Social comparison activity under threat: Downward evaluation and upward contacts. Psychological Review, 96, 569-75. Taylor, S. E., Pham, L. B., Rivkin, I. D., & Armor, D. A. (1998). Harnessing the imagination: Mental simulation, self-regulation, and coping. American Psychologist, 53, 429-39. Taylor, S. E., & Schneider, S. K. (1989). Coping and the simulation of events. Social Cognition, 1, 176-96. Taylor, S. E., & Thompson, S. C. (1982). Stalking the elusive vividness effect. Psycho­ logical Review, 89, 155-81.

298

LISA G. ASPINWALL, DOUGLAS L. HILL AND SAMANTHA L. LEAF

Taylor, S. E., Wood, J. V., & Lichtman, R. R. (1983). It could be worse: Selective evaluation as a response to victimization. Journal ofSocial Issues, 39(2), 19-40. Tesser, A. (1988). Toward a self-evaluation maintenance model of social behavior. In L. Berkowitz (Ed.), Advances in Experimental Social Psychology (Vol. 21, pp. 181227). Orlando, FL: Academic Press. Tesser, A., & Cornell, D. P. ( 1991). On the confluence of self processes. Journal of Experimental Social Psychology, 21, 501-26. Tesser, A., Crepaz, N., Beach, S. R. H., Cornell, D. & Collins, J. C. (200). Confluence of self-defense mechanisms: On integrating the self-zoo. Personality and Social Psy­ chology Bulletin, 26, 1476-89. Thornton, D. A., & Arrowood, A. J. (196). Self-evaluation, self-enhancement, and the locus of social comparison. Journal of Experimental Social Psychology, Suppl. 1 , 40-48.

Trope, Y., & Neter, E. (1994). Reconciling competing motives in self-evaluation: The role of self-control in feedback seeking. Journal of Personality and Social Psychol­ ogy, 66, 646-57. Trope, Y., & Pomerantz, E. M. (1998). Resolving conflicts among self-evaluative mo­ tives: Positive experiences as a resource for overcoming defensiveness. Motivation and Emotion, 22, 53-72. van der Zee, K., Buunk, B., & Sanderman, R. (1996). The relationship between social comparison processes and personality. Personality and lndividUll Differences, 20, 551-65. van der Zee, K., Buunk, B., & Sanderman, R. (1998). Neuroticism and reactions to social comparison information among cancer patients. Journal of Personality, 66, 177-93. Wenzlaff, R. M., & Beevers, C. G. (1998). Depression and interpersonal responses to others' moods: The solicitation of negative information about happy people. Person­ ality and Social Psychology Bulletin, 24, 386-98. Wheeler, L. (1966). Motivation as a determinant of upward comparison. Journal of Experimental Social Psychology, Suppl. 1 , 27-31 . Wheeler, L., Shaver, K . G., Jones, R . A., Goethals, G . R., Cooper, J., Robinson, J . E., Gruder, C. L., & Butzine, K. W. (1969). Factors determining choice of a comparison other. Journal of Experimental Social Psychology, S, 219-32. Wheeler, L., Martin, R., & Suls, J. (1997). The proxy model of social comparison for self-assessment of ability. Personality and Social Psychology Review, l, 54-61. Wills, T. A. (1981 ). Downward comparison principles in social psychology. Psychologi­ cal Bulletin, 90, 245-71 . Wills, T . A. (1997). Modes and families o f coping: A n analysis o f downward com­ parison in the structure of other cognitive and behavioral mechanisms. In B. P. Buunk & F. X. Gibbons (Eds), Health, Coping, and Well-Being: Perspectives from Social Comparison Theory (pp. 167-93). Mahwah, NJ: Erlbaum. Wood, J. V. (1989). Theory and research concerning social comparisons of personal attributes. Psychological Bulletin, 106, 231-48. Ybema, J. F., & Buunk, B. P. (1993). Aiming at the top? Upward social comparison of abilities after failure. European Journal of Social Psychology, 23, 627-45.

302

PAUL B. PAULUS ET A L .

Theoretical Explanations

A major focus of research on group brainstorming has been uncovering the basis for the production loss in groups. Reviews by Lamm and Trommsdorff (1973), Mullen et al. (1991), Stroebe & Diehl (1994), and Paulus, Larey, and Dzindolet (200) suggest that a number of social factors influence group brainstorming. Evaluation Apprehension It is emphasized in the brainstorming instructions that group members should not evaluate each others' contributions. Yet even though there may be no explicit evaluation in the group, there is likely to be much implicit evaluation. Most social contexts have a strongly evaluative component in which individ­ uals can compare performance with one another and with various standards, based on past experience (Cottrell, 1972; Sanna, 1992). These evaluative tend­ encies are not likely to be "shut off'' by simple instructions. So, even individ­ uals in brainstorming groups where judgment is deferred may still be concerned with the evaluative perceptions of other group members (Paulus & Murdoch, 1971). Several studies have examined whether variations in the evaluative context wil influence the brainstorming performance. Diehl and Stroebe (1987) manipulated the extent to which brainstorming performance could be evaluated later by external judges. The potential of external evalua­ tion reduced performance when participants were individually accountable, but it did so for both groups and individuals. Similarly, research on electronic brainstorming, in which individuals share ideas by means of a computerized system, has found that groups in which individuals cannot be identified with their performance (anonymous conditions) tend to perform better than groups in which ideas can be linked to specific individuals (Cooper, Gallupe, Pollard, & Cadsby, 1998). It is evident from this research that individuals are sensitive to the evaluative cues in their context and that these can inhibit the idea-generation process. However, it is not clear that this factor is responsible for the production loss in brainstorming groups. It is difficult to completely eliminate from a face-to-face interactive setting in order to demonstrate that such an elimination will over­ come the production-loss effect. An alternative strategy is to select individuals based on their sensitivity to evaluation in a group context. Some individuals appear to be much less sensitive to such cues (low interaction anxiety) than others (high interaction anxiety) (Leary & Kowalski, 1993). When brainstorm­ ing groups are composed of low-interaction-anxious individuals, the typical production loss is much reduced (Camacho & Paulus, 1995). Therefore, it appears that concern about evaluation of one's ideas in a group context is one factor that may inhibit the productivity of brainstorming groups.

304

PAUL B. PAULUS ET A L .

talking. In this situation, blocking was introduced because only one group member could speak at a time. This procedure produced performance similar to that of a group condition. Hearing the ideas generated by others in addition to the light procedure did not lead to a further reduction in performance. These results suggest that not being able to talk while others are talking is an important factor in the productivity loss in group brainstorming. The fact that type of group condition accounts for considerably more variance than manip­ ulations of evaluation apprehension and identifiability led Diehl and Stroebe (1987, 1991) to conclude that blocking is the main factor responsible for production loss in groups. Some research has used computer-based idea exchange to overcome the effects of production blocking in the sharing process (e.g., Gallupe, Bas­ tianutti, & Cooper, 1991). With this technology, participants do not have to coordinate their idea generation with other group members, since they can enter ideas into a common shared file at any time. The ideas generated by the group are typically available on the top half of the computer screen. Small groups using this technology perform better than verbal brainstorming groups and about the same as electronic brainstormers who perform the task alone. When various procedures are used to introduce blocking into computer brain­ storming (such as delays after idea generation, and tum-taking), this type of brainstorming is no longer more effective than verbal brainstorming ( Gallupe, Cooper, Grise, & Bastianutti, 1994).

SOCIAL COMPARISON PROCESSES IN GROUP BRAINSTORMING

It is clear from the literature on group brainstorming and on group perfor­ mance in general that evaluation apprehension, motivation losses, and pro­ duction blocking play a significant role in limiting the ability of groups to reach their performance potential. However, group interaction also provides an opportunity for individuals to make social comparisons of their perfor­ mance with those of the other group members. These types of comparisons could play a role in increasing the production losses in groups, or could motivate groups to overcome the inhibiting factors. We will summarize the basic elements of a social comparison perspective for group brainstorming and related empirical support. Social comparison theory is concerned primarily with selection of com­ parison persons to evaluate our opinions, abilities, and emotions (Sols & Miller, 1977). In the absence of objective indicators, people are presumed to be motivated to use comparison others as reference points in self-evaluations (Festinger, 1954). Although there is a general tendency to compare ourselves to others who are similar in attributes (Goethals & Darley, 1987), with regard

306

PAUL B. PAULUS ET A L.

performance level, partly because there are likely to be many pauses during their session as they try to think of new ideas. Downward Social Comparison

Yet how can a social comparison perspective explain production losses in groups? Paulus and Dzindolet (1993) presume that brainstorming groups may foster a tendency toward downward comparison. Since groups start their session by performing at a relatively low level, there should be a tendency for high performers to feel like "deviants". As a result they may move their performance in the direction of the low group standard. They may be partly motivated by a desire to not look foolish or "play the sucker" by taking the task more seriously than other group members (Kerr, 1983). Downward com­ parison processes have also been demonstrated in achievement situations. In these contexts individuals may be motivated to compare downward in order to enhance their self-esteem. However, continued comparisons of this type may lower one's performance standards or goals (Gibbons & Gerrard, 1991; Gibbons, Blanton, Gerrard, Buunk & Eggleston, 200; Larey & Paulus, 1995). Several studies have provided evidence for downward comparison in group brainstorming. Paulus and Dzindolet (1993) found that the initial perfor­ mance level of the two lowest performing members predicted the perfor­ mance of a group of four toward the end of their session, but the initial performance level of the two highest performing members did not (Study 3). Camacho and Paulus (1995) investigated brainstorming productivity in groups in which members' levels of interaction anxiety varied (Leary & Kowalski, 1993), and found that groups composed of all high-anxious individuals per­ formed significantly more poorly than groups composed of all low-anxious individuals (see Table 10. 1 ). Mixed groups which contained two high- and two low-anxious individuals performed close to the level of the high-anxious groups, suggesting that the low-anxious individuals matched their perfor­ mance to that of the low-performing high-anxious individuals. The downward comparison effect is also seen in goal setting in groups. Brainstorming groups set much lower goals for the number of ideas they will generate per individual than do individual brainstormers (Larey & Paulus, 1995). Table 10.1

Number of ideas in a brainstorming task as a function of group composition

M SD

Low anxious

High anxious

Mixed

85.5 14.5

40.8 13.0

53.2 15.9

Adapted from Camacho & Paulus, 1995.

308

PAUL B. PAULUS ET A L .

doing better than their partner or who did not receive any feedback. Coskun (200) found similar results using intergroup comparisons. Individuals and groups were given feedback after one performance session that they had performed better or worse than some comparison individual or group. The individuals and groups who received negative feedback showed enhanced performance in the second session, compared to the positive feedback and no feedback conditions. Paulus and Dzindolet (1993) found that providing groups or individuals with a high comparison standard at the beginning of the brainstorming session also had a strong positive effect on performance. In the experiment, some groups were instructed, prior to their brainstorming ses­ sion, that typical groups tend to generate about 100 ideas in a 25-minute session while other groups were not. The rate of 100 ideas in 25 minutes is actually 2.5 times greater than the rate usually achieved by brainstorming groups and represents a high comparison standard. The results indicated that groups generated more ideas when they received social comparison informa­ tion than when they did not receive social comparison information (see Table 10.2). In another context, Blanton, Buunk, Gibbons, and Kuyper (199) found that the academic performance of Dutch school children tended to improve if they compared their examination grades with high-performing students. Table 10.2

task as a information

M SD

Number of ideas in a brainstorming function of social comparison No information

Information

39.9 14. 1

62.4 25.9

Adapted from Paulus et a/., 1993.

Evidence is thus clear that, with brainstorming and other tasks, feedback that someone else has performed better seems to motivate individuals to higher levels of performance. Of course, if there is a very large discrepancy in performance, individuals may not use the other person as a relevant com­ parison person, and thus demonstrate little change in performance (Seta, 1 982). Interestingly, performing better than others is not associated with sub­ sequent increased levels of performance. This overall pattern of results is consistent with the upward comparison bias presumed to exist for ability comparisons (Festinger, 1954; Gibbons et a/., 200). Sumary A social comparison/matching perspective suggests a variety of possible outcomes for group interaction--convergence, downward matching, and up-

SOCIAL AND COGNITIVE INFLUENCES

309

ward matching. Evidence exists for each of these three patterns, but further research is required to delineate clearly the conditions under which they will occur. Our discussion suggests some likely possibilities. Convergence to some model level of performance should be typical for all groups, given the pressures for uniformity and conformity that may exist in such groups. Al­ though we have observed such effects in short-term laboratory groups, they would seem even more likely for long-term work or decision-making groups that are quite cohesive (Janis, 1 982; Roethlisberger & Dickson, 1939; Rom­ metveit, 1 955). The normative pressures may go in an upward or downward direction, depending on the social context. Under competitive conditions, when individuals are individually accountable, when self-improvement mo­ tives are dominant, upward matching processes may dominate (Gibbons et al., 2000). When there is little personal or group incentive, individual contri­ butions are dispensable, and efforts are costly (Shepperd, 1 993), or self­ enhancement motives are dominant (concern for evaluation, avoidance of embarrassment or playing the sucker), a downward matching process may be dominant. Upward Social Comparison versus Blocking

One interesting implication of a social comparison perspective is that the presence of highly productive members in a group could motivate increased performance by the other group members if the downward comparison pro­ cess could be counteracted or minimized. Such an increase in performance, associated with high activity levels of co-workers, would be in contrast to predictions from a blocking perspective. Highly productive group members should block opportunities for brainstorming of their co-workers. We have conducted a series of studies to assess these contrasting predictions. In these studies we manipulated the activity level of one pair of participants and assessed the influence on the performance level of the other two participants (Dzindolet, 1992; Roland, 1 994). Several studies using confederates who either performed at a very low level or not at all demonstrated no effects on the other two group members. The two naive participants seemed to ignore the inactive pair and function essen­ tially as a dyad. In another study, one pair was given prior experience on the brainstorming task for 20 minutes. They then joined a pair of inexperienced participants to brainstorm for another 20 minutes. In one condition the expe­ rienced pair was allowed to use any ideas generated in the prior session and in another condition they were not. Participants who generated a high number of ideas in the first session also generated more new ideas in the second session. Moreover, inexperienced pairs who were matched with productive experienced pairs in the second session had higher performance levels. This suggests a positive social influence effect of the productive pairs. The second

310

PAUL B . PAULUS ET A L.

session data were also coded in terms of how much of the time inexperienced pairs were blocked by the idea generation of the experienced pair, and how many ideas were generated by the experienced pairs. It was found that inex­ perienced pairs exposed to high-performing experienced pairs had a signifi­ cantly higher rate of performance. However, inexperienced partners who received a high amount of blocking did not differ in performance from those who received a low amount of blocking. The results of this research provide some additional support for the role of social comparison factors. However, the studies did not attain their goal of providing an unambiguous test of the relative importance of social com­ parison and blocking factors. It is likely that both factors play a role in group brainstorming and their effects may be counteracting one another. Product­ ive partners may increase motivation to perform, but their high rates of idea generation may limit opportunities for others to contribute. This was dem­ onstrated in a study in which students typed ideas on a computer (Paulus et al., 1996) . In this study, group members either verbalized or did not verbal­ ize their ideas as they typed them. Additionally, group members were either provided or not provided with performance feedback about the other group members. In the performance feedback conditions, individuals announced to the group how many ideas they generated every 5 minutes. The results indicated that performance feedback increased performance compared to no feedback, but verbalization inhibited performance relative to the no verbalization condition (see Table 10.3). This study demonstrated both the potential positive effects of social comparison and the interfering effects of oral idea sharing. The key to overcoming the negative effects of idea sharing may be for productive partners to make their contributions in an efficient manner, so as to leave sufficient time/opportunities for others to make their contribution (Brown & Paulus, 1 996). For example, Dzindolet (1992) found that the most productive inexperienced pairs were those that were matched with highly productive experienced pairs, who took very little time to ex­ press their ideas. Table 10.3

Number of ideas in a brainstorming task as a function of performance feedback and verbalization Performance feedback

No performance feedback

Verbalization

M SD

73.60 19.22

65.29 1 2.49

No verbalization

M SD

102.40 43.25

75.77 19.04

Adapted from Paulus et a/., 1996.

312

PAUL B. PAULUS ET A L.

these have been shown to be influential in different studies and their impact may depend on the nature of the manipulation, context, and participants (e.g., Larey & Paulus, 1999). Social comparison processes may lead to a con­ vergence in the rate of idea generation among participants and to the form­ ation of group performance norms. Because of the inhibitory effects of motivation losses, evaluation apprehension, and production blocking, groups may converge to rather low levels or norms. This tendency to a low rate of performance may be enhanced by a downward comparison process and illu­ sions that the group is actually doing quite well (Paulus et al., 1993). The resulting low performance norms may be maintained in subsequent group sessions (Paulus & Dzindolet, 1993). The social convergence processes may also lead groups to generate ideas in a limited set of conceptual categories, rather than considering a broad range of categories (Lamm & Trommsdorff, 1973; Larey & Paulus, 1999). Performance in brainstorming groups can be enhanced by various social and cognitive influences. Group interaction can enhance motivation to per­ form at high levels because of social facilitation (heightened arousal; Zajonc, 1980), by increasing accountability, inducing a sense of competition, and by producing a tendency toward upward comparison. Upward comparison is likely when individuals are individually accountable and there is some incen­ tive to perform as well as, or better than, others (Shepperd, 1993). Cognitive stimulation should result from exposure to novel ideas or category labels that can prime additional ideas or categories (Brown, Tumeo, Larey, & Paulus, 1998). Cognitive stimulation requires some degree of attention to, and mem­ ory for, the shared ideas. Some degree of diversity in idea pools of the group members is likely to facilitate generation of novel combinations (Diehl, 1992). Benefits of group cognitive stimulation may be most evident after some period of solitary reflection or incubation. The social and cognitive influence model suggests a number of ways in which production losses can be reduced and production gains can be demonstrated. We will review evidence for such outcomes. REDUCING PRODUCTION LOSSES

The social/cognitive influence model suggests that group performance can be improved by counteracting the negative forces that occur during group inter­ action. Making individuals accountable (Shepperd, 1993), increasing the value of task performance (Karau & Williams, 1993), reducing evaluation ap­ prehension (Camacho & Paulus, 1995; Cooper et al., 1998), and limiting pro­ duction blocking (Diehl & Stroebe, 1991) are all useful strategies to increase performance in brainstorming groups. Encouraging the development of high group norms, providing high performance standards or highly productive and

316

PAUL B. PAULUS ET A L.

ideas generated in brainstorming sessions can be organized into categories. For example, brainstorming about ways to improve their university could lead students to consider categories of ideas such as classes, instruction, and stu­ dent life. Each category can then be further subdivided into subcategories (Larey & Paulus, 1999). This clustering of ideas may actually represent the associative structure or semantic network related to the issue (Collins & Loftus, 1975). There are several simple presumptions that seem basic to a cognitive model of group brainstorming. First, cognitive stimulation by idea exchange pre­ sumes that individuals attend carefully to the ideas and category information exchanged during the group session. This may be difficult when participants are attempting to generate their own ideas and listen to the ideas generated by others at the same time. Second, ideas from others should be most useful if they do not overlap precisely with ideas readily accessible to the participant. The most stimulating ideas should be those that remind individuals of catego­ ries of ideas or specific ideas/knowledge that they might not have considered on their own. Therefore, the stimulation value of other brainstormers is likely to depend on the number of ideas they generate, the uniqueness of those ideas, and the diversity of ideas or categories. Although there may be much potential stimulation as a result of idea ex­ change, the benefit of exposure may not be fully evident in a group brain­ storming session. In such a session, individuals are likely to be preoccupied with both generating their own ideas and the ongoing social interaction pro­ cesses. To derive full benefit from associations generated by the idea sharing process, it may be necessary to provide an opportunity for incubation (Csikszentmihalyi & Sawyer, 1995; Torrance, 1972). That is, it may take some time for individuals to process (consciously or unconsciously) the group­ induced stimulation. Providing a solitary brainstorming session after group brainstorming may be one means of tapping the potential cognitive stimu­ lation experienced in the group exchange process. In our current research program, we have been examining both the social and cognitive factors that may enable brainstorming groups to demonstrate production gains relative to nominal groups. Evidence for Production Gains in Brainstorming Groups

One of the aims of our studies was to determine whether one brainstormer could process the information provided by another brainstormer. So far there has been no evidence for such an effect. Diehl and Stroebe ( 1987) found no difference between those who heard others' ideas while they were brain­ storming and those who did not. Electronic brainstormers exposed to ideas from eight or fewer other brainstormers do not demonstrate any benefit from such exposure, compared to individual electronic brainstormers (e.g., Gallupe

320

PAUL B. PAULUS ET A L .

memory instructions, as in our prior studies or some other techniques, might facilitate this process. Groups consisting of individuals with diverse experi­ ences or knowledge bases are most likely to allow groups to develop a wide range of ideas by linking ideas from these diverse sources (Diehl, 1992). Finally, group members may need some time to process the shared ideas and to connect them to their various knowledge bases. This type of incubation effect has been demonstrated in our studies by using an individual brain­ storming session after an exposure or exchange session (Dugosh Paulus

&

Yang,

20).

et al.,

200;

IMPLICATIONS AND FUTURE DIRECfiONS Our summary and analysis of the brainstorming literature suggest that inter­ active brainstorming can be both beneficial and harmful. Which of these outcomes occurs depends on a range of variables that influence the social and cognitive processes in groups. Therefore, the "optimistic" practitioners and the "pessimistic" groups researchers may both be right. Groups left to their own devices are not likely to be very productive, particularly if they are involved in a verbal exchange of ideas. When group interaction is structured so as to limit the inhibitory processes and facilitate the social and cognitive stimulation potential, groups can achieve high levels of idea generation and be effective means of increasing innovation and productivity in teams and organ­ izations. We agree with Osborn (1963), that a mixture of individual and group brainstorming will be optimal. The same principle probably applies to other performance domains, such as intellectual teamwork (e.g., Csikszentmihalyi

&

Sawyer, 1995; Dunbar, 1997). Intellectual progress and creativity requires

both the exchange of information and ideas among knowledgeable and di­ verse collaborators and personal reflection and analysis of the various im­ plications of the shared information. Although there is a tendency to attribute "lone genius" characteristics to many who have achieved great eminence, a careful analysis of their lives often reveals the important role of social factors in the development of the many creative contributions of eminent scholars and artists (Simonton, 1999). Although we have proposed a model of the social and cognitive factors that contribute to group creativity, there is still much research required to evaluate the relative importance of various components of this model. Moreover, much of our work has been restricted to laboratory studies of college students. Theory-based studies with populations in various work, educational, and therapeutic contexts are needed to assess the generality of our findings and theoretical notions (Ford, 199; Paulus, 2001 ; Paulus

&

Paulus, 1997). If the

laboratory research on idea generation does generalize to these other con­ texts, investigators should find that rather minor modifications in the group

SOCIAL A ND COGNITIVE INFLUENCES

321

process or task structure can have a major impact on the effectiveness of groups. Hopefully, this will lead to a practice with groups and teams based on scientific findings, rather than illusions about the effectiveness of groups or teams (Thompson, 2000). REFERENCES Bandura, A. (1989). Social Cognitive Theory. Englewood Cliffs, NJ: Erlbaum. Bennis, W., & Biederman, P. W. (1997). Organizing Genius: The Secrets of Creative Collaboration. Reading, MA: Addison Wesley. Blanton, H., Buunk, B. P., Gibbons, F. X., & Kuyper, H. (1999). When better-than­ others compare upward: choice of comparison and comparative evaluation as inde­ pendent predictors of academic performance. Journal of Personality and Social Psychology, 76, 420-30. Brown, V., & Paulus, P. (1996). A simple dynamic model of social factors in group brainstorming. Small Group Research, 21, 91-1 14. Brown, V., Tumeo, M., Larey, T. S., & Paulus, P. B. (1998). Modeling cognitive interactions during group brainstorming. Small Group Research, 29, 495-526. Bouchard, T. J., & Hare, M. (1970). Size, performance, and potential in brainstorming groups. Journal ofApplied Psychology, 54, 51-5. Camacho, L. M. (1991) . Understanding productivity loss in brainstorming groups: A new approach (unpublished doctoral dissertation). The University of Texas at Arlington. Camacho, L. M., & Paulus, P. B. (1995) . The role of social anxiousness in group brainstorming. Journal of Personality and Social Psychology, 68, 1071-80. Clawson, V. K., Bostrom, R. P. & Anson, R. (1993). The role of the facilitator in computer-supported meetings. Small Group Research, 24, 547-65. Cohen, S. G., & Bailey, D. E. (1997). What makes teams work: Group effectiveness research from the shop floor to the executive suite. Journal ofManagement, 23, 23990. Collins, A. M. & Loftus, E. F. (1975). A spreading-activation theory of semantic processing. Psychological Review, 82, 407-28. Connolly, T., Routhieaux, R. L., & Schneider, S. K. ( 1993). On the effectiveness of group brainstorming: Test of one underlying cognitive mechanism. Small Group Research, 24, 490-503. Cooper, W. H., Gallupe, R. B., Pollard, S., & Cadsby, J. (1998). Some liberating effects of anonymous electronic brainstorming. Small Group Research, 29, 147-78. Coskun, H. (200). The effects of outgroup comparison, social context, intrinsic motivation, and collective identity in brainstorming groups (unpublished doctoral dissertation). The University of Texas at Arlington. Coskun, H., Paulus, P. B., Brown, V., & Sherwood, J. J. (200). Cognitive stimulation and problem presentation in idea generation groups. Group Dynamics: Theory, Research, and Practice, 4, 307-29. Cottrell, N. B. (1972). Social facilitation: In C. G. McClintock (Ed.), Experimental Social Psychology. New York: Holt. Csikszentmihalyi, M., & Sawyer, K. ( 1 995) . Creative insight: The social dimension of a solitary moment. In R. J. Sternberg & J. E. Davidson (Eds), The Nature of lnsight (pp. 329-63). Cambridge, MA: MIT Press. Dennis, A. R., & Valacich, J. S. (1993) Computer brainstorms: More heads are better than one. Journal ofApplied Psychology, 78, 531-7.

322

PAUL B. PAULUS ET A L .

Diehl, M. (1992). Production losses in brainstorming groups: The effects of group composition on fluency and flexibility of ideas. Paper presented at the Joint Meeting of the European Association of Experimental Social Psychology and the Society for Experimental Social Psychology, Leuven/Louvain-la-Neuve, Belgium. Diehl, M., & Stroebe, W. (1987). Productivity loss in brainstorming groups: Toward the solution of a riddle. Journal of Personality and Social Psychology, 53, 497-509. Diehl, M., & Stroebe, W. (1991). Productivity loss in idea-generating groups: Tracking down the blocking effect. Journal of Personality and Social Psychology, 61, 392-403. Dugosh, K. L., Paulus, P. B., Roland, E. J., & Yang, H.-C. (200). Cognitive stimu­ lation in brainstorming. Journal of Personality and Social Psychology, 19, 722-35. Dunbar, K. ( 1 995). How scientists really reason: Scientific reasoning in real-world laboratories. In R. J. Sternberg & J. E. Davidson (Eds), The Nature of Insight (pp. 365-95). Cambridge, MA: MIT Press. Dunbar, K. (1997). How scientists think: On-line creativity and conceptual change in science. In T. B. Ward, S. M. Smith & J. Vaid (Eds), Creative Thought: An Investiga­ tion of Conceptual Structures and Processes. (pp. 461-93) Washington, DC: Ameri­ can Psychological Association. Dzindolet, M. T. (1992). An assessment of blocking and social influence processes in brainstorming (unpublished doctoral dissertation). University of Texas at Arlington. Festinger, L. (1950). Informal social communication. Psychological Review, 51, 271-82. Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 1 17-40. Ford, C. M. (1999). Thinking big about small groups in the real world: Comment on Craig & Kelly (1999). Group Dynamics: Theory, Research, and Practice, 3, 257-62. Gallupe, R. B., Bastianutti, L. M., & Cooper, W. H. (1991 ). Unblocking brainstorms. Journal of Applied Psychology, 76, 137-42. Gallupe, R. B., Cooper, W. H., Grise, M. L., & Bastianutti, L. M. ( 1 994). Blocking electronic brainstorms. Journal of Applied Psychology, 19, 77-86. Gibbons, F. X., Blanton, H., Gerrard, M., Buunk, B., & Eggleston, T. (200). Does social comparison make a difference? Optimism as a moderator of the relation between comparison level and academic performance. Personality and Social Psy­ chology Bulletin, 26, 637-48. Gibbons, F. X., & Gerrard, M. (1991). Downward comparison and coping. In J. M. Suls & T. A. Wills (Eds), Social Comparison: Theory and Practice (pp. 317-45). HiJisdale, NJ: Erlbaum. Goethals, G. R., & Darley, J. M. (1987). Social comparison theory: Self-evaluation and group life. In B. Mullen & G. R. Goethals (Eds), Theories of Group Behavior (pp. 22-47). New York: Springer-Verlag. Greene, R. J. (1987). Brainwriting-An effective way to create more ideas. The Quality Circles Journal, 10, 33-6. Harkins, S. G. (1987). Social loafing and social facilitation. Journal of Experimental Social Psychology, 23, 1-18. Janis, I. (1982). Groupthink (2nd Edn). Boston: Houghton Mifflin. Jessup, L. M., & Valacich, J. S. (1993) (Eds), Group Support Systems. New York: Macmillan. Karau, S. J., & Williams, K. D. (1993). Social loafing: A meta-analytic review and theoretical integration. Journal of Personality and Social Psychology, 65, 681-706. Kayser, T. A. (1995). Mining Group Gold: How to Cash in on the Collaborative Brain Power of a Group. Chicago, IL: Irwin. Kerr, N. L. (1983). Motivation losses in small groups: A social dilemma analysis. Journal of Personality and Social Psychology, 45, 819-28. Kramer, M., Kuo, C. L., & Dailey, J. C. (1997). The impact of brainstorming tech­ niques on subsequent group processes: Beyond generating ideas. Small Group Re­ search, 28, 21 8-42.

324

PAUL B. PAULUS ET AL.

Paulus, P. B., Dzindolet, M. T., Poletes, G., & Camacho, L. M. (1993). Perception of performance in group brainstorming: The illusion of group productivity. Personality and Social Psychology Bulletin, 19, 78--89. Paulus, P. B., Larey, T. S., & Dzindolet, M. T. (200). Creativity in groups and teams. In M. Turner (Ed.), Groups at Work: Advances in Theory and Research, pp. 319-38. Hillsdale, NJ: Erlbaum. Paulus, P. B., Larey, T. S., & Ortega, A. H. (1995). Performance and perceptions of brainstormers in an organizational setting. Basic and Applied Social Psychology, 17, 249-65. Paulus, P. B., Larey, T. S., Putman, V. L., Leggett, K. L., & Roland, E. J. (1996). Social influence process in computer brainstorming. Basic and Applied Social Psychology, 18, 3-14. Paulus, P. B., & Murdoch, P. (1971). Anticipated evaluation and audience presence in the enhancement of dominant responses. Journal of Experimental Social Psychol­ ogy, 7, 280-91. Paulus, P. B., & Paulus, L. E. (1997). Implications of research on group brainstorming for gifted education. Roeper Review, 19, 225-9. Paulus, P. B., & Yang, H. C. (200). Idea generation in groups: A basis for creativity in organizations. Organizational Behavior and Human Decision Processes, 82, 76-87. Pinsonneault, A., Barki, H., Gallupe, R. B., & Hoppen, N. (1 99). Electronic brain­ storming: The illusion of productivity. Information Systems Research, 10, 1 10-33. Price, K. H. (1993). Working hard to get people to loaf. Basic and Applied Social Psychology, 14, 329-4. Roethlisberger, F., & Dickson, W. (1939). Management and the Worker. Cambridge, MA: Harvard University Press. Roland, E. ( 1 994). The impact of co-worker performance on brainstorming (un­ published honors thesis). University of Texas at Arlington. Rommetveit, R. (1955). Social Norms and Roles: Explorations in the Psychology of Enduring Social Pressures. Minneapolis: University of Minnesota Press. Roy, M. C., Gauvin, S., & Limayem, M. (1996). Electronic group brainstorming: The role of feedback on productivity. Small Group Research, 27, 215-47. Sanna, L. J. (1992). Self-efficacy theory: Implications for social facilitation and social loafing. Journal of Personality and Social Psychology, 62, 774-86. Seta, J. J. (1982). The impact of comparison processes on co-actors' task performance. Journal of Personality and Social Psychology, 42, 281-91. Shepherd, M. M., Briggs, R. 0 . , Reinig, B. A, Yen, J., & Nunamaker, J. F. Jr (19951996) Invoking social comparison to improve electronic brainstorming: Beyond anonymity. Journal of Management Information Systems, 12, 155-70. Shepperd, J. A. (1993). Productivity loss in performance groups: A motivation analysis. Psychological Bulletin, 113, 67-81. Simonton, D. K. (1999). The Origins of Genius: Darwinian Perspectives on Creativity. New York: Oxford University Press. Stroebe, W., & Diehl, M. (1994). Why groups are less effective than their members: On productivity losses in idea-generating groups. In W. Stroebe & M. Hewstone (Eds), European Review of Social Psychology (Vol. 5, pp. 271-303). Chichester: Wiley. Stroebe, W., Diehl, M., & Abakoumkin, G. (1992). The illusion of group effectivity. Personality and Social Psychology Bulletin, 18, 643-50. Suls, J. M., & Miller, R. L. (Eds) (1977). Social Comparison Processes: Theoretical and Empirical Perspectives. Washington, DC: Hemisphere. Sutton, R. 1., & Hargadon, A. (1996). Brainstorming groups in context. Administrative Science Quarterly, 41, 685-718. Taylor, D. W., Berry, P. C., & Block, C. H. (1958). Does group participation when using brainstorming facilitate or inhibit creative thinking? Administrative Science Quarterly, 3, 23-47.

This page intentionally left blank

Author Index

Page numbers in italic refer to tables or figures. Aarts, H., 13, 18, 27, 28 Abakoumkin, G., 301 Abele, A. E., 129 Abelson, R. P., 156, 192, 202-3, 206-7, 212, 221, 222, 225 Abraham, C., 2, 4, 16, 22, 23, 206 Adler, N. E., 3 Ager, J., 21 6 Agnew, C. R., 74, 240, 260 Ahn, W. K., 224 Ahrens, A. H., 276 Aiken, L. S., 54 Ainslie, G., 39-40, 53 Ainsworth, M. D., 241 Ajzen, I., 1, 2, 3, 5, 7, 10, 1 1 , 12, 17, 19, 20, 21, 40, 46, 50, 53, 54, 60, 73, 82 Akst, L. M., 75, 76 Alicke, M. D., 278, 285 Allen, M., 214 Alloy, L. B., 276 Allport, G. W., 37, 41 Amabile, T. M., 244 Anderson, C. A., 127, 170, 173 Anderson, L. R., 74 Anderson, N., 178 Anderson, N. B., 178 Anderson, N. H., 84 Anderson, P. A., 217 Anson, R., 313 Antaki, C., 217 Argyle, M., 253

Arkin, R., 238 Arkin, R. M., 242 Armitage, C. J., 3, 4, 1 1 , 12, 18, 49, 5 1, 52, 54, 55, 58-9, 60-1 , 63 Armor, D. A., 273 Armstrong, T. L., 79, 85 Arndt, J., 238 Aron, A., 240, 257 Aron, E. N., 240, 257 Aronson, J., 176 Arriaga, X. B., 240 Arrowood, A. J., 277 Asch, S. E., 125, 151 Aspinwall, L. G., 268, 269, 271 , 273, 275, 276, 278, 279, 285-8, 289, 290, 293 Atkinson, J. W., 150 Axelrod, R., 257 Baggett, W. B., 218 Bagozzi, R. P., 3, 17, 18, 20, 21, 80 Bailey, D. E., 300 Bainbridge, J., 61 Baltes, P. B., 214 Banaji, M. R., 41, 156, 192, 281 Bandura, A., 150, 173, 189, 273, 290, 305 Bargh, J. A., 25, 26, 52, 61, 63, 184 Bargh, J. H., 228 Barki, H., 314 Baron, R. M., 28, 54, 55, 56 Bassili, J., 93, 94

328

Bassili, J. N., 3, 20, 41 , 42, 43, 52, 60, 61, 63 Bastianutti, L. M., 304 Bat-Chava, Y., 184 Baumeister, R. F., 1 50, 177, 240 Baumgartner, H., 2 1 Baxter, J. C . , 74 Bayton, J. A., 84 Beach, S. R. H., 240 Beardsworth, A., 55 Beattie, M., 185 Beaufoy, S. L., 75, 77, 78, 79, 81, 83 Beckman, L., 242 Beckmann, J ., 21 Beehr, T. A., 268 Beevers, C. G., 276 Bell, D. W., 44, 60, 62, 76, 79, 86, 87, 88, 90-1, 92-3 Bell, R., 3, 44 Bern, D. J., 93 Benbow, C. P., 278 Bennis, W., 300 Berent, M. K., 63, 74 Berg, C. A., 268 Bernard, J., 241 Berry, P. C., 301 Berscheid, E., 251 Bertson, G. G., 46 Betz, A. L., 238 Biciotti, E., 74 Biddle, S. J. H., 19 Biederman, P. W., 300 Biek, M., 89 Biernat, M., 103, 104, 1 06, 107-8, 109, 1 10, 1 1 1 , 1 12-13, 1 14, 1 15, 1 16, 1 17, 120, 1 22, 1 23, 126, 127, 1 30, 191, 268 Biesanz, J. C., 5 1 Billings, L . S., 268 Blackstone, T., 176 Blaine, B. E., 178 Blanton, H., 268, 278, 280, 281 , 284, 289, 290, 306, 308 Bless, H., 128 Block, C. H., 301 Blumberg, H. H., 241 Bodenhausen, G. V., 104, 152 Boney McCoy, S., 275 Bonfield, E., 3 Boninger, D. S., 63, 74 Borden, V. M. H., 257 Borgida, E., 41, 87, 104, 108, 292

AUTHOR INDEX

Bostrom, R. P., 313 Bouchard, T. J., 301 Bowlby, J., 1 50, 177 Boyd, K., 212 Bradburn, N ., 40 Bradbury, T. N., 243, 259 Braly, K., 80 Brandstatter, V., 13 Brandtstadter, J., 292 Branscombe, A. R., 223 Branscombe, N. R., 167, 169, 171-2, 173, 176, 178, 179-80, 182-3, 184-5, 186, 187, 190 Breckler, S. J., 37, 39, 41 , 45, 62, 72, 74, 80, 87, 88 Breinlinger, S., 189 Brekke, N., 104 Brewer, M. B., 1 1 7, 130, 140, 141, 158, 163, 210, 257, 280 Briggs, R. 0., 307, 313 Brim, G. 0., 292 Britt, T. W., 182, 241 Broadnax, S., 178 Brock, T. C., 74 Bromer, P., 38, 54, 61, 62, 63, 64 Brown, J. D., 174, 251 , 280 Brown, R. J., 140 Brown, V., 305, 310, 312 Burchill, S. A. L., 241 Burrows, L., 25 Busemeyer, J. R., 1 1 Buss, A . R., 225 Buswell, B. N., 259 Butler, J. L., 240 Butler, J. S., 1 13 Butler, R., 273, 278, 279, 290 Buunk, B. P., 268, 277, 278, 289, 290, 306, 308 Bybee, J ., 222 Bylsma, W. H., 241 , 278 Byrne, D., 251 Cacioppo, J. T., 21, 46, 49, 74, 80 Cadsby, J., 302 Calantone, R., 3 Camac, C., 260 Camacho, L. M., 30, 302, 303, 305, 306, 312 Campbell, D. T., 64, 84, 106, 140, 141 , 142, 148, 163

329

AUTHOR INDEX

Campbell, K. W., 238, 242, 244, 245-8, 250, 251 , 259 Campbell, W. K., 241 , 248-52, 253-5 Cantor, N., 104 Carini, B., 141 Carlston, D. E., 158 Carnot, C. G., 63 Carver, C. S., 268, 289 Castelli, L., 145-6 Chaiken, S., 37, 38, 39, 41 , 52, 54, 59, 60, 63, 71 , 72, 73, 74, 85, 88, 95 Chapman, M., 214 Chartrand, T. L., 25 Chatzisarantis, N. L. D., 19 Chavous, T. M., 184 Chen, M., 25 Cheng, P. W., 223, 224 Christensen, A., 243 Chuang, Y. C., 63 Chung, T., 280 Cialdini, R. B., 280 Claire, T., 176 Clark, M. S., 240, 248, 253, 260 Clark, R., 178 Clark, R. A., 150 Clark, V. R., 178 Clawson, V. K., 313 Clore, G. L., 44 Close, S., 27 Clow, K. A., 83-4 Coats, S., 104 Cogan, J. C., 178 Cohen, G. L., 126 Cohen, J., 4-5, 14 Cohen, L. L., 172 Cohen, S., 241 Cohen, S. G., 300 Cole, E., 152 Coleman, L. M., 104 Collins, A., 44 Collins, A. M., 316 Collins, R. L., 277, 278, 283 Commins, B., 190 Conner, M., 3, 4, 9, 1 1 , 12, 13, 18, 23, 27, 29, 41, 42, 43, 44, 45, 46, 47-8," 49, 5 1 , 52, 53, 54, 55, 56, 57-9, 60-1 , 63 Connolly, T., 314, 315, 317 Converse, P. E., 50, 53 Conville, R. L., 74 Conway, L. G. III, 104 Cooley, C. H., 177

Cooper, D. A., 285-8 Cooper, H., 238 Cooper, W. H., 302, 304, 312, 3 14 Cornell, D. P. 276 Cornell, S. D., 241 Cornwell, B., 190 Coskun, H., 305, 308, 313 Cote, J. A., 10 Cottrell, N. B., 302 Courneya, K. S., 5 Cox, C. L., 240 Coyne, J. C., 241 Cozzarelli, C., 177, 185 Cramer, L., 259 Crandall, C. S., 106, 1 12-13, 123, 182 Crandall, R., 292 Crawford, M. T., 157, 158, 159, 162 Crites, S. L., 81, 84 Crocker, J., 140, 170, 175, 176, 178, 181-3, 184, 185, 188, 190, 192, 278, 280, 281 Croizet, J. C., 176 Cronen, V. M., 74 Crosby, F. J., 180, 185 Cross, S., 216 Cross, W. Jr., 178 Crowe, D. H., 242 Csikszentmihalyi, M., 3 16, 320 Cunha, D. A., 260 ,

Dailey, J. C., 315 Dakof, G. A., 278, 279 D'Andrade, R., 40 Danks, J. H. , 291 Darley, J. M., 304, 305 Dasgupta, N ., 156, 192 David, S. E., 173 Davids, A., 184 Davidson, A. R., 3 Davis, F. D., 3, 12 De Bruin, E. M. N., 260 De Vos, G. A., 189 De Vries, N., 23, 82, 189 De Vries, N. K., 22, 27, 43, 44 Deaux, K., 163, 259 Deci, E. L., 19 DeHarpport, T., 260 Delbecq, A. L., 313 Dennis, A. R., 314 Densley, L., 214

330

DePaulo, B. M., 240 Depret, E., 185 Derlega, V. J., 248 Devine, P. G., 80, 82, 84 Diab, L. N., 1 19 Dickson, W., 309 Diehl, M., 38, 54, 299, 300, 301 , 302, 303-4, 307, 312, 316, 320 Dijksterhuis, A., 13, 25 Dion, K. L., 178, 180, 185 Ditto, P. H., 291 Doll, J., 17, 20, 50 Dollard, J ., 305 Donakowski, D. W., 77, 79 Doosje, B., 104, 120 Dorval, K. B., 300 Dovidio, J. F., 85, 86, 91 Downing, J ., 39 Downs, D. L., 177 Drigotas, S. M., 240, 241 Druley, J. A., 291 Duan, C., 1 14 Dugosh, K. L., 317, 318, 320 Dunbar, K., 300, 320 Dunning, D., 126, 238, 279 Dunton, B. C., 80 Dweck, C. S., 279 Dzindolet, M. T., 300, 302, 303, 305, 306, 308, 309, 310, 312, 313 Eagly, A. H., 37, 38, 39, 41, 54, 59, 63, 71, 72, 73, 74, 75, 76, 78, 79, 80, 81 , 82-3, 85, 92, 95 Earn, B. M., 178, 185 Edwards, K., 72 Eggleston, T., 278, 306 Einhorn, H., 222 Eiser, J. R., 1 14, 121 Ellemers, N., 104, 1 19, 120, 188, 189, 190, 191 Elliot, A. J., 80, 82, 84, 248-52, 253-5 Elliot, A. T., 245-8, 250 Emmons, R. A., 39 Engels, F., 191 Erb, H.-P., 223 Erber, M. W., 44, 46, 50, 51, 53, 61 Esses, V. M., 44, 46, 60, 62, 72, 73, 75, 76, 77, 78, 79, 80, 81 , 82, 83, 84, 85, 86, 87, 88, 90-3 Etcoff, N. L., 104, 146 Ethier, K. A., 163

AUTHOR INDEX

Evans, M. G., 83 Fabrigar, L. R., 81 Faes, S., 13 Fairfield, M., 172 Farris, E., 259 Fazio, R. H., 53 Fazio, R. J ., 80 Feaganes, J. R., 42 Feather, N., 150 Fein, S., 104 Feldman, S., 40, 51 Felson, R. B., 241 Ferdman, B. M., 191 Festinger, L., 268, 304, 305, 308 Fiedler, K., 202, 221 Fincham, F. D., 92, 243, 259 Finger, J. R., 313 Finlay, K., 19 Finlay, K. A., 11 Fishbein, M., 1, 5, 10, 17, 40, 46, 53, 54, 60, 73, 82, 83 Fiske, D. W., 64 Fiske, S. T., 38, 104, 140, 141, 146, 185, 229 flesch, D., 51, 54, 55 fletcher, G., 202, 224, 228 fletcher, G. J. 0., 259 flowerman, S., 84 Foddy, M., 1 1 7 Folkman, S . , 268 Ford, C. E., 280 Ford, C. M., 300, 315, 320 Ford, T. E., 75, 104 Forsterling, F., 202 Foster, C. A., 240 Foster, M. D., 185 Frable, D. E. S., 176, 177, 184 Francke, L. B., 1 13 Frazier, L. E., 285-8 Freud, S., 40 Frey, D., 247 Frey, K. S., 273, 279, 290, 292, 293 Friend, R., 277 Fuegen, K., 1 15 Fuhrmann, A., 21, 24 Gaertner, L., 141, 163 Gallois, C., 7, 28-9 Gallupe, R. B., 302, 304, 314, 316 Gangestad, S., 50

332

Hofstee, G., 46-7 Hogarth, R., 222 Hogg, �. t\., 103, 104, 1 19, 122, 150, 185 Holmes, J. G., 240 Holzberg, 1\. D., 126 Honore, T., 203, 204-5, 221 , 223, 224 Hopkins, N., 72 Hoppen, N., 314 Hornstein, H. 1\., 240 Hovland, C. 1., 121, 292 Hume, D., 204 Hunt, W. 1\., 106 Hunter, J. E., 4, 4, 5 Hutchinson, K. J., 141 Hyde, J. S., 259 Hyers, L. L., 172 Insko, C. 1\., 74 Isaksen, S. G., 30 Ishida, L., 220 Islam, �- R., 74 Izumi, H., 300 Jackman, �- R., 191 Jackson, J., 178 Jackson, J. R., 80 Jackson, L. �-. 79, 85 Jalbert, N., 260 James, R., 41 , 43, 47-8, 49, 54, 55 Janis, 1., 309 Janoff-Bulman, R., 238 Janssen, H. J. W., 47 Jaspars, J., 207, 210, 212 Jennings, L., 74 Jessup, L. �-. 314 Jetten, J., 185 Johnson, C., 299, 300 Johnson, J., 212, 223 Johnson, �-. 2 Johnson, �- K., 148 Johnston, L., 26 Johnston, S., 242 Johnston, W. 1\., 244 Joireman, J. 1\., 260 Jonas, K., 38, 41, 42, 43, 45, 52, 54, 59, 61, 62, 63, 64 Jones, E. E., 202, 226, 229 Jones, L. E., 1 1 Joreskog, K . G., 52 Jost, J. T., 191, 28 1 Jourden, F. J., 290

AUTHOR fNDEX

Joyce, �-. 3 Judd, C. �-. 1 1 Jussim, L., 104 Kahan, J. P., 242 Kahana, B., 291 Kahana, E., 291 Kahneman, D., 208 Kaiser, C. R., 174 Kalish, C. W., 224 Kallgren, C. 1\., 59, 89 Kao, C. F., 80 Kaplan, K. J., 38, 43, 45, 85 Kappen, D. � 169 Karafa, J. 1\., 177, 185 Karau, S. J., 303, 312 Kashima, Y., 28-9, 224, 225, 229 Kasser, T., 19 Katz, D., 80 Katz, I., 38, 79, 85, 86, 87, 191 Kayser, T. 1\., 30 Kelley, H. H., 130, 170, 171, 202, 207, 209-10, 212, 225, 243 Kelly, C., 189 Kelly, J., 189 Kendzierski, D., 21, 22, 24 Kenny, D. 1\., 28, 54, 55, 56, 240 Kerr, N. L., 306 Kessler, T., 176, 189 Kiger, G., 79 Kim, M. S., 4, 5 King, C. E., 243 Kirkland, S., 104 Klandermans, B., 189 Kleck, R. E., 180-1 Klein, J. L., 218 Klem, 1\., 50 Kling, K. C., 259 Klink, 1\., 190 Klonoff, E. 1\ . , 178 Knibbs, C. S., 22fr.7 Knobe, J., 214, 216, 224 Knott, I. C., 212, 218, 219, 223 Ko Man Fung, J ., 79 Kobrynowicz, D., 106, 109, 1 1 1 , 1 12, 1 13, 1 15, 1 16, 127, 130, 169, 172, 178, 179, 187 Koehler, D. J., 1 17 Kok, G., 3, 19, 54 Kolditz, T., 238 Koomen, W., 104, 120 .•

334

Major, B., 140, 170, 174, 175, 180, 181, 182, 183, 184, 185, 188, 190, 192, 241 , 278, 281 , 288 Malle, B. F., 202, 203, 214, 216, 219, 224, 225, 229 Mallery, P., 280 Mandel, D. R., 214, 215, 216, 221 Manis, M., 103, 104, 106, 107-8, 109, 1 10, 1 12, 1 13, 1 15, 121, 241 Manning, V., 178 Manstead, A. S., 104, 120 Manstead, A. S. R., 22, 27, 44 Manstead, C. N., 3 Marcus-Newhall, A., 202, 205, 206, 221, 224 Margulis, S. T., 248 Markus, H., 22, 216, 274 Marrs, S., 152 Marsh, H. W., 277 Martens, R., 314 Martin, J., 190 Martin, R., 282 Maruyama, G., 108 Marx, K., 191 Matheson, K., 185 McBroom, W. H., 6, 27 McCall, M., 260 McCamish, M., 28-9 McClelland, D. C., 150 McClelland, G. H., 1 1 McClure, J. L., 170, 206, 207, 208-9, 210-1 1 , 212-13, 214, 215, 216, 217, 218-21, 223, 224, 225, 226, 227-8, 229 McConahay, J. B., 79, 80 McConnell, A. R., 152, 154-5, 157, 158, 162 McCullough, J., 10 McGarty, C., 1 14, 1 19, 141 McGill, A. L., 212, 218, 219, 223 McGillis, D., 226 McGrath, J. E., 268 McGregor, 1., 62 Mcintosh, P., 176 McKenna, K. Y. A., 184 McKiman, D. J., 191 McMillan, B., 9 McMillan, D., 1 1 8 Mead, G . H., 177 Meadow, A., 301 Medin, D. L., 224

AUTHOR INDEX

Meek, G. A., 19 Mennecke, B. E., 314 Merisca, R., 222 Merton, R., 40 Metts, S., 248 Meyerowitz, J. A., 126 Michela, J. L., 207 Mickelson, K. D., 279, 280 Midden, C., 13 Mielke, R., 190 Milesi, A., 1 17 Mill, J. S., 202 Millar, K. U., 72 Millar, M. G., 72, 152 Millberg, S., 260 Miller, C. T., 174 Miller, D. T., 163, 242, 280 Miller, G. A., 105 Miller, L. C., 129 Miller, N. E., 305 Miller, R. L., 304 Miller, R. S., 242 Mills, J., 240, 248, 260 Milne, A. B., 104 Milne, S., 3, 4, 8, 13, 14, 15 Mischel, W., 104 Mittleman, D. D., 313 Mizrahi, K., 163 Mladinic, A., 72, 79 Mlicki, P. P., 120 Moghaddam, F. M., 191 Montalcini, F., 74 Montano, D. E., 9 Monteith, M. L., 74 Moonen, A., 18 Moore, J. C., 241 Moore, K. A., 291 Moore, M., 54 Moore, P. J ., 272 Moreland, R. L., 140 Morf, C., 172 Morris, M. W., 202, 210, 223 Morrison, D. M., 3 Morse, S., 271, 277 Moskos, C. C., 1 13 Mullen, B., 3, 86, 238, 299, 300, 301, 302, 303, 315 Mullin, B. A., 150 Mulvey, P. W., 307 Mummendey, A., 176, 189 Munford, M. B., 184

336

Pufall, A., 1 80 Putman, V . L., 3CJ7 Pyszczynski, T., 104, 1 77, 238 Quellette, R., 260 Quinn, D., 170 Quinn, D. M., 176, 1 90 Randall, D . M., 4, 1 7 Rasinski, K. A . , 40 Raymond, P., 25 Read, S. J., 1 29, 202, 205, 206, 207, 212, 221 , 224 Reardon, K., 279 Reed, H., 292 Reed, M. B., 275 Reeder, G., 245-52, 253-5 Reeder, G. D., 1 17, 210 Regalia, C., 92 Reicher, S., 72 Reicher, S. D., 103, 104, 1 1 9 Reid, A., 163 Reid, F. W., 6, 27 Reilly, M., 10 Reinig, B. A., 3CJ7 Reis, H. T., 240, 257 Reis-Bergan, M., 284, 293 Rempel, J. K., 71, 72, 73, 74, 93, 240 Reno, R. R., 260 Rhodes, N., 89 Rhodewalt, F., 1 72 Ribbens, B. A., 307 Richard, R., 22 Richards, J. M., 280 Riordan, C. A., 238 Rips, L. J., 217 Rivkin, I. D., 273 Rizzo, A., 214 Robertson, S. P., 217 Robins, C. J., 1 73 Roeber, S., 141 Rodin, J ., 277 Roefs, M., 120, 1 88 Roese, N. J., 1 18, 2 1 6 Roethlisberger, F., 309 Rogers, R. W., 1 Rogers, S., 104 Rogier, A., 141 Roland, E., 309 Roland, E. J., 307, 317 Rollins, J. H., 185

AUTHOR INDEX

Romero, A. A., 74 Rommetveit, R., 309 Rosen, S., 241 Rosenberg, M., 88, 177, 186 Rosenblatt, A., 104 Rosenfield, D., 259 Rosenthal, R., 4, 14 Ross, L., 277 Ross, M., 238, 242, 243 Rothbart, M., 84, 1 17 Rothman, A. J . 292 Routbieaux, R. L., 314 Rowley, S. J., 184 Roy, M. C., 305, 3CJ7, 315 Rozell, R. M., 74 Rubin, D. B., 4 Ruble, D. N., 273, 279, 290, 292, 293 Ruderman, A. J., 104, 146 Ruggiero, K. M., 173, 175-6, 180, 181, 182 Rusbult, C. E., 240, 241 , 253, 254, 257 Russell, D., 171 Ryan, M. J., 3 Ryan, R. M., 19 Ryf, C. D., 292 ,

Sachs, P. R., 282-3, 289, 290 Saenger, G., 84 Salas, E., 299, 30 Salovey, P., 277, 292 Sanchez, D. T., 173 Sanderman, R., 277 Sanna, L. J., 302 Sansone, C., 268 Sarason, B. R., 241 Sarason, I. G., 241 Sawyer, K., 316, 320 Schachter, S., 150, 273 Schadron, G., 141 Schaller, C., 104 Schaller, M., 104, 120 Schank, R. C., 202-3, 206, 221 Scheier, M. F., 268, 289 Scherbaum, C., 176 Schiff, B. B., 85 Schiffbauer, K., 169 Schifter, D. E., 21 Schimel, J ., 238 Schlenker, B. R., 238, 241 , 242 Schmader, T., 159, 160-1, 184 Schmidt, F. L., 4

AUTHOR INDEX

Schmitt, M. T., 1 67, 1 69, 1 71-2, 173, 1 78, 1 79, 182-3, 185, 1 87 , 1 90 Schneider, S. K . , 273, 314 Schooler, T. Y., 42 Schopler, J., 141, 1 63 Schulz, R., 293 Schwartz, J. C., 104 Schwartz, S. H., 53 Schwarz, N., 128, 203, 207, 212, 217, 218, 220, 223, 230 Schwarzer, R., 268 Scott, W. A., 38 Sedek, G . , 174 Sedikides, C., 42, 238, 240, 241 , 242, 244, 245-52, 253-5, 257, 258, 259 Sellers, R. M . , 1 84 Semin, G., 1 17 Semin, G. R., 221 Semin-Goossens, A., 260 Seta, J. J., 307, 308 Shaeffer, D. E., 1 73 Shannon, L., 238 Shavitt, S., 74 Sheeran, P., 2, 3, 4, 5 , 6, 7-8, 9, 12, 13, 14, 15, 16, 17, 1 8, 19, 20, 2 1 , 22, 23, 24, 26, 29, 5 1 Sheeran, R. , 1 1 Sheldon, K . M., 1 9 Shepherd, M . M., 307, 3 1 5 Shepherd, R., 40, 41, 43, 47-8, 49, 54, 55 Sheppard, B. H., 3, 4, 12, 54 Shepperd, J. A., 303, 307, 309, 3 1 2 Sherif, C. W., 93 Sherif, M., 93, 121 , 292 Sherlock, K., 3, 5 1 Sherman, D. A . , 1 26-7 Sherman, S. J., 1 15, 1 4 1 , 145-6, 149, 151-2, 155, 1 56 , 1 57, 1 58, 1 59, 162, 163 Sherman, S. L., 1 42-5 Sherman-Williams, B., 125, 127 Sherwood, J. J., 305 Shoda, Y., 104 Shore, W. J., 1 77 Showers, C. J., 259 Shrum, L. J., 38 Sicoly, F., 243 Sidanius, J., 1 72 Siegel, J. M., 174 Sigelman, L., 1 69, 1 72 Silverman, M . , 1 3 , 1 5

337 Simon, A. F., 1 63 Simon, B., 185 Simons, C., 120, 1 88 Simonton, D. K., 320 Sinclair, L., 104, 125 Skinner, E. A., 214 Skowronski, J. J., 1 58, 238 Slovik, L. F., 254 Slugoski, B. R., 226 Slusher, M. P., 127 Smedley, J. W., 84 Smedslund, J., 224 Smetana, J. G., 3 Smith, E. R., 104 Smith, L., 3, 54 Smith, M. A., 184 Smith, P. M., 1 85 Smith, W. P., 282-3, 289, 290 Smithson, M., 1 17 Smucker, W. D., 291 Snyder, C. R., 1 72, 280 Snyder, M., 50 Snyder, R. C., 180 Solomon, S., 104, 1 77 Sorbom, D., 52 Sparks, P., 3, 27, 40, 4 1 , 43, 46, 47-8, 49, 50, 54, 55, 56 Spears, R., 104, 1 17, 120, 1 76, 185 Spellman, B. A., 221 Spencer, S., 184 Spencer, S. J., 104, 276 Spencer, S. L., 1 76 Spicer, C. V., 74 Srull, T. K., 152 Stahlberg, D., 1 1 7 Stangor, C . , 75, 79, 80, 82, 83, 84, 104, 1 14, 1 18 Stanton, B. F., 6, 7 Steele, C., 140, 1 70 Steele, C. M., 176, 238, 275, 276 Steemers, G. E. M., 260 Stephan, W. G., 259 Stephens, A., 13 Stern, L. D., 152 Stewart, T. L., 1 73 Stiles, W. B., 241 Stillwell, A. M., 240 Stipek, D. J., 1 7 1 Stocker, M., 40 Story, A. L., 1 26 Strack, F., 25 , 106, 212

338

Stradling, S. G., 22 Strauss, L., 178 Strenta, A., 1� 1 Stroebe, W., 1 14, 299, 300, 301 , 302, 303-4, 307, 312, 316 Strube, M. J ., 238, 240 Suarez-Orozco, M. M., 189 Sullaway, M., 243 Sullivan, E., 74 Sullivan, L. A., 75 Suls, J ., 268, 282 Suls, J. M., 304 Susskind, J., 1 52-3, 154, 156, 162 Sutton, R., 227-S Sutton, R. 1., 300, 315 Sutton, S., 4, 5, 7, 9, 12 Swallow, S. R., 276 Sweeney, P. D., 1 73 Swim, J., 108 Swim, J. K., 172 Swinehart, D. M., 217 Tajfel, H., 103, 1 14, 1 19, 121, 150, 171, 1 77, 188, 189 Tambor, E. S., 177 Taplin, S. H., 9 Taylor, D. M., 173, 175-6, 180, 182, 191 Taylor, D. W., 301 Taylor, S. E., 104, 146, 251 , 268, 269, 273, 275, 278, 279, 283, 289, 290, 292 Terdal, S. K., 177 Tesser, A., 53, 240, 241 , 276, 277, 278 Testa, M., 170, 241 , 278 Tetlock, P. E., 242 Thagard, P., 103, 104, 124, 127, 128, 1 29-30, 131 Thakkar, V., 1 56-7, 158, 162 Theno, S. A., 191 Thompson, C. P., 238 Thompson, E. R., 1 14 Thompson, L., 260, 321 Thompson, M. M., 38, 39, 41 , 42, 44, 45, 47, 49, 59, 62, 87, 88 Thompson, S. C., 243, 292 Thomsen, C. J ., 41 Thomson, C. J., 87 Thornton, D. A., 277 Tice, D. M., 177 Tice, T. M., 240 Timko, C., 238 Tordesillas, R. S., 60

AUTHOR INDEX

Torrance, E. P., 316 Tourangeau, R., 40, 42 Townsend, A., 185 Trafimow, D., 3, 4, 1 1 , 17, 18, 19, 29, 51 Treffinger, D. J., 300 Triandis, H. C., 1 , 2 Trommsdorff, G., 300, 301 , 302, 303, 312 Trope, Y., 104, 275-6 Tudor, M., 240 Tumeo, M., 312 Turnbull, W., 217, 220 Turner, J., 171, 177 Turner, J. C., 103, 104, 1 19, 121, 122, 131, 150, 185 Turner, P. J., 185 Tversky, A., 1 1 7, 208 Uhles, A., 142-5 Upshaw, H. S., 105, 121 Valacich, J. S., 314 Valencia, L., 169 Van de Ven, A., 313 Van den Putte, B., 4 , 54 van der Pligt, J., 22, 27, 43, 44, 46, 61, 82 van der Zee, K., 277 van Eekelen, S. A. M., 3 Van Gundy, A. B., 317 van Harreveld, F., 27, 42, 43, 44 van Knippenberg, A., 18, 27, 104, 1 1 9, 120, 189, 190 van Knippenberg, C., 27 Van Lange, P. A. M., 240, 251 , 253, 260 van Rijswijk, W., 120, 188 Van Yperen, N. W., 278 Vassar, P. M., 173 Veeder, M., 104 Veenstra, K., 189 Verette, J., 254 Vernberg, D., 3 Verplanken, B., 13, 18, 20, 21, 27, 28, 46-7 Vescio, T. K., 104, 109, 120, 122, 191 Vinokur-Kaplan, D., 3 Voelkl, K., 170 Volkmann, J., 105, 121 Volpato, C., 243 von Hecker, U., 174 von Hippel, W., 72, 152 Wachter, R., 314

339

AUTHOR INDEX

Waenke, M., 212 Walster, E., 25 1 Walther, E., 202 Wankel, L. M., 315 Warren, R., 27 Warshaw, P. R., 3, 4, 12, 54 Warshaw, R. P., 3 Waterman, M., 48 Weary-Bradley, G., 238, 242 Weber, J. G., 141, 280 Webster, D. M., 50 Wegener, D. T., 39, 64, 128 Wegner, D. M, 24-5, 26 Wei, J., 176 Weiner, B . , 168, 170, 171 , 173, 174, 210, 214, 225, 229 Welbourne, J. L., 152, 155, 156, 157 Welch, S., 169, 172 Wellman, H. M., 202 Wenzlaff, R. M., 276 West, M. A., 300 West, S. G., 5 1 , 54, 260 Wetherell, M. S., 103, 104, 1 19 Whalen, P., 180 Wheatley, T., 24-5, 26 Wheeler, B. C., 314 Wheeler, L., 277, 282, 283, 290 Whitaker, D. J., 22, 24 White, D., 13 White, P. A., 217, 220 White, R. W., 150 Whitney, G. A., 254 Whitton, S. W., 241 Wieczorkowska, G., 142-5 Wieselquist, J., 240, 241 , 254 Wigboldus, D., 117 Wiggins, E. C., 72, 74, 80 Wilder, D., 163 Wilensky, R. W. , 208, 209, 212, 216, 229 Wilke, H., 189, 190 Wilkes, A. L., 1 21 Williams, C. J., 80

Williams, D. R., 178 Williams, K., 303 Williams, K. D., 177, 303, 312 Willis, H., 79, 91-2 Wills, T. A., 238, 241, 268, 275, 278, 289, 305 Wilson, T. D., 42, 44, 51 Winter, J. P., 313 Wirt, R., 184 Witcher, B. S., 240 Wojciszke, B., 1 17 Wolfe, C., 184 Wolff, J. A., 4, 17 Wolosin, R. J., 244 Wood, J. V., 277, 278, 305 Wood, W., 3, 18, 26, 27, 28, 59, 74, 89 Wright, S. C., 191 Wurf, E., 22 Wyer, R. S., 247 Wyer, R. S. Jr., 152 Yang, H.-C., 317, 319, 320 Ybema, J. F., 278 Yee, P. H. N., 185 Yen, J., 307 Yi, Y., 17, 18, 20, 21 Young, H., 176 Young, L. V., 106 Yu, Y., 178 Yzerbyt, V. Y., 141 , 157 Zabbini, S., 1 17 Zajonc, R. B., 312 Zaller, L., 40, 51 Zanna, M. P., 38, 41, 44, 46, 49, 57, 60, 62, 71, 72, 73, 74, 75, 76, 77, 78, 79, 81, 83, 85, 87, 92, 93, 1 18 Zarate, M. A., 190 Zerbst, J., 278 Zhang, S., 278 Zigler, E., 222 Zimmerman, R. S., 3 Zuckerman, M., 238, 242

This page intentionally left blank

Subject Index abnormal conditions focus theory, 226 accentuation theory, 121 action control, 21 action explanations see goal-based explanations affective component of attitudes, measurement, 72, 74-5, 94-5 applications, 91 , 92, 93-4 assessment, 82-5 attitude embeddedness, 89 attitude inconsistency, 88-9, 91 attitudinal ambivalence, 85-8 breadth of use, 79 cognitive component separation, 80-2 ease of completion, 78 evaluative implications of terms, 84-5 idiosyncratic responses, 78, 94-5 quantitative analysis, 76 affective states, proactive social comparison, 275-7 affiliation, group entitativity, 150-1 affiliative comparison activity, 279 aging, proactive social comparison, 291-3 ambivalence, and attitudes see attitudinal ambivalence anti-Semitism, 178, 185 anticipated regret, 22, 24 attitude-behavior relations attitudinal ambivalence, 40, 53-5, 57-9, 62 open-ended assessment, 92-3 attitude-behavior theory, 1, 2-3 attitude components, 71-2

affective see affective component of attitudes closed-ended measures, 72, 73, 94-5 assessing salient component, 82-3, 84 component separation, 80 content analyses, 77 cultural specificity, 79 evaluative implications of terms, 84 time-specificity, 80 validity, 76 cognitive see cognitive component of attitudes open-ended measures, 72-6, 94-5 assessing salient component, 82-5 attitude inconsistency, 88-9, 91 benefits, 76-85 breadth of use, 79-80 component separation, 80-2 conceptual framework, 73-4 content analyses, 76-7 cultural breadth, 79 ease of completion, 78 evaluative implications of terms, 84-5 idiosyncratic responses, 78, 94-5 quantitative analyses, 76-7 reliability, 76 time-specificity, 80 validity, 76 stereotypes see stereotypes attitude embeddedness, open-ended rneasures, 89, 90-1, 93-4

342

attitude inconsistency, open-ended measures, 88-9, 90-1 , 93-4 attitude-intention relations attitudinal ambivalence, 53-7, 58-9, 62 intention-behavior gap, 19-20, 21 attitude properties ambivalence see attitudinal ambivalence closed-ended measures, 93-4, 95 open-ended measures, 85-9, 90-4, 95 attitudinal ambivalence, 37-9 ambivalence judgment content, 44-5 attitude extremity, 59 attitude-behavior relations, 53-5, 57-9, 62 attitude-intention relations, 53-7, 58-9, 62 belief-based measures, 43-4, 46-7 consequences, 50-63, 64 definitions, 38, 39-41 direct measures, 41-2, 46 elicitors, 47-50, 64 future research, 63-4 Griffin formula, 45 indirect measures, 41-2, 43, 46 information processing, 61-2, 63 intercomponent, 86, 87-8 intracomponent, 85-7 measurement, 41-7, 57, 63-4, 85-8, 90-1, 93-4 open-ended measures, 85-8, 90-1 , 93-4 operative indices of attitude strength, 43 overall evaluation measures, 43 pliability, 59--61, 63 split semantic differential measures, 43, 46, 47, 49-50, 57 temporal stability, 50-2, 63 attributions, 201 , 202 and folk psychology, 225 integrating goal-based explanations, 225-8, 231 prejudice see attributions to prejudice self-evaluation, 238 see also self-serving bias attributions to prejudice, 167, 168 attributional ambiguity, 175-7, 181 attributional dimensions, 169-75 color-blind policies, 191

SUBJECT INDEX

consequences of recognizing prejudice, 178-83 controllability, 173-5 coping with pain of devaluation, 183-91, 193 discounting hypothesis, 170, 172, 179-82 discrimination experience severity, 169 dominant ideologies, 190-1 external locus, 170-1, 172 future research, 191-2 internal locus, 170-2 legitimate prejudices, 190 local relevance, 168, 173, 193 minimization, 1 76 Rejection-Identification Model, 184-91 self-protective consequences, 170-3, 180-3, 192-3 social implications, 177 social mobility, 189, 190 stability, 172-3 total relevance, 168-77 augmentation principle, 130 automatic activation effect, 53 automaticity, predicting behavior, 25, 2&-7 balance theory, relationships, 240 behaviors ambivalence towards, 40 see also attitudinal ambivalence and intention see intention-behavior relations open-ended measures, 92-3 belief-based measures, attitudinal ambivalence, 43-4, 46-7 beliefs ambivalence, 40 folk psychology, 224, 225 goal-based explanations, 224, 225 open-ended measures of attitude, 75--6 assessing salient component, 82-3 attitude embeddedness, 89 attitudinal component separation, 80-2 attitudinal inconsistency, 89 behavior prediction, 91 breadth of use, 79, 80 content analysis, 77 ease of completion, 78 idiosyncratic responses, 78

34 discounting theory attributions to prejudice, 170, 1 72, 1 79-82 goal-based actions, 209-10, 225 discrimination, perceived see attributions to prejudice dispositional inferences, 202 dyads, relational closeness see relationship closeness electronic group brainstorming, 302, 304, 313-15, 316--17 emotions see affective component of attitudes; affective states entitativity of groups, perceived, 139-42, 161-3 antecedents, 142-51 , 162 collective responsibility, 1 59-61, 162-3 consequences, 1 5 1-61, 162-3 cues to, 142-3, 162 definitions, 141 functions, 150-1 future research, 163 group types, 143-50, 162 integrative processing, 1 51-5, 162 intimacy groups, 143, 144, 150 judgment polarization, 155-7, 162 meaning for groups, 149-50 perception of group members, 157-9, 163 social categories, 143, 144, 150 spontaneous group typology, 145-9, 162 task-oriented groups, 143, 144, 1 50 equifinality theory, 203-4, 205, 216--17, 222 ethnic groups perceived discrimination see racial discrimination, attributions to prejudice stereotyping see racial stereotypes evaluation apprehension, group brainstorming, 302, 303, 31 1-12, 313, 314 evaluative language open-ended measures of attitudes, 84-5 shifting standards, 1 16, 1 23, 131 expectancy behaviors, shifting standards, 1 17-18

SUBJECT INDEX

expectancy-value models, attitude measures, 73-4 expectations, behavioral, 12 explanation theory, 202 attribution, 202, 2� goal-based see goal-based explanations facilitators, group brainstorming, 313, 315 folk psychology, and explanations, 224-5, 229 gay men, attributions to prejudice, 178, 185 gender diferences, self-serving bias, 250, 251 , 259 gender discrimination, attributions to prejudice attributional ambiguity, 176--7 consequences of recognizing prejudice, 178-80 controllability, 173-4 coping with pain of devaluation, 184, 185, 187 discounting hypothesis, 172, 180 discrimination experience severity, 169 locus of attribution, 171-2 Rejection-Identification Model, 185, 186-7 stability, 1 72-3 gender stereotypes, shifting standards contextual variations, 1 13-18 parallel-constraint satisfaction model, 126 shifting judgment standards, 105-13 goal achievement automaticity, 25 group entitativity, 150--1 implementation intentions, 16, 23, 24 proactive social comparisons, 290- I goal-based explanations, 201-3 causal chains, 221-2 causal questions, 217-21, 225, 230 causation criteria, 222-4 causation in the law, 204-5 competing goals, 208-9 conjunctions, 207--8, 212, 213 controllability, 214-17, 225 and covariational theories, 202, 226--8, 231 difficult actions, 209-10, 212

346

SUBJECT INDEX

intention-behavior relations, gap between (cont.) alternative governors of behavior,

24-9 anticipated regret, 22, 24 attitudinal control, 19-20, 21 automaticity, 25, 26-7 behavior type, 9-1 1 , 24 cognitive variables, 20-3, 24 conflicting intentions, 23, 24 control issues, 9-1 1 degree of intention formation,

18-19, 21, 24 expectations, 12 factors determining, 9-24 habit, 26, 27-8 implementation intentions,

178, 185 12-16,

23, 24

188-9 group entitativity, 163 open-ended measures, 74, 75-6 applications, 90-2 assessing salient component, 83-4 attitude embeddedness, 89, 90-1 attitude inconsistency, 88-9, 90-1 attitudinal ambivalence, 85-8, 90-1 attitudinal component separation,

81-2 content analysis, 77 cultural breadth, 79 ease of completion, 78 evaluative implications of terms, quantitative analysis,

77

104-5, 124-30, 131 104-5, 119-24, 131 shifting standards, 104-18, 120-4, 125-30, 131 social identity theory, 1 19, 120-1 . 124 interpersonal theory of actions, 202 intimacy groups, entitativity, 143, 144, 150 self-categorization,

Jewish people, attributions to prejudice,

intention accessibility, 20 intention certainty, 20 intention type, 12-16, 24 measurement artifacts, 5 moderator analyses, 28-9 normative control, 19-20, 21 past behavior, 25-6, 27-9 personality variables, 20-1 , 24 reasons for, 5-9 self-schemas, 22, 24 temporal stability, 17-18, 24 models, 2-3 intentional action theory, 202-4, 205-6 and folk psychology, 224-5 interdependence theory, 240 intergroup attitudes attributions to prejudice, 168-77,

84-5

time-specificity, 80 stereotypes in context, 103-4 parallel-constraint satisfaction,

judgments ambivalence, content of, 44-5 group entitativity integrative processing, 154-5 polarization, 155-7, 162 shifting standards see shifting standards model knowledge structures, goal-based explanations, 206-9, 224, 226-7 labels, intergroup attitudes, 77 legal implications, causal questions, 230 lesbians, attributions to prejudice, 178 measurement artifacts, 5 men perceived discrimination see gender discrimination, attributions to prejudice self-serving bias, 250, 251, 259 stereotyping see gender stereotypes meta-judgmental measures, 93, 94 moderator analyses, intention-behavior,

28-9 moods attitude measures see affective component of attitudes proactive social comparisons, 275-7 motivation behavioral intention as index of, 2 group brainstorming, 303, 308, 309,

310, 31 1-13, 314-15, 317 group entitativity, 150-1 self-categorization theory,

123-4

348

relationship closeness, and self-serving bias (SSB) (cont. ) social value orientation, 260 task importance, 244-5 relationship closeness induction task (RCIT), 249 relationships-as-bound hypothesis, 240, 245-52, 257-8 relationships-as-enabler hypothesis, 240-1 , 245-8, 257-8, 259 responsibility, collective, 159-61, 162-3 savings-in-relearning paradigm, 158 schematics, intention-behavior relations, 22, 24 self-categorization theory (SCf), 104-5, 1 19-24, 131 self-concept, and social comparison, 274-5 self-concept enhancing tactician (SCENT) model, 240 self-enhancement, 238, 239 and relationship closeness see selfserving bias self-evaluation maintenance model, 240 self-identity, group entitativity, 150 self-protection attributions to prejudice, 17�3. 180-3, 192-3 relationships-as-enabler hypothesis, 241 self-regulation, 268 proactive comparisons experienced targets, 272-3 functions, 273-4, 288, 289-90 proactive coping, 268-9, 273 self-schemas, intention-behavior relations, 22, 24 self-serving bias (SSB), and relationship closeness, 237-9 anticipated interaction, 248--50 attributional expectancies, 253-7, 258 attributional reciprocity, 254-7, 258-9 cognitive explanations, 242, 243 communal orientation, 260 expectancy violation, 254-7, 258 gender differences, 250, 251 , 259 interdependent outcomes tasks, 242, 243-5, 259-60 mechanisms of SSB reduction, 248--52 motivational explanations, 242-3

SUBJECT INDEX

participant selection, 248-50 partner impressions, 251-2, 258 reasons for SSB augmentation, 253-7 reasons for SSB reduction, 253-7, 258-60 relationship closeness induction task, 249 relationships-as-bound hypothesis, 240, 245-52, 257-8 relationships-as-enabler hypothesis, 240-1, 245-8, 257-8, 259 relativistic attribution measures, 243, 259 research issues, 242-3 social value orientation, 260 task importance, 244-5 sexism see gender discrimination sexual orientation, attributions to prejudice, 178, 185 shifting standards model (SSM), 104-6, 131 category combinations, 1 14 confirmatory standards, 1 17 contextual variations, 1 13-18 expectancy behaviors, 117-18 ingroup/outgroup status, 115-16 norms, 1 13 and parallel-constraint satisfaction, 125-30 and self-categorization, 12� setting judgment standards, 116-18 shifting judgment standards, 106-13, 128-30 subjective language, 1 16 trait concepts, 1 14, 1 17-18, 126-8 social categories, group entitativity, 143, 144, 150 sociaJ/cognitive influence model, group brainstorming, 31 1-12, 32�1 cognitive component, 315-20 social component, 312-15 social comparison convergence effect, 305, 308-9 downward, 277-80, 282, 285-8, 289 group brainstorming, 306, 308-9, 312 group brainstorming, 304-10, 312 proactive, 267-8 affiliative, 279 and aging, 291-3 coping, 268-9, 273 downward, 277-80, 282, 285-8, 289