Multiple Social Categorization

Multiple Social Categorization “Ethnic cleansing”, “institutional racism”, and “social exclusion” are just some of the...

Author: Richard Crisp

50 downloads 1374 Views 2MB Size Report

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

Report copyright / DMCA form

DOWNLOAD PDF

Multiple Social Categorization

“Ethnic cleansing”, “institutional racism”, and “social exclusion” are just some of the terms used to describe one of the most pressing social issues facing today’s societies: prejudice and intergroup discrimination. Invariably, these pervasive social problems can be traced back to diﬀerences in religion, ethnicity, or countless other bases of group membership: the social categories to which people belong. Social categorization, how we classify ourselves and others, exerts a profound inﬂuence on our thoughts, beliefs, feelings, and behaviors. In this volume, Richard Crisp and Miles Hewstone bring together a selection of leading ﬁgures in the social sciences to focus on a rapidly emerging, but critically important, new question: how, when, and why do people classify others along multiple dimensions of social categorization? The volume also explores what this means for social behavior, and what implications multiple and complex perceptions of category membership might have for reducing prejudice, discrimination, and social exclusion. Topics covered include: • • •

The cognitive, motivational, and aﬀective implications of multiple categorization The crossed categorization and common ingroup methods of reducing prejudice and intergroup discrimination The nature of social categorization among multicultural, multiethnic, and multilingual individuals

Multiple Social Categorization: Processes, Models, and Applications addresses issues that are central to social psychology and will be of particular interest to those studying or researching in the ﬁelds of Group Processes and Intergroup Relations. Richard J. Crisp is a Reader in Social Psychology at the University of Birmingham. His research focuses on cognitive, motivational, and aﬀective models of social categorization, group processes, and intergroup relations. He is a past winner of the British Psychology Society’s award for Outstanding Doctoral Research Contribution to Psychology, and the Society for the Psychological Study of Social Issues Louise Kidder Early Career Award for his work on multiple social categorization. In 2006 he was awarded the British Psychological Society’s Spearman Medal for outstanding early career contributions to psychological research. Miles Hewstone is Professor of Social Psychology and Fellow of New College, Oxford. He has published widely on the topics of attribution theory, social cognition, stereotyping, social inﬂuence, and intergroup relations. He is co-founding editor of the European Review of Social Psychology, a former editor of the British Journal of Social Psychology, and a past winner of the British Psychology Society’s Spearman Medal and Presidents’ Award. He is a Fellow of the British Academy.

Multiple Social Categorization Processes, Models, and Applications

Edited by Richard J. Crisp and Miles Hewstone

First published 2006 by Psychology Press 27 Church Road, Hove, East Sussex BN3 2FA Simultaneously published in the USA and Canada by Psychology Press 270 Madison Avenue, New York, NY 10016 This edition published in the Taylor & Francis e-Library, 2007.

“To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.” Psychology Press is an imprint of the Taylor & Francis Group, an informa business © 2006 Psychology Press All rights reserved. No part of this book may be reprinted or reproduced or utilized in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. The publisher makes no representation, express or implied, with regard to the accuracy of the information contained in this book and cannot accept any legal responsibility or liability for any errors or omission that may be made. This publication has been produced with paper manufactured to strict environmental standards and with pulp derived from sustainable forests. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data Multiple social categorization : processes, models, and applications / [edited by] Richard J. Crisp and Miles Hewstone. p. cm. Includes bibliographical references and index. ISBN 1-84169-502-5 1. Social groups. 2. Diﬀerentiation (Sociology) 3. Group identity. 4. Intergroup relations. 5. Categorization (Psychology) 6. Social psychology. I. Crisp, Richard J., 1973– II. Hewstone, Miles. HM716.M85 2006 303.3′85–dc22 2006008222 ISBN 0-203-96922-7 Master e-book ISBN

ISBN 13: 978-1-84169-502-0 ISBN 10: 1-84169-502-5 (Print Edition)

To the memory of Brian Mullen (1955–2006)—a true scholar on multiple dimensions

Contents

List of ﬁgures List of tables List of contributors Preface

ix xi xii xiv

PART I

Introduction 1 Multiple social categorization: Context, process, and social consequences

1

3

RICHARD J. CRISP AND MILES HEWSTONE

PART II

Multiple category representation 2 Hierarchies and minority groups: The roles of salience, overlap, and background knowledge in selecting meaningful social categorizations from multiple alternatives

23

25

CRAIG MGARTY

3 Multiply categorizable social objects: Representational models and some potential determinants of category use

50

ELIOT R. SMITH

PART III

Multiple categorization and social judgment 4 Recategorization and crossed categorization: The implications of group salience and representations for reducing bias JOHN F. DOVIDIO, SAMUEL L. GAERTNER, GORDON HODSON, BLAKE M. RIEK, KELLY M. JOHNSON, AND MISSY HOULETTE

63

65

viii

Contents

5 Commitment and categorization in common ingroup contexts

90

RICHARD J. CRISP

6 Self-concept threat and multiple categorization within groups

112

MICHAEL A. HOGG AND MATTHEW J. HORNSEY

PART IV

Cross-cutting categorization and evaluation 7 The crossed categorization hypothesis: Cognitive mechanisms and patterns of intergroup bias

137

139

THERESA K. VESCIO, CHARLES M. JUDD, AND POH-PHENG CHUA

8 Explaining the eﬀects of crossed categorization on ethnocentric bias

160

NORMAN MILLER, JARED B. KENWORTHY, CARRIE J. CANALES, AND DOUGLAS M. STENSTROM

9 Gender among multiple social categories: Social attraction in women but interpersonal attraction in men

189

RAMADHAR SINGH

PART V

Societal and political perspectives

209

10 Multiple social categorization and identity among multiracial, multiethnic, and multicultural individuals: Processes and implications

211

JEAN S. PHINNEY AND LINDA L. ALIPURIA

11 Political institutions and multiple social identities

239

NEAL A. CARTER

PART VI

Conclusion

269

12 Multiple social categorization: Integrative themes and future research priorities

271

MILES HEWSTONE, RHIANNON N. TURNER, JARED B. KENWORTHY, AND RICHARD J. CRISP

Author index Subject index

311 321

Figures

2.1 3.1 4.1 4.2 5.1 5.2 7.1

7.2 8.1 8.2 9.1

9.2 9.3 9.4 9.5 9.6 9.7 9.8

A tree diagram showing three levels of a hierarchy Example of a hierarchical representational scheme The common ingroup identity model The mutual intergroup diﬀerentiation model Ingroup favoritism as a function of categorization and identiﬁcation Ingroup favoritism as a function of categorization and identiﬁcation Categorization (within-group minus between-group errors) in simple versus crossed conditions, Experiments 1 and 2 Categorization (within-group minus between-group errors) within crossed conditions, Experiments 1 and 2 Preference ratings for targets (± SE) as a function of mood condition and target type Mean aﬃliative target evaluations (± SE) on composite index as a function of induced aﬀective state Mean agreement with man versus woman and ingroup– outgroup categorizations of gender groups by men and women in India Mean factor scores of men and women in Singapore Mean attraction of men as a function of race and gender of the targets Mean attraction as a function of nationality or race and gender of the targets Mean attraction and intelligence ratings of men and women targets in three designs of Experiment 2 Mean attraction of men (left graph) and women (right graph) as a function of race and gender of the targets Mean intelligence as a function of nationality and gender of the targets Mean attraction as a function of nationality and gender of the targets

30 51 68 72 97 101

149 150 171 173

192 193 195 198 199 199 201 201

x

Figures

10.1 10.2

10.3 11.1 11.2 11.3 11.4 11.5

Identity statuses deﬁned by presence or absence of exploration and commitment Path model showing the prediction of outgroup attitudes from ethnic identity and ingroup attitudes among black and Latino adolescents Strength of positive orientation toward other groups in terms of ethnic identity status Basic model of social identities Brewer’s model of optimal distinctiveness Models of nested games Model of competing bases of identiﬁcation Combining levels and bases of identiﬁcation

217

221 222 246 246 249 259 260

Tables

1.1 1.2 4.1

5.1 7.1 7.2 7.3 10.1

A priori predictions of the six main outcome patterns of evaluation resulting from crossed categorization Multiple categorization: Current trends Group representations and bias as a function of a function of common group (same dress), original categories (diﬀerent-dress), and crossed categorization (mixed-dress) conditions Evaluation as a function of cross categorization and identiﬁcation Illustration of the moderation hypothesis in a race × gender crossed context Possible memory errors in a race × gender crossed categorization condition Bias toward subgroups of crossed conditions compared to ingroups and outgroups of simple conditions Ethnic/racial identity formation and intergroup attitudes of minority group members

12 15

75 98 143 146 151 220

Contributors

Linda L. Alipuria, Department of Psychology, College of Natural and Social Sciences, California State University, Los Angeles, 5151 State University Drive, Los Angeles, CA 90032-4226, USA Carrie J. Canales, Psychology Department, Los Angeles City College, 855 North Vermont Avenue, Los Angeles, CA 90029, USA Neal A. Carter, Box 37, Department of Political Science, St. Bonaventure University, St. Bonaventure, NY 14778, USA Poh-Pheng Chua, Department of Psychology, Pennsylvania State University, University Park, PA 16802, USA Richard J. Crisp, School of Psychology, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK John F. Dovidio, 406 Babbidge Road, Unit 1020, University of Connecticut, Stoors, Connecticut 06269-1020, USA Samuel L. Gaertner, Department of Psychology, University of Delaware, Newark, DE 19716-2577, USA Miles Hewstone, Department of Experimental Psychology, University of Oxford, South Parks Road, Oxford OX1 3UD, UK Gordon Hodson, Department of Psychology, Brock University, St. Catharines, Ontario L2S 3A1, Canada Michael A. Hogg, School of Behavioral and Organizational Sciences, Claremont Graduate University, 123 East Eighth Street, Claremont, CA 91711, USA Matthew J. Hornsey, School of Psychology, University of Queensland, Brisbane, QLD 4072, Australia Missy Houlette, College of Mount St. Joseph, Department of Business Administration/Behavioral Sciences, Cincinnati, OH 45233-4270, USA

Contributors

xiii

Kelly M. Johnson, Department of Psychology, University of Delaware, Newark, DE 19716-2577, USA Charles M. Judd, Department of Psychology, University of Colorado, Boulder, CO 80309-0345, USA Jared B. Kenworthy, Department of Psychology, Life Science Building, Box 19528, University of Texas at Arlington, Arlington, Texas 76019-0528, USA Craig McGarty, School of Psychology, The Australian National University, Canberra, ACT 0200, Australia Norman Miller, Department of Psychology, University of Southern California, Los Angeles, CA 90089-1061, USA Jean S. Phinney, Department of Psychology, College of Natural and Social Sciences, California State University, Los Angeles, 5151 State University Drive, Los Angeles, CA 90032-4226, USA Blake M. Riek, Department of Psychology, University of Delaware, Newark, DE 19716-2577, USA Ramadhar Singh, Department of Social Work and Psychology, National University of Singapore, BLK AS6, 11 Law Link, Singapore 117570, Republic of Singapore Eliot R. Smith, Department of Psychology, Indiana University, 1101 East Tenth Street, Bloomington, IN 47405, USA Douglas M. Stenstrom, Department of Psychology, University of Southern California, Los Angeles, CA 90089-1061, USA Rhiannon N. Turner, School of Psychology, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK Theresa K. Vescio, Department of Psychology, Pennsylvania State University, University Park, PA 16802, USA

Preface

As we begin the 21st century, the arrival of global multiculturalism provides familiar old problems, but also new opportunities. While delineation of cultural and societal boundaries has undoubtedly enriched our social world, this same increased proximity of diverse ethnic, cultural, or religious groups has sometimes accentuated old problems of conﬂict and prejudice between social groups. Whether it is “ethnic cleansing” in the Balkans or “institutional racism” in the UK, we have far from succeeded in banishing the specter of intergroup discrimination. When we examine instances of conﬂict a common link emerges: Almost universally these conﬂicts can be traced to diﬀerences in religion, ethnicity, or countless other bases for group membership; in other words, the social categories to which people belong. The role of social categorization in deﬁning many previous, current, and probable future conﬂicts has long been a central concern for social psychologists (e.g., Allport, 1954). Fifty years of research has now established the psychological process of social categorization as an integral component in intergroup relations. In this edited book, leading ﬁgures in the social sciences focus on a set of rapidly emerging, but critically important, new questions in this ﬁeld: how, when, and why do people classify others along multiple dimensions of social categorization? What does this mean for social behavior? What implications might multiple and complex perceptions of category membership have for understanding group processes, intergroup relations, and social systems? How do we get from existing work on social categorization to our focus in this book on multiple social categorization? Social categorization is widely accepted as a key psychological determinant of social behavior and a substantive literature has emerged that has explored the eﬀects and implications of the categorization process. Over the last 30 years a more “qualiﬁed” social psychological account of social categorization has emerged, which has taught us a great deal about when social categories become salient, and what underlying processes are involved. While situations involving just one (selfincluding) ingroup and one (self-excluding) outgroup may be common, recent work has found that in many cases people attend to more than one dimension of group membership. Thus, while a pervasive social distinction is made on the basis of ethnicity (e.g., Black, White, or Asian), this is by no

Preface

xv

means the only possible way to categorize others. We are all also “female” or “male”, “young” or “elderly”, “liberal” or “conservative”, even “Manchester United” or “Everton” football team supporters. The emerging ﬁndings that people can, and do, attend to more than one dimension of categorization at a time has profound implications for existing theories of group processes and intergroup relations. Work on multiple categorization is therefore an essential qualiﬁcation to our current understanding of social categorization, and one which may radically change the extant theories of how group membership, identity, stereotyping, and prejudice interrelate and interact. Collectively this wide variety of work—ranging from cognitive models of the categorization process, to interventions designed to promote social harmony, to complex social system analysis from the social sciences more broadly—focuses on multiple social categorization. This edited book brings together, for the ﬁrst time, a collection of the most important and inﬂuential scholars in this area, to review past research and set the agenda for future work. We hope that it will not only help to illuminate the importance of considering the multitude of categorization opportunities available to perceivers, but also demonstrate how an integrative understanding of multiple categorization phenomena can have signiﬁcant implications in today’s increasingly multicultural, multiethnic, and multiracial societies.

REFERENCES Allport, G. W. (1954). The nature of prejudice. Garden City, NY: Doubleday.

Part I

Introduction

1

Multiple social categorization Context, process, and social consequences Richard J. Crisp and Miles Hewstone

Think about how important your diﬀerences are to you. Think about how we all organise our lives in little boxes: man, woman, British, American, Muslim, Christian, Jew, Tory, Labour, New Labour . . . how could you navigate life if you didn’t know the diﬀerence between a child and an adult, an African and an Indian, a scientist and a lawyer? Bill Clinton, The Dimbleby Lecture 2001, 14 December 2001 At any one time we have access to many identities, including race, sexuality, gender, nationality, class and religion. Far from being neutral, these identities are rooted in material conditions that confer power and privilege in relation to one another. These power relations, however, are not ﬁxed. They are ﬂuid in character, dynamic by nature and, therefore, complex in practice. The decisions as to which identities we assert, when we want to assert them and what we want to do with them are ours. But those decisions do not take place in a vacuum. They are shaped by circumstance and sharpened by crisis. We have a choice about which identities to give the ﬂoor to; but at speciﬁc moments they may also choose us. Gary Younge, The Guardian, 21 January 2005

Categorization is integral and essential to human social interaction (Allport, 1954; Chaiken & Trope, 1999; Fiske & Neuberg, 1990). We classify and categorize without, for the most part, giving it a second thought—but it is crucial for our successful navigation of social life. As journalist Gary Younge notes above, our multiple identities are not simple, neutral, passive deﬁnitions of our existence; they are laden with implied status, power and value, and therefore of profound consequence for our lives. Categorization is dynamic and ﬂuid—and we can both choose to be categorized, or categories can be chosen for us. Whether we are talking about being black or white, male or female, young or old, such social classiﬁcations have signiﬁcant implications for how we think about ourselves and form impressions of others. In any given situation the categories that deﬁne ourselves, and others, can depend on various factors, such as context (e.g., Oakes, Turner, & Haslam, 1991) and motivation (e.g., Sinclair & Kunda, 1999) and increasingly it is apparent that

4

Multiple social categorization

in many contexts multiple bases for social categorization can be salient, combined and used simultaneously (e.g., see Brewer, Ho, Lee, & Miller, 1987; Stangor, Lynch, Duan, & Glass, 1992). This volume discusses how our classiﬁcation of ourselves, and others, along these multiple criteria, can impact on psychological and social processes. In the following chapters the authors discuss how social categorization— the classiﬁcation of self and others into broad social groupings—has an impact on how we perceive, think, and behave. Social categories can be thought about with no reference to self-inclusion—Asian, British, young—or they can be classiﬁed by membership. That is, we can deﬁne social contexts by the groups to which we belong (“ingroups”) and the groups to which we do not belong (“outgroups”). In such intergroup contexts, ingroup and outgroup membership can have signiﬁcant implications for social judgment and behavior (Hewstone, Rubin, & Willis, 2002) and the chapters in this volume discuss much about how categorization aﬀects intergroup bias—the extent to which someone evaluates their own group more positively than other groups. The term “multiple categorization” refers to any intergroup context that involves perceiving more than a single basis for social classiﬁcation. Examples abound from everyday life. In the 1983 British General Election a Conservative Party poster depicted a young black man, with the caption “Labour says he’s black. We say he’s British.” Writing about the debate within Britain on what it means to be “British”, Home Secretary Jack Straw returned to this poster and wrote, “Their suggestion that one had to make a choice—that one could not be both—was absurd . . .” (Straw, 2000). More recently, Abdolkarim Sorush, a leading Iranian intellectual, oﬀered the view that every Iranian has three identities: Shia, Persian, and Western.1 Research into multiple categorization is appealing because it qualiﬁes and reﬁnes existing approaches to the study of social identiﬁcation and social categorization. Typically, work in intergroup relations has focused on single ingroups and outgroups. In increasingly multicultural, multiethnic, and multireligious societies, there are now, however, multiple ways in which we can be diﬀerent from others. This complexity of intergroup aﬃliation presents an intriguing challenge to theoretical accounts of group processes and intergroup relations. In this introductory chapter, we provide a broad overview of the current state of research across these varied perspectives.

FOUNDATIONS That people can be aﬃliated to multiple social categories is not a new idea, nor exclusively an idea explored by social psychologists. Anthropologists, sociologists, and political scientists were the ﬁrst to observe some unique

1

Cited by Armstrong, K. (2003), “Faith and freedom”. The Guardian, 8 May.

Crisp and Hewstone 5 consequences associated with the apparent use of multiple categorization in societal contexts. Evans-Pritchard (1940) and Murphy (1957) both document cases of reduced conﬂict in cultures with cross-cutting social structures (see also Coser, 1956; Deutsch, 1973; Gluckman, 1955; Lipset, 1960; Rae & Taylor, 1970). LeVine and Campbell (1972) suggested that group members actively make use of cross-cutting aﬃliations because they ensure security and stability (it is more diﬃcult, for instance, to have conﬂictual relations with a group based on territory that is simultaneously an ally according to common ancestry). In a famous example, sociologist Georg Simmel (1950) observed that in the Australian Aborigine culture the whole population is divided into ﬁve “gentes”, with members of the various gentes found across many diﬀerent tribes. Such cross-cutting contexts provide divergent bases for group membership—tribe does not correspond to gente (group memberships neither converge nor correlate). Research has conﬁrmed that cross-cutting ties can be important on broad societal levels (Cairns & Mercer, 1984; Ross, 1985; see Phinney & Alipuria, this volume) and such category structures have emerged as an important concept for political scientists (see Horowitz, 1985; Wallace, 1973; see Carter, this volume). Based on these observations, models have been developed that focus on the perception and use of multiple categorization. Examples of work in this area focus on varied phenomena such as selection of multiple alternative categories in person perception (Macrae, Bodenhausen, & Milne, 1995; van Rijswijk & Ellemers, 2002), the emergence of novel stereotypes following category combinations (Smith & DeCoster, 1999), or the causal reasoning processes accompanying the emergence of those stereotypes (Kunda, Miller, and Claire, 1990). Other examples include the use of hierarchically ordered subtypes in memory (Stangor et al., 1992) or how multiple categories are represented and used in social judgment (Crisp, Ensari, Hewstone, & Miller, 2002; Klauer, Ehrenberg, & Wegener, 2003) as well as broader sociological, anthropological, and political science perspectives (e.g., Evans-Pritchard, 1940; Horowitz, 1985; LeVine & Campbell, 1972; Murphy, 1957; Wallace, 1973). The chapters in this volume address these very issues. There are now an increasing number of psychological perspectives on multiple categorization which range from accounts of categorical structuring of knowledge about diﬀerent social groups (see McGarty, this volume), and the representation of stereotypic attributes and how these attributes are combined (see Smith, this volume), through to the implications of multiple categorization for social judgment (see chapters in this volume by Crisp; Dovidio et al.; Hogg & Hornsey; Miller et al.; Singh; and Vescio et al.). We brieﬂy review these diﬀerent models of category representation and evaluation in terms of two speciﬁc criteria: (a) in terms of the conjunctions formed by considering multiple criteria (i.e., “combined categories”) and (b) in terms of the separable components of targets formed by considering multiple criteria (i.e., “constituent categories”).

6

Multiple social categorization

A TAXONOMY OF MULTIPLE CATEGORIZATION There are many ways in which multiple social categories can be perceived, represented, and used. One of the key distinctions in the literature is between models of multiple categorization at varying levels of inclusiveness (where inclusiveness is deﬁned as the property of one category, e.g., European, subsuming another, e.g., Britain). For instance, a considerable body of work has examined the eﬀects on evaluations of making a superordinate social classiﬁcation salient instead of, or in addition to, a subordinate intergroup dichotomy (Gaertner & Dovidio, 2000; Gaertner, Dovidio, Anastasio, Bachman, & Rust, 1993; Gaertner, Mann, Dovidio, Murrell, & Pomare, 1990; Gaertner, Mann, Murrell, & Dovidio, 1989; see also Crisp; Hogg & Hornsey; and Dovidio et al., this volume). This experimental research has shown, beyond doubt, that the introduction of a superordinate categorization can reduce bias. Whether, or for how long, it does so, however, in extra-laboratory conﬂicts is another matter. National sporting success is sometimes seen as a means for superordinate national categorizations to dominate more conﬂictual subordinate ethnic groupings. Thus, when the Rwandan national football team beat favorite Uganda to qualify for the African Nations Cup in February 2003, the victory was seen as having signiﬁcance way beyond the world of sport, because the Hutu and Tutsi communities (caught up in genocide in 1994) celebrated together. Yet the optimism surrounding South Africa’s Rugby World Cup success in 1995 and France’s Football World Cup victory in 1998 was only brief, despite initial celebrations. South Africa’s sole black player, winger Chester Williams, opined that the “one team, one country” image was “a lie”,2 and one of France’s star black players, Thierry Henry, expressed the view, just four year later, that “when we won the World Cup for three months everyone was French. No one was saying someone was black, white, Algerian. It was nice but of course it did not last.”3 Studies of “crossed categorization” explore the eﬀects on intergroup evaluations when two dichotomous dimensions of group membership are simultaneously salient (Brewer et al., 1987; Crisp & Hewstone, 2000a; 2001; Deschamps & Doise, 1978; Ensari & Miller, 1998; Hewstone, Islam, & Judd, 1993; for reviews see Crisp & Hewstone, 1999a; 2000b; Migdal, Hewstone, & Mullen, 1998; Mullen, Migdal, & Hewstone, 2001; Urban & Miller, 1998). Although similar conceptually to the common ingroup identity model, work in crossed categorization makes no speciﬁcation of the inclusiveness of the cross-cutting categorizations, and examines the eﬀects of both convergent and divergent categorization. Convergent categorization refers to targets that are designated outgroup on multiple dimensions simultaneously. For example,

2 3

Colquhoun, A. (2003). Spokesman reignites Springbok race row”. The Guardian, 3 September. The Independent, 28 September 2002.

Crisp and Hewstone 7 when gender and age dimensions are crossed, one of the composite groups created is convergent (young females are most diﬀerent categorically from old males—from whom they diﬀer according to two criteria). Divergent categorization refers to the comparison of two targets who share membership on one dimension, but who belong to diﬀerent groups on another. At the time of Indian independence the philosopher Bertrand Russell said to Indian political leader Pandit Nehru that they had one thing in common, the fact that they were both atheists. “Yes,” replied Nehru, “but never forget, Russell, that you are a Christian atheist, and I am a Hindu one.” Like work on the common ingroup identity model, cross-cutting categorization has also been found to moderate evaluations (e.g., Crisp, Hewstone, & Rubin., 2001; Vanbeselaere, 1987; 1991; see Singh; Miller et al. and Vescio et al., this volume). Other research has focused on the eﬀects of “subgrouping” (considering multiple categories that are less inclusive than initial target groups; Brewer, Dull, & Lui, 1981; Maurer, Park, & Rothbart, 1995; Park, Ryan, & Judd, 1992; for a review see Richards & Hewstone, 2001). For example, within universities there are typically subgroups that deﬁne diﬀerent areas of study, such as natural sciences, humanities, etc. Within these broad groupings there are further subgroups, so within natural sciences one can study physics, chemistry, etc. Interestingly, while the common ingroup identity and crossed categorization models have focused on the shared and overlapping nature of multiple memberships, subgrouping research has been more concerned with intragroup eﬀects, and has not examined the intercategory implications of shared and overlapping subgroup categorization. Whereas the common ingroup identity and crossed categorization models have been almost exclusively concerned with intergroup evaluations, work into subgrouping has been almost exclusively concerned with stereotyping (typically speciﬁc nonevaluative beliefs about groups) and perceived group variability (typically, the extent to which all members of a group are perceived as similar or diﬀerent from one another; Park & Judd, 1990; Park, Judd, & Ryan, 1991). Subgrouping not only has a moderating inﬂuence on stereotyping, but also seems the natural way people structure their intracategory representations (Park et al., 1992). Ingroup information can be organized into subgroups, which can be encoded, recalled, and applied easily and accurately, while there generally appears insuﬃcient information to organize outgroup representations in the same way (see Richards & Hewstone, 2001). While subgrouping has beneﬁcial eﬀects on stereotyping and perceived group variability, its eﬀects on broader positive or negative intergroup evaluations are relatively unexplored. Although the diﬀerential inclusiveness of social categories has undoubtedly revealed distinct eﬀects in the subgrouping compared to the common ingroup identity and crossed categorization literature, this appears to some extent to be due to the diﬀerent intra- and intergroup foci of the diﬀerent research traditions (stereotyping versus evaluations, respectively). We have recently examined the predictive power of inclusiveness in diﬀerentiating

8

Multiple social categorization

models of multiple categorization. We found common representational and evaluative consequences irrespective of whether categories combined were of diﬀerent levels of inclusiveness (Crisp, Beck, & Hewstone, 2004). This suggests that, at least with respect to the phenomena considered in this volume, diﬀerential inclusiveness does not have a critical moderating role. More crucial is a focus on how categories are represented and used when perceived simultaneously, and it is this focus that we adopt. REPRESENTING MULTIPLE CATEGORIES

Category conjunctions When multiple dimensions of social categorization are simultaneously salient, one task for the social perceiver is to integrate the multiple sources of information to form a holistic impression of the target person. The ﬁrst class of models we consider have explored the psychological processes and implications of such category conjunctions. The ﬁrst way that perceivers can deal with integrating multiple categories is, in fact, to avoid any integration at all and simply focus on a single basis for classiﬁcation. Put another way, if one of the constituent categories is for some reason perceptually dominant, a second category will simply be ignored. It is well documented that when one category is more salient (e.g., Macrae et al., 1995), or more meaningful with respect to the context (Oakes, Haslam, & Turner, 1994), then that particular category will dominate the subsequently formed impressions. For instance, Van Twuyver and van Knippenberg (1999), using a memory-based paradigm (“who-said-what”; Taylor, Fiske, Etcoﬀ, & Ruderman, 1978), found evidence supportive of a “negative interdependence” between multiple categories applicable to particular targets (see also van Knippenberg & Dijksterhuis, 2000). Speciﬁcally, the use of a gender categorization was negatively correlated with the use of a student/teacher categorization. Such competitive activation of categories and stereotypes may be functional and meaningful in many situations (enabling eﬃciency gains in person perception; Dijksterhuis & van Knippenberg, 1998). Despite this apparent mutual exclusivity of category use in some contexts, this is not always the case. There is also evidence that people can and do process, represent, and use multiple dimensions of social categorization simultaneously in a variety of real intergroup settings. Importantly, it is in the absence of any contextual dominance of one category over the other that perceivers seem to notice and structure memory according to multiple criteria (Crisp & Hewstone, 2001; see also Brewer et al., 1981; Stangor, Lynch, Duan, & Glass, 1992). For example, we found that in Northern Ireland participants did not only spontaneously remember people’s religion, but also their gender (Crisp, Hewstone, & Cairns, 2001), and in Bangladesh, both nationality and religion structured judgments about target persons (Hewstone et al., 1993; for a review see Crisp et al., 2002).

Crisp and Hewstone 9 How then are categories combined when there is no clearly dominant constituent? Some seminal work by Kunda and Hastie suggests an answer to this question. Kunda et al. (1990) used the example of a “Harvard-educated carpenter” and asked whether such a combination would be perceived as possessing characteristics associated with being “Harvard-educated” or characteristics associated with “carpenters”, or some additive or interactive combination of these constituent categories (see also Hastie, Schroeder, & Weber, 1990). This research illustrated that when multiple categories are perceived, then the resulting representation can be either a simple additive, or more complex interactive, combination of the constituent category stereotypes. Whether it is additive or interactive depends on the nature of the context and the relationship between the categories involved. An additive combination results in a representation based upon a simple linear combination of attributes associated with both constituents (see Anderson, 1965; Singh, Yeoh, Lim, & Lim, 1997). This appears to be the case particularly when the combination is unsurprising (e.g., a male mechanic). When, however, the combination is surprising (e.g., a female mechanic), there appears to be a more complex interactive process involved. The nature of a particular combination of social categories (surprising or unsurprising) can thus determine whether an additive or interactive combination of categories deﬁnes the resulting impression formed. An interactive combination of categories is unique in that it results in the generation of “emergent attributes”, which are characteristics applied to the combination that are not typically applied to either constituent in isolation (Hastie et al., 1990; Kunda et al., 1990). For example, a disabled athlete may be perceived as particularly headstrong, which may not be such a salient characteristic associated with either disabled people or athletes independently. Such a pattern may be likely when the attributes associated with the two constituents conﬂict with one another (Wilkenﬁeld & Ward, 2001). Kunda et al. (1990), for example, suggested that a “Harvard-educated carpenter” might be perceived as “nonmaterialistic (an emergent attribute) because having attended Harvard conﬂicts with carpentry as a chosen occupation. The emergent attribute is thus the result of an attempt to resolve the conﬂict and create a coherent impression of the target person. This reasoning echoes in Uleman, Newman, and Moskowitz’s (1996) suggestion that when information relating to a particular target is not already available from memory (which, by deﬁnition, it is not when perceiving surprising combinations), impressions will be constructed in a spontaneous manner. Similar ﬁndings have also been observed in studies of combining nonsocial concepts; it is the surprising or incongruent nature of the perceived concepts that appears to produce emergent attributes (Hampton, 1997; Murphy, 1988; Wattenmaker, 1995; Wilkenﬁeld & Ward, 2001). A further consequence of perceiving surprising combinations of social categories is an inhibition of constituent stereotypes that accompanies the generation of emergent attributes. For instance, Hutter and Crisp (2005) found that characteristics

10

Multiple social categorization

associated with females and mechanics were perceived as less applicable to the combination “female mechanic” than when either females or mechanics were considered independently of one another. These ﬁndings illustrate how multiple categorization can have an impact not only on the way that categories are combined, but also on the representation of the constituents that comprise such combinations.

Category constituents How does considering multiple criteria for social categorization aﬀect the way we think about the categories that comprise any combination, independent of that combination? For example, if gender and age are salient, then a young female perceiver may think of others as a combination of categorizations along both salient criteria (e.g., comparing herself with an elderly male). However, what happens when context requires the perceiver to revert to focusing on the separable components of this combination (i.e., being either just young or just female)? Will there be any carry-over eﬀects on the representation of social categories after one has become aware that people can be deﬁned by multiple, cross-cutting categories and conjunctions of categories? Would our young female’s perception of the elderly per se change after contact with an elderly female with whom she shares a group membership? These questions form the basis for work that has examined whether multiple categorization can change intergroup relations and reduce intergroup bias. Doise’s (1978) category diﬀerentiation model outlines the expected eﬀects of multiple categorization on perceived similarity and diﬀerentiation (see also the meta-contrast process outlined by self-categorization theory; Turner, 1985; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). This model suggests that considering multiple categories that cross-cut one another will lead to the operation of counteracting cognitive processes. The act of (simple) categorization leads to an accentuation of diﬀerences between, and similarities within, categories (Campbell, 1956; Tajfel, 1959; Tajfel & Wilkes, 1963). However, in a context of crossed categorization, where the dimensions of categorization are diametrically and orthogonally opposed, these processes should work against one another (Deschamps & Doise, 1978). The accentuation of diﬀerences on an initial dimension of classiﬁcation will be accompanied by a simultaneous and counteracting accentuation of similarities on a second salient classiﬁcation. Thus, age can cause an accentuation of diﬀerences between young and elderly group members. When, however, the young and elderly group members are also female, then young (females) and elderly (females) will be perceived as more similar. These proposed diﬀerentiation-reducing eﬀects of crossed categorization have been reliably demonstrated many times with respect to both non-social (Crisp & Hewstone, 1999b; Deschamps, 1977) and social (Arcuri, 1982; Deschamps & Doise, 1978; van Twuyver & van Knippenberg, 1999) stimuli.

Crisp and Hewstone 11 Work on the common ingroup identity model (Gaertner & Dovidio, 2000) has also outlined the diﬀerentiation-reducing properties of multiple categorization. Gaertner et al. (1990; 1993) specify how a range of causal factors can have implications for the cognitive representation of ingroups and outgroups (see also Gaertner & Dovidio, 1986). Goal relations and group interactions aﬀect intergroup attitudes apparently by altering the salience of relevant category distinctions (Doise, 1978; Turner, 1982; Worchel, 1979). Cooperative interdependence (i.e., common task or fate; see Brown & Abrams, 1986; Brown & Wade, 1987; Gaertner et al., 1989; 1993; Sherif, 1966) can transform members’ cognitive representation of group boundaries from “us” and “them” to a more inclusive “we”. In sum, multiple social categorization appears to have signiﬁcant implications not only for how conjunctions of categories are represented, but also for the constituents that make up the combinations. In the next section we consider research that has focused on how these diﬀerent representational consequences of multiple categorization aﬀect social judgment and, in particular, intergroup evaluations. EVALUATING MULTIPLE CATEGORIES

Category conjunctions When categories are considered as combined entities, this leads to some interesting implications for intergroup evaluations. These eﬀects have been examined speciﬁcally in the crossed categorization literature (see also work on simultaneous categorization related to the common ingroup identity model; Hornsey & Hogg, 2000). In this paradigm (Deschamps & Doise, 1978), two orthogonal dimensions of categorization are crossed to form four new “crossed category” groups. Instead of considering only females versus males or young versus elderly, in crossed categorization situations perceivers evaluate targets with respect to both of these dimensions and respond to composite groups such as young females, young males, elderly females and elderly males. In terms of ingroup/outgroup relations there are four groups that are similar and diﬀerent from the perceiver in distinct ways. If our perceiver is a young female, then other young females are double ingroup members (sharing group membership with the perceiver on both dimensions of categorization), young males and elderly females are mixed category members (being partially ingroup and partially outgroup), and elderly males are double outgroup members (being diﬀerent from the perceiver on both dimensions of categorization). How evaluations vary across the four category conjunctions formed by crossing categories has been the focus of several extensive research programs. A number of diﬀerent patterns of evaluations across the composite groups have been observed (Brewer et al., 1987; Hewstone et al., 1993) and there are six main patterns that have been identiﬁed (see Table 1.1; also Miller et al. and

12

Multiple social categorization

Table 1.1 A priori predictions of the six main outcome patterns of evaluation resulting from crossed categorization. (adapted from Crisp et al., (2002), reproduced with permission from Taylor & Francis)

Pattern 1. Additive 2. Dominance 3. Social inclusion 4. Social exclusion 5. Hierarchical acceptance 6. Hierarchical rejection

Dimension 1: Ingroup

Dimension 1: Outgroup

Dimension 2: Ingroup

Dimension 2: Ingroup

Dimension 2: Outgroup

Dimension 2: Outgroup

ii ii ii

> = =

io io io

= > =

oi oi oi

> = >

oo oo oo

ii

>

io

=

oi

=

oo

ii

>

io

>

oi

=

oo

ii

=

io

>

oi

>

oo

Vescio et al., this volume, for more details). These patterns have been observed across a number of studies, and reviews of the literature have conﬁrmed the most common among them in varied antecedent contexts (Crisp & Hewstone, 1999a; Migdal et al., 1998; Urban & Miller, 1998). Subsequent experimental work has also ascertained the moderating conditions that predict when different patterns will be observed (e.g., Crisp & Hewstone, 2000a; 2001; Crisp et al., 2003; Ensari & Miller, 1998; Kenworthy, Canales, Weaver, & Miller, 2003; Urada & Miller, 2000). Recently, Crisp et al. (2002) outlined a dualroute model to predict and explain the processes involved in evaluating category combinations. Crisp et al.’s (2002) model of crossed categorization eﬀects focuses on the apparent aﬀective nature of ingroups and outgroups. Ingroup and outgroup membership seems to carry an acquired aﬀective valence (Dovidio, Evans, & Tyler, 1986; Fazio, Jackson, Dunton, & Williams, 1995; Perdue, Dovidio, Gurtman, & Tyler, 1990). Ingroup categories are inherently positive in connotation, and outgroup categories inherently negative. Whether this association develops via a process of associative learning with real groups over time (e.g., consistently hearing ingroups being referred to in positive terms, outgroups in negative terms, throughout development, Bargh, 1997; Perdue et al., 1990; see also Cacioppo, Marshall-Goodell, Tassinary, & Petty, 1992; Das & Nanda, 1963; Staats & Staats, 1958) or automatic generalization of self-positivity with novel groups (i.e., perceiving the ingroup to have positive traits because it includes the self; Cadinu & Rothbart, 1996; Otten & Wentura, 1999), there seems at least to be a strong positive default whereby ingroups are perceived as more positive than outgroups (Maass & Schaller, 1991) that can account for ingroup favoritism. Activation of positive and negative aﬀect with respect

Crisp and Hewstone 13 to the ingroup and outgroup respectively can be expected to inﬂuence judgments of ingroups and outgroups. A tendency to feel more positively toward ingroups (and more negatively toward outgroups) will lead to the baseline additive pattern (where evaluation occurs as a linear function of ingroup membership; see Table 1.1) in conjunctive category contexts. Evidence that consideration of multiple social criteria can change evaluations of combined category targets comes from work showing that mood can moderate the pattern of crossed categorization observed. If in-/outgroup status has acquired aﬀect (Otten & Wentura, 1999; Perdue et al., 1990), then positive or negative mood should lead to changes in the salience of the aﬀective connotations of ingroup and outgroup membership. Urada and Miller (2000) suggest that positive mood will eﬀectively cue positive material in memory (Isen, Shalker, Clark, & Karp, 1978; Teasdale & Russell, 1983), and have a priming eﬀect, whereby attention is directed primarily to other positive aspects in the environment (Higgins & King, 1981; see also Miller et al., this volume). Evaluations of combined groups will be modiﬁed in an aﬀective context that enhances the positive (ingroup) characteristics of combined groups. There is indeed growing evidence that positive and negative mood can lead to shifts from the baseline additive pattern to more complex interactive patterns of evaluation (Crisp & Hewstone, 2000a; Ensari & Miller, 1998; Kenworthy et al., 2003; Urada & Miller, 2000). In sum, when people evaluate targets deﬁned by a combination of categories, then intergroup evaluations appear to vary depending on the aﬀect associated with the relative composition of ingroup (positive) and outgroup (negative) constituents.

Category constituents As well as category combinations, from the work reviewed above it is apparent that multiple categorization can have an impact on whether constituent ingroups and outgroups are perceived to be similar or distinct, and we know from existing work on intergroup relations that similarity exerts a signiﬁcant impact on evaluations. For instance, Tajfel’s work with the minimal group paradigm (e.g., Tajfel, Flament, Billig, & Bundy, 1971) illustrated how even with no prior contact, anonymity and, with meaningless social categories, the knowledge that “they” are diﬀerent from “us” can translate into evaluative diﬀerentiation (see Brewer, 1979; Mullen, Brown, & Smith, 1992). In other words, similarity and diﬀerentiation can deﬁne intergroup attitudes. Models have been proposed to explain the relationship between intergroup similarity and intergroup evaluation. The work discussed above by Campbell (1956) and Tajfel (1959; Tajfel & Wilkes, 1963) has been expanded in various theoretical accounts such as Doise’s (1978) category diﬀerentiation model and Turner et al.’s (1987) self-categorization theory that we discussed earlier. While the emphasis of these accounts varies, what is common to all is the notion that categorization provides a psychological basis for understanding

14

Multiple social categorization

“them” to be diﬀerent from “us”, and it is the emergence of this distinction between ingroups and outgroups that provides the prerequisite for intergroup discrimination. Deschamps and Doise argue that ingroup favoritism will be positively correlated with cognitive diﬀerentiation. Put another way, as groups become less diﬀerent, they will also be evaluated less diﬀerently. Gaertner and Dovidio’s work on the common ingroup identity model has provided an extensive program of empirical work supporting this idea (Gaertner & Dovidio, 2000; Gaertner et al., 1989, 1990, 1993, 1994) and there is consistent mediational evidence that the formation of a superordinate membership (Gaertner et al., 1989, 1990) and increased intergroup similarity (Marcus-Newhall, Miller, Holtz, & Brewer, 1993) explain reductions in ingroup favoritism. As “us” and “them” psychologically merge into each other, the cognitive prerequisite for bias will be removed and evaluations of ingroups and outgroups will converge. In sum, similarity induced by considering shared categorizations appears to exert a moderating eﬀect on evaluation (although this diﬀerentiation-discrimination link may not always be a simple one, see Crisp; Vescio et al., this volume).

INTEGRATION AND FRAMEWORK From the above review we can summarize a broad framework that can provide a reference point when reading the chapters within this volume (see Table 1.2). We would argue that studies fall into two broad (and cross-cutting) bases for classiﬁcation. First, most research has tended to focus either on representation or evaluation, that is, the implications of multiple categorization for how people think about, and use, social categories in social judgment. Second, work on the representation and evaluation of multiple categories can focus on the conjunctions formed by combining social categories, or the implications for individual constituents once devolved from any context that makes multiple categories temporarily salient. We can thus classify the extant approaches to the study of multiple categorization according to a focus on representation or evaluation, and, orthogonal to this, a focus on conjunctions or constituents. As a caveat, it is notable that some recent work has focused on the simultaneous study of both representation and evaluation, and of both conjunctions and constituents, and this is reﬂected in some of the chapters in this volume. Each chapter does, however, focus on at least one of these four perspectives as summarized in Table 1.2. For instance, McGarty (Chapter 2) and Smith (Chapter 3) focus on the formation, representation, and use of multiple categorization. Dovidio et al. (Chapter 4), Crisp (Chapter 5), and Hogg and Hornsey (Chapter 6) all consider the implications of multiple categorization for constituent diﬀerentiation and evaluation. Vescio et al. (Chapter 7), Miller et al. (Chapter 8), and Singh (Chapter 9) consider how category conjunctions are evaluated. The ﬁnal section of the book includes

Crisp and Hewstone 15 Table 1.2 Multiple categorization: Current trends Outcome focus Categorization focus

Representation

Evaluation

Conjunctions

Chapter 2: McGarty Chapter 3: Smith Chapter 7: Vescio et al. Chapter 10: Phinney & Alipuria Chapter 11: Carter

Chapter 7: Vescio et al. Chapter 8: Miller et al. Chapter 9: Singh

Constituents

Chapter 4: Dovidio et al. Chapter 6: Hogg & Hornsey

Chapter 4: Dovidio et al. Chapter 5: Crisp Chapter 6: Hogg & Hornsey

two chapters in which the relevance of these foci in psychological work can be applied to wider societal issues. Phinney and Alipuria (Chapter 10) discuss the relevance of multiple categorization for sociological conceptualizations of multiculturalism, while Carter (Chapter 11) discusses how a focus on multiple categorization can contribute to a better understanding of political structures and processes. Both of these chapters consider how conjunctions of multiple categories are represented at broader societal levels. Consideration of the full spectrum of work in multiple categorization is beyond the remit of one volume and we are concerned primarily with some speciﬁc consequences of considering multiple memberships. Although the above is but one possible framework for considering multiple categorization, it oﬀers the potential for furthering integration of work on the cognitive, motivational, and behavioral consequences of such phenomena. We discuss the potential for such integration, and oﬀer a future agenda, in the ﬁnal chapter of this volume. In this ﬁnal chapter we consider the research discussed in the preceding chapters, as it relates to recent developments within the social-psychological literature, as well as with other areas of psychology and indeed broader disciplines in the social sciences. For instance, we consider how Roccas and Brewer’s (2002) concept of social identity complexity has elaborated on conceptualizations of multiple categorization and how work on essentialism, including judgments of ingroup homogeneity, group polarization, self-anchoring, and entitativity (see Castano, Yzerbyt, & Bourguignon, 2003; Yzerbyt, Estrada, Corneille, Seron, & Demoulin, 2004), has important implications for future studies of multiple categorization. The ﬁnal chapter further considers issues raised by Phinney and Alipuria (Chapter 10), and Carter (Chapter 11), and how multiple categorization can contribute to an understanding of biculturalism (see Berry, 1986; LaFromboise, Coleman, & Gerton, 1993; Phinney, Lochner, & Murphy,

16

Multiple social categorization

1990), a topic that has until recently received relatively little attention from social psychologists. On the basis of these discussions the ﬁnal chapter outlines seven themes that we believe underlie much of the thinking outlined in this volume. We note them only brieﬂy here, as an appetizer, in order to prepare the reader for what we see as the most important issues arising from this work. We discuss how some of the basic principles of categorization theory link all the chapters: the assumption that categorization is necessary to function in a highly complex social world (theme 1), but also how categorization can involve either involuntarily classifying others, or unconsciously deﬁning ourselves (theme 2). Categorization is not always implicit and involuntary; we can choose to self-categorize or categorize others to satisfy varied motives (theme 3). As well as the issue of choice, how we use categorization also changes over time, both within individuals and within cultures (theme 4). It is this dynamic, essential, and ﬂexible quality of categorization that makes it both a cause and potential cure of intergroup discrimination (theme 5), and understanding the cognitive and motivational processes (theme 6) that determine and deﬁne our use of categories is already providing increasingly sophisticated means for attenuating intergroup bias. Finally, however, we must be careful not to equate the psychological deﬁnition of intergroup bias with the broader notion of intergroup conﬂict. The latter is more complex and requires a multidisciplinary focus. To tackle conﬂict we must be prepared to integrate work from diverse perspectives on multiple categorization, incorporating historical, political, sociological, and anthropological perspectives (theme 7). Our social world is increasingly characterized by multiple and cross-cutting group aﬃliations, and, as such, the psychological processes and implications of such relations are becoming increasingly relevant. Work on multiple categorization has qualiﬁed in important ways our understanding of how social classiﬁcation aﬀects pressing social issues like prejudice and discrimination. Future and further clariﬁcation and reﬁnement of the extant phenomena associated with such complex group aﬃliation will likely yield more insights. What we now understand of the intricate links between mental representation and evaluation of social groups is contributing to the development of prejudice reduction and conciliation strategies. This is one example of where the application of models of multiple categorization may provide a valuable contribution to our understanding of social issues, and indeed, the means to tackle a pervasive social problem. We believe there may be many more, and we hope that the chapters in this volume will provide the grounding for multiple (and perhaps cross-cutting) research agendas for future work into these exciting and important social phenomena.

Crisp and Hewstone 17 ACKNOWLEDGEMENTS Preparation of this chapter was supported by British Academy (SG-31041 and SG-34045), and Leverhulme Trust (F/00094/H) grants to R. J. Crisp and an Economic and Social Research Council grant (R000239382) to R. J. Crisp and M. Hewstone.

REFERENCES Allport, G. W. (1954). The nature of prejudice. Garden City, NY: Doubleday. Anderson, N. H. (1965). Averaging versus adding as a stimulus-combination rule in impression formation. Journal of Experimental Psychology, 70, 394–400. Arcuri, L. (1982). Three patterns of social categorization in attribution memory. European Journal of Social Psychology, 12, 271–282. Bargh, J. A. (1997). The automaticity of everyday life. In R. Wyer Jr. (Ed.), The automaticity of everyday life: Advances in social cognition (Vol. 10, pp. 1–61). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Berry, J. W. (1986). Multiculturalism and psychology in plural societies. In L. H. Ekstrand (Ed.), Ethnic minorities and immigrants in a cross-cultural perspective (pp. 37–51). Lisse: Swets & Zeitlinger. Brewer, M. B. (1979). Ingroup bias in the minimal intergroup situation: A cognitivemotivational analysis. Psychological Bulletin, 86, 307–324. Brewer, M. B., Dull, V., & Lui, L. (1981). Perceptions of the elderly: Stereotypes as prototypes. Journal of Personality and Social Psychology, 41, 656–670. Brewer, M. B., Ho, H. K., Lee, J. Y., & Miller, N. (1987). Social identity and social distance among Hong Kong schoolchildren. Personality and Social Psychology Bulletin, 13, 156–165. Brown, R. J., & Abrams, D. (1986). The eﬀects of intergroup similarity and goal interdependence on intergroup attitudes and task performance. Journal of Experimental Social Psychology, 22, 78–92. Brown, R. J., & Wade, G. (1987). Superordinate goals and intergroup behaviour: the eﬀect of role ambiguity and status on intergroup attitudes and task performance. European Journal of Social Psychology, 17, 131–142. Cacioppo, J. T., Marshall-Goodell, B. S., Tassinary, L. G., & Petty, R. E. (1992). Rudimentary determinants of attitudes: Classical conditioning is more eﬀective when prior knowledge about the attitude stimulus is low than high. Journal of Experimental Social Psychology, 28, 207–233. Cadinu, M. R., & Rothbart, M. (1996). Self-anchoring and diﬀerentiation processes in the minimal group setting. Journal of Personality and Social Psychology, 70, 661–677. Cairns, E., & Mercer, G. W. (1984). Social identity in Northern Ireland. Human Relations, 37, 1095–1102. Campbell, D. T. (1956). Enhancement of contrast as a composite habit. Journal of Abnormal and Social Psychology, 53, 350–355. Castano, E., Yzerbyt, V., & Bourguignon, D. (2003). We are one and I like it: The impact of ingroup entitativity on ingroup identiﬁcation. European Journal of Social Psychology, 33, 735–754.

18

Multiple social categorization

Chaiken, S., & Trope, Y. (Eds.) (1999). Dual process theories in social psychology. New York: The Guilford Press. Coser, L. A. (1956). The function of social conﬂict. New York: Free Press. Crisp, R. J., Beck, S. R., & Hewstone, M. (2004) Crossed categorization, dual identiﬁcation, and the inclusiveness of common ingroup identities. Manuscript in preparation. Crisp, R. J., Ensari, N., Hewstone, M., & Miller, N. (2002). A dual-route model of crossed categorization eﬀects. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 13, pp. 35–74). Hove, UK & Philadelphia, PA: Psychology Press. Crisp, R. J., & Hewstone, M. (1999a). Diﬀerential evaluation of crossed category groups: Patterns, processes, and reducing intergroup bias. Group Processes and Intergroup Relations, 2, 303–333. Crisp, R. J., & Hewstone, M. (1999b). Subcategorization of physical stimuli: Category diﬀerentiation and decategorization processes. European Journal of Social Psychology, 29, 665–671. Crisp, R. J., & Hewstone, M. (2000a). Crossed categorization and intergroup bias: The moderating roles of intergroup and aﬀective context. Journal of Experimental Social Psychology, 36, 357–383. Crisp, R. J., & Hewstone, M. (2000b). Multiple categorization and social identity. In D. Capozza & R. Brown (Eds.), Social identity theory: Trends in theory and research (pp. 149–166). Beverly Hills, CA: Sage Publications. Crisp, R. J., & Hewstone, M. (2001). Multiple categorization and implicit intergroup bias: Diﬀerential category dominance and the positive-negative asymmetry eﬀect. European Journal of Social Psychology, 31, 45–62. Crisp, R. J., Hewstone, M., & Cairns, E. (2001). Multiple identities in Northern Ireland: Hierarchical ordering in the representation of group membership. British Journal of Social Psychology, 40, 501–514. Crisp, R. J., Hewstone, M., Richards, Z., & Paolini, S. (2003). Inclusiveness and crossed categorization: Eﬀects on co-joined category evaluations of in-group and out-group primes. British Journal of Social Psychology, 42, 25–38. Crisp, R. J., Hewstone, M., & Rubin, M. (2001). Does multiple categorization reduce intergroup bias? Personality and Social Psychology Bulletin, 27, 76–89. Das, J. P., & Nanda, P. C. (1963). Mediated transfer of attitudes. Journal of Abnormal and Social Psychology, 66, 12–16. Deschamps, J.-C. (1977). Eﬀects of crossing category membership on quantitative judgment. European Journal of Social Psychology, 7, 517–521. Deschamps, J.-C., & Doise, W. (1978). Crossed category memberships in intergroup relations. In H. Tajfel (Ed.), Diﬀerentiation between social groups (pp. 141–158). Cambridge: Cambridge University Press. Deutsch, M. (1973). The resolution of conﬂict. London: Yale University Press. Dijksterhuis, A., & van Knippenberg, A. (1998). Inhibition, Aberdeen, and other cloudy subjects. In R. S. Wyer, Jr. (Ed.), Advances in social cognition (Vol. 11, pp. 83–96). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Doise, W. (1978). Groups and individuals: Explanations in social psychology. Cambridge: Cambridge University Press. Dovidio, J. F., Evans, N., & Tyler, R. B. (1986). Racial stereotypes: The contents of their cognitive representations. Journal of Experimental Social Psychology, 22, 22–37. Ensari, N., & Miller, N. (1998). Eﬀect of aﬀective reactions by the outgroup on

Crisp and Hewstone 19 preferences for crossed categorization discussion partners. Journal of Personality and Social Psychology, 75, 1503–1527. Evans-Pritchard, E. E. (1940). The Nuer. London: Oxford University Press. Fazio, R. H., Jackson, J. R., Dunton, B. C., & Williams, C. J. (1995). Variability in automatic activation as an unobtrusive measure of racial attitudes: A bona ﬁde pipeline? Journal of Personality and Social Psychology, 69, 1013–1027. Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression formation, from category-based to individuating processes: Inﬂuences of information and motivation on attention and interpretation. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 23, pp. 1–74). New York: Random House. Gaertner, S. L., & Dovidio, J. F. (1986). Prejudice, discrimination, and racism: Problems, progress and promise. In J. F. Dovidio & S. L. Gaertner (Eds.), Prejudice, discrimination, and racism (pp. 315–332). Orlando, FL: Academic Press. Gaertner, S. L., & Dovidio, J. F. (2000). Reducing intergroup bias: The common ingroup identity model. Philadelphia, PA & Hove, UK: Psychology Press. Gaertner, S. L., Dovidio, J. F., Anastasio, P. A., Bachman, B. A., & Rust, M. C. (1993). The common ingroup identity model: Recategorization and the reduction of intergroup bias. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 2, pp. 247–278). Chichester: Wiley International. Gaertner, S. L., Mann, J. A., Dovidio, J. F., Murrell, A. J., & Pomare, M. (1990). How does co-operation reduce intergroup bias? Journal of Personality and Social Psychology, 59, 692–704. Gaertner, S. L., Mann, J. A., Murrell, A. J., & Dovidio, J. F. (1989). Reducing intergroup bias: The beneﬁts of recategorization. Journal of Personality and Social Psychology, 57, 239–249. Gaertner, S. L., Rust, M. C., Dovidio, J. F., Bachman, B. A., & Anastasio, P. A. (1994). The contact hypothesis: The role of a common ingroup identity on reducing intergroup bias. Small Group Research, 25, 224–249. Gluckman, M. (1955). Customs and conﬂict in Africa. London: Blackwell Publishing. Hampton, J. A. (1997). Emergent attributes in combined concepts. In T. B. Ward, S. M. Smith, & J. Vaid (Eds.), Creative thought: An investigation of conceptual structures and processes (pp. 83–111). Washington, DC: APA Books. Hastie, R., Schroeder, C., & Weber, R. (1990). Creating complex social conjunction categories from simple categories. Bulletin of the Psychonomic Society, 28, 242–247. Hewstone, M., Islam, M. R., & Judd, C. M. (1993). Models of crossed categorization and intergroup relations. Journal of Personality and Social Psychology, 64, 779–793. Hewstone, M., Rubin, M., & Willis, H. (2002). Intergroup bias. Annual Review of Psychology, 53, 575–604. Higgins, E. T., & King, G. A. (1981). Accessibility of social constructs: Information processing consequences of individual and contextual variability. In N. Cantor & J. F. Kihlstrom (Eds.), Personality, cognition, and social interaction (pp. 69–122). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Hornsey, M. J., & Hogg, M. A. (2000). Subgroup relations: a comparison of mutual intergroup diﬀerentiation and common ingroup identity models of prejudice reduction. Personality and Social Psychology Bulletin, 26, 242–256. Horowitz, D. L. (1985). Ethnic groups in conﬂict. Berkeley, CA: University of California Press. Hutter, R. R. H., & Crisp, R. J. (2005). The composition of category conjunctions. Personality and Social Psychology Bulletin, 31, 647–657.

20

Multiple social categorization

Isen, A. M., Shalker, T., Clark, M. S., & Karp, L. (1978). Aﬀect, accessibility of material and behavior: A cognitive loop? Journal of Personality and Social Psychology, 36, 1–2. Kenworthy, J. B., Canales, C. J., Weaver, K. D., & Miller, N. (2003). Negative incidental aﬀect and mood congruency in crossed categorization. Journal of Experimental Social Psychology, 39, 195–219. Klauer, K. C., Ehrenberg, K., & Wegener, I. (2003). Crossed categorization and stereotyping: Structural analyses, eﬀect patterns, and dissociative eﬀects of context relevance. Journal of Experimental Social Psychology, 39, 332–354. Kunda, Z., Miller, D. T., & Claire, T. (1990). Combining social concepts: The role of causal reasoning. Cognitive Science, 14, 551–577. LaFromboise, T., Coleman, H. L., & Gerton, J. (1993). Psychological impact of biculturalism: Evidence and theory. Psychological Bulletin, 114, 395–412. LeVine, R. A., & Campbell, D. T. (1972). Ethnocentrism: Theories of conﬂict, ethnic attitudes and group behaviour. New York: Wiley International. Lipset, S. M. (1960). Political man. Garden City, NY: Doubleday. Maass, A., & Schaller, M. (1991). Intergroup biases and the cognitive dynamics of stereotype formation. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 2, pp. 189–209). Chichester: Wiley International. Macrae, C. N., Bodenhausen, G. V., & Milne, A. B. (1995). The dissection of selection in person perception: Inhibitory processes in social stereotyping. Journal of Personality and Social Psychology, 69, 397–407. Marcus-Newhall, A., Miller, N., Holtz, R., & Brewer, M. B. (1993). Cross-cutting category membership with role-assignment: A means of reducing intergroup bias. British Journal of Social Psychology, 32, 125–146. Maurer, K. L., Park, B., & Rothbart, M. (1995). Subtyping versus subgrouping processes in stereotype representation. Journal of Personality and Social Psychology, 69, 812–824. Migdal, M., Hewstone, M., & Mullen, B. (1998). The eﬀects of crossed categorization in intergroup evaluations: A meta-analysis. British Journal of Social Psychology, 37, 303–324. Mullen, B., Brown, R., & Smith, C. (1992). Ingroup bias as a function of salience, relevance and status: An integration. European Journal of Social Psychology, 22, 103–122. Mullen, B., Migdal, M. J., & Hewstone, M. (2001). Crossed categorization vs. simple categorization and intergroup evaluations: A meta-analysis. European Journal of Social Psychology, 31, 721–736. Murphy, G. L. (1988). Comprehending complex concepts. Cognitive Science, 12, 529–562. Oakes, P. J., Haslam, S. A., & Turner, J. C. (1994). Stereotyping and social reality. Oxford: Blackwell Publishing. Oakes, P. J., Turner, J. C., & Haslam, S. A. (1991) Perceiving people as group members: The role of ﬁt in the salience of social categorizations. British Journal of Social Psychology, 30, 125–144. Otten, S., & Wentura, D. (1999). About the impact of automaticity in the minimal group paradigm: Evidence from aﬀective priming tasks. European Journal of Social Psychology, 29, 1049–1071. Park, B., & Judd, C. M. (1990). Measures and models of perceived group variability. Journal of Personality and Social Psychology, 59, 173–91.

Crisp and Hewstone 21 Park, B., Judd, C. M., & Ryan, C. S. (1991). Social categorization and the representation of variability information. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 2, pp. 211–246). Chichester: Wiley International. Park, B., Ryan, C. S., & Judd, C. M. (1992). Role of meaningful subgroups in explaining diﬀerences in perceived variability for in-groups and out-groups. Journal of Personality and Social Psychology, 63, 553–567. Perdue, C. W., Dovidio, J. F., Gurtman, M. B., & Tyler, R. B. (1990). “Us” and “them”: Social categorization and the process of intergroup bias. Journal of Personality and Social Psychology, 59, 475–486. Phinney, J. S., Lochner, B. T., & Murphy, R. (1990). Ethnic identity development and psychological adjustment in adolescence. In A. R. Stiﬀman & L. E. Davis (Eds.), Ethnic issues in adolescent mental health (pp. 53–72). Thousand Oaks, CA: Sage Publications. Rae, D. W., & Taylor, M. (1970). The analysis of political cleavages. New Haven, CT: Yale University Press. Richards, Z., & Hewstone, M. (2001). Subtyping and subgrouping: Processes for the prevention and promotion of stereochange. Personality Psychology Review, 5, 52–73. Roccas, S., & Brewer, M. B. (2002). Social identity complexity. Personality and Social Psychology Review, 6, 88–106. Ross, M. H. (1985). Internal and external conﬂict and violence: Cross-cultural evidence and a new analysis. Journal of Conﬂict Resolution, 29, 547–579. Sherif, M. (1966). Group conﬂict and cooperation. London: Routledge and Kegan Paul. Simmel, G. (1950). The sociology of Georg Simmel (Kurt Wolﬀ, Trans.). New York, NY: Free Press (original work published 1908). Sinclair, L., & Kunda, Z. (1999). Reactions to a black professional: Motivated inhibition and activation of conﬂicting stereotypes. Journal of Personality and Social Psychology, 77, 885–904. Singh, R., Yeoh, B. S. E., Lim, D. I., & Lim, K. K. (1997). Cross categorization eﬀects in intergroup discrimination: Adding versus averaging. British Journal of Social Psychology, 36, 121–138. Smith, E. R., & DeCoster, J. (1998). Knowledge acquisition, accessibility, and use in person perception and stereotyping: Simulation with a recurrent connectionist network. Journal of Personality and Social Psychology, 74, 21–35. Staats, A. W., & Staats, C. K. (1958). Attitudes established by classical conditioning. Journal of Abnormal and Social Psychology, 57, 37–40. 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, 207–218. Straw, J. (2000). Blame the Left, not the British. The Observer, 15 October. Tajfel, H. (1959). Quantitative judgment in social perception. British Journal of Psychology, 50, 16–29. Tajfel, H., Flament, C., Billig, M., & Bundy, R. F. (1971). Social categorization and intergroup behaviour. European Journal of Social Psychology, 1, 149–178. Tajfel, H., & Wilkes, A. L. (1963). Classiﬁcation and quantitative judgment. British Journal of Psychology, 54, 101–113. 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, 778–793. Teasdale, J. D., & Russell, M. L. (1983). Diﬀerential aspects of induced mood on

22

Multiple social categorization

the recall of positive, negative, and neutral words. British Journal of Clinical Psychology, 22, 163–171. Turner, J. C. (1982). Towards a cognitive redeﬁnition of the social group. In H. Tajfel (Ed.), Social identity and intergroup relations. Cambridge: Cambridge University Press. Turner, J. C. (1985). Social categorization and the self-concept: A social cognitive theory of group behavior. In E. J. Lawler (Ed.), Advances in group processes (Vol. 2, pp. 77–122). Greenwich, CT: JAI Press, Inc. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. Oxford: Blackwell Publishing. Uleman, J. S., Newman, L. S., & Moskowitz, G. B. (1996). People as ﬂexible interpreters: Evidence and issues from spontaneous trait inference. Advances in Experimental Social Psychology, 28, 211–279. Urada, D., & Miller, N. (2000). The impact of positive mood and category importance on crossed categorization eﬀects. Journal of Personality and Social Psychology, 78, 417–433. Urban, L. M., & Miller, N. M. (1998). A theoretical analysis of crossed categorization eﬀects: A meta-analysis. Journal of Personality and Social Psychology, 74, 894–908. van Knippenberg, A., & Dijksterhuis, A. (2000). Social categorization and stereotyping: A functional perspective. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 11, pp. 105–144). Chichester: Wiley International. van Rijswijk, W., & Ellemers, N. (2002). Context eﬀects on the application of stereotype content to multiple categorizable targets. Personality and Social Psychology Bulletin, 28, 90–101. Van Twuyver, M., & van Knippenberg, A. (1999). Social categorization as a function of relative group size. British Journal of Social Psychology, 38, 135–156. Vanbeselaere, N. (1987). The eﬀect of dichotomous and crossed social categorizations upon intergroup discrimination. European Journal of Social Psychology, 17, 143–156. Vanbeselaere, N. (1991). The diﬀerent eﬀects of simple and crossed categorizations: A result of the category diﬀerentiation process or of diﬀerential category salience? In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 2, pp. 247–278). Chichester: Wiley International. Wallace, M. D. (1973). Alliance polarization, cross-cutting, and international war, 1815–1964. Journal of Conﬂict Resolution, 17, 575–604. Wattenmaker, W. D. (1995). Knowledge structures and linear separability: Integrating information in object and social categorization. Cognitive Psychology, 28, 274–328. Wilkenﬁeld, M. J., & Ward, T. B. (2001). Similarity and emergence in conceptual combination. Journal of Memory and Language, 45, 21–38. Worchel, S. (1979). Cooperation and the reduction of intergroup conﬂict: Some determining factors. In W. Austin & S. Worchel (Eds.), The social psychology of intergroup relations (pp. 262–273). Monterey, CA: Brooks/Cole. Yzerbyt, V. Y., Estrada, C., Corneille, O., Seron, E., & Demoulin, S. (2004). Subjective essentialism in action: Self-anchoring and social control as consequences of fundamental social divides. In V. Y. Yzerbyt, C. M., Judd, & O. Corneille (Eds.), The psychology of group perception: Perceived variability, entitativity, and essentialism (pp. 101–124). Hove, UK: Psychology Press.

Part II

Multiple category representation

2

Hierarchies and minority groups The roles of salience, overlap, and background knowledge in selecting meaningful social categorizations from multiple alternatives Craig McGarty

It is easy to start with the truism that all things can be categorized in multiple ways. Unlike some other truisms that can be made about the remarkable and ubiquitous process of categorization, such an opening statement risks serious understatement. Beyond the simple observation that people are capable of using many diﬀerent categorizations of the same thing, we can recognize that unless people were able to categorize any given object not just in many diﬀerent ways, but in an inﬁnite number of diﬀerent ways, this process would not be anywhere near as useful as it is. Indeed, we can interpret the opening statement as being anything from a bland truism to a grand (and potentially controversial) statement about categorization. The power of categorization emerges primarily because things can be categorized in many diﬀerent ways. That is, at two diﬀerent times an object can be assigned to diﬀerent categories (implying we need to understand category selection or application), and at any given time, that object may be simultaneously assigned to more than one category, all of which may aﬀect the way we perceive that object (implying we need to understand how categories are combined or category conjunction; see chapters by Crisp & Hewstone; Smith, this volume). Both of these points suggest that beyond understanding individual categories we need to understand systems of categories. This chapter is about ways of understanding the structure and nature of social categorical systems. The enormous ﬂexibility of the categorization process comes at an apparent cost. If we accept that categorization involves making sense of multiple things by ﬁnding the way in which they are alike (or can be treated as the same, or be seen to share an identity), then the multiplicity of categorization can seem like an embarrassment of riches. Surely the beneﬁt of categorization, the glittering prize it oﬀers to our knowledge of any given state of aﬀairs, is when it allows us to be speciﬁc? In other words, should we not be most convinced that categorizing is useful when we see the single best way to

26

Hierarchies and groups

categorize some set of things rather than the vast array of competing alternatives available to us? The short answer to this question is a resounding “No”. There are several reasons why; some of these relate to the categorization of objects of all types, but others have their most immediate impact for social psychology. To take the most general point ﬁrst: the multiplicity of categorization provides the basis for varying and selective perception. Without putting too ﬁne a point on it, we are able to see things in diﬀerent ways because we are able to categorize them in diﬀerent ways (see McGarty, 1999; McGarty, Spears, & Yzerbyt, 2002; Oakes, Haslam, & Turner, 1994; Oakes & Turner, 1990). If we were unable to change our categorizations, we would be trapped with unvarying perceptions, and as self-categorization theorists (and others) have long asserted, perception is variable and context-dependent even in the domain of stereotyping where those perceptions are conventionally misconstrued as rigid and invariant (see Haslam, Turner, Oakes, McGarty, & Hayes, 1992; Haslam, Oakes, Turner, & McGarty, 1995; for exemplar and distributed models of context-dependency, see Smith, this volume). A chilling example of the power of novel and changeable categorizations of people and other objects is provided by the ingenuity of the murderous actions taken by the attackers on 11 September 2001. If those attackers had attempted to launch guided missile attacks at the United States using conventional military means, there was very little prospect that they would have succeeded. As is now widely understood, the success of the attacks rested on the ability of the attackers to conceive a fuel-laden jet as a missile and themselves as guidance systems. Beyond this, however, the attackers had to perceive an item of stationery (a box cutter) as a weapon, and more importantly to convince passengers that they were hostages (whose lives were of some value to the attackers) and not murder victims. One response of security oﬃcials has been to classify a broader range of objects as weapons (including screwdrivers, but extending to nail clippers and Allen wrenches). The point of this example is simply to show that acts that had massive destructive consequences, and attempts to forestall repetitions of these acts, both rest on imaginative and novel categorizations. However, the beneﬁts of multiple categorizations of the same object are not limited to enabling new and diverging perceptions (see McGarty, 1999). Even though it might seem that multiple categorization creates nothing but the prospect of endless disagreement, it is also true that multiple categorizations allow us to resolve disagreements. That is, if we take the case of two people who recognize that they categorize the same thing in diﬀerent ways, then it is impossible for them to reach agreement unless they are able to adopt a new categorization of that object. Thus, multiple categorization enables variable and ﬂexible perception, variable and ﬂexible perception enables social inﬂuence, and social inﬂuence enables consensus. Paradoxically, then, variable perception enables consensus because we could have no true consensus about the way to categorize

McGarty

27

something unless we were truly able to change our perceptions of it. Otherwise we would be forever locked in irreconcilable disagreement (for a more detailed presentation of this idea see McGarty, 1999). Even though the foregoing ideas are regarded as extremely important, it is nevertheless fair to say that they do not map onto the central concerns that psychologists bring to the idea of multiple categorization. Essentially, it is suggested that when psychologists are confronted by the idea of multiple categorizations, the sort of question they are most keenly interested in is “which of these many diﬀerent categorizations is the single categorization that is the most appropriate to apply?” This is supplemented by a more subtle question that is perhaps of especial interest with social categories and is part of the inspiration for many of the chapters of this book, “how are multiple simultaneous category memberships blended together to aﬀect the perception of people and things?” Famous examples of attempts to answer the ﬁrst question are easy to ﬁnd in cognitive and social psychology. Bruner’s (1957) paper “On perceptual readiness” was dedicated to solving this question. He famously argued that the category that was applied in any particular situation was that which maximized the interaction of perceptual readiness (the degree to which that category was accessible to the perceiver on the basis of past experience and current goals) and ﬁt to the speciﬁcations of the category. These ideas were adapted to social psychology by self-categorization theorists (e.g., Oakes, 1987) and the idea of accessibility has underscored the work of Higgins and many others (e.g., Higgins, 1996; Higgins & King, 1981). The second idea, the inﬂuence of blended multiple categorizations, has been of especial interest to social psychologists because of the existence of cross-cutting social categorizations (Doise, Deschamps, & Meyer, 1978). This work has long held open the possibility that such multiple social categorizations may reduce intergroup hostility (see Crisp, Hewstone, & Rubin, 2001, and many of the chapters in this volume). It is probably impossible to understand modern attempts to explain category application and category conjunction without considering another idea that has a long and respected tradition in science. This is the idea that categories can be hierarchically ordered. This is because one circumstance in which the same exemplar can be thought of as belonging to more than one category is when we conceive those categories as being contained within more abstract, general (or simply larger) categories. To take a famous example, George W. Bush is both a Texan and a citizen of the United States, and both of these categories apply to him because one of these categories is (at least at ﬁrst glance) wholly contained within (i.e., it is a subset of) the other. This containment relation is, of course, also a special (extreme or maximal) case of overlap or conjunction between categories. It is almost impossible to understate the centrality of the idea of hierarchies for speciﬁc sciences such as biological taxonomy (which is, after all, about little else) but also for logic and science in general.

28

Hierarchies and groups

Given this central importance, the idea of hierarchically organized categories is considered in detail in the next section. Next, the conceptual and logical problems that are entailed by the idea that all categories are hierarchically organized are discussed, and then some alternatives to the idea that all categories are hierarchically organized are considered, with a particular focus on the relevance of the idea of categories formed around shared opinions for social psychology.

HIERARCHICALLY ORGANIZED CATEGORIES IN COGNITIVE PSYCHOLOGY The idea of a hierarchically organized system of knowledge is enormously attractive simply because such networks are so useful. Their usefulness applies especially to the process of making logical inferences. If we know that all birds are susceptible to a disease and that a robin is a type of bird, then we know with absolute mathematical certainty that robins are susceptible to the disease, but we can be similarly certain about other types of birds that we have never heard of. All we need to know is that they are birds. It is not surprising therefore that Rosch et al. focused to such a degree on the hierarchical nature of category structure (Rosch, 1978; Rosch, Mervis, Gray, Johnson, & Boyes-Braem, 1976). The form of hierarchical organization that Rosch et al. focused on is well known to most psychologists. Rosch (1978) suggested that categorical structures can be distinguished for convenience into three levels of abstraction (of increasing inclusiveness): the subordinate, the intermediate, and the superordinate. For any given categorical system it was further argued that one of the levels tended to be the basic level at which perception tended to be located and where objects would be spontaneously named. Thus for pieces of furniture the basic level might be at the level of table and chair rather than less abstract (dining table) or more abstract (furniture or artifact). In addition to this hierarchical (or vertical) structure of categorical systems, Rosch et al. made important contributions to the idea of a horizontal structure of categories. Their approach speciﬁed that there was a variation within categories in terms of the degree to which the members were representative or prototypical of the category. This idea was important due to the lack of evidence for what had come to be known as the classical view of categories: the view that categories had all-or-none deﬁning features and hence had ﬁxed boundaries that separated them. Thus it was shown by Rips (1975) that robins were regarded as more typical of the category bird than were ducks and so a disease that aﬀected robins was judged to be more likely to be transmitted to ducks than a disease that aﬀected ducks was likely to be transmitted to robins. There is a curious tension between the two central ideas embodied in this prototype view of categorization. On the one hand, the hierarchical structure of categories means that common higher-order category membership enables

McGarty

29

powerful logical inferences to be made about all members of lower-order categories. On the other hand, members of categories are supposed to vary in their prototypicality and logical inferences are actually made more easily about these prototypical members. As seen in a later section, the evidence for this is at best patchy and this suggests that, psychologically speaking, hierarchies can be rather fuzzy. Nevertheless, hierarchical organization remains an important idea in cognitive work on categorization. As Barsalou (1993, p. 176) suggests, it is not surprising that cognitive scientists “typically model the human categorizer as an intuitive taxonomist”. However, although Barsalou suggests that taxonomies are important for developing models of the world, he also argues that many categories exist for current goal-directed cognitive functions that have little to do with taxonomies. Having given the barest taste of this idea in cognitive psychology, it is now timely to return to consider social-psychological applications of related ideas. In order to do this we ﬁrst focus on the self-categorization theory of Turner, Hogg, Oakes, Reicher, and Wetherell (1987). In focusing on this theory, it should not be assumed that the problems with using hierarchical structures are distinctive to it. The same problems apply to all hierarchical formulations that make similar assumptions. HIERARCHICALLY ORGANIZED CATEGORIES IN SOCIAL PSYCHOLOGY

The case of self-categorization theory Self-categorization theory borrows extensively but not uncritically from the work of Rosch (1978) in this domain. In particular the idea of multiple levels of abstraction and variations in prototypicality (vertical and horizontal structure) are endorsed but the idea of a basic level of categorization is rejected. As already mentioned, self-categorization theorists elevate the idea of ﬂexible and context-dependent perception to a ﬁrst principle of cognition. In terms of the hierarchical organization idea, self-categorization theory is strictly Roschian orthodox. Following Rosch (1978), Turner et al. (1987) assert that some set of objects can only be compared at some level of abstraction by seeing them as identical at the immediate higher level of abstraction that includes them all. That is, we can only explore the ways in which two things are similar to or diﬀerent from each other when we acknowledge their sameness at some higher level of abstraction. Thus, apples and oranges cannot be compared as apples and oranges (because apples are not oranges) but apples and oranges can be compared with each other as types of fruit. Chalk and cheese would need to be compared in terms of some higher-order common category such as they are both substances (e.g., cheese is a soft substance that is good to eat and chalk is a hard substance that is not good to eat).

30

Hierarchies and groups

When applied to social perception, a comparison between two individuals is an intragroup comparison: in order for them to be compared, they need to be seen to be members of the same group. A particularly interesting extension of this idea is that the lower-level members are themselves categories. In other words, people are essentially low-level categories. To date, however, the implications of this idea have not been fully elaborated in the literature but Mavor et al. have reported progress in this area (Mavor, 2001; Mavor & Reynolds, 2002). The clearest illustration of these points can be seen in the presentation of self-categorization theory concepts in the form of tree diagrams. An example of this is shown in Figure 2.1 (based on McGarty, 1999, which is itself adapted from Oakes et al., 1994). Here there are three levels of abstraction: the level of individual people, a national group, and a language group. Thus, Diana and Fred can be compared as members of the category Australians but Scott and Tina can only be compared as members of a higher-order group (English speakers).

THE BENEFITS OF HIERARCHIES

Identity and cooperation This exploration of the idea of hierarchical organization of categories in cognitive and social psychology is very useful because it highlights the two principal beneﬁts that are derived from using hierarchies. Hierarchies provide solutions to two long-standing and vexing problems. At the cognitive level hierarchical organization provides an answer to the problem of identity. That is, we can say that two diﬀerent things are the same in some way (i.e., they are identical) if we can establish that they share a superordinate category membership.

Figure 2.1 A tree diagram showing three levels of a hierarchy.

McGarty

31

We can see the struggle with this idea in the cognitive domain. Medin, Goldstone, and Gentner (1993) focus at length on similarity and these and other authors investigate a variety of ways in which theory might serve to constrain similarity. That is, it was diﬃcult to see how two things could be said to be similar without invoking some knowledge about the relations between them. Hierarchical organization was a one-stop shop for providing the sort of knowledge that could be used to bind diﬀerent entities together conceptually. It is easy to see the attraction of this idea for a social identity theorist such as Turner, who was developing his ideas independently of this emerging trend in the cognitive psychology of categorization (the ﬁrst complete statement of self-categorization theory by Turner appeared in the same year, 1985, as Murphy and Medin’s paper on theory-based coherence). The socialpsychological processes speciﬁed in Tajfel and Turner’s (1979) social identity theory rested upon the mutual recognition of shared identity but the theory was unclear about how this recognition might be achieved. In adopting the idea of hierarchical organization of self-categories in the subsequent formulation of self-categorization theory, Turner et al. explained how the enormous inferential power that hierarchical structures enable can be harnessed to self-perception. Cognitively speaking, the existence of a hierarchical structure of self-perception enables people to assert in a logically plausible way that two diﬀerent people share an identity (i.e., they are identical or are the same in some relevant way). This is, of course, a logical platform for tackling what Asch (1952) called the master problem of social psychology: how are we to conceptualize the relation between individuals and society? The answer that rapidly emerges from self-categorization theory is that those psychological aspects of society rested on this recognition of sameness or commonality at particular levels that applied at particular times. We can thus examine the social-psychological basis of moral assertions or slogans of the form “All men and women are brothers and sisters”. According to self-categorization theory, such a statement will be seen to be more valid from the perspective of the perceiver when he or she is thinking of people as human beings rather than as individuals or group members (because at the human level of categorization both men and women will tend to see themselves as more interchangeable with men and women). The hierarchy idea receives considerable attention in a variety of other related formulations such as Brewer’s (1991, 1993) optimal distinctiveness theory. Perhaps the most obvious example though appears in Gaertner, Dovidio, Anastasio, Bachman, and Rust’s (1993; Gaertner & Dovidio, 2000) common ingroup identity model. This model rests upon the idea that causing two separate groups to recategorize so that they perceive themselves both to be contained within the superordinate group will reduce the bias against the other group. The common ingroup identity model assumes that the development of a common identity (the perception that both groups really are contained within the larger superordinate) motivates group members to

32

Hierarchies and groups

perceive the members of the other group in a more positive light. The common identity thus induces cooperation (and can be induced by cooperation). Evidence supporting the model in a range of domains is presented in the review by Anastasio, Rust, Gaertner, and Dovidio (1997; see also Dovidio et al., this volume). A diﬀerent view has been put by Hewstone and Brown (1986; see Hewstone, 1996), who proposed a mutual intergroup diﬀerentiation model. This model proposes that cooperative intergroup relations are maintained by preserving group boundaries and identities. This is because attempts to dispense with subgroup identities and move to a superordinate creates a threat to those identities. More recently, Mummendey and Wenzel (1999) adopted what could be considered to be a more elaborated hierarchical model of cooperation and conﬂict between groups. These authors distinguish between two diﬀerent responses that can be shown toward outgroups: discrimination and tolerance. They argue that the response adopted will depend on the way in which the two groups are conceived in relation to some superordinate. If the superordinate is perceived to have characteristics that are prototypical of the ingroup, then the outgroup will be judged to be inferior, deﬁcient, and non-normative and will be discriminated against. If, on the other hand, the superordinate is conceived to include attributes of the outgroup, then it is likely to be tolerated. Up to this point the analysis is strictly hierarchical. It is nevertheless more elaborated than the common ingroup identity model due to its suggestion that the mere existence of a superordinate identity is not suﬃcient for a positive view of the other group. If the outgroup is perceived to belong to a superordinate but is deﬁcient in terms of that superordinate, then that outgroup can expect discrimination precisely because the outgroup members fail to meet the standards expected at the superordinate level. Importantly, though, Mummendey and Wenzel’s analysis is extended further to suggest that tolerance can also arise from a nonhierarchical conceptualization. That is, where two groups are not perceived to share a superordinate group, tolerance may also emerge because it is more diﬃcult to deﬁne standards that show the outgroup to be deﬁcient. As McGarty (1999) contends, this general argument has considerable plausibility at some level. Serbs and Croats in the 1990s may have felt more hostility to each other as citizens of the former Yugoslavia, but be relatively tolerant of Inuits and Zulus. However, discrimination can also occur in particularly virulent forms between groups who contest the existence of any shared superordinate identity. Nevertheless, Mummendey and Wenzel have advanced the analysis of this issue by pointing to the importance of meaning of the groups in determining relations between groups. Hornsey and Hogg (2000; see also Hogg & Hornsey, this volume) suggest a diﬀerent form of hierarchical model for addressing relations between subgroups of a superordinate. They suggest that the most important variable standing in the way of cooperation (social harmony) is identity threat.

McGarty

33

Factors that threaten the identity of the subgroups are most likely to promote conﬂict between groups, and harmony can be achieved by maintaining subgroup identities (an idea that is parallel to Hewstone and Brown’s, 1986, mutual intergroup diﬀerentiation model, which is now incorporated into Gaertner and Dovidio’s, 2000, common ingroup identity/dual identity model; and see Crisp, this volume). These authors’ integrative model of subgroup relations stresses the beneﬁts of dual identiﬁcation for positive intergroup relations. If people simultaneously identify with a superordinate group and their subgroup, there are two beneﬁts. People who would otherwise have been seen as outgroup members come to be seen as ingroup members (by virtue of a shared superordinate categorization). Additionally, though, distinctiveness threat (as discussed in the mutual intergroup diﬀerentiation model) is also minimized by maintaining the salience of the subgroups. Eggins et al. (see Eggins, Haslam & Reynolds, 2002; Eggins, Reynolds, & Haslam, 2003; Haslam, 2001) have made a very similar argument based more closely on self-categorization theory. These authors (as do Hornsey & Hogg, 2000) argue that these ideas are consistent with genuine multiculturalism (see Berry, 1984), where both the subgroup and superordinate group are valued. The particular emphasis in the work of Eggins et al. is on the role of maintaining or restoring positive and distinct social identities for subgroups and the explicit recognition by all parties of the distinct role of the subgroups. In other words, dual identiﬁcation is about the crafting of an understanding (or developing a meaning) of the superordinate that recognizes and values the subgroups. We can see from this short summary of this range of social-psychological approaches that, as in cognitive psychology, part of the appeal of the idea of hierarchies for social psychology is for providing solutions to the scientiﬁc problem of establishing identity (deciding when two things are the same). The additional dividend that ﬂows from this ﬁrst idea, though, is the solution to the social problem of cooperation. If people perceive themselves to be the same as others in some way, then the possibility exists that there is an increased likelihood that they will act in a cooperative manner (or at least not in a conﬂictual manner). The self-categorization theory analysis of social behavior rests on this very idea. Seeing oneself as sharing a group membership with others (deﬁned as perceiving themselves to share an identity with those people) at some instant is a precursor to seeing oneself as interchangeable with those other people, and this is the precondition for cooperation and other forms of group behavior. Exactly the same idea is contained within the common ingroup identity model and Mummendey and Wenzel’s formulation. Higher-order sameness is a precondition for cooperation or speciﬁc psychological states that predispose people toward positive social behavior (though, as noted above, Mummendey and Wenzel add the nuance that a shared superordinate categorization can also be a basis for hostility).

34

Hierarchies and groups

It is also readily apparent that the social-psychological literature has started to move beyond a simple acceptance of the idea that categories are hierarchically organized to recognize that the idea that a hierarchical relationship can exist can be contested by members of social categories.

THE PROBLEMS WITH HIERARCHIES We have gone to some lengths to explain the burden that hierarchical organization bears in the conceptual world that social psychology inhabits. The fact that hierarchies do not work the way they should is therefore very problematic. Some relevant evidence here has been provided by Sloman et al. and is discussed in detail by Sloman (1998) and by McGarty (1999). The ﬁrst problem is that hierarchically organized categories are not particularly common in the world. The world is full of exceptions where there is ambiguity about whether all members of some class of objects are included within a higher level of a category. There are very few examples of naturally occurring categories that have the relatively neat hierarchical, inclusive structure of the famous biological taxonomy. However, even where the hierarchical structure does apply, there is strong evidence (Sloman, 1998) that people do not use hierarchical structures even where they appear to have a correct understanding of the relevant categorical relations. For example, adult reasoners might accept that “A car headlight is a kind of lamp” and that “A lamp is a kind of furniture” but (not unreasonably) reject the inference that “A car headlight is a kind of furniture.” Sloman has identiﬁed that reasoners prefer inferences that involve subordinate categories that are similar to, or typical of, the superordinate. For example, they prefer to make inferences about birds based on knowledge about robins rather than about ducks. They also prefer inferences where the premises are of similar speciﬁcity. Thus, people prefer to apply an inference to robins on the basis of information about all birds than about all animals. Logically speaking, neither of these should make any diﬀerence. Given the paucity of strictly hierarchical structures in the world, it is not surprising that human information processing should be sensitive to the material conditions that it operates within. People may be averse to using the logic of hierarchies simply because they do not encounter many systems that can be understood as hierarchies. It was noted above that the inference that a car headlight is not an item of furniture is not unreasonable. Of course, it is problematic only so long as the other premises are held to be true; in this case the second premise could be argued to be false (not all lamps are types of furniture, but rather only those typical lamps that are evoked in the context of a sentence mentioning furniture are really items of furniture). The fuzzy nonhierarchical nature of the world (in this technical sense of hierarchy, of course, there are all sorts of hierarchical power structures

McGarty

35

that exist in social institutions, of which most military organizations provide excellent examples) could hardly be more obvious in relation to human social category memberships, and the danger comes in insisting that people’s knowledge of the world is structured hierarchically. For example, the categories in Figure 2.1 are actually contrived to look as if they are hierarchical. Of course, the categories “American” and “English speaking” do not stand in any strict hierarchical relation to each other. Most Americans speak English but millions (e.g., recent immigrants, people with severe brain injuries, and newborn children) do not. We could try to adjust the categories so that there was a true hierarchical relationship. We might refer to “American” and “citizens of countries where the sole oﬃcial language is English” but in doing so we would be going to a lot of trouble (and excluding bilingual nations such as New Zealand and Canada from that categorical system). If we assume, as self-categorization theory does (see assumption A.5 of Turner et al., 1987, p. 45), that each lower-level category is entirely contained only within one category at the next higher level, then we can see where the problems begin. People can be citizens of more than one country, and the structure of our self-perception could hardly fail to acknowledge and capitalize on the existence of such overlapping category memberships. Furthermore, because of the existence of overlap, many social categories are ambiguous in terms of their level of inclusiveness. In some contexts the attribute “American” is less inclusive than the attribute “English speaking” and in some contexts it is more inclusive. Rarely will there be enduring, ﬁxed levels of inclusiveness that can be used to deﬁne the position of a category in relation to some hierarchy and the available evidence suggests that even where there do appear to be hierarchies they often fall apart on close examination. For example, Texans might be thought to be wholly contained within the category “Americans” but if Texan is deﬁned as “a resident of Texas” and American as “a citizen of the United States of America”, then this is clearly not true (due to the existence of hundreds of thousands of illegal immigrants and other noncitizens). Of course these issues and arguments recall the ideas of fuzzy sets (Smithson, 1987; Zadeh, 1965). Many categories, including conceptual categories, do not have clear all- or nondeﬁning features or boundaries. This point was keenly recognized by Rosch (1978) and was in fact the inspiration for the prototype approach to categorization. However, insisting on ﬁxed hierarchies simply relocates the problem of ﬁxed boundaries to a higher level. Barsalou (1987) had noted that some categorical systems appeared not to have any hierarchical structure and simply represented lists of things. He suggested that these were likely to be goal-directed categories, categories that were developed for a particular purpose (the famous example being children, pets, insurance policies, and photo albums that comprise the category of things to remove from a burning house). Medin, Lynch, Coley, and Atran (1997) made similar points about the categorical reasoning of tree experts.

36

Hierarchies and groups

The categories that park maintenance workers used, for example, were groupings based on particular problems that these workers needed to solve in their work (e.g., whether they were deciduous or dropped dangerous branches). On the other hand, biologists applied the biological taxonomy, a type of categorization that Medin et al. suggested was not goal-directed. We (McGarty, 1999) have argued that these authors were mistaken in this regard. The biological taxonomy is also a goal-directed categorization but here the goal is scientiﬁc description. It works very well for this setting because organisms are well described by such a hierarchical system. An extension of the argument in the previous paragraph is to suggest that systems of relations between categories are not based on ﬁxed diﬀerences in inclusiveness, but rather that perceptions of inclusiveness and overlap emerge from the process of categorization. Where we come to understand that one category is wholly contained within another, we may develop a hierarchical understanding of the categorical relations (as biologists do with organisms) but where we come to perceive categories as overlapping we will not develop such an understanding because it does not reﬂect reality from our current vantage point. Of course these ideas are utterly consistent with the spirit and intent of self-categorization theory in relation to a variety of other points (see Oakes et al., 1994; Turner, Oakes, Haslam, & McGarty, 1994). Self-categorization theorists argue that other features of categorization are highly contextdependent. The point here is that the formal statement of the theory (which after all is merely representative of a much wider range of applications of these old hierarchical ideas about categorizations) explicitly argues that hierarchical organizations are omnipresent. An alternative view is that hierarchies are possible outcomes of categorical systems but they represent but one possible form of outcome, and there is no sense in which they can be seen to be utterly and uniformly present in all forms of categorical knowledge. Up to this point my critique of the hierarchy metaphor has taken the form of a logical argument by analogy. We have argued that the hierarchy metaphor is not sustainable in logical terms because psychological and object categories tend not to be organized in this hierarchical manner and also that social categories strongly tend not to be organized in this way. The further logical extension is that if categories in the world are not organized in this way, then it is impossible for self-categories to veridically reﬂect the world. One possible conclusion is that self-categorization theory is wrong is assuming that self-categories do veridically reﬂect the world in which the perceiver exists. My own view is that self-categorization is broadly correct in its emphasis on veridical perception, but that it is broadly wrong in following Rosch in assuming that categories are hierarchically organized. It is worth asking, however, what the practical implications of this argument are. Assuming, for the sake of argument, that self-categories are not hierarchically organized in the way that self-categorization theory (and other related accounts) assume, what are the consequences of making this

McGarty

37

mistake? Below are details of some empirical evidence of the problems with logical inference that occur with the hierarchy assumption. The bigger problem though is a strategic issue for the ﬁeld. The hierarchy metaphor is held out as the basis of a solution to the problem of engaging social cooperation and promoting positive intergroup relations. If we want to get people to cooperate with or feel less negatively about some other group, then we need to have those people perceive themselves as part of the same group or at least not part of clearly diﬀerent groups (through establishing cross-cutting category memberships, but see Crisp et al., 2001, for contrary evidence). We believe that this assumption is entirely correct. The hierarchy metaphor, however, also leads us to assume that this common ingroup identity should be a superordinate identity that includes all members of the subordinate group. We believe that this assumption is deeply problematic and we address these problems in the next section. An unpublished thesis by Ross (2000) investigated some of the problems raised by Sloman and others about hierarchical inference in a socialpsychological context. A key question here was the extent to which hierarchical inference would be equally strong with self-categories, other (non-self) social categories, and categories not containing people. Much of the research on categorical inference requires inferences to be made with blank predicates, that is, categories about which the participants have no knowledge (and no expectancies). That is, participants are given logical premises of the form “All birds have wings” and “An exron is a bird” and then are asked to assess how likely it is that an exron would have wings. The purpose of the use of blank predicates is to control for the eﬀect of expectancies and prior knowledge. In the case of self-categories, however, blank predicates make less sense. Self-categories are those categories that are currently perceived to apply to the self. It is diﬃcult (but not impossible, as the minimal group paradigm of Tajfel, Flament, Billig, & Bundy, 1971, reveals) to have self-categories that are unknown to participants, and it is especially diﬃcult to construct multiple unknown categories. For this reason Ross (2000) used existing categories with known subcategories. These were selected to be categories of people and categories of other things. The categories of people were selected so as to include a plausible range of self and non-self categories (to make sure of this, Ross asked participants to rate the degree to which each category applied to that person). Examples of self-categories were men and women, university participants, and Australians; examples of non-self categories included Spanish people, accountants, and soldiers. The logical premises used were based on the following structure: A. Members of category X have property Y. B. Members of category Z have/do not have property Y. C. Category Z is a subcategory of category X.

38

Hierarchies and groups

In this format statements must all be ones that the participant accepts as being true (or as largely true). After being exposed to statements B and C, the participants were asked how true they thought statement A was now. The information about the subcategories was selected so as to challenge or support the hierarchical relationship. Thus, information that Americans are patriotic, that soldiers are patriotic, and that some Americans are soldiers supports a hierarchy; information that Italians are emotional, that accountants are not, but that some Italians are accountants challenges it. Information that challenges or supports a hierarchy should have a larger eﬀect where hierarchical reasoning is being used. The ﬁrst question therefore was whether there was a bigger eﬀect in some domains than others. For example, although self-categorization theory assumes that all categorization is hierarchical, it makes this argument most explicitly about self-categories. Self-categories then should certainly show no less evidence of hierarchical reasoning than other types of categories. There are, however, some other methodological choices necessary for the eﬀects to be observed. First, any subcategory used could be the result of an overlap between the two categories (American soldiers) or it could, in practical terms, be wholly contained within the superordinate (Maori New Zealanders). Arguably, information that challenges the hierarchy should be particularly powerful in disconﬁrming the hierarchy when it relates to a subcategory that is wholly contained within the category. This is the case because wholly contained subcategories provide unambiguous evidence of exceptions to a hierarchical rule. Ross included equal numbers of conjunctive and wholly contained categories (though, in fact, this was found to make little diﬀerence to the results). Second, for superordinate self-categories, the subgroups can themselves be self-categories or non-self categories. The importance of this distinction is that it allows us to determine whether categorical inferences are aﬀected by self-categorical boundaries. Ross used two dependent measures: the rating of the truthfulness of the original statement about the superordinate and the degree to which it was believed that there were important exceptions to the statement. On the truthfulness measure Ross found a signiﬁcant information type eﬀect for all four category types (self-self, self-non-self, non-self-non-self, and non-human). This indicated that, as expected, challenging information about subcategories was associated with weaker endorsement of the hierarchy than supporting information. More interestingly, the size of this eﬀect varied enormously, so that there were very large eﬀects for self-self categories (η2 = 0.65), dropping to smaller eﬀects for non-human (η2 = 0.38) and non-self-non-self (η2 = 0.37) categories, and a still smaller eﬀect for self-non-self categories (η2 = 0.15). A similar but stronger pattern was found for the importance of exceptions ratings with the strongest diﬀerence being shown for self-self (η2 = 0.69), then non-human (η2 = 0.40), then non-self-non-self (η2 = 0.35) with a nonsigniﬁcant eﬀect of small size being shown for self-non-self categories. These results can be interpreted as showing that hierarchy-consistent

McGarty

39

reasoning is easier to accept for categorical systems that are applicable to the self at the superordinate and subordinate levels but hierarchy-inconsistent reasoning is also easier to reject under the same circumstances. This diﬀerentiation is much less strong for self-non-self categories. When it is pointed out that some subcategory has a quality that is inconsistent with the hierarchy, this matters a lot when the perceiver is a member of the subcategory and very little when the subcategory represents an outgroup (see also Brewer, Ho, Lee, & Miller, 1987; Hewstone, Islam, & Judd, 1993, for a related point). More broadly, however, the results of Ross’s study suggest that, as was found by Sloman (1998), there is variation in the applicability of hierarchically based inferences. In particular, the group membership of the subcategory member had powerful moderating eﬀects on the extent to which logical inferences using a hierarchical structure could be made. In other words, the use and applicability of hierarchical structures in reasoning is itself a variable that depends on context. Thus, although it is certainly impossible to reconcile all self-categorization theory assumptions with these data, they do strongly support its key contentions. Perceivers use hierarchical reasoning in relation to self-perception but (predictably) their reasoning is not utterly constrained by the limitations of the tree diagram metaphor. More interestingly, though, the use of hierarchical reasoning is powerfully constrained by self-categorical boundaries. Outgroup members whose properties violate or support a hierarchical relationship are paid less heed than ingroup members, but only especially so where the superordinate relates to self. Thus, the acceptance of the proposition that people in developed countries (self-relevant superordinate) have access to good health services was unaﬀected by the statement that many indigenous Australians (non-self subgroup for these participants) do not have such access, but an equivalent proposition that New Zealand culture (non-self) developed from a British heritage was undermined by pointing to the existence of Maori culture (non-self). What role should we ascribe to hierarchies as conceptual structures then? In general it seems wise to move away from the limitations of the tree diagram metaphor. Overlap between categories is too pervasive (and too responsive to the process of categorization) to be banished from a consideration of the knowledge that constrains the way categories are formed and used. More generally, however, it may be sensible to move away from considering hierarchies as the underpinning of categorical systems to consider them instead as a particular type of outcome of the categorization process (and one that is particularly useful for some sorts of reasoning and descriptive tasks, as the case of the biological taxonomy shows). Given the vast complex network of possible categorizations, it might seem that this view of categorical knowledge casts us adrift from the simplicity of the hierarchical organization. That is, strict hierarchies may not capture reality all that well but at least they are straightforward. On the other hand, though, merely pointing to elements of a superordinate category

40

Hierarchies and groups

membership may not be a strong basis for convincing people to have more favorable attitudes to some group, especially given the full range of relevant categorizations that may be known in relation to the target group. If we were to convince employed people that unemployed people share a group membership in terms of presence in the job market, nationality, or even the human race, are we likely to promote lasting positive attitudes toward that other subgroup? Given the enormous diversity of the process of categorization, it can appear diﬃcult to get a handle on the ways in which many diﬀerent categorizations will be resolved. One solution rests on the idea that categories are more likely to be shared with other people where they deﬁne themselves as members of common groups. In this case the groups we focus on, however, rather than being superordinates deﬁned in terms of broad category memberships, are opinion-based groups. CRAFTING SPECIFIC IDENTITIES

The idea of opinion-based groups In all the approaches to dual identity that we considered earlier, the focus was on ﬁnding some common identity that provides a positive basis for cooperative relations. As Kessler and Mummendey (2001) demonstrate, however, common ingroup identiﬁcation can have its own problematic consequences in that it can serve to unite the group in opposition to some outgroup (so that the antipathy of East and West Germans toward foreigners increased when they identiﬁed in terms of the common ingroup of Germans). What happens when there is no convenient, ready-made superordinate identity to hand or where we may have doubts about the eﬃcacy of the superordinate as a valuable identity? Thus, indigenous people such as New Zealand Maori, Australian Aborigines or American Indians may exist in a society with nonindigenous people but the superordinate identity may be deeply problematic as a basis for that identity due to the legacy of colonization. If there is no ready-made identity that provides a basis for positive relations, then one option is to recraft or rebuild that identity. We mentioned above how Eggins et al. (2002) had suggested that new superordinate identities (which built on, recognized, and valued subgroup identities) could be crafted within a hierarchical framework. The same goal can be pursued without making hierarchical assumptions. According to Bliuc, McGarty, Musgrove, Reynolds, and Hendres (2004; see also McGarty & Bliuc, 2004), an opinion-based group is a group in the sense used by Turner (1982) but where the social identity is deﬁned by a shared opinion. Such groups are widely studied in social psychology (e.g., in studies of social inﬂuence) but their properties are not always recognized. Perhaps the most important property of opinion groups is that they tend to

McGarty

41

be less ambiguous than other groups are in relation to behavior. This is because the existence of a shared opinion implies the existence of related behavioral norms about which the opinion-based group members might reasonably be expected to reach consensus. By deﬁning oneself as a member of an opinion-based group, a person is committing to a view about an issue, and certain other things ﬂow from that. We can expect, for example, that they will be more likely (when compared to people who belong to other opinion-based groups) to support action consistent with that view and to share related views. To take the issue of abortion as an example, we cannot be certain that pro-life supporters will attend a pro-life rally, but we would be very surprised if they marched in a pro-choice rally. When we compare these opinion-based groups to groups that are deﬁned on the basis of mere membership in a social category, we ﬁnd that the latter are generally less well speciﬁed in relation to associated opinions and behaviors. An enormous range of positions and behaviors can be reconciled with being a woman, but a much more restricted range can be reconciled with an opinion-based group such as support for gender equality. Indeed, female proponents and opponents of gender equality might deﬁne their views as being utterly consistent with what it means to be a woman (or even to express strong identiﬁcation with the group). Schisms of the form studied by Sani and Reicher (1998) seem most likely to occur for opinion-based groups where two diﬀerent positions are contested as representing the core or true nature of the group. McGarty and Bliuc (2004) made this argument in relation to the experience of collective guilt. In a series of studies we had found almost no connection between the level of collective guilt experienced by nonindigenous Australians over the past treatment of indigenous Australians and the strength of respondents’ identiﬁcation with Australia. This was surprising because we had expected that people who strongly identiﬁed with Australia would be less willing to experience a negative emotion in relation to their group’s actions. Rather it seems that identiﬁcation with this social category was complex (in the same way that identiﬁcation with a category such as women is complex), as was the relationship of identiﬁcation to feelings of past behavior. Indeed, high social identiﬁcation can itself be understood as a position about which opinion-based groups can form. Bliuc et al. (2004) showed clear evidence of extremely strong relationships between political opinion-based group memberships and intention to take part in collective action. In two studies it was shown that the degree to which people identiﬁed with political opinion-based groups was a very strong predictor of intention to take part in collective action to advance that opinion (e.g., attend rallies and meetings) but this relationship was fully mediated by the salience of that group membership (measured here as the degree to which the perceiver was certain that the group membership was self-deﬁning). The importance of this research is that it shows that the salience of opinion-based groups is a strong predictor of collective behavioral intentions.

42

Hierarchies and groups

It holds out the prospect that by manipulating the salience of these groups we can produce changes in behavior. As these opinion-based group appear relevant to collective behavior, they would seem to be important ones to focus on. If we are interested in collective action by women, then it makes more sense to look at the degree to which those women identify with (for example) the feminist movement than with women. This is a point that is very much in keeping with what we know from the work of Simon and Klandermans (2000) and Kelly and Breinlinger (1995). The empirical work we have done on this topic is especially encouraging. We have shown that self-perception as an opinion-based group member is an excellent predictor (R2 > 0.5) of intentions to take collective action. The opinion-based group idea represents a clear alternative to the strictly hierarchical approaches to facilitating cooperation between subgroups that we discussed earlier. Rather than just pointing to or emphasizing commonality at some superordinate level (by stressing that we are all part of the same organization or we are all in the same team), it should often be possible to emphasize membership of a relevant opinion-based group that may cut across organizational boundaries. This might mean, for example, that rather than seeking to promote cooperation between East and West Germans because they are both part of a superordinate Germany, we might seek to stress common membership of opinion-based groups that are relevant for cooperation (e.g., being in favor of promoting opportunities for all Germans). It is easy to argue that, implicitly at least, this is already being done in work of this form (see Mummendey & Wenzel, 1999). When researchers suggest that groups have goals or follow strategies, this is generally a shorthand for the formation of opinion-based groups within some broad social category. Although it is convenient to say that blacks in America in the 1960s had a goal of ending segregation, in reality the process involved was one where an antisegregation opinion group formed and rapidly gained support within the black (and other) communities. At the same time there were numerous other black people who never came to deﬁne themselves in this way, and (presumably) others who favored segregation. Given that the former group was not politically active and the second group was politically unsuccessful, they are often forgotten, making it easy to equate the broad social category with the social movement. This is almost inevitably a mistake that researchers such as Klandermans (2000) would caution us against. REFINING CATEGORY MEMBERSHIP

The group-type distinction There is one ﬁnal idea we will introduce in this chapter. It relates to a distinction introduced by Khalaf and McGarty (2002; see also McGarty, 2002) between various forms of social categories as being either groups or types. We

McGarty

43

will explain that the distinction helps us to get a handle on a diﬀerent aspect of social categorical multiplicity. We are referring here to the multiple interpretations that are possible of a labeled category. Haslam, Rothschild, and Ernst (2000) had developed the distinction between the degree to which social categories had the quality of being natural kinds or being entitative (or coherent). Lickel, Hamilton, Wieczorkowska, Lewis, Sherman, and Uhles (2000), however, have also noted a distinction between social categories and dynamic groups. For Khalaf and McGarty (2002) the diﬀerences between these distinctions suggested that the twodimensional structure of Haslam et al. was underspeciﬁed in that it did not take account of the collective qualities (such as goals, purposes, structure, sense of identity, interdependence, etc.) that social psychologists had long noted to be properties of groups. We had participants rate diﬀerent types of social category (deﬁned by shared opinions, ethnicity, activities, physical medical conditions, mental disorders, and psychological features) on a set of 18 dimensions (reﬂecting naturalness, entitativity, and collective qualities), and we found that the largest diﬀerences between types of social category occurred on collective quality dimensions. Of particular interest here, however, was that mental disorder categories (anxiety, depression, schizophrenia) were seen as being almost utterly devoid of collective qualities, whereas opinion-based groups were strongly associated with these qualities. This suggested to us a distinction among social categories between types and groups. Any social category might be thought of as a type or a group, depending on whether the category is perceived to have undergone group formation, that is, whether the category members were believed to be acting as group members. A social category that is not associated with collective qualities (i.e., is not perceived to have undergone group formation) is referred to as a type. Some key implications of this idea relate to stereotype change and prejudice reduction. If some category is perceived to be a type, then there is less prospect that the qualities of the group will be perceived to be changeable through collective action on the part of the group. If some opinion or ethnic group is perceived to be hostile and untrustworthy, then it is always possible for that group to be perceived to be capable of changing its behavior (for better or worse) through collective action. In the same way that Meenes (1943) noted that American stereotypes of the Japanese changed so that they were no longer seen as artistic and progressive but as sly, treacherous, and deceitful following the surprise attack on Pearl Harbor, so an organized group can be perceived to be capable of altering its qualities. If people with schizophrenia are perceived to be unpredictable and violent (but these qualities are understood to be part of their essential psychological makeup rather than as arising from the operation of social norms), then there is little prospect for expecting change in their behavior. A perception of a category as a group (which equates to having particular types of mechanism knowledge

44

Hierarchies and groups

about how the category forms and operates) enables the members of that category to take control of the way they are perceived. Perhaps the clearest example of such a transition in perception has been that achieved by gay men and lesbians in Western countries over the last few decades. For many homosexuals gay has come to be deﬁned as part of a political and social identity rather than as a moral failing, a mental or physical disorder, or a biological or personality trait (just to mention some of the ways homosexuality has been characterized over time). By deﬁning sexual orientation as a social movement, gay men and lesbians have been able to challenge legal restrictions, and to contest stereotypes. It is worth asking whether the same improvements can be achieved for people with mental disorders. In a follow-up study Khalaf manipulated perceptions of group membership by pointing to the collective qualities that people with mental disorders actually exhibit. To do this she provided participants with information about the activities of mental health advocacy networks. These networks involve people with mental disorders and their supporters working for the beneﬁt of other people with these disorders. The manipulation produced a strong positive change in beliefs about mental disorders. In a further longitudinal study participants were asked to sign on to mental health advocacy networks, that is, to become active members of an opinionbased group devoted to countering negative stereotypes of people with mental disorders (this involved attending a series of monthly meetings where they discussed strategies for countering negative stereotypes of people with mental disorders on campus). The results demonstrated extremely large positive changes in terms of beliefs, emotional responses, and behavioral responses that were maintained over a three-month period. This set of ﬁndings has two important implications for the idea of multiplicity. First, it seems that social categories are themselves open to (at least) two competing broad interpretations (as groups or types) depending on the nature of the background knowledge about the categories. Second, it seems that particular opinion-based identities can be used as a means of creating positive changes in social stereotypes that are diﬃcult to achieve or maintain through the conventional approach of stressing dual or superordinate identity.

CONCLUSION The central point of this chapter can be summarized quite easily. The ways in which people are perceived are indeed diverse and complex and this is in large part true because of the multiplicity of categorization. One approach to addressing the issues posed by multiple categorization is to consider the idea that categorical systems are based on hierarchies. This straightforward and compelling idea was a cornerstone of the initial focus of self-categorization theory and a number of related approaches.

McGarty

45

The problem with this conceptualization is that it does not reﬂect the way that many (perhaps most) real social (or natural) structures are organized. The idea of hierarchies where elements at a lower level are uniquely contained within elements at the higher level has little relevance to actual social categorical systems. On the contrary, real social categorical systems are characterized by fuzzy boundaries and overlapping memberships that shift in response to context (a point that is, of course, fully consistent with the general intent of self-categorization theory). The idea of a hierarchical system can tend to lock us into solutions that are less than optimal when it comes to considering a means of promoting cooperation and positive relations with other social groups. The most common solution is to consider the two groups as subgroups nested within some superordinate group and then to attempt to convince the subgroup members to endorse (identify with) this superordinate membership. We pointed out that this solution is fraught with diﬃculties, especially where the superordinate membership was problematic for one or the other group, or where the superordinate group is vague (e.g., humanity) or otherwise unlikely to become psychologically signiﬁcant for members of one or more subgroup. We therefore suggested an alternative solution that involved crafting a new group identity based on shared opinions. Such an opinion-based group need not be problematic in either sense. To illustrate the potential of this idea we used the case of changing negative stereotypes toward people with mental disorders held by people without such disorders. The evidence so far strongly suggests that the creation of a salient identity based on a shared opinion involving commitment to a common cause can produce large and longlasting positive changes. The point here is that rather than just suggesting that people with and without mental disorders are all people (a category that is, of course, dominated by those without mental disorders), we can move beyond the banality of this common ingroup to form new groups that cut across the categories that we might bring in some ready-made fashion to some particular situation and seek to create new categories (and potentially groups). In so doing we can provide a clear basis for social organization and action by those who identify in this way. It is important to realize therefore that the idea of the hierarchy is a great place to start a treatment of knowledge about categories but it is a bad place to ﬁnish. This follows quite naturally from the idea that all things can be categorized in many diﬀerent ways and that much of the diversity and variability of knowledge about categories (especially social categories) that people can have, and need to have in order to grasp the complexities of human social relations, is potentially lost when we insist that that knowledge is hierarchically organized.

46

Hierarchies and groups

REFERENCES Anastasio, P. A., Rust, M. C., Gaertner, S. L., & Dovidio, J. F. (1997). Categorization, recategorization and common ingroup identity. In R. Spears, P. J. Oakes, N. Ellemers, & S. A. Haslam (Eds.), The social psychology of stereotyping and group life (pp. 236–256). Oxford: Blackwell Publishing. Asch, S. E. (1952). Social psychology. Englewood Cliﬀs, NJ: Prentice-Hall. Barsalou, L. (1987). The instability of graded structure. In U. Neisser (Ed.), Concepts and conceptual development: Ecological and intellectual factors in categorization (pp. 101–140). Cambridge: Cambridge University Press. Barsalou, L. (1993). Challenging assumptions about concepts. Cognitive Development, 8, 169–180. Berry, J. W. (1984). Cultural relations in plural societies: Alternatives to segregation and their sociopsychological implications. In N. Miller & M. B. Brewer (Eds.), Groups in contact: The psychology of desegregation (pp. 11–27). New York: Academic Press. Bliuc, A.-M., McGarty, C., Musgrove, L., Reynolds, K. J., & Hendres, D. (2004). Understanding the roles of social identiﬁcation with and salience of political opinionbased groups in predicting commitment to take collective action. Unpublished manuscript, The Australian National University. Brewer, M. B. (1991). The social self: On being the same and diﬀerent at the same time. Personality and Social Psychology Bulletin, 17, 475–482. Brewer, M. B. (1993). Social identity, distinctiveness, and in-group homogeneity. Social Cognition, 11, 150–164. Brewer, M. B., Ho, H., Lee, J., & Miller, N. (1987). Social identity and social distance among Hong Kong schoolchildren. Personality and Social Psychology Bulletin, 13, 156–165. Bruner, J. S. (1957). On perceptual readiness. Psychological Review, 64, 123–152. Crisp, R. J., Hewstone, M., & Rubin, M. (2001). Does multiple categorization reduce intergroup bias? Personality and Social Psychology Bulletin, 27, 76–89. Doise, W., Deschamps, J.-C., & Meyer, G. (1978). The accentuation of intercategory similarities. In H. Tajfel (Ed.), Diﬀerentiation between social groups. London: Academic Press. Eggins, R. A., Haslam, S. A., & Reynolds, K. J. (2002). Social identity and negotiation: Subgroup representation and superordinate consensus. Personality and Social Psychology Bulletin, 28, 887–899. Eggins, R. A., Reynolds, K. J., & Haslam, S. A. (2003). Working with identities: The ASPIRe model of organisational planning, negotiation and development. In S. A. Haslam, D. van Knippenberg, M. J. Platow, & N. Ellemers (Eds.), Social identity at work: Developing theory for organisational practice. Philadelphia: Taylor & Francis. Gaertner, S. L., & Dovidio, J. F. (2000). Reducing intergroup bias: The common ingroup identity model. Hove, UK: Psychology Press. Gaertner, S. L., Dovidio, J. F., Anastasio, P. A., Bachman, B. A., & Rust, M. C. (1993). The common ingroup identity model: Recategorization and the reduction of intergroup bias. European Review of Social Psychology, 4, 1–26. Haslam, N. O., Rothschild, L., & Ernst, D. (2000). Essentialist beliefs about social categories. British Journal of Social Psychology, 39, 113–127. Haslam, S. A. (2001). Psychology in organizations: The social identity approach. London: Sage Publications.

McGarty

47

Haslam, S. A., Oakes, P. J., Turner, J. C., & McGarty, C. (1995). Social categorization and group homogeneity: Changes in the perceived applicability of stereotype content as a function of comparative context and trait favourableness. British Journal of Social Psychology, 34, 139–160. Haslam, S. A., Turner, J. C., Oakes, P. J., McGarty, C., & Hayes, B. K. (1992). Context dependent variation in social stereotyping 1: The eﬀects of intergroup relations as mediated by social change and frame of reference. European Journal of Social Psychology, 22, 3–20. Hewstone, M. (1996). Contact and categorization: Social psychological interventions to change intergroup relations. In C. N. Macrae, C. Stangor, & M. Hewstone (Eds.), Stereotypes and stereotyping (pp. 323–368). New York: The Guilford Press. Hewstone, M., & Brown, R. (1986). Contact is not enough: An intergroup perspective. In M. Hewstone & R. Brown (Eds.), Contact and conﬂict in intergroup encounters (pp. 1–44). Oxford: Blackwell Publishing. Hewstone, M., Islam, M. R., & Judd, C. M. (1993). Models of crossed categorization and intergroup relations. Journal of Personality and Social Psychology, 64, 779–793. Higgins, E. T. (1996). Knowledge activation: Accessibility, applicability, and salience. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles (pp. 133–168). New York: The Guilford Press. Higgins, E. T., & King, G. (1981). Accessibility of social constructs: Information processing consequences of individual and contextual variability. In N. Cantor, & J. F. Kihlstrom (Eds.), Personality, cognition, and social interaction (pp. 69–121). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Hornsey, M. J., & Hogg, M. A. (2000). Assimilation and diversity: An integrative model of subgroup relations. Personality and Social Psychology Review, 4, 143–156. Kelly, C., & Breinlinger, S. (1995). Identity and injustice: Exploring women’s participation in collective action. Journal of Community and Applied Psychology, 5, 41–57. Kessler, T., & Mummendey, A. (2001). Is there any scapegoat around? Determinants of intergroup conﬂicts at diﬀerent categoriztion levels. Journal of Personality and Social Psychology, 81, 1090–1102. Khalaf, A., & McGarty, C. (2002). Natural kinds, entitative categories and collective qualities: Comparing mental disorders with other social categories. Unpublished manuscript, The Australian National University. Klandermans, B. (2000). Identity and protest: How group identiﬁcation helps to overcome collective action dilemmas. In M. Van Vugt, M. Snyder, T. Tyler, & A. Biel (Eds.), Cooperation in modern society: Promoting the welfare of communities, states and organizations (pp. 162–183). London: Routledge. Lickel, B., Hamilton, D. L., Wieczorkowska, G., Lewis, A., Sherman, S. J., & Uhles, A. N. (2000). Varieties of groups and the perception of group entitativity. Journal of Personality and Social Psychology, 78, 223–246. Mavor, K. I. (2001). Homogeneity at the individual and group level: Ingroup-outgroup homogeneity and the actor-observer eﬀect. Paper presented at the 9th Brisbane Symposium on Social Identity, Brisbane, Australia. Mavor, K. I., & Reynolds, K. J. (2002). Categorisation, homogeneity, and attribution: Out-person homogeneity and the actor-observer eﬀect. Poster presented at the 3rd Annual conference of the Society for Personality and Social Psychology, Los Angeles, CA. McGarty, C. (1999). Categorization in social psychology. London: Sage Publications. McGarty, C. (2002). Stereotype formation as category formation. In C. McGarty,

48

Hierarchies and groups

V. Y. Yzerbyt, & R. Spears (Eds.), Stereotypes as explanations: The formation of meaningful beliefs about social groups (pp. 16–37). Cambridge: Cambridge University Press. McGarty, C., & Bliuc, A.-M. (2004). Reﬁning the meaning of the “Collective” in collective guilt: Harm, guilt and apology in Australia. In N. R. Branscombe & B. Doosje (Eds.), Collective guilt: International perspectives (pp. 112–129). New York: Cambridge University Press. McGarty, C., Spears, R., & Yzerbyt, V. Y. (2002). Conclusion: Stereotypes are selective, variable and contested explanations. In C. McGarty, V. Y. Yzerbyt, & R. Spears (Eds.), Stereotypes as explanations: The formation of meaningful beliefs about social groups (pp. 186–199). Cambridge: Cambridge University Press. Medin, D. L., Goldstone, R. L., & Gentner, D. (1993). Respects for similarity. Psychological Review, 100, 254–278. Medin, D. L., Lynch, E. B., Coley, J. D., & Atran, S. (1997). Categorization and reasoning among tree experts: Do all roads lead to Rome? Cognitive Psychology, 32, 49–96. Meenes, M. (1943). A comparison of racial stereotypes in 1935 and 1942. Journal of Social Psychology, 17, 327–336. Mummendey, A., & Wenzel, M. (1999). Social discrimination and tolerance in intergroup relations: Reactions to intergroup diﬀerence. Personality and Social Psychology Review, 3, 158–174. Murphy, G. L., & Medin, D. L. (1985). The role of theories in conceptual coherence. Psychological Review, 92, 289–316. Oakes, P. J. (1987). The salience of social categories. In J. C. Turner, M. A. Hogg, P. J. Oakes, S. D. Reicher, & M. S. Wetherell (Eds.), Rediscovering the social group: A self-categorization theory (pp. 117–141). London: Blackwell Publishing. Oakes, P. J., & Turner, J. C. (1990). Is limited information processing capacity the cause of social stereotyping? European Review of Social Psychology, 1, 111–135. Oakes, P. J., Haslam, S. A., & Turner, J. C. (1994). Stereotyping and social reality. Oxford, UK and Cambridge, MA: Blackwell Publishing. Rips, L. J. (1975). Inductive judgments about natural categories. Journal of Verbal Learning and Verbal Behavior, 14, 665–681. Rosch, E. H. (1978). Principles of categorization. In E. H. Rosch & B. Lloyd (Eds.), Cognition and categorization (pp. 28–49). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Rosch, E. H., Mervis, C. B., Gray, W. D., Johnson, D. M., & Boyes-Braem, P. (1976). Basic objects in natural categories. Cognitive Psychology, 8, 382–439. Ross, L. (2000). Hierarchical structure as a basis for inference in reasoning about nonhuman, social and self-categories. Unpublished honours thesis, The Australian National University. Sani, F., & Reicher, S. D. (1998). When consensus fails: An analysis of the schism within the Italian Communist Party (1991). European Journal of Social Psychology, 28, 623–645. Simon, B., & Klandermans, B. (2000). Politicized collective identity: A social psychological analysis. American Psychologist, 56, 319–331. Sloman, S. A. (1998). Categorical inference is not a tree: The myth of inheritance hierarchies. Cognitive Psychology, 35, 1–33. Smithson, M. J. (1987). Fuzzy set analysis for behavioral and social sciences. New York: Springer-Verlag.

McGarty

49

Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conﬂict. W. G. Austin & S. Worchel (Eds.), The social psychology of intergroup relations (pp. 33–47). Monterey, CA: Brooks/Cole. Tajfel, H., Flament, C., Billig, M. G., & Bundy, R. F. (1971). Social categorization and intergroup behavior. European Journal of Social Psychology, 1, 149–177. Turner, J. C. (1982). Towards a cognitive redeﬁnition of the group. In H. Tajfel (Ed.), Social identity and intergroup relations. Cambridge: Cambridge University Press. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. Oxford: Blackwell Publishing. Turner, J. C., Oakes, P. J., Haslam, S. A., & McGarty, C. (1994). Self and collective: Cognition and social context. Personality and Social Psychology Bulletin, 20, 454–463. Zadeh, L. A. (1965). Fuzzy sets. Information and Control, 8, 338–353.

3

Multiply categorizable social objects Representational models and some potential determinants of category use Eliot R. Smith

Although it is well recognized that individual people and other social objects are multiply categorizable, common conceptions of mental representation in social psychology do not easily allow for this property. In this chapter two more ﬂexible types of model that treat representations as transient states rather than as static things are reviewed: exemplar and distributed representations. Both of these can accommodate multiple categorizations. If such types of representation are assumed, questions naturally arise about the factors that inﬂuence which of many available categorizations will be used. To complement the emphasis within the social identity tradition (Tajfel & Turner, 1979; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987) on the social context as a determinant (see McGarty, this volume), two factors involving the perceiver are suggested as potentially important, based on prior research. The perceiver’s practice or experience in categorizing a particular individual in one way or another, as well as the strength of the perceiver’s attitudes toward potential categories, ought to aﬀect the likelihood of categorizing a person in one way or another.

CATEGORY REPRESENTATION The theme of this volume is that every individual person is simultaneously a member of multiple categories. This situation contrasts sharply with many common nonsocial objects whose representation has been studied by cognitive psychologists. Since the pioneering work on object representations by Collins and Quillian (1969) and others, cognitive psychologists have focused on common noun concepts, and have developed models of the representation of well-structured hierarchical categories. For example, a tree-like structure could be created to represent the facts that Fluﬀy is a canary; a canary is a bird; a robin is a bird; a bird is a living thing; and so on. Nodes representing these individual concepts would be connected by special category-inclusion links arranged to reﬂect the conceptual hierarchy. Thus, canary and robin

Smith

51

(and other nodes) would be subordinate to the bird node, while the Fluﬀy node would be subordinate to canary. Properties that pertain to each speciﬁc concept would also be attached to the concept nodes. Figure 3.1 illustrates this idea. Hierarchical representations were originally proposed by Collins and Quillian (1969) as a means of economizing on storage by allowing the cognitive system to look up properties at any hierarchical level. For example, the fact that canaries are yellow is assumed to be stored at the canary level of the hierarchy (for it does not apply to other birds). But the fact that birds have wings is stored at the “bird” level, yielding great economy because it does not have to be stored separately as a fact about canaries, robins, crows, etc. If one wants to retrieve the answer to the question of whether Fluﬀy has wings, one can follow the object hierarchy upwards to obtain the answer. Note that this involves traversing one more node compared to answering the question of whether Fluﬀy is yellow, giving rise to obvious response-time predictions that were successfully tested in studies by Collins and Quillian (1969) and others. Hierarchical representations can even be used to store “exception” properties, such as the fact that ostriches do not ﬂy—this would

Figure 3.1 Example of a hierarchical representational scheme (adapted from Collins & Quillian, 1969). Note: solid arrows represent class-inclusion (“isa”) links; dashed arrows represent property links.

52

Representational models

be stored with the “ostrich” node and would override the more general fact that birds ﬂy. Despite the seeming advantages of this hierarchical type of representation, it has problems. Most notably, it does not allow for multiple cross-cutting categorizations. A concept is stored at a particular place in the tree-structured hierarchy, making it a member of one and only one superordinate category (and of course, indirectly, a member of the higher-level categories above that one). Cross-cutting “categories” such as “yellow things” could, of course, be postulated, but they are not represented directly in this type of scheme. Thus, their exemplars could be retrieved and enumerated only by traversing the tree and examining every single concept node for the feature “yellow” (see McGarty, this volume, for a related critique of hierarchical representation). Besides hierarchical tree-structured models of concept representation, other representational models that are prominent within social psychology also assume that each concept is categorized in just a single way. Wyer and Srull (1989), in their highly inﬂuential person memory model, proposed that social as well as nonsocial memory is structured as a set of storage bins. A representation of a person (for example, George W. Bush) would thus be stored in one bin (perhaps for Republicans) and could not be stored simultaneously in other bins (e.g., male, white, Texan, United Methodist). Even beyond formal representational models, the underlying assumption that an object is a member of only one category seems to be part of our everyday intuitions about social objects (such as people). We frequently ask questions like “How did the observer categorize the target?”, which carry the assumption that the target was “stored under” a single category. Interpretations of data patterns in empirical studies typically make the same assumption. For example, Taylor and Falcone (1982) and many others have used “who said what?” paradigms in which perceivers’ memory errors are used to draw inferences about “how the targets were categorized” (e.g., by gender, by race). But if we acknowledge (as the chapters in this volume argue) that many category memberships can simultaneously apply to a given person, we need to discard this intuition and this type of language. Persons and other objects are not represented in memory as members of only a single category. This means that we need to seek alternative representational models. In models like the hierarchical tree or storage bin, representations are treated as static, or metaphorically as “things”. Just as Abelson and Prentice (1989) described beliefs as being like possessions, metaphors for memory that involve the notions of storage, search, and retrieval elicit images of ﬁle cabinets or storehouses ﬁlled with thing-like records of knowledge (Smith, 1998). Like a piece of paper in a ﬁle cabinet, the knowledge is stored at a particular location, then searched for and retrieved when needed, at which point it is expected to be the same as when it was stored. Tree-structured hierarchical category representations and storage bins presuppose this way of thinking. An alternative way of thinking of representations is as states rather than things (Smith, 1998). A representational state is deﬁned at a particular

Smith

53

moment in time, and will change as the person shifts beliefs or focuses on new information. The notion of a state, in contrast to that of static thing-like representations, means that we cannot talk of “retrieving” a static representation (as from a ﬁle cabinet) but must instead talk in terms of “reconstructing” a previous state (Clark, 1993; Gilbert, 1993). Perhaps most important is that a currently active representation is quite diﬀerent in form from a representation that is not currently active, but that the system can reconstruct in appropriate circumstances. This contrasts with the “thing” picture, in which mental representations are considered to be essentially the same whether or not they are currently active (just as a paper stays the same whether it is stored in the ﬁle cabinet or retrieved for use). Interestingly, the notion of reconstruction is intuitive and natural when we think about emotions. Nobody would ever ask, “Where is your anger when you are not feeling it?” because we naturally think of anger as a state. However, “Where is your representation of person X stored when you are not thinking about it?” seems a perfectly reasonable question because we are used to thinking of knowledge representations as things. To start to build up intuitions about how representations can be considered as states, consider the perception of ambiguous ﬁgures such as the Necker cube or the familiar duck/rabbit drawing. In such cases, multiple perceptual interpretations of the ﬁgure are possible, although they typically do not coexist, but rather switch back and forth because interpretations of parts of the ﬁgure constrain other parts. For example, seeing one part of the ﬁgure as the rabbit’s ears means that another speciﬁc part must be the rabbit’s mouth, etc. Thus the multiple interpretations compete and cannot be held onto simultaneously; the interpretation seems more like a state than a static thing. A similar phenomenon at a more abstract level involves the way sentence or narrative contexts can select and make salient particular features of concepts (Barsalou, 1982). For example, as a perceiver comprehends a sentence like “The movers lifted the piano”, the attribute heavy becomes activated, while in the sentence “The performer sat down at the piano”, attributes like keys or sound are more active. The current interpretation of even a seemingly stable concept like piano is thoroughly shaped by context, acting like a state rather than a thing.

MODELS OF REPRESENTATIONS AS STATES How can we go beyond intuitions to develop speciﬁc, theoretically coherent models of representation that capture their state-like (rather than thing-like) properties, such as context-sensitive, on-line reconstruction? It turns out that there are two general types of representational theory that have these properties: exemplar-based and distributed representations. Smith (1998) gives a detailed overview of these, including illustrative applications within social psychology, and also describes conceptual relationships between the exemplar and distributed frameworks.

54

Representational models

Exemplar models In brief, exemplar models assume that conceptual knowledge is based on a large number of representations of individual exemplars (instances) of the concept. For example, our knowledge about birds would be underlain by a large number of representations of speciﬁc birds that one has witnessed, been told about, imagined, etc. When general knowledge about birds as a category is required, such as to answer the question of whether birds have wings, the individual exemplars are aggregated on-line to provide the general response. The important feature is that in this aggregation process, the current context, goals, moods, or other factors can aﬀect the likelihood of retrieval of individual exemplars (Hintzman, 1986; Medin & Schaﬀer, 1978; Smith & Zárate, 1992). For example, as Barsalou (1987) has demonstrated, thinking of birds in a barnyard context results in chicken and similar exemplars being more retrievable (and so being judged as more central to the resulting reconstructed context), while in a suburban backyard context, robin is more central. Similar contextual variation has been demonstrated for social concepts such as stereotypes. Bodenhausen, Schwarz, Bless, & Waenke (1995) found that when participants had been recently exposed in a previous task to well-liked exemplars of African Americans (e.g., Bill Cosby), their attitudes regarding African Americans in general became more positive. The most straightforward interpretation of this ﬁnding is that the primed exemplars were preferentially retrieved when the participants were generating their overall concept of African Americans (Smith & Zárate, 1992). In summary, under exemplar models, a conceptual representation is a state, constructed on-line by retrieval and aggregation processes operating on underlying exemplar representations, potentially biased and inﬂuenced by current contextual factors. With regard to multiple categorization speciﬁcally, exemplar models furnish a ready account, as outlined by Smith and Zárate (1992). The eﬀect rests on the basic principle that attention to a particular dimension tends to exaggerate the impact of diﬀerences along that dimension, as documented by Medin and Schaﬀer (1978) in their seminal context model of exemplar-based categorization. Thus, imagine a perceiver thinking about a white woman. If the perceiver pays attention to race as a dimension, this exaggerates diﬀerences along this dimension, making exemplars of black or Hispanic women less similar to the target (in comparison to other white exemplars). As a result, the woman will be judged, evaluated, and categorized on the basis of the most similar exemplars retrieved from memory, other whites—we could say that the perceiver sees her as a white person. In contrast, attention to the gender dimension would have the opposite eﬀects: male exemplars would become less similar (less likely to be retrieved) but other female exemplars would be retrieved and used as a basis for judgments and categorizations. The same underlying stock of social exemplars can thus be ﬂexibly used, resulting in diﬀering reactions to a given target (e.g., a white female) based on whatever

Smith

55

contextual or perceiver factors alter relative attention to the stimulus dimensions. Such predictions have been recently tested and conﬁrmed (van Rijswijk & Ellemers, 2002) and, indeed, are consistent with one of the key components of self-categorization theory’s account of multiple categorization (see McGarty, this volume).

Distributed models Besides exemplar models, distributed models of concept representation also allow for ﬂexible, state-like reconstruction. Distributed models, which are also called parallel distributed processing (PDP) or neural-network or connectionist models, were popularized within psychology by Rumelhart, McClelland, and others (Rumelhart & McClelland, 1986; see also McClelland & Rumelhart, 1986; McClelland, Rumelhart, & Hinton, 1986). Brief introductions for social psychologists can be found in Smith (1996; 1998). In these models a concept is represented by a pattern of activation across a number of atomic processing units termed “nodes”, which are at least metaphorically based on biological neurons. Although each individual node has no ﬁxed conceptual meaning, a set of nodes can represent a very large number of diﬀerent patterns, just as the set of pixels on a TV screen or computer monitor can represent many diﬀerent pictures or texts by taking on diﬀerent color or brightness values. Nodes are connected and send signals to each other based on their current activation levels, which are assumed to change rapidly with time. The connections in turn are modiﬁable by experience, changing (more slowly) to alter the overall information-processing properties of the network of nodes. Distributed representations have a number of unique and desirable properties, but most important for our purposes is that active representations (patterns of activation) are reconstructed by ﬂows of activation within the network. A pattern that is not currently active is not “stored” anywhere as a static copy of the pattern, but is implicitly represented in the set of connection strengths or weights that enable the network to recreate the pattern upon presentation of the appropriate input values (McClelland, Rumelhart, & Hinton, 1986, p. 31). As with exemplar representations, the reconstruction of an activation pattern representing an object or concept is context-sensitive, dependent on other cues or sources of activation that are present at the time as well as the speciﬁc target stimuli. Clark (1993) describes this process as entailing that “there need be no context-independent, core representation for [a concept]. Instead, there could be a variety of states linked merely by a relation of family resemblance” (p. 24). In summary, distributed representations (like exemplar-based ones) oﬀer the type of context-sensitivity and ﬂuidity that is required to account for processes like multiple categorization of social objects. Both of these types of models portray representations as transient, reconstructed states rather than as things, and both take us beyond the relative

56

Representational models

inﬂexibility of models in which objects are “stored” in one or another mental “location” or “bin” based on static category membership.

Conjunctive categories Thus far, we have been discussing the situation where a given object can be seen ﬂexibly as a member of one category or another (rather than rigidly being “categorized as” a member of only a single category). Another aspect of crossed categories is also important: the situation where a conjunctive category eﬀectively becomes a new category of its own (see Crisp & Hewstone, this volume, for a detailed discussion of these diﬀerent research foci in multiple categorization work). Particularly interesting is the possibility that the conjunctive category acquires novel “emergent” properties, instead of simply inheriting the properties of its parent categories. A classic example is “pet ﬁsh” (Barsalou, 1987)—the best example of this conjunctive category (a goldﬁsh, perhaps, or a guppy) does not possess many of the typical properties of either a “pet” or a “ﬁsh”. Social category conjunctions also have been found to possess emergent properties (Kunda, Miller, & Claire, 1990). For example, a “Harvard-educated carpenter” might be assumed to be nonmaterialistic, a property not particularly associated with either Harvard-educated people or carpenters. How can emergent properties like this example be explained by exemplar or distributed representations? Under an exemplar model, one can in two categories (such as pet ﬁsh) oﬀer cues that are speciﬁc enough to cue the retrieval of particular exemplars (goldﬁsh) that then become salient in deﬁning the conjunctive category. These examplars may not be highly salient or typical exemplars of either parent category, particularly if the category overlap is relatively small. But what if the category overlap is so small that one is unlikely to even have any exemplars of the conjunctive category (such as the Harvard-educated carpenter example)? Using distributed representations, emergent properties can arise as eﬀects of multiple representations in memory (Clark, 1993; Smith & DeCoster, 1998). Imagine that the person knows several separate facts: (a) Harvard-educated people are qualiﬁed for highpaying occupations, (b) carpenter is a relatively low-paying occupation, and (c) people might take up a low-paying occupation if they are nonmaterialistic (among other possible reasons). Because distributed representations have the property of simultaneously drawing on all knowledge represented in memory as they actively reconstruct a state-like representation (instead of retrieving a single, ﬁxed existing representation), they can combine these separate bits of knowledge. The result may be an emergent Harvard-carpenter category that has, among other properties, the property nonmaterialistic. In contrast, the same distributed memory would not tend to construct nonmaterialistic as a property of either parent category. Thus, state-like representational models (including both exemplar and distributed representations) can generate representations that are ﬂexible in two

Smith

57

distinct ways. A multiply categorizable object can be seen, from one moment to the next, as a member of one category or the other. In addition, a conjunctive category can come to function as a category in its own right, a process that is particularly signiﬁcant when it acquires new emergent properties.

Summary At the level of behaviorally observable phenomena, researchers in the social identity theory and self-categorization theory traditions have long recognized the great ﬂexibility in the ways people think of themselves and others (e.g., McGarty, this volume). Thinking of oneself as a member of one or another group (e.g., by age versus gender versus nationality), for example, makes a diﬀerence not only in one’s behavior but in the way one thinks about and treats other people as they shift from ingroup to outgroup members. “The concept of the self as a separate mental structure does not seem necessary, because we can assume that any and all cognitive resources—long-term knowledge, implicit theories . . . and so forth—are recruited, used, and deployed when necessary” (Turner, Oakes, Haslam, & McGarty, 1994). To account for this type of ﬂexibility, we need to draw on representational models (such as exemplar or distributed models) that allow for on-line reconstruction of representations as states, rather than restricting ourselves to more static models that assume objects are categorized once and then stored, inﬂexibly, in a way that is based on that categorization.

DETERMINANTS OF ON-LINE CATEGORIZATION Exemplar and distributed models of representation allow for concepts or objects to be considered as multiply categorizable, and for state-like representations to shift depending on current circumstances. But what factors determine which of many potential categories will be applied in any given instance? Social identity theory and self-categorization theory have emphasized situational factors that aﬀect the categorization of multiply categorizable individuals. For example, a white female target in a group of black and white men might be readily categorized as a woman, while the same person in a group of black men and women would be more likely to be seen as white (e.g., van Rijswijk & Ellemers, 2002). Besides such situational factors, attributes of the perceiver can also aﬀect categorization tendencies, but they have been much less researched in this area. Two such factors may be inﬂuential, based on my own research: the perceiver’s past experience (or practice) in categorizing a particular person or other social object, and the strength of the perceiver’s attitudes toward the relevant categories. The investigations of practice or experience stemmed from research interest in the automatization or proceduralization of cognitive processes (Smith, 1989). This research showed in general that when people repeatedly practice a

58

Representational models

categorization process (such as categorizing behaviors into trait categories) they become faster or more eﬃcient at doing so. The increased speed is evident as much as 24 hours later. A study by Smith, Stewart, and Buttram (1992) applied this general ﬁnding to look at eﬀects on the way people categorize multiply categorizable objects. In this study, we used behaviors that could be seen as reﬂecting two diﬀerent traits. For example, “tried to ﬁx his friend’s refrigerator but ended up making it worse” could reﬂect either friendliness or unintelligence. Based on previous ﬁndings in the literature on proceduralization, we predicted that when someone had previously judged this behavior on one trait, they would be likely to again see it as reﬂecting that trait at a later time—even after the original encounter with the behavior had been forgotten. To assess which trait people used to categorize this and other multiply categorizable behaviors, we measured their overall evaluation of the behavior. Presumably, seeing this behavior as friendly would lead to evaluating it more positively than seeing it as unintelligent. In the study, participants judged a large number of diﬀerent behaviors in two experimental sessions, on either friendliness or intelligence. Among the behaviors were embedded 12 multiply categorizable behaviors similar to the example. The target trait for these judgments was manipulated between subjects. Each behavior appeared in turn on a computer screen, and the participant pressed a “yes” or “no” key to indicate whether or not the behavior reﬂected the target trait. The two experimental sessions were either 1 day or 7 days apart. After making the trait judgments in the second session, participants were given the list of 12 multiply categorizable behaviors and asked to indicate their overall evaluation of those behaviors (under the cover story that the researchers were collecting pretest ratings of behaviors that might be used as materials in similar judgment studies in the future). Results showed that when participants had previously judged the behavior on one or the other trait, their evaluation was more in line with the behavior’s implications for that trait. This was true even when the initial judgment had been made 24 hours or 7 days previously (in the ﬁrst experimental session). Remarkably, the eﬀect was evident even for behaviors that the participant did not recognize as having been seen before at all. This study, then, demonstrates the strength and durability of a type of item-speciﬁc practice eﬀect, as well as its independence from conscious or explicit recollection. Applied to the categorization of people, this eﬀect suggests that once perceivers see people in a given context as members of a particular category, they will be likely to categorize them in the same way on a later occasion (even if the ﬁrst meeting has been forgotten). For example, suppose the ﬁrst meeting with a white woman is in a social context that (as in the example above) biases the perceiver to see her as a woman. On meeting the same person in a neutral context, perhaps many days or even weeks later, the results of the Smith et al. (1992) study would suggest that the perceiver would be likely to see her again as a woman rather than as white. As a result, it might be expected that she should be judged in ways that reﬂect that categorization, such as on traits

Smith

59

reﬂecting the female stereotype. Indeed, Castelli, Zogmaister, Smith, & Arcuri (2004) recently demonstrated that evaluations based on a previous categorization of an individual can be automatically retrieved, even when the categorization is not itself remembered. A second factor that we have found to aﬀect the categorization of multiply categorizable social objects is the strength of the perceiver’s attitudes toward the relevant categories. This research was based on work by Roskos-Ewoldsen and Fazio (1992), who found that objects toward which people have highly accessible attitudes or strong attitudes (Fazio, Sanbonmatsu, Powell, & Kardes, 1986) tend to attract attention when they are included in an array of visually presented objects. Reasoning by analogy, we argued that a multiply categorizable object oﬀers an array of potential category memberships and therefore that a category toward which the perceiver had a highly accessible attitude might also attract attention. This would aﬀect the likelihood of categorizing the object on that dimension. In three experiments, Smith, Fazio, and Cejka (1996) tested and conﬁrmed this hypothesis. Stimulus materials in this study included both social objects (e.g., Pete Rose, categorizable as a baseball player or gambler) and nonsocial objects (e.g., yogurt: health food or dairy product). Both the strength of pre-existing attitudes (in a correlational study) and manipulations of attitude strength through repeated attitude expression aﬀected the way people categorized multiply categorizable objects. Thus, participants whose attitudes toward gamblers were stronger than their attitudes toward baseball players (either because of natural, nonlaboratory experiences or because of a laboratory manipulation of attitude accessibility) would be more likely to see Pete Rose as a gambler than as a baseball player. One study, for example, used a cued recall measure. Participants had their attitudes toward one or the other category strengthened by a repeated attitude expression manipulation, while they made other semantic (but nonattitudinal) judgments on the alternative category. Finally, they were presented with the multiply categorizable object as a retrieval cue and asked to write down whichever of the previously studied categories they could recall. Out of 36 trials, cued recall averaged 16.51 of the categories toward which the attitude had been strengthened, compared to only 10.32 of the alternative categories. This indicates that the multiply categorizable object was more likely to cue the retrieval of the category associated with a stronger attitude. Speculatively, this ﬁnding oﬀers insight into the worldview of the extreme bigot. Someone with a very strong and accessible negative attitude toward a particular group will be expected (on the basis of these experimental ﬁndings) to categorize all group members together and neglect their other potential cross-cutting category memberships. This process, then, furnishes a basis for treating members of the despised group as “all alike” and for ignoring potentially positive individuating information that might lead to a reduction in prejudice.

60

Representational models

SUMMARY Social categorization is extremely ﬂexible and adaptive. We generally respond to other people and social groups in ways that are attuned to current situations and to our own momentary or enduring goals. In the case of multiply categorizable social objects—such as individual persons—this ﬂexibility means that we will sometimes categorize them one way and sometimes another way. Tendencies in categorization can be driven by speciﬁc past experiences, even those that have been forgotten (Smith et al., 1992) as well as by the perceiver’s accessible attitudes toward the relevant categories (Smith et al., 1996). To understand these and other categorization processes, we will ultimately need to develop representational models that allow contextsensitivity and ﬂexibility. Models that postulate static hierarchical trees or storage bins make ﬂexibility diﬃcult to achieve. Better are models, such as exemplar-based or distributed models, that treat mental representations as states. Such models will facilitate further research on how social context shapes mental representations of social objects in ways that facilitate ﬂexible and adaptive social behavior.

ACKNOWLEDGEMENT This chapter was prepared with support from NIMH grant K02 MH01178 and NSF grant 0091807-BCS.

REFERENCES Abelson, R. P., & Prentice, D. A. (1989). Beliefs as possessions: A functional perspective. In A. R. Pratkanis & S. J. Breckler (Eds.), Attitude structure and function (pp. 361–381). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Barsalou, L. W. (1982). Context-independent and context-dependent information in concepts. Memory and Cognition, 10, 82–93. Barsalou, L. W. (1987). The instability of graded structure: Implications for the nature of concepts. In U. Neisser (Ed.), Concepts and conceptual development (pp. 101–140). Cambridge: Cambridge University Press. Bodenhausen, G. V., Schwarz, N., Bless, H., & Waenke, M. (1995). Eﬀects of atypical exemplars on racial beliefs: Enlightened racism or generalized appraisals? Journal of Experimental Social Psychology, 31, 48–63. Castelli, L., Zogmaister, C., Smith, E. R., & Arcuri, L. (2004). On the automatic evaluation of social exemplars. Journal of Personality and Social Psychology, 86, 373–387. Clark, A. (1993). Associative engines: Connectionism, concepts, and representational change. Cambridge, MA: The MIT Press. Collins, A. M., & Quillian, M. (1969). Retrieval time from semantic memory. Journal of Verbal Learning and Verbal Behavior, 8, 240–247.

Smith

61

Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C., & Kardes, F. R. (1986). On the automatic activation of attitudes. Journal of Personality and Social Psychology, 50, 229–238. Gilbert, D. T. (1993). The assent of man: Mental representation and the control of belief. In D. M. Wegner & J. W. Pennebaker (Eds.), Handbook of mental control (pp. 57–87). Englewood Cliﬀs, NJ: Prentice-Hall. Hintzman, D. L. (1986). “Schema abstraction” in a multiple-trace memory model. Psychological Review, 93, 411–428. Kunda, Z., Miller, D. T., & Claire, T. (1990). Combining social concepts: The role of causal reasoning. Cognitive Science, 14, 551–577. McClelland, J. L., & Rumelhart, D. E. (1986). Parallel distributed processing (Vol. 1). Cambridge, MA: The MIT Press. McClelland, J. L., Rumelhart, D. E., & Hinton, G. E. (1986). The appeal of parallel distributed processing. In D. E. Rumelhart & J. L. McClelland (Eds.), Parallel distributed processing (Vol. 1, pp. 3–44). Cambridge, MA: The MIT Press. Medin, D. L., & Schaﬀer, M. M. (1978). Context theory of classiﬁcation learning. Psychological Review, 85, 207–238. Roskos-Ewoldsen, D. R., & Fazio, R. H. (1992). On the orienting value of attitudes: Attitude accessibility as a determinant of an object’s attraction of visual attention. Journal of Personality and Social Psychology, 63, 198–211. Rumelhart, D. E., & McClelland, J. L. (1986). Parallel distributed processing (Vol. 2). Cambridge, MA: The MIT Press. Smith, E. R. (1989). Procedural eﬃciency: General and speciﬁc components and eﬀects on social judgment. Journal of Experimental Social Psychology, 25, 500–523. Smith, E. R. (1996). What do connectionism and social psychology oﬀer each other? Journal of Personality and Social Psychology, 70, 893–912. Smith, E. R. (1998). Mental representation and memory. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), Handbook of social psychology (4th ed., Vol. 1, pp. 391–445). New York: McGraw-Hill. Smith, E. R., & DeCoster, J. (1998). Knowledge acquisition, accessibility, and use in person perception and stereotyping: Simulation with a recurrent connectionist network. Journal of Personality and Social Psychology, 74, 21–35. Smith, E. R., Fazio, R. H., & Cejka, M. A. (1996). Accessible attitudes inﬂuence categorization of multiply categorizable objects. Journal of Personality and Social Psychology, 71, 888–898. Smith, E. R., Stewart, T. L., & Buttram, R. T. (1992). Inferring a trait from a behavior has long-term, highly speciﬁc eﬀects. Journal of Personality and Social Psychology, 62, 753–759. Smith, E. R., & Zárate, M. A. (1992). Exemplar-based model of social judgment. Psychological Review, 99, 3–21. 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. 33–47). Monterey, CA: Brooks/Cole. Taylor, S. E., & Falcone, H.-T. (1982). Cognitive bases of stereotyping: The relationship between categorization and prejudice. Personality and Social Psychology Bulletin, 8, 426–432. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. Oxford: Blackwell Publishing.

62

Representational models

Turner, J. C., Oakes, P. J., Haslam, S., & McGarty, C. (1994). Self and collective: Cognition and social context. Personality and Social Psychology Bulletin, 20, 454–463. van Rijswijk, W., & Ellemers, N. (2002). Context eﬀects on the application of stereotype content to multiply categorizable targets. Personality and Social Psychology Bulletin, 28, 90–101. Wyer, R. S., & Srull, T. K. (1989). Memory and cognition in its social context. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.

Part III

Multiple categorization and social judgment

4

Recategorization and crossed categorization The implications of group salience and representations for reducing bias John F. Dovidio, Samuel L. Gaertner, Gordon Hodson, Blake M. Riek, Kelly M. Johnson, and Missy Houlette

Prejudice and discrimination are intergroup, as well as interpersonal, phenomena. Consequently, some recent approaches have considered the role of basic group processes, such as social categorization, for understanding the nature of these biases and ways to reduce them. From a social categorization perspective, one universal facet of human perception essential for eﬃcient functioning is the ability to sort people, spontaneously and with minimum eﬀort or awareness, into a smaller number of meaningful categories (Brewer, 1988; see also Fiske, Lin, & Neuberg, 1999; see also McGarty, this volume). Given the centrality of the self in social perception, social categorization further involves a basic distinction between the group containing the self, the ingroup, and other groups, the outgroups—between the “we’s” and the “they’s” (see social identity theory, Tajfel & Turner, 1979; self-categorization theory, Turner, Hogg, Oakes, Reicher, & Wetherell, 1987; see also Hogg & Hornsey, this volume). In this chapter, we examine how interventions designed to alter the way people categorize others as members of their ingroup or outgroup can critically shape the nature and extent of social prejudice and bias. First, we describe the ways in which social categorization inﬂuences interpersonal and intergroup responses. Second, we review diﬀerent approaches aimed at altering the nature of social categorization to reduce bias. Third, we discuss potential limitations of interventions designed to create a single, inclusive social categorization. In the fourth section, we explore two diﬀerent approaches to recategorization, crossed categorization, and dual identities, within the context of the common ingroup identity model. The concluding section summarizes the ﬁndings and implications of the work.

66

Recategorization

SOCIAL CATEGORIZATION AND BIAS Social identity theory (Tajfel & Turner, 1979) and self-categorization theory (Turner, 1985; see also Onorato & Turner, 2001) posit that how a person conceives himself or herself, in terms of personal or collective identity, at any point in time is critical in determining the person’s response to others. Speciﬁcally, when personal identity is salient, a person’s individual needs, standards, beliefs, and motives primarily determine behavior. In contrast, when social identity is salient, “people come to perceive themselves as more interchangeable exemplars of a social category than as unique personalities deﬁned by their individual diﬀerences from others” (Turner et al., 1987, p. 50). Under these conditions, collective needs, goals, and standards are primary as group properties are deﬁned by diﬀerences from other groups. Illustrating the dynamics of this distinction, Verkuyten and Hagendoorn (1998) found that when individual identity was primed, individual diﬀerences in authoritarianism were the major predictor of the prejudice of Dutch students toward Turkish migrants. In contrast, when social identity (i.e., national identity) was made salient, ingroup stereotypes and standards primarily predicted prejudiced attitudes. When collective identity is salient, the distinction between ingroup and outgroup members as a consequence of social categorization has a profound inﬂuence on social perception, aﬀect, cognition, and behavior. Perceptually, when people or objects are categorized into groups, actual diﬀerences between members of the same category tend to be minimized (Tajfel, 1969) and often ignored in making decisions or forming impressions, while between-group diﬀerences tend to become exaggerated (Abrams, 1985; Turner, 1985). Emotionally, people spontaneously experience more positive aﬀect toward other members of the ingroup than toward members of the outgroup (Otten & Moskowitz, 2000), particularly toward those ingroup members who are most prototypical of their group (Hogg & Hains, 1996). Cognitively, people retain more information in a more detailed fashion for ingroup members than for outgroup members (Park & Rothbart, 1982), they have a better memory for information about ways in which ingroup members are similar to and outgroup members are dissimilar to the self (Wilder, 1981), and remember less positive information about outgroup members (Howard & Rothbart, 1980). In terms of behavioral outcomes, people are more helpful toward ingroup than toward outgroup members (Dovidio, Gaertner, Validzic, Matoka, Johnson, & Frazier, 1997), and they work harder for groups identiﬁed as ingroups than outgroups (Worchel, Rothgerber, Day, Hart, & Butemeyer, 1998). When ingroup–outgroup social categorizations, rather than personal identities, are salient, people tend to behave in a more greedy and less trustworthy way toward members of other groups than if they were reacting to each other as individuals (Insko et al., 2001). Thus, although the functional nature of the relations between groups can further inﬂuence the degree to

Dovidio et al.

67

which discrimination is manifested (Campbell, 1965; Sherif, 1966), the process of social categorization itself provides the basis for social biases to develop and to be maintained. Social categorization is a dynamic process, however, and people possess many diﬀerent group identities and are capable of focusing on diﬀerent social categories (Allport, 1954; Brewer, 2000). By modifying a perceiver’s goals, motives, perceptions of past experiences, expectations, as well as factors within the perceptual ﬁeld and the situational context more broadly, there is opportunity to alter the level of category inclusiveness that will be primary or most inﬂuential in a given situation. This malleability of the level at which impressions are formed is important because of its implications for altering the way people think about members of ingroups and outgroups, and consequently about the nature of intergroup relations. Attempts to combat these biases can therefore be directed at altering the nature of social categorization.

ALTERING SOCIAL CATEGORIZATION Since the mid-20th century, the contact hypothesis (Allport, 1954; see also Dovidio, Gaertner, & Kawakami, 2003; Pettigrew, 1998) has been a prominent strategy for reducing intergroup bias and conﬂict. This hypothesis proposes that simple contact between groups is not necessarily suﬃcient to improve intergroup relations; rather, certain prerequisite features must be present. The characteristics of contact that lead to more positive relationships include equal status between groups, cooperative interaction, opportunities for personal acquaintance between the members (particularly involving nonstereotypic elements), and norms that support interaction both within and outside of the contact situation (Cook, 1985; Pettigrew, 1998). Both laboratory and ﬁeld research have provided substantial documentation of improvement in intergroup relations when these prerequisite criteria have been established (see Pettigrew & Tropp, 2000). Recent research, however, has moved beyond specifying what conditions moderate the reduction of bias to understanding what underlying processes, such as those relating to social categorization, may be involved (see Pettigrew, 1998). From a social categorization perspective, the issue to be addressed is how intergroup contact can be structured to alter inclusive–exclusive collective representations of others. Two of the approaches that have been proposed involve decategorization and recategorization. Decategorization refers to inﬂuencing whether people identify themselves primarily as group members or as distinct individuals (Tajfel & Turner, 1979; see also Brewer, 1988; Brewer & Miller, 1984; Fiske et al., 1999). Recategorization, in contrast, is not designed to reduce or eliminate categorization, but rather to restructure or redeﬁne the nature of group categorization to reduce intergroup bias (Allport, 1954, p. 43; see Gaertner & Dovidio, 2000). In each case, reducing the salience of the original inclusive–exclusive group

68

Recategorization

boundaries is expected to decrease intergroup bias. With decategorization, group boundaries are degraded, inducing members of diﬀerent groups to conceive themselves and others as separate individuals (Wilder, 1981) and encouraging more personalized interactions. When personalized interactions occur, people “attend to information that replaces category identity as the most useful basis for classifying each other” (Brewer & Miller, 1984, p. 288; see also Miller, 2002), and category-based biases are further reduced (see also Miller et al., this volume). With recategorization, as proposed by the common ingroup identity model (Gaertner & Dovidio, 2000; Gaertner, Dovidio, Anastasio, Bachman, & Rust, 1993), the goal is to alter systematically the perception of intergroup boundaries, redeﬁning who is conceived as an ingroup member, to reduce a speciﬁed form of bias. In the common ingroup identity model (see Figure 4.1), we outline speciﬁc potential antecedents and outcomes of direct or symbolic intergroup contact, as well as identify possible mediating social categorization processes. As illustrated in Figure 4.1, we hypothesize that the diﬀerent types of intergroup interdependence and cognitive, perceptual, linguistic, aﬀective, and environmental factors can either independently or in concert alter individuals’ cognitive representations of the aggregate. Perceptions of common ingroup identity can be achieved by increasing the salience of existing common superordinate memberships (e.g., a school, a company, a nation) or by making salient new elements that redeﬁne group relations (e.g., common goals or shared fate). The diﬀerent cognitive representations (i.e., one group, two subgroups within one group, two groups, or separate individuals) are then proposed to produce speciﬁc cognitive, aﬀective, and overt behavioral

Figure 4.1 The common ingroup identity model.

Dovidio et al.

69

consequences (listed on the right). For instance, once others who were originally perceived to be members of another group are recategorized within a shared one-group representation, people will process information about them more deeply and fully, will experience less anxiety with them, and will be more open and prosocial in their behavior toward them. In general, with a common ingroup identity, the cognitive and motivational processes that initially produced ingroup favoritism are redirected to beneﬁt the common ingroup, including former outgroup members, which produces more positive responses to former outgroup members and thereby reduces bias toward them. Allport (1954, 1958) was aware of the beneﬁts of a common ingroup identity, although he regarded it as a catalyst rather than as a mediator of the eﬀect of contact on prejudice reduction: To be maximally eﬀective, contact and acquaintance programs should lead to a sense of equality in social status, should occur in ordinary purposeful pursuits, avoid artiﬁciality, and if possible enjoy the sanction of the community in which they occur. While it may help somewhat to place members of diﬀerent ethnic groups side by side on a job, the gain is greater if these members regard themselves as part of a team. (Allport, 1958, p. 454) We have examined the diﬀerent ways that decategorization and recategorization strategies can reduce bias. In one experiment (Gaertner, Mann, Murrell, & Dovidio, 1989), members of two separate laboratory-formed groups were induced through various structural interventions (e.g., seating arrangement) either to decategorize themselves (i.e., conceive themselves as separate individuals) or to recategorize themselves as one superordinate group. Supporting the proposed value of altering the level of category inclusiveness, both of these changes in the perceptions of intergroup boundaries reduced intergroup bias. Furthermore, as expected, these strategies reduced bias in diﬀerent ways. Decategorizing members of the two groups reduced bias by decreasing the attractiveness of former ingroup members. In contrast, recategorizing ingroup and outgroup members as members of a more inclusive group reduced bias by increasing the attractiveness of the former outgroup members. Consistent with self-categorization theory, “the attractiveness of an individual is not constant, but varies with the ingroup membership” (Turner, 1985, p. 60). Moreover, across a number of studies we have found consistent evidence that many of the key features speciﬁed by the contact hypothesis (e.g., cooperative interdependence, egalitarian norms) operate, at least in part, through the mechanism of transforming people’s representations of others from “us” vs. “them” to a more inclusive “we”. For instance, in a laboratory experiment, Gaertner, Mann, Dovidio, Murrell, and Pomare (1990) directly tested and found strong support for the hypothesis that the relation between

70

Recategorization

intergroup cooperation and enhanced favorable evaluations of outgroup members would be mediated by the extent to which members of both groups perceived themselves as one group. In addition, recent evidence suggests that intergroup interaction of even the most vacuous sort that is not hostile also induces recategorization and consequently reduces intergroup bias (Gaertner, Dovidio, Rust et al., 1999). In other studies we have demonstrated that interventions that successfully establish a common ingroup representation eﬀectively reduce biases not only in attitudes (Gaertner, Dovidio, Nier, Ward, & Banker, 1999) but also in intergroup behaviors, such as helping and self-disclosure (Dovidio et al., 1997). Moreover, the generalizability of this eﬀect has been supported by a series of survey studies conducted in natural settings across very diﬀerent intergroup contexts: bankers experiencing corporate mergers, students in a multiethnic high school, and college students from blended families (see Gaertner, Dovidio, & Bachman, 1996). Although manipulations that successfully produce a common ingroup identity yield consistent improvement in attitudes toward members of other groups and consequent reductions in bias, we caution that, depending on contextual and historical issues, not all interventions designed to create a common identity will work in the intended fashion. As we consider in the next sections, interventions that emphasize a common group identity may, under some conditions, be limited in their eﬀectiveness or even can exacerbate rather than reduce bias.

COMMON GROUPS, DISTINCTIVENESS, AND DIFFERENTIATED IDENTITIES Based on the assumptions of social identity theory (Tajfel & Turner, 1979), introducing interventions that emphasize common group identity or group similarity may threaten the integrity of members’ separate group identities and needs for group distinctiveness under some conditions. Consequently, to re-establish positive and distinctive group identities, members of these diﬀerent groups may maintain relatively high levels of intergroup bias (Brown & Wade, 1987) or even show increased levels of bias (Deschamps & Brown, 1983). This increase in bias may be likely to occur when the diﬀerent group identities and the superordinate group identity are incompatible or when adopting a superordinate identity would require groups to abandon valued, more speciﬁc identities, such as racial or ethnic identity (Dovidio, Gaertner, Niemann, & Snider, 2001). Consistent with this position, Hornsey and Hogg (2000) found that a condition that emphasized students’ common university membership produced even higher levels of bias between humanities and math-science students, groups often competing for resources and status within the institution, than did a condition that emphasized their separate group identities (see also Crisp; Hogg & Hornsey, this volume).

Dovidio et al.

71

Because recategorization and decategorization approaches both rely on reducing or eliminating the salience of intergroup diﬀerentiation, they may involve interventions that can challenge or threaten existing social identities. Under circumstances that are likely to arouse this kind of threat, other approaches may be more eﬀective. The mutual intergroup diﬀerentiation model (Hewstone & Brown, 1986), for instance, emphasizes maintaining original group boundaries but restructuring the nature of the intergroup context within which group members relate. In particular, Hewstone and Brown (1986) recommended introducing cooperative activities in which members of the diﬀerent groups have distinct but complementary roles to contribute toward common goals. In this way, both groups can maintain positive distinctiveness while recognizing and valuing mutual assets and weaknesses within the context of an interdependent cooperative task or common, superordinate goals. Thus, the mutual intergroup diﬀerentiation model does not seek to change the basic category structure of the intergroup contact situation, but instead it aims to change intergroup relations, potentially from perceived incompatibility or threat to resources or identity, to positive interdependence within the intergroup contact context. Evidence in support of this approach comes from the results of an experiment by Brown and Wade (1987) in which work groups composed of students from two diﬀerent faculties engaged in a cooperative eﬀort to produce a twopage magazine article. When the representatives of the two groups were assigned separate roles in the team task (one group working on ﬁgures and layout, the other working on text), the contact experience had a more positive eﬀect on intergroup attitudes than when the two groups were not provided with distinctive roles (see also Deschamps & Brown, 1983). We propose that the common ingroup identity model and the mutual intergroup diﬀerentiation model can be viewed as complementary rather than as competing frameworks. The mutual intergroup diﬀerentiation model identiﬁes conditions that moderate reductions in bias, whereas the common ingroup identity speciﬁes how group representations mediate these reductions in bias. A study by Dovidio, Gaertner, and Validzic (1998) illustrates this point. In this experiment, two three-person groups, composed of participants randomly assigned as overestimators or underestimators, ﬁrst worked separately on a group task about survival after a plane crash in a remote area during the winter. In particular, each three-person group ranked ten items (e.g., cigarette lighter, knife, chocolate bars, parachute) in terms of their survival value for group members. The groups were asked to determine the best solution for ranking the items based on the instructions that, to maximize their chances for survival, they should stay with the plane to be rescued, or that they should abandon the plane and attempt to hike to safety. Two manipulations were introduced: (1) same/diﬀerent dimension and (2) equal/diﬀerent status. In the same dimension condition, each three-person group was informed that the six participants would be asked to ﬁnd the best solution based on the same assumption (stay or hike) that both groups

72

Recategorization

worked under previously. That is, the condition in which intergroup contact occurred involved interaction on a task for which each group had previous but separate experience and developed their own group solutions. In the diﬀerent dimensions condition, the members of the three-person groups were told that the six participants would be asked to come up with the best overall solution considering that the survivors could choose to stay with the plane or hike to safety (the separate alternatives that each group previously discussed). Thus, each group brought unique and complementary experience (a solution based on the assumption that survivors should stay with the crashed plane and a solution based on the assumption that survivors should hike to safety) to address an assigned task in the intergroup situation in which each contribution was important. Status was manipulated by providing the three-person groups with false feedback about their performance and the performance of the other group on the initial problem-solving task. In the unequal status condition, one group was informed that it had scored in the 91st to 93rd percentile compared to all other groups in the quality of its initial solution, whereas the other group was told that it had scored in the 52nd to 54th percentile. In the equal status condition, both groups were informed that they had scored in the 91st to 93rd percentile or in the 52nd to 54th percentile. Supportive of hypotheses based on Hewstone and Brown’s (1986) mutual intergroup diﬀerentiation model, as illustrated in Figure 4.2, interaction between equal status groups reﬂected low levels of intergroup bias only when the groups had distinct areas of experience. The levels of bias in the other three conditions were equivalent and signiﬁcantly higher. In addition, consistent with the processes proposed in the common ingroup identity model, the representation reﬂecting that the six-person aggregate felt like one group

Figure 4.2 The mutual intergroup diﬀerentiation model.

Dovidio et al.

73

was higher in the condition in which groups had equal status but diﬀerent areas of experience than in each of the other three conditions (i.e., equal status-same task, unequal status-diﬀerent task, and unequal status-same task). Furthermore, this one-group representation mediated the relation between the experimental manipulations of both status and task perspective on intergroup bias. An additional study using a diﬀerent paradigm similarly revealed the importance of maintaining group distinctiveness when equal status groups interact cooperatively. In this study, following up on our research with bankers who had actually experienced mergers (see Gaertner et al., 1996), we varied relative group status among actual employees of many diﬀerent companies by asking them to imagine that their current organization was about to merge with another (Mottola, 1996). In one condition, participants’ present company was described as higher in status than the other company in terms generating greater sales and greater proﬁts. In another condition, their company was lower in status on both dimensions. In a third condition, both companies were described as having equal status in terms of both sales and proﬁt. In a fourth condition, their company was described as higher in status on one dimension (e.g., proﬁt), while the other company was higher on the other dimension. Consistent with the hypothesis and the Dovidio et al. (1998) experiment, participants in the fourth condition in which one company had higher proﬁts and the other higher sales anticipated that they would more strongly identify with the merged organization, that it would have higher levels of organizational unity and enjoy greater success, than did participants in each of the other three conditions (which did not diﬀer from one another). Thus, when groups have equal status and each group can maintain positive distinctiveness, we can anticipate greater acceptance of a superordinate identity from the members of both groups and more successful intergroup contact. The mutual intergroup diﬀerentiation model addresses the potential inﬂuence of motivations to achieve positive distinctiveness that might be aroused by decategorization or recategorization interventions by varying the relation between groups to create positive interdependence. Other recategorization approaches, however, do not solely emphasize a single, superordinate identity but instead make multiple identities salient. Such approaches may be similarly eﬀective in reducing bias without arousing the countervailing forces of needs for positive distinctiveness. We consider these two approaches, crossed categorization and the induction of a dual identity, in the next section.

SOCIAL CATEGORIZATION AND MULTIPLE IDENTITIES In the crossed categorization approach, group boundaries are restructured such that the newly established boundaries cross-cut the original group

74

Recategorization

boundaries, redeﬁning who is an ingroup member and who is an outgroup member. The dual identity approach is a particular form of crossed categorization and recategorization in which the original group boundaries are maintained but within a salient superordinate group identity that represents a higher level of inclusiveness. We consider both of these approaches within the context of the common ingroup identity model.

Crossed categorization The crossed categorization strategy for reducing intergroup bias involves crisscrossing category memberships by forming new subgroups or increasing the salience of additional existing subgroups, each composed of members from former groups (Brewer, Ho, Lee, & Miller, 1987; Deschamps & Doise, 1978; Vanbeselaere, 1987). For example, during interaction among political liberals and conservatives attention may be directed to the fact that each group is composed of two men and two women. This type of intervention thus changes the pattern of who’s “in” and who’s “out”, or the degree to which participants are ingroup or outgroup members when both subgroup categories are considered simultaneously. That is, some members are ingroup members on one dimension but outgroup members on the other, while others are ingroup or outgroup members on both dimensions (see reviews by Crisp, Ensari, Hewstone, & Miller, 2003; Migdal, Hewstone, & Mullen, 1998; Urban & Miller, 1998). Crossed categorization strategies have proven to be eﬀective at reducing biases toward members of other groups, relative to the original simple group categorization (Mullen, Migdal, & Hewstone, 2001), across a broad range of situations (see also Crisp & Hewstone; Miller et al.; Singh; Vescio et al., this volume). A key question, from the perspective of the common ingroup identity model, involves how this intervention reduces bias. We speciﬁcally explored the processes associated with crossed categorization in a study (Dovidio & Gaertner, 2003) in which, following the basic procedures employed in our earlier studies (Gaertner et al., 1989, 1990), two groups of three participants worked independently (without knowledge of the existence of the other group) on a problem-solving task. The groups were then brought together under conditions that visually: (1) reinforced the original diﬀerent group memberships by having the two original groups dress diﬀerently, one group wearing blue laboratory coats and the other group wearing white laboratory coats (Vanbeselaere, 1987; Worchel, Axsom, Ferris, Samaha, & Schweitzer, 1978), ostensibly to enhance the image for videorecording (“original categories” condition); (2) made salient the potential for comprising one group by having the six members of both original groups dress the same, all wearing either the white or the blue laboratory coats (“common group” condition); or (3) emphasized cross-cutting group categories through mixed dress, by

Dovidio et al.

75

having a total of three members across both original groups (one from one group and two from the other) wear white coats and three members across both original groups (two from one group and one from the other) wear blue coats (“crossed categorization” condition). Participants then rated the extent to which the six-person aggregate felt most like one group, two diﬀerent groups, two subgroups within a superordinate group, or separate individuals, as well as how much they liked each person in the session. Using the group as the unit of analysis, the measures of one group, two subgroups in one group, diﬀerent groups, and separate individuals representations were the percentages of participants in each group who selected the representation as most strongly characterizing the six-person aggregate. The dependent measures of liking for ingroup and outgroup members were the mean of liking the two other ingroup members (i.e., excluding the possibility of self-ratings) and the three original outgroup members, averaged across the six participants in each session. These ratings could potentially range from one to seven. Bias was computed as the diﬀerence in liking between ingroup and outgroup members (see Gaertner et al., 1989). The means for each of these measures are presented in Table 4.1, with percentages for the representation reﬂecting the average percentage of participants across the session selecting the representation as most characteristic of the six-person group. We predicted that more bias in liking between original ingroup and outgroup members would be obtained in the condition that reinforced original membership through diﬀerent colored laboratory coats than in the common group (i.e., all dressed in white or blue) or crossed categorization (i.e., dressed in colored laboratory coats that cut across original group boundaries) conditions. In addition, we expected that the lower level of bias in the common group condition relative to the original categories condition would, as in previous work, be related to more inclusive group (i.e., stronger one-group and lower diﬀerent-groups) representations. Table 4.1 Group representations and bias as a function of common group (same dress), original categories (diﬀerent-dress), and crossed categorization (mixed-dress) conditions

Common group (same-dress) Original categories (diﬀerent-dress) Crossed categorization (mixed-dress)

One group

Two Diﬀerent Separate subgroups in groups individuals one group

Ingroup liking

Outgroup liking

Bias in liking

61%

25%

6%

8%

5.54

5.25

0.29

44%

29%

21%

6%

5.72

5.06

0.66

64%

15%

15%

5%

5.37

5.17

0.20

76

Recategorization

These measures of representation also allowed us to assess the process by which crossed categorization reduces bias. One possibility is that the crossed categorization manipulation would have its eﬀect through decategorization, rendering the earlier categorization less salient (Brown & Turner, 1979) and reducing the representation of participants as members of two diﬀerent groups and increasing the extent to which participants are seen as separate individuals. For instance, when cross-cutting group memberships appear incompatible or cognitive demands make it diﬃcult for people to consider two dimensions of group memberships simultaneously (Vanbeselaere, 1991), crossed categorization can lead to an abandonment of social categorization as a basis for evaluation, leading to a reliance on decategorized (i.e., separate individuals) representations. A second possibility is that the crossed categorization manipulation would primarily create new salient group boundaries, in which the redeﬁned ingroup is composed of former ingroup and outgroup members. As former outgroup members come to be viewed as part of the ingroup, the forces of ingroup favoritism could enhance liking. To the extent that former ingroup members are recategorized as members of the outgroup, liking for them will tend to diminish. Thus bias based on original group membership could be reduced but replaced by a diﬀerent type of bias based on new group boundaries that are deﬁned by dress. Supportive of previous work (Gaertner & Dovidio, 1986), emphasizing the salience of group memberships visually through dress signiﬁcantly inﬂuenced intergroup bias in liking. As expected, bias was greater in the condition that reinforced original category boundaries, in which original group memberships were maintained by having members of the diﬀerent original groups wear diﬀerent colored laboratory coats during the six-person problem-solving session, than in the common group (all dressed the same) and the crossed categorization (mixed dress) conditions (p < 0.03; see Table 4.1 for means). The levels of bias in the common group and crossed categorization conditions did not diﬀer signiﬁcantly from one another. Thus, compared to the condition that reinforced original category distinctions, the interventions designed to create common group identity and crossedcategory representations were eﬀective at reducing intergroup bias. With regard to the processes underlying the diﬀerence in bias between the common group and the original categories conditions, participants in the common group condition had stronger inclusive group representations and weaker diﬀerent group representations than did those in the original categories condition (see Table 4.1). Moreover, supportive of the common ingroup identity model, stronger one-group representations and weaker diﬀerentgroup representations predicted lower levels of intergroup bias in liking (rs = −0.38 and 0.56) and mediated the eﬀect of the same- (common group) vs. diﬀerent-dress (original categories) manipulation on bias. Taken together, these results conceptually replicate previous research supporting the common ingroup identity model (Gaertner & Dovidio, 1986; Gaertner et al., 1989).

Dovidio et al.

77

In terms of the processes involved in the way that the crossed categorization condition reduced bias relative to the original categories condition, there was little support for the role of decategorization. The two conditions were comparable in the extent to which the six-person aggregate was seen to be composed of separate individuals (p > 0.82), and across these conditions ratings of the separate individuals’ representation did not signiﬁcantly predict level of bias (r = −0.17). Instead, several ﬁndings point to a recategorization process. First, although the strength of the one-group identity was comparable between the common group (same dress) and crossed categorization (mixed dress) condition (61% vs. 64%; see Table 4.1), there tended to be diﬀerences in the extent to which the aggregate was seen as two diﬀerent groups across conditions (p < 0.08). The strength of the diﬀerent-groups representation was comparable in the crossed categorization and the original categories conditions (F < 1), indicating that participants showed similar levels of intergroup diﬀerentiation. The diﬀerent-groups representation was weaker in the common group condition than in these other two conditions (p < 0.05). Thus, participants in the crossed categorization condition did tend to diﬀerentiate people systematically based on group membership. Second, although there was not signiﬁcant ingroup–outgroup bias based on original group classiﬁcations in the crossed categorization condition (Ms = 5.37 vs. 5.17, p = 0.14; see Table 4.1), when the data were reanalyzed and participants were reclassiﬁed based on their group membership based on dress, signiﬁcant ingroup–outgroup bias in liking was observed (Ms = 5.56 vs. 4.98, p = 0.002) that was comparable to bias in the diﬀerent-groups condition. Participants liked those wearing the same color laboratory coat as they signiﬁcantly more than those wearing a diﬀerent color laboratory coat. Thus, even with a backdrop of a level of common group representations that was comparable to that of the common group condition, the manipulation of diﬀerent dress still produced social categorization processes that generated diﬀerent group representations and corresponding intergroup bias. And third, within the crossed categorization condition, weaker one-group representations and stronger diﬀerent-group representations were substantially related to group bias on the basis of dress (rs = 0.57 and −0.45). Thus, the bias associated with original group memberships was not just eliminated; it was replaced by bias based on a diﬀerent dimension. This overall pattern of ﬁndings indicating that crossed categorization can reduce original intergroup bias by shifting bias from one dimension to another is consistent with research, guided by social identity theory, demonstrating more generally that environmental factors can inﬂuence the salience of social identities, which, in turn, can inﬂuence responses to others based on their perceived group membership. People possess multiple social identities and can be categorized on the basis of multiple dimensions (Brewer, 2000). The salience of these identities and dimensions are typically inﬂuenced by context and can vary as a function of interventions, such as our manipulation

78

Recategorization

of the color of laboratory coats that people were wearing. Turner, Oakes, Haslam, and McGarty (1994) argue that “self-categorizing is inherently variable, ﬂuid, and context dependent, as self categories are social comparative and are always relative to a frame of reference” (p. 454). Similarly, perceptions of the group membership of others is also a function of the social context, salient features of groups, and others’ perceived “ﬁt” with social categories (see McGarty; Smith, this volume). Thus, although aﬀective processes may be a factor in how crossed categorization can inﬂuence evaluations of others, as when the positive aﬀect associated with an important ingroup categorization overrides the negative aﬀect associated with a person’s outgroup membership on a less important dimension (Urada & Miller, 2000), our results suggest that shifting bases of social categorization may play a fundamental role. Our ﬁndings and this reasoning parallel recent work by Kessler and Mummendey (2001). In a longitudinal survey study, Kessler and Mummendey found that the strength of respondents’ subgroup identity as “East German” was related to intergroup tensions (support for public protest) within Germany, whereas a shift to a salient superordinate identity as a “German” was unrelated to public protest but did predict negative orientations toward other groups (xenophobia). Nevertheless, to the extent that the immediate problem to be addressed is bias-rooted in a particular type of group membership, crossed categorization strategies and recategorization based on redeﬁnition of group boundaries can eﬀectively be employed to reduce this type of bias, tension, or conﬂict—but not without some costs involving intergroup attitudes toward other groups. Although crossed categorization reduced original group bias in our study through a redirection of bias based on new group boundaries, this does not necessarily imply that other processes are unimportant. For example, it is possible that under other conditions crossed categorization can still produce decategorized representations instead of creating redeﬁned boundaries. In this study, the basis of group membership was visible and direct. It is possible that, under more cognitively demanding conditions and with diﬀerent types of categorizations, people could abandon their reliance on group memberships and treat others on the basis of more individualized, decategorized representations (Vanbeselaere, 1991). These representations could then facilitate more personalized interaction, which could further reduce bias (Bettencourt, Brewer, Rogers-Croak, & Miller, 1992; Marcus-Newhall, Miller, Holtz, & Brewer, 1993; Miller, 2002, see also Miller et al., this volume). With personalized interaction, an individual’s characteristics rather than their group memberships become primarily salient and can thereby undermine the validity of group stereotypes (Brewer & Miller, 1984; Miller, 2002), producing more positive attitudes toward their group as a whole. Even when recategorization is the initial process, other processes may be subsequently involved in complementary ways. Although the more favorable impressions of outgroup members as a function of recategorization within a

Dovidio et al.

79

common ingroup are not likely to be ﬁnely diﬀerentiated, at least initially (see Mullen & Hu, 1989), more elaborated, personalized impressions can soon develop within the context of a common identity because the newly formed positivity bias is likely to encourage more open communication and greater self-disclosing, personalized interaction with former outgroup members (Dovidio et al., 1997). Thus, as hypothesized by Hewstone (1996) and Pettigrew (1998), the nature of intergroup contact can inﬂuence group representations in diﬀerent ways sequentially across time.

Dual identity As we have previously proposed within the context of the common ingroup identity model (Gaertner & Dovidio, 2000), the development of a common ingroup identity does not necessarily require groups to forsake their original identities. People possess multiple identities related to diﬀerent types of social categorization (Brewer, 2000). These social categories are hierarchically organized, with higher-level categories (e.g., nations) being more inclusive of lower-level ones (e.g., cities), and identities at diﬀerent levels of inclusiveness can be salient simultaneously. In this study on dual identity, we focus on a particular type of crossed categorization in which separate group identities are embedded within a more inclusive superordinate identity. This research shows that stronger dual identity representations have sometimes been related to lower levels of bias, as in school settings (Gaertner, Rust, Dovidio, Bachman, & Anastasio, 1996), and have sometimes been associated with higher levels of bias, as in stepfamilies (Banker & Gaertner, 1998) and corporate mergers (Anastasio, Bachman, Gaertner, & Dovidio, 1996). The impact of dual identities on intergroup bias is therefore likely moderated by a number of factors. In this section of the chapter, we consider three: (1) the relative salience of diﬀerent-group and common identities; (2) the statuses of the groups; and (3) the opportunities for intergroup comparisons. Relative salience As we noted earlier, highly salient and important separate group identities can limit the eﬀectiveness of trying to induce a common ingroup identity. Under these circumstances, attempting to increase the salience of a common group identity can threaten people’s sense of collective identity and can exacerbate rather than reduce bias. In one study, we examined how a target person’s emphasis on their common group membership, diﬀerent group memberships, or their dual membership inﬂuenced how exposure to a single group member aﬀected attitudes toward the target person’s racial group as a whole (Dovidio, Gaertner, & Johnson, 1999). Race, the subgroup identity in this experiment, is a basic and important form of social categorization and identity in the United States. In

80

Recategorization

this study, white college students from Colgate University viewed a videotape that portrayed a black student being interviewed on a campus TV show. The presentation of the black student, a confederate, was designed to make a positive impression. After the initial presentation of the confederate, an interviewer asked on the videotape, “And how do you see yourself ?” The response was constructed to reﬂect one of three representations outlined in the common ingroup identity model: (1) “I see myself primarily as a Colgate student” (common group identity); (2) “I see myself primarily as a black person” (diﬀerent group identity); or (3) “I see myself primarily as a black Colgate student [or a Colgate student who is black]” (dual identity). Because this response followed the main introduction to the student, we expected that it would have little eﬀect on the evaluation of and attitudes toward this student. It did not; across all of the conditions, white participants responded to the person equivalently favorably. Our hypothesis, though, was that this information would aﬀect the generalization of this positive response to other members of the group as a whole. Of particular interest was whether the emphasis solely on a common-group identity or a dual identity involving the highly salient diﬀerent-group racial category, would be more eﬀective at improving white students’ attitudes toward blacks as a whole. The results revealed that the emphasis on common group identity was the most eﬀective strategy. Bias, expressed in terms of prejudice toward blacks in general, was lowest in the common-group identity condition, and it was slightly higher in the dual identity than in the diﬀerent-group condition. The salience of diﬀerent social categories and identities is often malleable, however. Thus, the relative eﬀectiveness of interventions that activate both superordinate and subgroup identities may be moderated by factors that alter the relative salience of these identities. For example, Dovidio, Gaertner, Isen, Rust, and Guerra (1998) found that intergroup bias was stronger when separate group and common group identities were made simultaneously salient (i.e., Colgate University liberals, Colgate University conservatives) by using badges visible to participants. In this same research, though, we also found that the introduction of manipulations that produced positive aﬀect, which increased the salience of inclusive categorizations (Isen, Niedenthal, & Cantor, 1992), reversed this eﬀect. That is, whereas people in neutral moods showed more bias in the dual identity than in the diﬀerent-groups condition, people who were experiencing positive aﬀect showed less bias when common and separate group identities were salient than when only separate group identities (liberals, conservatives) were available. Thus, the salience of the common-group component, relative to the diﬀerent-group component, can be a critical moderator of the eﬀectiveness of a dual identity for reducing bias. The salience of a superordinate identity can also potentially moderate the eﬀectiveness of crossed categorization interventions for reducing intergroup bias. Rust (1997), for example, investigated this possibility in a study in which she independently varied (1) whether two diﬀerent groups of college students, ﬁrst-year students (F) and sophomores (S), participated in homogeneous

Dovidio et al.

81

(FF and SS) two-person groups or cross-cut (FS and FS) groups, and (2) whether a superordinate university identity was made salient. Replicating the ﬁndings of Marcus-Newhall et al. (1993), bias toward ingroup and outgroup members present was lower in the cross-cut than in the homogeneous group conditions. Although the manipulation of a superordinate identity did not directly inﬂuence bias toward outgroup members present or outgroup members more generally, it was involved in how participants’ ratings of their experience related to their attitudes toward members of the outgroup. Speciﬁcally, the greater extent to which participants perceived themselves as “crossed subgroups within a superordinate entity”, the more positive were their evaluations of outgroup members present and outgroup members more generally. In contrast, the representation involving crossed subgroups without a superordinate connection did not relate to outgroup attitudes at all. These ﬁndings therefore suggest that the recognition of a common group identity may be a valuable catalyst in reducing bias through crossed categorized interactions. Relative status Another important factor that moderates the eﬀectiveness of a dual identity on bias is the status of the groups with which people identify. Members of majority and minority group members exhibit very diﬀerent perspectives on intergroup relations that can shape perceptions of and reactions to the nature of the contact (Islam & Hewstone, 1993). Whereas minority group members often tend to want to retain their cultural identity, majority group members favor the assimilation of minority groups into one single culture (a traditional “melting pot” orientation)—a process that reaﬃrms and reinforces the values of the dominant culture. Van Oudenhoven, Prins, and Buunk (1998), for instance, found in the Netherlands that Dutch majority group members preferred an assimilation of minority groups (in which minority group identity was abandoned and replaced by identiﬁcation with the dominant Dutch culture), whereas Turkish and Moroccan immigrants most strongly endorsed integration (in which they would retain their own cultural identity while also valuing the dominant Dutch culture). With samples in the United States, we have found that white college students most prefer the one-group, assimilation approach, whereas racial and ethnic minorities favor a “same team”, pluralistic integration model (Dovidio, Gaertner, & Kafati, 2000). Furthermore, we have found that these preferred types of intergroup relations for majority and minority groups (a one-group representation for whites, and a dual-identity representation for people of color) also diﬀerentially mediate the consequences of intergroup contact for the diﬀerent groups (see Dovidio et al., 2000). For white college students, more positive perceptions of intergroup contact related to stronger perceptions of students consisting of one group, diﬀerent groups on the same team (a dual identity), and separate individuals, as well as to weaker perceptions of diﬀerent-group

82

Recategorization

identities. However, when considered simultaneously, only the one-group representation mediated commitment to the university. Conditions of contact also signiﬁcantly predicted each of the representations for students of color, but the mediating representation was diﬀerent. In contrast to the pattern for whites, it was the diﬀerent groups on the same-team representation (dual identity), and not the one-group representation, that mediated commitment to the university. In general, these eﬀects were stronger for people higher in racial ethnic identiﬁcation, both for whites and people of color. Thus, recategorized representations—but diﬀerent ones—were critical mediators of institutional commitment for white students and students of color. Additional analyses, using intergroup attitudes as the outcome measure of interest, rather than commitment to the university, produce a generally similar pattern of results. The one-group representation was the strongest predictor of positive attitudes toward minorities for whites, whereas the same-team representation was the strongest predictor of positive attitudes toward whites among students of color. Complementing these results for white students and students of color, a study by Johnson, Gaertner, and Dovidio (2001) found that, independent of race, status moderated the relationship between a dual identity and bias. Among low- and high-status university students (i.e., regular students and students in the prestigious honors program, respectively) who expected to perform the same tasks within a superordinate workgroup, the relationship between perceptions of the aggregate as two subgroups within a group (a dual identity) and bias depended on the status of the group. For low-status, regular students, higher perceptions of a dual identity predicted signiﬁcantly less bias (beta = −0.31), whereas for higher-status honors students a stronger dual identity (which preserves the status diﬀerential) predicted a greater bias (beta = 0.46). Intergroup comparison According to Mummendey and Wenzel’s (1999) theory of ingroup projection, when subgroups are included within an inclusive superordinate entity, comparisons between these groups become likely. Through these comparisons, members of each subgroup can perceive their subgroup as more typical of the superordinate group. In essence, members of each subgroup “project” their subgroup’s attributes onto the superordinate group. Subsequent social judgments become more biased because members regard their subgroup as superior to the other. As the ﬁndings we have reported suggest, this process may be more likely to occur for high-status than for low-status groups. However, as we have shown, a dual identity does not always lead to higher levels of bias, even when relative status is not a factor (Gaertner, Rust et al., 1996). Another explanation for the diﬀerent eﬀects of a dual identity may therefore involve the relationship between the subgroups and the speciﬁc superordinate identity that is emphasized. Recent studies by Mummendey,

Dovidio et al.

83

Meiser, and Waldzus (2002) and by Riek and Gaertner (2002) indicate that the relevance of the superordinate to the subgroup identity may be critical in determining the eﬀects of dual identiﬁcation. When the superordinate identity is relevant to the subgroups, there is a high degree of cognitive association between the subgroups and the superordinate entity, and the superordinate identity readily provides important dimensions on which the subgroups can be compared. In these studies by Mummendey et al. (2002) and Riek and Gaertner (2002), a relevant superordinate identity (e.g., America for students who identiﬁed with either the Democratic or Republican Party) led to higher ingroup projection, which, in turn, related to higher levels of intergroup bias compared to when the superordinate identities were less relevant to the subgroups (e.g., when the University of Delaware served as the superordinate identity among students who identiﬁed with these political parties).

CONCLUSIONS AND IMPLICATIONS In this chapter we have examined how diﬀerent types of interventions can reduce intergroup bias within the framework of the common ingroup identity model (Gaertner & Dovidio, 2000; Gaertner et al., 1993). Rooted in the social categorization perspective of intergroup behavior (Tajfel & Turner, 1979), the common ingroup identity model recognizes the central role of social categorization and group representations in reducing as well as in creating intergroup bias. The model proposes that if members of diﬀerent groups are induced to conceive themselves more as a single, superordinate group than as two separate groups, attitudes toward former outgroup members will become more positive through processes involving pro-ingroup bias. We have reviewed research that shows how interventions creating a sense of common identity are eﬀective at reducing intergroup bias. However, additional research also demonstrates that it is not always possible, and often not desirable, to induce a common ingroup identity in the absence of other salient identities. We note, however, that the development of a common ingroup identity does not necessarily require groups to forsake their original identities. Recognizing both diﬀerent and common group membership, a more complex form of a common ingroup identity than a simple one-group representation, may also reduce intergroup bias. The focus of the chapter was thus on other techniques that maintain the salience of original group identities but alter the context of group relations (the mutual intergroup diﬀerentiation model); that emphasize common, more inclusive social categorization (a dual identity); or that re-engineer group memberships and identities (crossed categorization). We propose that understanding the factors that moderate the development of inclusive and diﬀerent group identities and understanding how diﬀerent group representations mediate attitudes, orientations, and behaviors toward members of other groups have both theoretical and practical value. Studying

84

Recategorization

how the salience and importance of separate group identities moderate the eﬀectiveness of interventions emphasizing a common group identity contributes to a conceptual understanding of the relationship among multiple possible identities and their translation into thought, aﬀect, and action. Pragmatically, this knowledge can aid practitioners in determining what kind of intervention (one-group identity, crossed categorization, decategorization, or recategorization) is most likely to be eﬀective in a given situation, as well as which approaches are likely to be counterproductive. Moreover, developing a more comprehensive understanding of the mediating mechanisms that underlie how diﬀerent interventions produce changes in attitudes and bias highlights ways that alternative approaches and literatures are conceptually related to one another. For example, as we illustrated (Dovidio, Gaertner, & Validzic, 1998), structuring intergroup contact in the way speciﬁed by the mutual intergroup diﬀerentiation model produced decreases in intergroup bias that were mediated by stronger representations of the memberships as one group, as speciﬁed by the common ingroup identity model. The study of crossed categorization (Dovidio & Gaertner, 2003) also showed how creating new deﬁnitions of group boundaries that cross-cut previous group memberships can reduce the original group bias, again by altering representations of the groups. This focus on group representations also revealed that crossed categorization reduced bias by creating a new, alternative form of bias. Therefore, although diﬀerent approaches may focus on diﬀerent interventions, group representations may play a common mediating role across these diﬀerent contexts and frameworks. The challenge for the future is thus to move beyond demonstrations about which technique is most eﬀective, or even beyond a consideration of which technique is more eﬀective in what circumstances, to consider how these diﬀerent psychological forces relate and can combine to reduce bias maximally. We reported, for instance, that bias was reduced signiﬁcantly for participants in crossed categorized groups relative to those in homogeneous groups, but that among the diﬀerent representations it was only perceptions of crossed categorization within a superordinate identity that predicted lower levels of bias (Rust, 1997). Findings such as these imply the promise of more comprehensive and theoretically integrative work. In conclusion, then, the common ingroup identity model is not a model that only advocates simple categorization in terms of a single group identity. Rather, it is a model that: (1) recognizes the centrality of threat to collective identity for motivation, attitudes, and action; (2) acknowledges that recategorization can take many, alternate forms; (3) notes the importance of multiple social categories and identities in social, as well as in individual, behavior. As the data we have reported illustrate, we believe that the common

Dovidio et al.

85

ingroup identity model can help illuminate the psychological processes underlying crossed categorization eﬀects, and that the literature on crossed categorization eﬀects can oﬀer new perspectives on when and how the development of a common ingroup identity can reduce intergroup bias.

ACKNOWLEDGEMENT Preparation of this chapter was supported by NIMH Grant MH 48721 to the ﬁrst and second authors.

REFERENCES Abrams, D. (1985). Focus of attention in minimal intergroup discrimination. British Journal of Social Psychology, 24, 65–74. Allport, G. W. (1954). The nature of prejudice. Cambridge, MA: Addison-Wesley. Allport, G. W. (1958). The nature of prejudice (abridged). Garden City, NY: Doubleday. Anastasio, P. A., Bachman, B. A., Gaertner, S. L., & Dovidio, J. F. (1996). In R. Spears, P. J. Oakes, N. Ellemers, & S. A. Haslam (Eds.), The social psychology of stereotyping and group life (pp. 236–256). Oxford: Blackwell Publishing. Banker, B. S., & Gaertner, S. L. (1998). Achieving stepfamily harmony: An intergroup-relations approach. Journal of Family Psychology, 12, 310–325. Bettencourt, B. A., Brewer, M. B., Rogers-Croak, M., & Miller, N. (1992). Cooperation and the reduction of intergroup bias: The roles of reward structure and social orientation. Journal of Experimental Social Psychology, 28, 301–319. Brewer, M. B. (1988). A dual process model of impression formation. In T. S. Srull & R. S. Wyer (Eds.), Advances in social cognition: A dual process model of impression formation (Vol. 1, pp. 1–36). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Brewer, M. B. (2000). Reducing prejudice through cross-categorization: Eﬀects of multiple social identities. In S. Oskamp (Ed.), Reducing prejudice and discrimination (pp. 165–183). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Brewer, M. B., Ho, H., Lee, J., & Miller, N. (1987). Social identity and social distance among Hong Kong schoolchildren. Personality and Social Psychology Bulletin, 13, 156–165. Brewer, M. B., & Miller, N. (1984). Beyond the contact hypothesis: Theoretical perspectives on desegregation. In N. Miller & M. B. Brewer (Eds.), Groups in contact: The psychology of desegregation (pp. 281–302). Orlando, FL: Academic Press. Brown, R. J., & Turner, J. C. (1979). The criss-cross categorization eﬀect in intergroup discrimination. British Journal of Social and Clinical Psychology, 18, 371–383. Brown, R. J., & Wade, G. (1987). Superordinate goals and intergroup behavior: The eﬀect of role ambiguity and status on intergroup attitudes and task performance. European Journal of Social Psychology, 17, 131–142. Campbell, D. T. (1965). Ethnocentric and other altruistic motives. In D. Levine (Ed.), Nebraska symposium on motivation (Vol. 13, pp. 283–311). Lincoln, NE: University of Nebraska Press.

86

Recategorization

Cook, S. W. (1985). Experimenting on social issues: The case of school desegregation. American Psychologist, 40, 452–460. Crisp, R. J., Ensari, N., Hewstone, M., & Miller, N. (2002). A dual-route model of crossed categorization eﬀects. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 13, pp. 35–74). Hove, UK & Philadelphia, PA: Psychology Press. Deschamps, J. C., & Brown, R. (1983). Superordinate goals and intergroup conﬂict. British Journal of Social Psychology, 22, 189–195. Deschamps, J. C., & Doise, W. (1978). Crossed-category membership in intergroup relations. In H. Tajfel (Ed.), Diﬀerentiation between social groups (pp. 141–158). London: Academic Press. Dovidio, J. F., & Gaertner, S. L. (2003). Unpublished data, Colgate University, Hamilton, NY. Dovidio, J. F., Gaertner, S. L., Isen, A. M., Rust, M., & Guerra, P. (1998). Positive aﬀect, cognition, and the reduction of intergroup bias. In C. Sedikides, J. Schopler, & C. Insko (Eds.), Intergroup cognition and intergroup behavior (pp. 337–366). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Dovidio, J. F., Gaertner, S. L., & Johnson, J. D. (1999, October). New directions in prejudice and prejudice reduction: The role of cognitive representations and aﬀect. Symposium paper presented at the annual meeting of the Society for Experimental Social Psychology, St. Louis, MO. Dovidio, J. F., Gaertner, S. L., & Kafati, G. (2000). Group identity and intergroup relations: The common in-group identity model. In S. R. Thye, E. J. Lawler, M. W. Macy, & H. A. Walker (Eds.), Advances in group processes (Vol. 17, pp. 1–34). Stamford, CT: JAI Press, Inc. Dovidio, J. F., Gaertner, S. L., & Kawakami, K. (2003). The contact hypothesis: The past, present, and the future. Group Processes and Intergroup Relations, 6, 5–21. Dovidio, J. F., Gaertner, S. L., Niemann, Y. F., & Snider, K. (2001). Racial, ethnic, and cultural diﬀerences in responding to distinctiveness and discrimination on campus: Stigma and common group identity. Journal of Social Issues, 57, 167–188. Dovidio, J. F., Gaertner, S. L., & Validzic, A. (1998). Intergroup bias: Status, diﬀerentiation, and a common in-group identity. Journal of Personality and Social Psychology, 75, 109–120. Dovidio, J. F., Gaertner, S. L., Validzic, A., Matoka, K., Johnson, B., & Frazier, S. (1997). Extending the beneﬁts of re-categorization: Evaluations, self-disclosure and helping. Journal of Experimental Social Psychology, 33, 401–420. Fiske, S. T., Lin, M., & Neuberg, S. L. (1999). The continuum model: Ten years later. In S. Chaiken & Y. Trope (Eds.), Dual process theories in social psychology (pp. 231–254). New York: The Guilford Press. Gaertner, S. L., & Dovidio, J. F. (1986). Prejudice, discrimination, and racism: Problems, progress and promise. In J. F. Dovidio & S. L. Gaertner (Eds.), Prejudice, discrimination, and racism (pp. 315–332). Orlando, FL: Academic Press. Gaertner, S. L., & Dovidio, J. F. (2000). Reducing intergroup bias: The common ingroup identity model. Philadelphia, PA: Psychology Press. Gaertner, S. L., Dovidio, J. F., Anastasio, P. A., Bachman, B. A., & Rust, M. C. (1993). The common ingroup identity model: Recategorization and the reduction of intergroup bias. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 4, pp. 1–26). New York: Wiley International. Gaertner, S. L., Dovidio, J. F., & Bachman, B. A. (1996). Revisiting the contact

Dovidio et al.

87

hypothesis: The induction of a common ingroup identity. International Journal of Intercultural Relations, 20 (3 & 4), 271–290. Gaertner, S. L., Dovidio, J. F., Nier, J., Ward, C., & Banker, B. (1999). Across cultural divides: The value of a superordinate identity. In D. Prentice & D. Miller (Eds.), Cultural divides: Understanding and overcoming group conﬂict (pp. 173–212). New York: Russell Sage Foundation. Gaertner, S. L., Dovidio, J. F., Rust, M. C., Nier, J., Banker, B., Ward, C. M., Mottola, G. R., & Houlette, M. (1999). Reducing intergroup bias: Elements of intergroup cooperation. Journal of Personality and Social Psychology, 76, 388–402. Gaertner, S. L., Mann, J. A., Dovidio, J. F., Murrell, A. J., & Pomare, M. (1990). How does cooperation reduce intergroup bias? Journal of Personality and Social Psychology, 59, 692–704. Gaertner, S. L., Mann, J. A., Murrell, A. J., & Dovidio, J. F. (1989). Reduction of intergroup bias: The beneﬁts of recategorization. Journal of Personality and Social Psychology, 57, 239–249. Gaertner, S. L., Rust, M. C., Dovidio, J. F., Bachman, B. A., & Anastasio, P. A. (1996). The contact hypothesis: The role of a common ingroup identity on reducing intergroup bias among majority and minority group members. In J. L. Nye & A. M. Brower (Eds.), What’s social about social cognition? (pp. 230–260). Newbury Park, CA: Sage Publications. Hewstone, M. (1996). Contact and categorization: Social psychological interventions to change intergroup relations. In C. N. Macrae, C. Stangor, & M. Hewstone (Eds.), Stereotypes and stereotyping (pp. 323–368). New York: The Guilford Press. Hewstone, M., & Brown, R. J. (1986). Contact is not enough: An intergroup perspective on the “contact hypothesis”. In M. Hewstone & R. Brown (Eds.), Contact and conﬂict in intergroup encounters (pp. 1–44). Oxford: Blackwell Publishing. Hogg, M. A., & Hains, S. C. (1996). Intergroup relations and group solidarity: Eﬀects of group identiﬁcation and social beliefs on depersonalized attraction. Journal of Personality and Social Psychology, 70, 295–309. Hornsey, M. J., & Hogg, M. A. (2000). Subgroup relations: A comparison of mutual intergroup diﬀerentiation and common ingroup identity models of prejudice reduction. Personality and Social Psychology Bulletin, 26, 242–256. Howard, J. M., & Rothbart, M. (1980). Social categorization for in-group and out-group behavior. Journal of Personality and Social Psychology, 38, 301–310. Insko, C. A., Schopler, J., Gaertner, L., Wildschut, T., Kozar, R., Pinter, B., Finkel, E. J., Brazil, D. M., Cecil, C. L., & Montoya, M. R. (2001). Interindividualintergroup discontinuity reduction through the anticipation of future interaction. Journal of Personality and Social Psychology, 80, 95–111. Isen, A. M., Niedenthal, P. M., & Cantor, N. (1992). An inﬂuence of positive aﬀect on social categorization. Motivation and Emotion, 16, 65–78. Islam, M. R., & Hewstone, M. (1993). Dimensions of contact as predictors of intergroup anxiety, perceived outgroup variability and outgroup attitude: An integrative model. Personality and Social Psychology Bulletin, 19, 700–710. Johnson, K. M., Gaertner, S. L., & Dovidio, J. F. (2001). The eﬀect of equality of job assignment on ingroup identity and bias for low and high status groups. Unpublished data, Department of Psychology, University of Delaware, Newark, DE. Kessler, T., & Mummendey, A. (2001). Is there any scapegoat around? Determinants of intergroup conﬂict at diﬀerent categorization levels. Journal of Personality and Social Psychology, 81, 1090–1102.

88

Recategorization

Marcus-Newhall, A., Miller, N., Holtz, R., & Brewer, M. B. (1993). Cross-cutting category membership with role assignment: A means of reducing intergroup bias. British Journal of Social Psychology, 32, 125–146. Migdal, M. J., Hewstone, M., & Mullen, B. (1998). The eﬀects of crossed categorization on intergroup evaluations: A meta-analysis. British Journal of Social Psychology, 37, 303–324. Miller, N. (2002). Personalization and the promise of contact theory. Journal of Social Issues, 58, 387–410. Mottola, G. (1996). The eﬀects of relative group status on expectations of merger success. Unpublished Ph.D. dissertation, Department of Psychology, University of Delaware, Newark, DE. Mullen, B., & Hu, L. T. (1989). Perceptions of ingroup and outgroup variability: A meta-analytic integration. Basic and Applied Social Psychology, 10, 233–252. Mullen, B., Migdal, M. J., & Hewstone, M. (2001). Crossed categorization vs. simple categorization and intergroup evaluations: A meta-analysis. European Journal of Social Psychology, 31, 721–736. Mummendey, A., Meiser, T., & Waldzus, S. (2002). On the blessings and perils of a common ingroup for intergroup relations. Unpublished manuscript, University of Jena, Germany. Mummendey, A., & Wenzel, M. (1999). Social discrimination and tolerance in intergroup relations: Reactions to intergroup diﬀerence. Personality and Social Psychology Review, 3, 158–174. Onorato, R. S., & Turner, J. C. (2001). The “I,” “me,” and the “us”: The psychological group and self-concept maintenance and change. In C. Sedikides & M. B. Brewer (Eds.), Individual self, relational self, collective self (pp. 147–170). Philadelphia, PA: Psychology Press. Otten, S., & Moskowitz, G. B. (2000). Evidence for implicit evaluative in-group bias: Aﬀect-based spontaneous trait inference in a minimal group paradigm. Journal of Experimental Social Psychology, 36, 77–89. Park, B., & Rothbart, M. (1982). Perception of out-group homogeneity and levels of social categorization: Memory for the subordinate attributes of in-group and out-group members. Journal of Personality and Social Psychology, 42, 1051–1068. Pettigrew, T. F. (1998). Intergroup contact theory. Annual Review of Psychology, 49, 65–85. Pettigrew, T. F., & Tropp, L. R. (2000). Does intergroup contact reduce prejudice? Recent meta-analytic ﬁndings. In S. Oskamp (Ed.), Reducing prejudice and discrimination (pp. 93–114). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Riek, B. M., & Gaertner, S. L. (2002). Reconciling disparate ﬁndings in dual identiﬁcation: The role of superordinate relevance. Unpublished manuscript, University of Delaware, Newark, DE. Rust, M. C. (1997). Social identity and social categorization. Unpublished Ph.D. dissertation, Department of Psychology, University of Delaware, Newark, DE. Sherif, M. (1966). Group conﬂict and cooperation: Their social psychology. London: Routledge and Kegan Paul. Tajfel, H. (1969). Cognitive aspects of prejudice. Journal of Social Issues, 25 (4), 79–97. 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. 33–48). Monterey, CA: Brooks/Cole.

Dovidio et al.

89

Turner, J. C. (1985). Social categorization and the self-concept: A social cognitive theory of group behavior. In E. J. Lawler (Ed.), Advances in group processes (Vol. 2, pp. 77–122). Greenwich, CT: JAI Press, Inc. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. Oxford: Blackwell Publishing. Turner, J. C., Oakes, P. J., Haslam, S. A., & McGarty, J. C. (1994). Self and collective: Cognition and social context. Personality and Social Psychology Bulletin, 20, 454–463. Urada, D. I., & Miller, N. (2000). The impact of positive mood and category importance on crossed categorization eﬀects. Journal of Personality and Social Psychology, 78, 417–433. Urban, L. M., & Miller, N. (1998). A theoretical analysis of crossed categorization eﬀects: A meta-analysis. Journal of Personality and Social Psychology, 74, 894–908. van Oudenhoven, J. P., Prins, K. S., & Buunk, B. (1998). Attitudes of minority and majority members towards adaptation of immigrants. European Journal of Social Psychology, 28, 995–1013. Vanbeselaere, N. (1987). The eﬀects of dichotomous and crossed social categorization upon intergroup discrimination. European Journal of Social Psychology 17, 143–156. Vanbeselaere, N. (1991). The diﬀerent eﬀects of simple and crossed categorizations: A result of the category diﬀerentiation process or of diﬀerential category salience? In W. Stroebe & M. Hewstone (Eds.), European Review of Social Psychology (Vol. 2, pp. 247–278). Chichester: Wiley International. Verkuyten, M., & Hagendoorn, L. (1998). Prejudice and self-categorization: The variable role of authoritarianism and in-group stereotypes. Personality and Social Psychology Bulletin, 24, 99–110. Wilder, D. A. (1981). Perceiving persons as a group: Categorization and intergroup relations. In D. L. Hamilton (Ed.), Cognitive processes in stereotyping and intergroup behavior (pp. 213–257). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Worchel, S., Axsom, D., Ferris, F., Samaha, C., & Schweitzer, S. (1978). Factors determining the eﬀect of intergroup cooperation on intergroup attraction. Journal of Conﬂict Resolution, 22, 429–439. Worchel, S., Rothgerber, H., Day, E. A., Hart, D., & Butemeyer, J. (1998). Social identity and individual productivity with groups. British Journal of Social Psychology, 37, 389–413.

5

Commitment and categorization in common ingroup contexts Richard J. Crisp

One of the aims of social psychologists interested in group processes is the development of intervention strategies designed to reduce prejudice, improve social inclusion, and create more harmonious intergroup relations (Gaertner & Dovidio, 2000; Pettigrew, 1998). Although the study of multiple social categorization is varied and wide-ranging, part of its appeal is its potential for attaining such goals. The recognition that multiple and cross-cutting social categorizations may reduce intergroup bias has, from early work, been a key motivation (for reviews see Crisp & Hewstone, 1999a; Crisp & Hewstone, this volume). In particular, the common ingroup identity model (Gaertner & Dovidio, 2000; Gaertner, Mann, Dovidio, Murrell, & Pomare, 1990; Gaertner, Mann, Murrell, & Dovidio, 1989) and the crossed categorization model (Crisp, Ensari, Hewstone, & Miller, 2002; Deschamps & Doise, 1978) have developed as potentially viable methods of encouraging more positive intergroup relations. Recent ﬁndings, however, suggest that such interventions do not invariably attenuate intergroup bias, and that there may be important moderating conditions that are yet to be fully speciﬁed. In this chapter a program of research that has focused on one such moderator, social identiﬁcation, is reviewed. I argue that social identiﬁcation is a key determinant of when and whether creating common ingroups can improve intergroup relations and, on the basis of this, suggests an agenda for future research seeking to maximize the bias-reducing potential of multiple categorization.

SOCIAL CATEGORIZATION AND CATEGORY DIFFERENTIATION Why should emphasizing multiple group memberships improve intergroup attitudes? The idea is based on what we know of the fundamental principles that govern intra- and intergroup perception. Sherif’s (1966) seminal summer camp studies illustrated the relative ease with which categorization can lead to ingroup favoritism, and this link has been an enduring aspect of work on intergroup relations. Tajfel’s later work with the minimal group

Crisp

91

paradigm (Tajfel, Flament, Billig, & Bundy, 1971) illustrated how even with meaningless social categories, anonymity, and no prior contact, when “they” are diﬀerent from “us”, then they are also evaluated less positively than us (see Brewer, 1979; Mullen, Brown, & Smith, 1992). The extent to which ingroups and outgroups are diﬀerentiated and distinct determines how positively such groups are evaluated. These ideas have been formalized in various theoretical accounts such as Doise’s (1978) category diﬀerentiation model and selfcategorization theory’s (SCT) (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987) “meta-contrast ratio” (Oakes, Haslam, & Turner, 1994; see also Campbell, 1956; Tajfel, 1959; Tajfel & Wilkes, 1963; see McGarty; Smith, this volume). These accounts all predict a positive correlation between category diﬀerentiation and evaluative discrimination: More intergroup diﬀerentiation means more ingroup favoritism. Multiple categorization oﬀers the potential to weaken category boundaries, increase intergroup similarity, and reduce categorical diﬀerentiation, and in so doing correspondingly reduce intergroup bias.

MODELS OF MULTIPLE CATEGORIZATION AND BIAS REDUCTION Gaertner and Dovidio’s (2000) common ingroup identity model (see also Dovidio et al., this volume) outlines the eﬀects of making a second, superordinate, and shared social categorization salient in the context of an existing ingroup–outgroup dichotomy. The model states that when perceivers in an initially dichotomous intergroup context are encouraged to form a common ingroup identity (i.e., representing members of both in- and outgroups as members of one superordinate category), ingroup favoring bias at the subgroup level will be reduced. The basis of the common ingroup identity model is the notion that category structure is variable. We know, for instance, that cognitive representations of social categories can be heterogeneous and diﬀerentiated or homogeneous and undiﬀerentiated (Park & Judd, 1990). On dichotomous dimensions of group membership the similarity of social categories can also vary (to the extent that the prototypical norms of the two social categories are similar versus dissimilar; e.g., Jetten, Spears, & Manstead, 1998). In the same way as the boundaries between representations of object categories can be clearly deﬁned or “fuzzy” (i.e., share overlapping characteristics with each other; Rosch, 1973), so too can the distinctions between social categories vary. When increased between-group similarity weakens the intergroup boundary, two groups will become perceived as a single superordinate entity. Thus, factors that encourage the perception of greater intergroup similarity help to blur original category boundaries and promote recategorization at a higher, more inclusive, level of categorization. Positive attitudes associated with the ingroup subgroup, following recategorization, can then be extended

92

Commitment and categorization

toward former outgroup members (via social projection, see Cadinu & Rothbart, 1996; Clement & Krueger, 2000). This is because the previous outgroupers have moved closer to the self and can be regarded as new ingroup members by virtue of the inclusive common ingroup (and thus receive the beneﬁts of ingroup aﬃliation; Gaertner, Dovidio, Anastasio, Bachman, & Rust, 1993). Gaertner et al. argue that when the cognitive distinction between the ingroup and outgroup is blurred in this way, ingroup bias, consistent with meta-contrast and social projection, will be attenuated. There is considerable evidence that merging intergroup boundaries can indeed reduce intergroup bias. For instance, integrated seating patterns can increase the perception of a common shared identity and reduce bias relative to segregated seating (Gaertner et al., 1989, 1990, 1993). Wearing the same colored laboratory coats (a superordinate shared identity) during intergroup interaction leads to less bias than wearing the diﬀerent colored coats (Worchel, Axsom, Ferris, Samaha, & Schweitzer, 1978). Positive mood can reduce bias via the formation of an (inclusive) common ingroup representation (Dovidio, Gaertner, Isen, & Lowrance, 1995; Dovidio, Gaertner, Isen, Rust, & Guerra, 1998). Gaertner and Dovidio (Gaertner et al., 1989, 1990) have also established a mediational path to the reduction in bias via the perceived formation of a common ingroup identity, and Marcus-Newhall, Miller, Holtz, and Brewer (1993) found that cooperative interaction can reduce bias via increased intergroup similarity. Work on crossed categorization has also investigated the eﬀects on intergroup evaluations when multiple group memberships are made salient in conjunction with an initial intergroup dichotomy (for reviews see Crisp et al., 2002; Crisp & Hewstone, 1999a; Migdal, Hewstone, & Mullen, 1998; Urban & Miller, 1998; see also Miller et al.; Vescio et al., this volume). Although similar in many ways to the common ingroup identity model, the crossed categorization model also examines what happens when multiple categories reinforce existing boundaries. For example, a white female may not feel quite so diﬀerent from a black female, but may feel considerably more diﬀerent from a black male when gender is salient alongside race. In the latter case the consideration of an additional criterion for classiﬁcation has reinforced the distinctiveness of the initial dichotomy; in the former cases, thinking about additional classiﬁcations has weakened the “us” versus “them” distinction. Deschamps and Doise (1978), on the basis of Doise’s (1978) category differentiation model, predict that crossing categories will reduce intergroup bias via the operation of counteracting cognitive processes. In simple categorization contexts, diﬀerences between, and similarities within, categories are accentuated (Tajfel, 1959; Tajfel & Wilkes, 1963). When two dimensions of categorization are simultaneously salient, and crisscross one another, these processes should work against each other (Deschamps & Doise, 1978). Diﬀerentiation on one dimension will be counteracted by assimilation on a second. For example, race categorization can cause an accentuation of diﬀerences between black and white group members. When, however, members of

Crisp

93

both black and white categories are also female, and gender is salient, then (relative to males), black females and white females will be perceived as more similar (that is, there is an accentuation of similarities within females, relative to males, which includes the membership attributes “black” and “white”). The diﬀerentiation-reducing (vis-à-vis bias-reducing) eﬀects of crossed categorization have been reliably demonstrated many times with respect to both physical (Crisp & Hewstone, 1999b; Deschamps, 1977) and social (Arcuri, 1982; Deschamps & Doise, 1978; van Twuyver & van Knippenberg, 1999) stimuli. Corresponding to this weakening of category distinctiveness, Deschamps and Doise argue that evaluations will also become less diﬀerentiated, and that bias will be reduced. The crossed categorization model provides an extended test of the category diﬀerentiation model in that it tests not only the prediction of reduced bias when boundaries are blurred by shared categorization, but also the corresponding prediction that reinforcing boundaries will increase intergroup bias. The bias-inducing eﬀectiveness of reinforcing category distinctiveness has received as much (if not more) support as the bias-reducing eﬀects in the crossed categorization literature (see Crisp, Hewstone, & Rubin, 2001; Experiment 1; Vanbeselaere, 1996). Crisp and Hewstone’s (1999a) review revealed evidence not only in support of the idea that crossed categorization could reduce ingroup favoritism (e.g., Deschamps & Doise, 1978, Study 2; Vanbeselaere, 1987, 1991), but also that convergent categorization (i.e., where category distinctions reinforce one another) could increase ingroup favoritism (see Diehl, 1990; Marcus-Newhall et al., 1993). In summary there is much evidence that blurring intergroup boundaries via multiple categorization can reduce intergroup bias. Some recent ﬁndings have, however, suggested that reducing categorical diﬀerentiation may not always reduce intergroup bias. Crisp and Hewstone (1999a) noted a number of studies, comparable to those that supported the bias-reduction hypothesis, that failed to oﬀer support for this hypothesis (e.g., Brown & Turner, 1979; Singh, Yeoh, Lim, & Lim, 1997, Studies 1 and 2; Vanbeselaere, 1996; see also Crisp et al., 2001, for mixed evidence). In their meta-analysis Mullen, Migdal, & Hewstone (2001) observed that crossed categorization could sometimes reduce bias, but sometimes it had either no eﬀect or even led to an increase in bias. Vescio et al. (this volume) have also recently observed ﬁndings inconsistent with the view that reducing category diﬀerentiation via crossed categorization can always reduce bias. Similarly a superordinate shared category does not always reduce bias. Dovidio, Gaertner, and Validzic (1998) and González and Brown (2003) observed that totally eroding subgroup distinctiveness was not always the most eﬀective way to reduce bias. Hornsey and Hogg (2000) found in the context of university faculty memberships that, relative to a baseline condition, shared superordinate categorization signiﬁcantly increased intergroup bias. Emphasizing shared, common and cross-cutting group memberships may therefore, in some cases, be either inconsequential for intergroup attitudes or

94

Commitment and categorization

may even increase intergroup bias. While this is counter to the predictions of the prevailing models in the multiple categorization literature, there are, in fact, strong theoretical reasons why such an outcome might be expected, given speciﬁc antecedent conditions. In particular, we may expect diverging eﬀects of multiple categorization when elements of the intergroup context trigger motivational processes associated with social identiﬁcation.

CATEGORIZATION AND MOTIVATION Work from the social identity perspective (Tajfel, 1982; Tajfel & Turner, 1979; Turner, 1982; Turner et al., 1987) argues that individuals are motivated to achieve a positive and distinct self-image. According to this perspective we are not just incidentally members of diﬀerent social categories, passively noting our multiple memberships. Rather, we are psychologically invested in some groups more than others, and as such how we deﬁne and feel about ourselves is intricately and intimately bound up with such group memberships. Individuals, by virtue of their aﬃliation, will have higher self-esteem when their ingroup is evaluated positively (e.g., Abrams & Hogg, 1998; Brown, 1984; Brown & Abrams, 1986). The self-evaluative beneﬁts of being a member of a positively valued group will thus motivate comparisons between the ingroup and outgroup that favor the former (e.g., for a review see Ellemers, Spears, & Doosje, 2002). Distinctive social categorization also clariﬁes and deﬁnes social situations, providing a means for predicting how outgroupers will behave, and providing a set of prescriptive ingroup norms to guide perceivers’ behavior. It is not just a desire for positivity that drives group members, but also a desire for distinctiveness. Group members are motivated to maintain diﬀerentiation from other groups in order to reduce subjective uncertainty in social relations (Hogg & Mullin, 1999; Mullin & Hogg, 2000). Hogg (2000, 2001) has illustrated how social categorization and positive diﬀerentiation processes are accentuated in conditions that promote uncertainty in the social context (fuzzy intergroup boundaries being one such context). The social identity perspective thus posits a motivational drive toward diﬀerentiation and distinctiveness—a drive that can manifest itself in the form of increased evaluative diﬀerentiation (ingroup favoritism). There is extensive evidence that increased similarity (weakened intercategory distinctiveness) can, indeed, provide the motivation for intergroup discrimination. Jetten, Spears, and Manstead (2001) observed a negative and linear relationship between distinctiveness and bias (see also Jetten et al., 1998). Similarly, Henderson-King, Henderson-King, Zhermer, Posokhova, and Chiker (1997) found that perceived intergroup similarities led to greater bias when the outgroup was perceived as a threat to a valued ingroup identity. In a school setting, Roccas and Schwartz (1993) found that children exhibited increased bias when distinctiveness was threatened (for a review of collected work on distinctiveness and discrimination, see Jetten & Spears, 2004). This

Crisp

95

perspective thus provides predictions that diverge considerably from existing cognitive models of multiple categorization and as such implies that there must be moderating conditions under which motivational drives will or will not have an impact on intergroup evaluations. Research suggests that this moderator is ingroup identiﬁcation.

INGROUP IDENTIFICATION Higher ingroup identiﬁcation, contextually or chonically deﬁned, appears to be the condition under which motivations will cause a negative reaction to blurred intergroup boundaries (Spears, Doosje, & Ellemers, 1997; see also Jetten, Spears, & Manstead, 1997). Higher identiﬁers (i.e., group members who feel strongly committed to a salient ingroup) appear to be motivated to diﬀerentiate their ingroup from similar outgroups, on relevant dimensions of comparison, in order to restore or maintain group distinctiveness and a positive social identity. For instance, British people can vary according to how much they feel that being British is an important part of their self-concept (illustrated in recent heated debates over whether Britain should enter the European single currency, giving up the national currency, sterling, in favor of the Euro). One way to achieve such positive distinctiveness is to be ingroup-favoring (Branscombe, Wann, Noel, & Coleman, 1993; Brown & Abrams, 1986; Doosje, Ellemers, & Spears, 1995; Hinkle & Brown, 1990; Hodson, Dovidio, & Esses, 2003; Hogg & Abrams, 1990; Jetten et al., 2001). There is consistent evidence that the importance that perceivers ascribe to group membership (how committed they are to it) is a critical predictor of whether increases in similarity will motivate a reactive increase in bias (for reviews see Jetten & Spears, 2004; Spears, Jetten, & Scheepers, 2002).

IDENTIFICATION AND MULTIPLE CATEGORIZATION Although the theoretical basis for higher identiﬁcation leading to negative reactions to intergroup similarity is well established, there have been few empirical tests of this idea in the bias-reduction literature. Surveying multiple categorization work, the applicability of the motivational model does appear justiﬁed. Bias-reduction predictions have commonly been conﬁrmed when using artiﬁcially created laboratory groups, groups that would not normally engender higher levels of commitment from their members. Observations of reduced bias following cross-cutting categorization have been provided by Vanbeselaere (1987, 1991, 1996) and Marcus-Newhall et al. (1993), who used ad hoc categorizations. Crisp et al. (2001) and Diehl (1990) also found that bias was lower when categories diverged than when they converged, and again they used artiﬁcial groups. Although there are also many real group studies where multiple categorization methods have successfully reduced bias, it is

96

Commitment and categorization

only when groups are real that some deviations from the basic bias-reduction predictions have been observed (e.g., Hornsey & Hogg, 2000; Singh et al., 1997). From the above discussion it is apparent that social identity motivations may provide an explanation as to why, as Hornsey and Hogg (2000) observed, shared, cross-cutting, or superordinate categorization may sometimes accentuate, rather than attenuate, intergroup bias. An assessment of individuals’ idiosyncratic identiﬁcation with, or commitment to, ingroups might therefore not only oﬀer a means for predicting when increases in bias might be expected, but also support the notion that social identity motivations have a part to play in understanding multiple categorization phenomena. This was the basis for a program of work carried out with colleagues testing the idea that commitment to ingroups may be an important determinant of the success of multiple categorization interventions. Below, this work is reviewed, along with further studies that have tested methods and means of avoiding the “problem” of higher identiﬁcation when boundaries are blurred.

PREDICTING THREATS FROM MULTIPLE CATEGORIZATION In a recent set of studies van Leeuwen, van Knippenberg, and Ellemers (2003) provided evidence that weakening intergroup boundaries could have divergent eﬀects for higher and low identiﬁers. Using minimal groups and a paradigm designed to test the eﬀects on group perception of a merger between the ingroup and outgroup (a design that modeled the real-life context of corporate mergers), van Leeuwen et al. found that pre-merger identiﬁcation predicted post-merger intergroup attitudes. Speciﬁcally, prior to the merger, identiﬁcation did not predict intergroup evaluations. Following the merger, however, identiﬁcation was positively correlated with ingroup favoritism. The higher the level of pre-merger identiﬁcation, the higher the post-merger ingroup favoritism. A similar program of work was carried out with colleagues, but focused on the implications of ingroup identiﬁcation for the eﬀectiveness of intervention strategies to reduce intergroup bias. Supporting the ﬁndings of van Leeuwen et al. (2003), Crisp, Stone, and Hall (2006) consistently found identiﬁcation to moderate the eﬀects of making a superordinate context salient. For instance, we found that compared to participants who were simply asked to evaluate the British and the French, when the experimental group were ﬁrst asked to read a paragraph advocating closer European integration and the dissolution of member states into a “United States of Europe” (making a common ingroup identity salient and threatening the distinctiveness of being British), bias was not reduced, and in fact increased for participants who were higher identiﬁers with their British category membership; see Figure 5.1. Employing other multiple categorization paradigms, we have also observed

Crisp

97

Figure 5.1 Ingroup favoritism as a function of categorization and identiﬁcation (adapted from Crisp et al., 2006; Experiment 2). Note: for illustrative purposes, means are presented using a median split on the identiﬁcation scale.

a moderating eﬀect of identiﬁcation. Bloom and Crisp (2004) asked participants to generate multiple cross-cutting or convergent categories. Thinking about cross-cutting categories should emphasize intergroup similarity, while thinking about converging categories should reinforce diﬀerences and increase intergroup diﬀerentiation. Doise’s (1978) category diﬀerentiation model would predict that emphasizing cross-cutting categories would reduce intergroup bias, while convergent categorization would increase intergroup bias. On the basis of the motivational predictions associated with social identity theory, we would, however, expect divergent consequences of crosscutting and convergent categorization. When perceivers are higher ingroup identiﬁers, then the pattern predicted by the category diﬀerentiation model should be reversed: cross-cutting categorization would represent a threat to distinctiveness and therefore lead to a reactive increase in ingroup favoritism. In contrast, convergent categorization would be particularly welcomed by higher identiﬁers as it would strengthen the intergroup boundary and maintain the distinctiveness of their valued ingroup. As such, higher identiﬁers generating convergent categories would be least ingroup-favoring. In this experiment the target groups were “University of Birmingham (ingroup)” and “Aston University (outgroup)” students, the latter being a local rival university and obvious source of social comparison for our participants. Identiﬁcation was manipulated by way of false feedback on a multiple choice task involving trivia questions about the ingroup university (e.g., “What year was the University of Birmingham founded?” with ﬁve options). Following false feedback informing participants that they were

98

Commitment and categorization

either lower or higher identiﬁers with their university student group, participants completed the cross-cutting or convergent categorization task, which involved participants ﬁrst generating, and then elaborating on, categories that either cross-cut or reinforced the Birmingham–Aston intergroup distinction. Following this, participants completed a measure of intergroup evaluation (a Likert scale rating of team member preference in a supposed future problem-solving task). The ﬁndings supported the predictions. Participants who had received feedback that they were lower identiﬁers were less biased following generation of shared (cross-cutting) categories compared to nonshared (converging) categories. In contrast, participants who had received feedback that they were higher identiﬁers were more biased following generation of cross-cutting categories compared to generation of convergent categories; see Table 5.1. Using a similar paradigm, Crisp and Beck (2005) asked participants to think about multiple characteristics that were shared between the ingroup and the outgroup as a means of reducing categorical diﬀerentiation. Participants thought about characteristics that students at their own and a local rival university shared; for example, being studious, sporty, enthusiastic, etc. In line with expectations, while identiﬁcation was unrelated to bias under control conditions, following completion of the shared characteristics task identiﬁcation was positively correlated with bias. Conﬁrming the ﬁndings described above, while lower identiﬁers were signiﬁcantly less biased after completing the shared characteristics task than control 5, higher identiﬁers were more biased after completing the shared characteristics task. The studies reviewed above support the idea that ingroup identiﬁcation can moderate the eﬀectiveness of diﬀerentiation-reducing interventions on intergroup attitudes. Using varied multiple classiﬁcation paradigms, and Table 5.1 Evaluation as a function of cross categorization and identiﬁcation (adapted from Bloom & Crisp, 2004) Categorization Cross-cutting

Convergent

Identiﬁcation

M

SD

Low Ingroup Outgroup Bias

7.15 6.31 0.846

1.21 1.84 1.21

7.85 6.38 1.46*

1.07 1.94 2.33

High Ingroup Outgroup Bias

7.92 6.62 1.31*

1.19 1.50 1.44

6.85 7.15 −0.308

1.68 1.28 1.70

Note: * indicates signiﬁcant intergroup bias at p < 0.05.

M

SD

Crisp

99

a range of target groups and contexts, we and others have consistently found that whereas lower identiﬁers respond positively to reduced intergroup diﬀerentiation, higher identiﬁers appear to react against the implied merging of boundaries and become more ingroup-favoring. Having established higher identiﬁers as the group most likely to react negatively to blurring intergroup boundaries, it becomes important to test for conditions under which such reactivity may be avoided. In the next section, we review work that has begun to develop modiﬁcations to existing diﬀerentiationreducing strategies as a means to improve and expand their potential for reducing intergroup bias.

PREVENTING THREATS FROM MULTIPLE CATEGORIZATION Since identiﬁcation with the subgroup category appears to inﬂuence ingroup favoritism in multiple category contexts, so too should identiﬁcation with the cross-cutting or superordinate group. Crisp, Beck, and Hewstone (2004) found in their ﬁrst experiment that making a cross-cutting categorization salient upon an existing intergroup dichotomy led to an increase in intergroup bias (similar to eﬀects observed by Dovidio et al., 1998; Hornsey & Hogg, 2000; see also Hogg & Hornsey, this volume). A post-test revealed that participants in this experiment identiﬁed to a signiﬁcantly higher level with the subgroup (university aﬃliation: Birmingham or Aston University) than the superordinate group (major studied: science or humanities). In two subsequent experiments we reversed the subgroup and superordinate groups such that participants identiﬁed to a greater extent with the superordinate group (now university aﬃliation) and to a lesser extent with the subgroup (major studied). Now bias was reduced after introduction of the shared categorization, and was mediated by increased intergroup similarity (measured by Gaertner et al.’s, 1989, 1990, measures of intercategory structure). These ﬁndings suggested that while higher identiﬁcation with a subgroup may lead to the type of reactive increases in bias reviewed above, simultaneous higher identiﬁcation with the superordinate or cross-cutting category could counteract this tendency. Stone and Crisp (2005) took this one step further and simultaneously measured identiﬁcation with the subgroup and superordinate group prior to a recategorization manipulation. We found that while recategorization (European) once again increased bias for higher subgroup (British) identiﬁers, this reactivity was attenuated when participants were simultaneously higher superordinate (European) identiﬁers. In other words, while the previous work had shown that higher subgroup identiﬁers were more biased following a recategorization manipulation, this eﬀect was limited to only those participants who were also lower identiﬁers with the cross-cutting superordinate group.

100

Commitment and categorization

These ﬁndings suggest that for many perceivers cross-cutting and superordinate categorization can be eﬀective at reducing integroup bias. If participants are lower identiﬁers at the subgroup level, or if they are higher subgroup identiﬁers but also higher identiﬁers with the cross-cutting or superordinate group, then intergroup evaluations can improve. Although this is promising, one issue is that higher subgroup identiﬁcation may, by deﬁnition, preclude higher superordinate group identiﬁcation. Higher British identiﬁers may, for instance, see European categorization as an inherent threat to “Britishness”, which is unlikely to induce higher European identiﬁcation. As such the number of higher subgroup identiﬁers who are also higher superordinate group identiﬁers might be relatively low. Other, more direct means of encouraging higher subgroup identiﬁers to adopt superordinate categories might therefore be warranted. The theoretical basis for expecting higher subgroup identiﬁers to resist recategorization is the implied loss of subgroup distinctiveness associated with such interventions (which is valued by higher, but not lower, identiﬁers). If this is the case, then it may be possible to avoid negative reactions to recategorization by preserving, or even encouraging, intergroup diﬀerentiation. Some previous work has established that preserving distinctiveness can be an important element in the reduction of intergroup bias. Indeed in Gaertner et al.’s (1993) original discussions of the common ingroup identity model, the authors argued that with real groups perceivers may be reluctant to give up their subgroup identities in favor of a superordinate entity (Gaertner & Dovidio, 2000; see also Hewstone, 1996; Hewstone & Brown, 1986, for similar warnings not to forsake subgroup identiﬁcation in the contact hypothesis literature). Dovidio et al. (1998) and González and Brown (2003) have both shown that introducing a superordinate categorization while simultaneously preserving the salience of the subgroup is more eﬀective than recategorization alone at reducing intergroup bias. Similarly, Hornsey and Hogg (2000) observed that relative to increases in bias following recategorization, simultaneous categorization can reduce intergroup bias. Although these previous studies did not measure subgroup identiﬁcation, on the basis of the work reviewed above we might suggest that simultaneous categorization will work best speciﬁcally for higher identiﬁers because it is these participants for whom superordinate categorization is a threat. Crisp et al. (2006, Experiment 4) replicated Hornsey and Hogg’s observation of increased bias following recategorization and decreased bias following simultaneous categorization, but also found that subgroup identiﬁcation qualiﬁed this eﬀect. The overall increase in bias observed in the recategorization condition was driven exclusively by higher subgroup identiﬁers, while the overall decrease in bias following simultaneous categorization was the consequence of decreased bias for higher identiﬁers (see Figure 5.2). Simultaneous categorization may thus be a way of reducing intergroup bias even for higher subgroup identiﬁers. Although simultaneous categorization appears to be successful for both

Crisp

101

Figure 5.2 Ingroup favoritism as a function of categorization and identiﬁcation (adapted from Crisp et al., 2006; Experiment 4). Note: for illustrative purposes, means are presented using a median split on the identiﬁcation scale.

lower and higher identiﬁers, there are some good theoretical reasons to remain a little tentative. It is clear why simultaneously preserving subgroup distinctiveness in the context of superordinate categorization reduces bias for higher identiﬁers. For lower identiﬁers, however, there is much evidence that diﬀerentiation is directly and positively correlated with ingroup favoritism. For lower identiﬁers, recategorization alone and the complete dissolution of category boundaries have been consistently eﬀective at reducing bias. Simultaneous categorization preserves subgroup boundaries, but if for lower identiﬁers the salience of the subgroup boundary is directly related to evaluative diﬀerentiation, then this approach will be less eﬀective for such perceivers. That is to say, for lower, but not higher, identiﬁers simultaneous categorization may be less eﬀective than strategies that aim to more thoroughly blend intergroup boundaries. An additional interesting speculation is that in some contexts, for instance where groups are poorly diﬀerentiated in the ﬁrst instance, heightened diﬀerentiation alone (i.e., with no implied shared categorization) may be even more eﬀective than simultaneous categorization. Bloom and Crisp (2004) found that for higher identiﬁers thinking about cross-cutting categories led to signiﬁcant ingroup favoritism, but this became nonsigniﬁcant when converging categories were generated. Here, then, increasing diﬀerentiation, with no implied common identity or similarity at all, was maximally eﬀective at reducing bias; see Table 5.1. This raises the intriguing possibility that in some speciﬁc situations actually aiming to increase diﬀerentiation may be an eﬀective strategy for reducing bias.

102

Commitment and categorization

A ﬁnal means of reducing intergroup bias, potentially even for higher identiﬁers, has recently been tested by Hall and Crisp (2005). The intervention was developed from Crisp et al. (2001) and consisted of participants generating unrelated versus related alternative categories to an initial target dichotomy. Unrelated categories are those that are uncorrelated with each other (e.g., gender and race). Related categories, in contrast, overlap with each other (e.g., race and socioeconomic status; see also Eurich-Fulcer & Schoﬁeld, 1995, for similar ideas). Hall and Crisp found that thinking about alternatives that were unrelated to the target dichotomy reduced intergroup bias, but that thinking about alternatives that were related to the target dichotomy did not. Superordinate categories may be regarded as “related alternatives” because by deﬁnition the inclusion of a subgroup within a superordinate group implies some representational overlap (e.g., British and European group membership). As such, they should be susceptible to the identiﬁcation-moderated responses discussed in this chapter. In contrast, when thinking of multiple categories that are alternatives to a target dichotomy, and which are unrelated (i.e., share nothing in common such as gender and race), the complexity of the category context may lead to decategorization, that is, a shift in mode of processing from categorical to individuated perception (see Brewer & Miller, 1984; Crisp et al., 2001; Miller, Brewer, & Edwards, 1985). Decategorized perception should focus on individuating characteristics, and, as such, group-deﬁned moderators, such as identiﬁcation, may no longer hold any predictive power. Related work supports this idea. Hutter and Crisp (2005) have recently observed evidence of increased individuation, and decreased stereotyping, when category combinations are surprising (i.e., share few stereotypic attributes in common such as “females” and “mechanics”), compared to unsurprising combinations of categories that have overlapping characteristics (e.g., “females” and “nurses”). Thinking about the former, more complex, type of category combination appears to require more cognitive resources, and results in a more detailed impression of the target person being formed (Hutter, 2005). Roccas and Brewer’s (2002) model of social identity complexity similarly emphasizes the importance of overlap and diﬀerentiation between multiple ingroup identities for predicting intergroup attitudes. If multiple categorization can therefore shift perception to the interpersonal level, it may then be that group-level moderators such as ingroup identiﬁcation may become less relevant. These are tentative ideas but initial evidence is promising that creating category complexity may constitute an alternative means to avoid identiﬁcation-related reactivity to blurred intergroup boundaries.

IMPLICATIONS AND FUTURE WORK An extensive literature has developed around the notion that emphasizing shared, superordinate or cross-cutting categorizations can reduce intergroup

Crisp

103

bias. While the extant multiple categorization literature oﬀers some support for the bias-reduction hypothesis, a number of studies have reported diﬀerent outcomes, including heightened ingroup favoritism. We have argued that diﬀerences in perceivers’ level of ingroup identiﬁcation can provide an explanation for why sometimes cross-cutting or superordinate categorization can lead to such divergent eﬀects. The research reviewed above supports this argument, and has found consistent evidence that the eﬀects of shared categorization on intergroup evaluations do vary as a function of ingroup identiﬁcation. We discuss below the theoretical and practical implications of this research for the development of strategies to reduce intergroup bias.

Theoretical implications Work into multiple categorization has tended to make little reference to the motivational processes that are understood to play a major part in social identiﬁcation and intergroup perception. In the research reviewed above it was consistently found that cross-cutting or superordinate categorization can lead to increased bias for participants who report higher levels of commitment with their ingroup, and this supports the notion of a key role for identity-related motivational processes (see Dovidio et al., 1998; González & Brown, 2003; Hornsey & Hogg, 2000). In line with the social identity perspective, weakening the intergroup boundary via consideration of shared categorizations reduced bias for lower identiﬁers, but increased bias, we argue, due to a perceived threat to positive distinctiveness for higher identiﬁers. It is important to note that the argument presented here does not dispute the operation of category diﬀerentiation processes, nor the meta-contrast related predictions from self-categorization theory. On the contrary, it is clear that the category diﬀerentiation model describes the basic cognitive eﬀects of multiple categorization (see Crisp & Hewstone; Vescio et al., this volume). Indeed, the ﬁndings that we observed for lower identiﬁers are precisely what are predicted according to the cognitive models. When social categories are less important to perceivers, then the weakening of category boundaries does appear to be reﬂected in evaluations. The work reviewed in this chapter instead argues for a qualifying process in some contexts based on perceivers’ motivations to maintain a positive distinction between important social categories. Notably, for higher identiﬁers, we have observed in many cases a signiﬁcant increase in bias, rather than just no eﬀect of shared categorization relative to lower identiﬁers. Had we consistently observed only that shared categorization was inconsequential for higher identiﬁers, then this in itself might not necessarily indicate a motivational process. It might simply be that higher identiﬁers are more practiced at thinking about ingroups and outgroups in comparative ways, and as such have developed a more solidiﬁed

104

Commitment and categorization

representation of intergroup diﬀerences that are more diﬃcult to modify by the type of manipulations used above. If higher identiﬁers were simply resistant to shared categorization, then this possibility could not be ruled out as an alternative to the motivational explanation. We did, however, observe signiﬁcant increases in bias for higher identiﬁers in many cases. This cannot be explained by the idea that higher identiﬁers simply have a ﬁrmer and more established representation of intergroup diﬀerences. The increase in bias must indicate a reaction to the shared category manipulation, and as such, this supports the notion that the eﬀects are due to motivational processes. Further work is needed to determine more precisely the mediating processes involved (for instance, measures of self-esteem or uncertainty motivations), but these ﬁndings are suggestive of the potential for an exciting uniﬁcation of cognitive and motivational theorizing within the context of improving intergroup relations.

Practical recommendations These ﬁndings have some important implications for eﬀorts to apply socialpsychological theorizing to work on intergroup bias and bias reduction. Many approaches developed to improve intergroup relations rest on the assumption that similarity, and often shared categorization (on some level), is the key to intergroup harmony. Contact (Brewer & Miller, 1984; Wright, Aron, McLaughlin-Volpe, & Ropp, 1997) and cooperation can lead to the formation of a common ingroup identity (Gaertner & Dovidio, 2000; Gaertner et al., 1993; Gaertner, Dovidio, Rust, Nier, Banker, Ward, Mottola, & Houlette, 1999)—a shared superordinate categorization. The ﬁndings reviewed above might suggest a modiﬁcation of this assumption. For lower identiﬁers similarity does appear to reduce intergroup bias. When identities are valued, however, we have found that emphasizing shared categorization can increase ingroup favoritism. How then can we avoid negative reactions from higher identiﬁers? One option may be to use a decategorization strategy, such as encouraging thought about alternative bases for categorization that are uncorrelated with the target distinction (Hall & Crisp, 2005). Although shifting perception from a category-based to individual-based mode of processing may militate against any negative inﬂuence of identiﬁcation (which is a category-based moderator), this intervention has a limitation. By shifting focus to individuating characteristics, it works well for one-to-one interactions, but not so well for changing broader beliefs about social groups. A strategy is needed that can avoid the eﬀects of higher identiﬁcation at the categorical level of perception. Some previous work has shown that protecting any loss of distinctiveness by maintaining subgroup salience within a superordinate group can reduce bias when recategorization alone does not (e.g., Crisp et al., 2006; Hornsey & Hogg, 2000). Others have shown that simultaneous categorization can, overall, be a more eﬀective strategy for bias reduction than recategorization

Crisp

105

(Dovidio et al., 1998; González & Brown, 2003). Maintaining the salience of original identities within a new superordinate common ingroup (a simultaneous subordinate/superordinate categorization) may avoid any identitythreatening consequences to subgroup members. If the driving force behind the negative reactions to shared categorization is a reluctance to forsake subgroup identities, then retaining the salience of those identities within a superordinate structure should obviate any such motivated perseverance of positive diﬀerentiation. While simultaneous categorization oﬀers a means for reducing bias for higher identiﬁers, there is, however, the possibility that it might be less eﬀective than recategorization alone for lower identiﬁers (our data suggest that lower identiﬁers readily accept the merging of subgroup identities and loss of distinctiveness implied by recategorization). One intriguing possibility is that recategorization may be the optimal strategy for lower identiﬁers, with simultaneous categorization the most eﬀective strategy for higher identiﬁers. Indeed, it is possible that diﬀerent approaches are needed for perceivers with diﬀering levels of psychological investment in their groups. Of course, how to practically target diﬀerent interventions to diﬀerent people presents something of a problem, and at present simultaneous categorization appears to oﬀer the best all-round solution. Recategorization may be more eﬀective for lower identiﬁers but it can cause increases in bias for higher identiﬁers. Simultaneous categorization, in contrast, may be less eﬀective for lower identiﬁers, but it would not increase bias, and it oﬀers a way to reduce bias for higher identiﬁers. How policy and practice will adapt to develop interventions that are optimally eﬀective across the range of people with diﬀering drives and motivations will certainly be a challenge. González and Brown (2003) argue that when we apply psychological interventions to real social contexts, a combination of diﬀerent approaches to suit the assessed psychological characteristics of the populations concerned is likely to be needed. The ﬁndings reviewed in this chapter appear to support this sentiment.

CONCLUSION The aim of this chapter was to outline a motivational model of multiple categorization. On a broad level, the reviewed ﬁndings call for an increased consideration of individual perceiver characteristics in application of models of bias reduction. The ﬁndings reported reinforce the notion that the same intervention can mean diﬀerent things to diﬀerent people. Optimal interventions to reduce intergroup bias should thus be targeted and tailored to particular populations. I argue that consideration of this perspective for work in multiple categorization is an important development, not least theoretically, but also with respect to its applied potential to inform policy and practice as they pertain to conciliation and intervention strategies to improve intergroup relations.

106

Commitment and categorization

Continued incorporation of extant social-psychological models of intergroup processes into models of multiple categorization will undoubtedly help the development of integrated, or targeted, and ﬂexible strategies for bias reduction and reconciliation in contexts of intergroup conﬂict. In a social world that appears to increasingly involve multiple and cross-cutting intergroup aﬃliations, there may be clear applicability of specifying a link between theoretical perspectives in intergroup relations, social identity, and social cognition to contexts involving multiple social classiﬁcation. Such an approach may provide a basis upon which to develop new models of bias reduction that explicitly draw on established links between distinctiveness and diﬀerentiation to promote, improve, and enhance intergroup relations.

ACKNOWLEDGEMENTS Preparation of this chapter was supported by British Academy (SG-31041 and SG-34045), and Leverhulme Trust (F/00094/H) grants to R. J. Crisp, and an Economic and Social Research Council grant (R000239382) to R. J. Crisp and M. Hewstone.

REFERENCES Abrams, D., & Hogg, M. A. (1998). Prospects for research in group processes and intergroup relations. Group Processes and Intergroup Relations, 1, 7–20. Arcuri, L. (1982). Three patterns of social categorization in attribution memory. European Journal of Social Psychology, 12, 271–282. Bloom, Z., & Crisp, R. J. (2004). Social identiﬁcation and intergroup bias in multiple membership contexts. Unpublished dataset, University of Birmingham. Branscombe, N. R., Wann, D. L., Noel, J. G., & Coleman, J. (1993). Ingroup or outgroup extremity: Importance of the threatened social identity. Personality and Social Psychology Bulletin, 19, 381–388. Brewer, M. B. (1979). In-group bias in the minimal intergroup situation: A cognitivemotivational analysis. Psychological Bulletin, 86, 307–324. Brewer, M. B., & Miller, N. (1984). Beyond the contact hypothesis: Theoretical perspectives on desegregation. In N. Miller & M. B. Brewer (Eds.), Groups in contact: The psychology of desegregation (pp. 281–302). Orlando, FL: Academic Press. Brown, R. J. (1984). The role of similarity in intergroup relations. In H. Tajfel (Ed.), The social dimension: European developments in social psychology (pp. 603–623). Cambridge: Cambridge University Press. Brown, R. J., & Abrams, D. (1986). The eﬀects of intergroup similarity and goal interdependence on intergroup attitudes and task performance. Journal of Experimental Social Psychology, 22, 78–92. Brown, R. J., & Turner, J. C. (1979). The criss-cross categorization eﬀect in intergroup discrimination. British Journal of Social and Clinical Psychology, 18, 371–383. Cadinu, M. R., & Rothbart, M. (1996). Self-anchoring and diﬀerentiation processes in the minimal group setting. Journal of Personality and Social Psychology, 70, 661–677.

Crisp

107

Campbell, D. T. (1956). Enhancement of contrast as a composite habit. Journal of Abnormal and Social Psychology, 53, 350–355. Clement, R. W., & Krueger, J. (2000). The primacy of self-referent information in perceptions of social consensus. British Journal of Social Psychology, 39, 279–299. Crisp, R. J., & Beck, S. R. (2005). Reducing intergroup bias: The moderating role of ingroup identiﬁcation. Group Processes and Intergroup Relations, 8, 173–185. Crisp, R. J., Beck, S. R., & Hewstone, M. (2004). Crossed categorization, dual identiﬁcation, and the inclusiveness of common ingroup identities. Unpublished manuscript, University of Birmingham. Crisp, R. J., Ensari, N., Hewstone, M., & Miller, N. (2002). A dual-route model of crossed categorization eﬀects. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 13, pp. 35–74). Hove, UK & Philadelphia, PA: Psychology Press. Crisp, R. J., & Hewstone, M. (1999a). Diﬀerential evaluation of crossed category groups: Patterns, processes, and reducing intergroup bias. Group Processes and Intergroup Relations, 2, 303–333. Crisp, R. J., & Hewstone, M. (1999b). Subcategorization of physical stimuli: Category diﬀerentiation and decategorization processes. European Journal of Social Psychology, 29, 665–671. Crisp, R. J., Hewstone, M., & Rubin, M. (2001). Does multiple categorization reduce intergroup bias? Personality and Social Psychology Bulletin, 27, 76–89. Crisp, R. J., Stone, C. H., & Hall, N. R. (2006). Recategorization and subgroup identiﬁcation: Predicting and preventing threats from common ingroups. Personality and Social Psychology Bulletin, 32, 230–243. Deschamps, J.-C. (1977). Eﬀects of crossing category membership on quantitative judgment. European Journal of Social Psychology, 7, 517–521. Deschamps, J.-C., & Doise, W. (1978). Crossed category memberships in intergroup relations. In H. Tajfel (Ed.), Diﬀerentiation between social groups (pp. 141–158). Cambridge: Cambridge University Press. Diehl, M. (1990). The minimal group paradigm: Theoretical explanations and empirical ﬁndings. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 1, pp. 263–292). Chichester: Wiley International. Doise, W. (1978). Groups and individuals: Explanations in social psychology. Cambridge: Cambridge University Press. Doosje, B., Ellemers, N., & Spears, R. (1995). Perceived intragroup variability as a function of group status and identiﬁcation. Journal of Experimental Social Psychology, 31, 410–436. Dovidio, J. F., Gaertner, S. L., Isen, A. M., & Lowrance, R. (1995). Group representations and intergroup bias: Positive aﬀect, similarity, and group size. Personality and Social Psychology Bulletin, 21, 856–865. Dovidio, J. F., Gaertner, S. L., Isen, A. M., Rust, M., & Guerra, P. (1998). Positive aﬀect, cognition, and the reduction of intergroup bias. In C. Sedikides, J. Schopler, & C. A. Insko (Eds.), Intergroup cognition and intergroup behavior (pp. 337–366), Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Dovidio, J. F., Gaertner, S. L., & Validzic, A. (1998). Intergroup bias: Status, diﬀerentiation and a common ingroup identity. Journal of Personality and Social Psychology, 75, 109–120. Ellemers, N., Spears, R., & Doosje, B. (2002). Self and social identity. Annual Review of Psychology, 53, 161–186.

108

Commitment and categorization

Eurich-Fulcer, R., & Schoﬁeld, J. W. (1995). Correlated versus uncorrelated social categorizations: The eﬀect on intergroup bias. Personality and Social Psychology Bulletin, 21, 149–159. Gaertner, S. L., & Dovidio, J. F. (2000). Reducing intergroup bias: The common ingroup identity model. Philadelphia, PA: Psychology Press. Gaertner, S. L., Dovidio, J. F., Anastasio, P. A., Bachman, B. A., & Rust, M. C. (1993). The common ingroup identity model: Recategorization and the reduction of intergroup bias. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 2, pp. 247–278). Chichester: Wiley International. Gaertner, S. L., Dovidio, J. F., Rust, M. C., Nier, J., Banker, B. S., Ward, C. M., Mottola, G. R., & Houlette, M. (1999). Reducing intergroup bias: Elements of intergroup cooperation. Journal of Personality and Social Psychology, 76, 388–402. Gaertner, S. L., Mann, J. A., Dovidio, J. F., Murrell, A. J., & Pomare, M. (1990). How does cooperation reduce intergroup bias? Journal of Personality and Social Psychology, 59, 692–704. Gaertner, S. L., Mann, J. A., Murrell, A. J., & Dovidio, J. F. (1989). Reducing intergroup bias: The beneﬁts of recategorization. Journal of Personality and Social Psychology, 57, 239–249. González, R., & Brown, R. J. (2003). Generalization of positive attitude as a function of subgroup and superordinate group identiﬁcations in intergroup contact. European Journal of Social Psychology, 33, 195–214. Hall, N. R., & Crisp, R. J. (2005). Considering multiple criteria for social categorization can reduce intergroup bias. Personality and Social Psychology Bulletin, 31, 1435–1444. Henderson-King, E., Henderson-King, D., Zhermer, N., Posokhova, S., & Chiker, V. (1997). Ingroup favoritism and perceived similarity: A look at Russian’s perceptions in the post-Soviet era. Personality and Social Psychology Bulletin, 23, 1013–1021. Hewstone, M. (1996). Contact and categorization: Social psychological interventions to change intergroup relations. In C. N. Macrae, C. Stangor, & M. Hewstone (Eds.), Stereotypes and stereotyping (pp. 323–368). New York & London: The Guilford Press. Hewstone, M., & Brown, R. (1986). Contact is not enough: An intergroup perspective on the ‘contact hypothesis’. In M. Hewstone & R. Brown (Eds.), Contact and conﬂict in intergroup encounters (pp. 1–44). Oxford: Blackwell Publishing. Hinkle, S., & Brown, R. (1990). Intergroup comparisons and social identity: Some links and lacunae. In D. Abrams & M. Hogg (Eds.), Social identity theory: Constructive and critical advances (pp. 48–70). Hemel Hempstead: Harvester Wheatsheaf. Hodson, G., Dovidio, J. F., & Esses, V. M. (2003). Ingroup identiﬁcation as a moderator of positive-negative asymmetry in social discrimination. European Journal of Social Psychology, 33, 215–233. Hogg, M. A. (2000). Subjective uncertainty reduction through self-categorization: A motivational theory of social identity processes. European Review of Social Psychology, 11, 223–255. Hogg, M. A. (2001). Self-categorization and subjective uncertainty resolution: Cognitive and motivational facets of social identity and group membership. In J. P. Forgas, K. D. Williams, & L. Wheeler (Eds.), The social mind: Cognitive and motivational aspects of interpersonal behaviour (pp. 323–349). New York: Cambridge University Press.

Crisp

109

Hogg, M. A., & Abrams, D. (1990). Social motivation, self-esteem and social identity. In D. Abrams & M. A. Hogg (Eds.), Social identity theory: Constructive and critical advances (pp. 28–47). New York: Springer-Verlag. Hogg, M. A., & Mullin, B. A. (1999). Joining groups to reduce uncertainty: Subjective uncertainty reduction and group identiﬁcation. In D. A. Abrams & M. A. Hogg (Eds.), Social identity and social cognition (pp. 249–279). Oxford: Blackwell Publishing. Hornsey, M. J., & Hogg, M. A. (2000). Subgroup relations: A comparison of mutual intergroup diﬀerentiation and common ingroup identity models of prejudice reduction. Personality and Social Psychology Bulletin, 26, 242–256. Hutter, R. R. H. (2005). Social category conjunctions: Cognitive processes and representational consequences. Unpublished doctoral dissertation, University of Birmingham. Hutter, R. R. H., & Crisp, R. J. (2005). The composition of category conjunctions. Personality and Social Psychology Bulletin, 31, 647–657. Jetten, J., & Spears, R. (2004). The divisive potential of diﬀerences and similarities: The role of intergroup distinctiveness in intergroup diﬀerentiation. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 14, pp. 203–241). Hove, UK & Philadelphia, PA: Psychology Press. Jetten, J., Spears, R., & Manstead, A. S. R. (1997). Distinctiveness threat and prototypicality: Combined eﬀects on intergroup discrimination and collective self-esteem. European Journal of Social Psychology, 27, 635–657. Jetten, J., Spears, R., & Manstead, A. S. R. (1998). Deﬁning dimensions of distinctiveness: Group variability makes a diﬀerence to diﬀerentiation. Journal of Personality and Social Psychology, 74, 1481–1492. Jetten, J., Spears, R., & Manstead, A. S. R. (2001). Similarity as a source of discrimination: The role of group identiﬁcation. European Journal of Social Psychology, 31, 621–640. Marcus-Newhall, A., Miller, N., Holtz, R., & Brewer, M. B. (1993). Cross-cutting category membership with role-assignment: A means of reducing intergroup bias. British Journal of Social Psychology, 32, 125–146. Migdal, M., Hewstone, M., & Mullen, B. (1998). The eﬀects of crossed categorization in intergroup evaluations: A meta-analysis. British Journal of Social Psychology, 37, 303–324. Miller, N., Brewer, M. B., & Edwards, K. (1985). Cooperative interaction in desegregated settings: A laboratory analog. Journal of Social Psychology, 22, 103–122. Mullen, B., Brown, R., & Smith, C. (1992). Ingroup bias as a function of salience, relevance, and status: An integration. European Journal of Social Psychology, 22, 103–122. Mullen, B., Migdal, M. J., & Hewstone, M. (2001). Crossed categorization vs. simple categorization and intergroup evaluations: A meta-analysis. European Journal of Social Psychology, 31, 721–736. Mullin, B. A., & Hogg, M. A. (2000). Motivations for group membership: The role of subjective importance and uncertainty reduction. Basic and Applied Social Psychology, 21, 91–102. Oakes, P. J., Haslam, S. A., & Turner, J. C. (1994). Stereotyping and social reality. Oxford: Blackwell Publishing. Park, B., & Judd, C. M. (1990). Measures and models of perceived group variability. Journal of Personality and Social Psychology, 59, 173–191.

110

Commitment and categorization

Pettigrew, T. F. (1998). Intergroup contact theory. Annual Review of Psychology, 49, 65–85. Roccas, S., & Brewer, M. B. (2002). Social identity complexity. Personality and Social Psychology Review, 6, 88–106. Roccas, S., & Schwartz, S. H. (1993). Eﬀects of intergroup similarity on intergroup relations. European Journal of Social Psychology, 23, 581–595. Rosch, E. H. (1973). Natural categories. Cognitive Psychology, 4, 328–350. Sherif, M. (1966). Group conﬂict and cooperation. London: Routledge and Kegan Paul. Singh, R., Yeoh, B. S. E., Lim, D. I., & Lim, K. K. (1997). Cross categorization eﬀects in intergroup discrimination: Adding versus averaging. British Journal of Social Psychology, 36, 121–138. Spears, R., Doosje, B., & Ellemers, N. (1997). Self-stereotyping in the face of threats to group status and distinctiveness: The role of group identiﬁcation. Personality and Social Psychology Bulletin, 23, 538–553. Spears, R., Jetten, J., & Scheepers, D. (2002). Distinctiveness and the deﬁnition of collective self. In A. Tesser, D. A. Stapel, & J. Wood (Eds.), Self and motivations: Emerging psychological perspectives (pp. 147–171). Washington, DC: American Psychological Association. Stone, C. H., & Crisp, R. J. (2005). Relative identiﬁcation and intergroup bias in common ingroup contexts. Manuscript submitted for publication. Tajfel, H. (1959). Quantitative judgment in social perception. British Journal of Psychology, 50, 16–29. Tajfel, H. (1982). Social psychology of intergroup relations. Annual Review of Psychology, 33, 1–39. Tajfel, H., & Wilkes, A. L. (1963). Classiﬁcation and quantitative judgment. British Journal of Psychology, 54, 101–113. Tajfel, H., Flament, C., Billig, M., & Bundy, R. F. (1971). Social categorization and intergroup behaviour. European Journal of Social Psychology, 1, 149–178. 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. 33–47). Monterey, CA: Brooks/Cole. Turner, J. C. (1982). Towards a cognitive redeﬁnition of the social group. In H. Tajfel (Ed.), Social identity and intergroup relations (pp. 15–40). Cambridge: Cambridge University Press. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. Oxford: Blackwell Publishing. Urban, L. M., & Miller, N. M. A theoretical analysis of crossed categorization eﬀects: A meta-analysis. Journal of Personality and Social Psychology, 74, 894–908. van Leeuwen, E., van Knippenberg, D., & Ellemers, N. (2003). Continuing and changing group identities: The eﬀects of merging on social identiﬁcation and ingroup bias. Personality and Social Psychology Bulletin, 29, 697–690. van Twuyver, M., & van Knippenberg, A. (1999). Social categorization as a function of relative group size. British Journal of Social Psychology, 38, 135–156. Vanbeselaere, N. (1987). The eﬀect of dichotomous and crossed social categorizations upon intergroup discrimination. European Journal of Social Psychology, 17, 143–156. Vanbeselaere, N. (1991). The diﬀerent eﬀects of simple and crossed categorizations: A result of the category diﬀerentiation process or of diﬀerential category salience?

Crisp

111

In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 2, pp. 247–278). Chichester: Wiley International. Vanbeselaere, N. (1996). The impact of diﬀerentially valued overlapping categorizations upon the diﬀerentiation between positively, negatively and neutrally evaluated social groups. European Journal of Social Psychology, 26, 75–96. Worchel, S., Axsom, K., Ferris, F., Samaha, C., & Schweitzer, S. (1978). Factors determining the eﬀect of intergroup attraction. Journal of Conﬂict Resolution, 22, 428–439. Wright, S. C., Aron, A., McLaughlin-Volpe, T., & Ropp, S. A. (1997). The extended contact eﬀect: Knowledge of cross-group friendships and prejudice. Journal of Personality and Social Psychology, 73, 73–90.

6

Self-concept threat and multiple categorization within groups Michael A. Hogg and Matthew J. Hornsey

Groups, such as nations, religions, organizations, or sports teams, are rarely homogeneous. Almost always they are internally diﬀerentiated, latently or explicitly, in multiple ways. For example, an organization can embrace diﬀerent professions, ethnicities, genders, sexual orientations, and speciﬁc and generic roles. There is an intrinsic tension between the overarching group and the multiple ways in which its members can be categorized. The resolution of this tension hinges on many factors, among which is the extent to which the group feels under threat. Groups can feel unassailable (strong, successful, prestigious, united, and eﬃcacious), or they can feel under threat (fearing failure, disunity, social disintegration, loss of prestige, and reduced eﬃcacy). A key question for social psychologists is how do groups respond to threat? Do they disintegrate into individuals or fragment into multiple categories, or do they regroup and forge ahead? The answer rests on how one deﬁnes a group, on what the nature and the source of the threat may be, on what motives are aroused by the threat, and on what types of response are contextually available and possible. In this chapter we adopt a social cognitive deﬁnition of the social group that is framed by the social identity perspective. Psychologically, a group is two or more people who share a social identity—they deﬁne and evaluate themselves in terms of the same fuzzy set of attitudes, behaviors, perceptions, and feelings that deﬁnes their group in contradistinction to other groups or to people who are not in their group. Because groups frame the collective selfconcept, the key feature of group threat is collective self-concept threat. In this chapter we focus upon the impact of self-concept threat on subgroup structure and on structural and category diﬀerentiation within groups. What causes a group to remain a single uniﬁed category, or to break down into subcategories, warring factions, or isolated or marginalized individuals, or to become a battleﬁeld on which larger intergroup conﬂicts are played out? In examining this question, we employ an extended social identity framework (e.g., Abrams & Hogg, 2001; Hogg, 2001a, 2003) and draw on a number of recent literatures, including work on uncertainty reduction motivation (e.g., Hogg, 2000), subgroup relations within groups (e.g., Hornsey & Hogg, 2000a), group leadership (e.g., Hogg, 2001b; Hogg & van Knippenberg, 2003), schisms

Hogg and Hornsey

113

within groups (e.g., Sani & Reicher, 2000), identity threat (e.g., Branscombe, Wann, Noel, & Coleman, 1993; Jetten, Spears, & Manstead, 1997), and deviance and marginalization (e.g., Marques, Abrams, Páez, & Hogg, 2001). At the heart of our analysis is the idea that a threat to the collective selfconcept (i.e., social identity as a group member) can undermine how we evaluate ourselves and/or how certain we feel about ourselves as group members. In response, people can reconsolidate the group as a tight-knit, cohesive, well-structured, and homogeneous all-embracing entity. Or people can behave in ways that split the group asunder, that marginalize and reject deviants, that consolidate sharp status and leadership diﬀerentials, or that accentuate existing subgroup or intergroup cleavages within the group. The accentuation of diﬀerences may take diﬀerent forms. It may introduce harmful and ultimately destructive intragroup conﬂict, but it may also celebrate diversity (of subgroups, roles, opinions,) and marshal diversity as a way of strengthening the group in the face of threat. In the past, social identity research has mainly focused on the consequences of intergroup threat for intergroup relations (e.g., Tajfel & Turner, 1986; also see Ellemers, 1993; Hogg & Abrams, 1988). In this chapter we focus on the consequences of threat for intragroup relations and structures—treating multiple categorization as a consequence of threat. We brieﬂy introduce the social identity perspective in order to say more about the twin motivations of self-enhancement and uncertainty reduction. We then discuss the general impact of valence threat (arousing self-enhancement) and entitativity threat (arousing uncertainty reduction) on intragroup structure, and focus on a number of speciﬁc structural outcomes: subgroups and cross-cutting categories; schisms and splinter groups; deviance, dissent, and marginalization; hierarchy and totalism. These outcomes are treated as identity management strategies in the face of valence and entitativity threat. We conclude with a discussion of ways to protect against threat-mediated outcomes that propel groups toward fragmentation or monolithic orthodoxy.

SOCIAL IDENTITY AND GROUP MEMBERSHIP Thirty years of research provides good evidence that social identity plays a pivotal role in group and intergroup phenomena. The social identity perspective (e.g., Hogg & Abrams, 1988; Tajfel & Turner, 1986; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987; for recent overviews see Abrams & Hogg, 2001; Hogg, 2001a, 2003; Turner, 1999) argues that people deﬁne and evaluate themselves in terms of the groups to which they belong. Groups provide people with a collective self-concept, a social identity, and people have as many distinct social identities as the groups to which they feel they belong. Social identity is diﬀerentiated from personal identity, which is tied to interpersonal relationships and idiosyncratic personal traits (see Hogg, 2001c; Hogg & Williams, 2000; cf. Brewer & Gardner, 1996).

114

Threat and multiple categorization

Because social identities deﬁne, prescribe, and evaluate who one is and how one should think, feel, and act, people seek to establish or maintain the evaluative superiority of their own group over relevant other groups—there is an intergroup struggle for evaluatively positive group distinctiveness. This struggle is, however, tempered by people’s understanding of the nature of the relations between their group and relevant outgroups. In particular, people pay attention to status diﬀerences and the stability and legitimacy of such diﬀerences, to the permeability of intergroup boundaries, and to the existence of achievable alternatives to the status quo. For social identity theory, group behaviors (conformity, stereotyping, ethnocentrism, ingroup favoritism, intergroup discrimination, ingroup cohesion, etc.), as distinct from interpersonal behaviors, occur when social identity is the contextually salient basis of self-conceptualization. The content of group behavior rests on the speciﬁc social identity that is salient. Social identity processes are cognitively generated by social categorization of self and others. People cognitively represent groups as prototypes— multidimensional fuzzy sets of attributes that describe and prescribe perceptions, thoughts, feelings, relationships, and actions that deﬁne the ingroup and distinguish it from relevant outgroups. Prototypes are formed, maintained, and modiﬁed to optimize the balance between accentuating similarities within the group and accentuating diﬀerences between the ingroup and the outgroup. Prototypes maximize metacontrast and category distinctiveness, and more broadly maximize category entitativity (Hogg, 2004). High entitativity categories are ones that are relatively homogeneous, have clear boundaries, are distinct from other categories, and/or are clearly structured internally (Campbell, 1958; Hamilton & Sherman, 1996). An important point to bear in mind is that high entitativity categories do not have to be highly homogeneous—entitativity can rest on good internal structure (Hamilton, Sherman, & Castelli, 2002). Social categorization transforms the basis of perception and evaluation, so that people are viewed through the lens of category membership. The relevant prototype is the perceptual standard and people are perceptually assimilated to the prototype. In this sense they are depersonalized (see Reicher, Spears, & Postmes, 1995)—they are not viewed as unique individuals, but as group members who appear, in varying degree, to match the relevant prototype. Applied to self, social categorization has the same eﬀect as the categorization of other people—it transforms self-conception so that people feel like group members, and it depersonalizes attitudes, feelings, and behaviors such that they conform to the ingroup prototype. Thus, self-categorization is responsible for conformity, normative behavior, and the rest of the ediﬁce of group and intergroup behaviors. Although depersonalization is a discrete consequence of social categorization, the extent of depersonalization is inﬂuenced by factors such as the enduring and/or situational subjective importance and relevance of the social category, and the perceived prototypicality of self and others.

Hogg and Hornsey 115 SOCIAL IDENTITY MOTIVATIONS What motivates social identity processes? People may identify with groups for a variety of reasons—for example, to achieve goals that can only be achieved through cooperative interaction (e.g., Sherif, 1958), to obtain consensual validation of one’s perceptions from similar others (e.g., Festinger, 1954), to buﬀer the fear of inevitable death (e.g., Greenberg, Solomon, & Pyszczynski, 1997), or simply to satisfy a need to belong (e.g., Baumeister & Learly, 1995). From a social identity perspective, there are two motivational processes that are particularly closely associated with collective self-conception, social identity, and group and intergroup phenomena: self-enhancement and uncertainty reduction.

Self-enhancement Intergroup behavior is motivated by a struggle between groups to promote or protect their evaluatively positive distinctiveness from one another, and thus secure a relatively favorable social identity. People engage in this struggle because, at the individual level, group membership mediates self-evaluation via social identity, and people tend to be motivated to feel good about themselves—to have a positive sense of self-esteem (e.g., Sedikides & Strube, 1997). In intergroup contexts self-esteem may motivate social identity processes, but how this motivation plays out is signiﬁcantly impacted by social conventions and social belief systems (see Abrams & Hogg, 1988; Rubin & Hewstone, 1998). For example, a direct intergroup struggle over group status, and thus the valence of the ingroup and self, may only occur if members of the lower-status group perceive their relative group status to be illegitimate and unstable, and they can see how they can eﬀectively, and without excessive risk, pursue change (e.g., Tajfel & Turner, 1986).

Uncertainty reduction Intergroup relations also, of course, rest on social categorization. A core motive for social categorization is a need to structure our subjective environment in contextually meaningful ways that reduce uncertainty and allow us to predict people’s behavior, plan our own actions, and locate and deﬁne ourselves relative to other people (e.g., Hogg, 2000, 2001d). People are more likely to identify with groups when they are faced by self-conceptual uncertainty. When such uncertainty is very high they may seek out totalist groups that are highly orthodox, have simple and consensual prototypes, high entitativity, and strong charismatic leaders, and that engage in extreme forms of intergroup behavior (e.g., Hogg, 2004). This analysis allows us to understand why disadvantaged groups may acquiesce in their position rather than struggle for change—change may improve status and group valence, but it will also introduce uncertainty (also

116

Threat and multiple categorization

see Jost & Kramer, 2003). It also allows us to understand why groups whose distinctiveness is threatened may react in ways that are aimed, not primarily at improving status, but primarily at raising entitativity and cohesion in order to provide a secure and clear social identity (e.g., Hamilton & Sherman, 1996), perhaps through rejection of deviates (e.g., Marques & Páez, 1994). One way in which entitativity can be raised is by accentuating intragroup homogeneity. One problem, however, is that extreme intragroup homogeneity threatens people’s sense of individual distinctiveness and may lead back to subgroup or individual diﬀerentiation within the group to achieve an optimal level of distinctiveness (e.g., Brewer, 1991). However, this hydraulic equilibrium process is not inevitable. Although high entitativity can entail increased intragroup homogeneity, it does not necessarily have to do so. A high entitativity group that is eﬀective in reducing self-conceptual uncertainty may also deﬁne itself as an entity that tolerates or indeed celebrates constructive relations among a diverse range of individuals, roles, or subgroups. A group like this may have very high entitativity and be distinctive, but not be a homogeneous monoculture.

MOTIVATIONAL RESPONSES TO THREAT How a group responds or reacts to social identity threat will be inﬂuenced by what motivations are aroused by the threat. In the context of the discussion above, one might, for example, expect diﬀerent outcomes if the threat is only to group/identity valence (arousing self-enhancement concerns), or only to group/identity entitativity or certainty (arousing uncertainty reduction concerns). Valence threat targets the status, value, prestige, and social standing of the group. Entitativity threat targets what the group stands for and what it does, its internal integrity and structure, its distinctiveness, its cohesiveness, and the degree to which there is internal agreement. In reality, both threats may co-occur, and quite often the reaction to threat is the same irrespective of which motivation is aroused.

Responding to valence threat Where the group’s status, prestige or reputation is threatened, there is a threat to self-esteem—people are concerned about how they and others feel about them. In an intergroup context people can respond in a number of ways that are outlined by social identity theory (e.g., Tajfel & Turner, 1986; also see Ellemers, 1993; Hogg & Abrams, 1988). Where people are not strongly identiﬁed with the ingroup, and believe that intergroup boundaries are permeable and that social mobility is possible, they disidentify from the group and identify with other more prestigious groups that mediate positive social identity. The endpoint of this process, if successful, is that the original group ceases to exist psychologically—everyone has disidentiﬁed. However, this rarely

Hogg and Hornsey 117 happens, because group boundaries are rarely completely permeable, especially ones to do with important social divisions based on race, ethnicity, gender, socioeconomic status, and so forth (see Crocker, Major, & Steele’s, 1998, discussion of stigma). In reality, people are more likely to claim that their own personal self-esteem is unaﬀected by group valence threat (e.g., Crocker & Major, 1994; Crosby, 1984). Or they may be creative in trying to improve the status of their group by redeﬁning the valence of existing groupdeﬁning dimensions, by choosing other group-deﬁning dimensions, or by comparing their group to less prestigious outgroups. People can also engage the outgroup source of valence threat head on, in an attempt to change the actual status relations between groups. Intergroup valence threat can, thus, lead to group disintegration or to various forms of group consolidation. However, this analysis tends to underplay the fact that most groups are structured into latent, or not so latent subgroups, roles, cross-cutting categories, and so forth (see McGarty, this volume). Not surprisingly, a common response to a threat to the overall group’s prestige is for its members to proclaim their membership of more prestigious subgroups, or roles within the group, or with more prestigious cross-cutting categories. The original group loses solidarity and entitativity and becomes a context for other intergroup relations to be played out. For example, many organizations are socially diverse in terms of membership— having members that diﬀer in terms of race, ethnicity, gender, socioeconomic status, and so forth. A common response to a threat to the prestige of the organization as a whole is for members to disengage psychologically, and to recreate within the organization the often troublesome ethnic, racial, and so forth relations that exist in the wider society within which the organization is located (e.g., Brewer, 1996; Brewer, von Hippel, & Gooden, 1999).

Responding to entitativity threat Where a group’s entitativity or prototype clarity and distinctiveness are threatened, there is a threat to self-conceptual certainty (Hogg, 2000, 2001d, 2004). People become unsure of what it means to be a member of the group, of what the prescribed attitudes, feelings, perceptions, and behaviors are, and of how they should interact with fellow members or members of other groups. There are a number of reactions to this kind of threat. People can disidentify with the group and pursue or accentuate membership of other groups (outgroups, subgroups, cross-cutting categories) that have higher entitativity and are more self-conceptually clear, or they can “remodel” the existing group and its deﬁning features so that it once again mediates a secure and certain self-concept. Remodeling can take diﬀerent forms, which we discuss in detail below. For example, the prototype can be refocused by marginalizing and rejecting individuals or groups who do not contribute to a clear and consensual ingroup prototype; or a strong leader can emerge or be endorsed to act as a focus for a

118

Threat and multiple categorization

consensual ingroup prototype; or a speciﬁc outgroup can be targeted in order to accentuate intergroup distinctiveness and to clarify the ingroup prototype. Research suggests that although intergroup distinctiveness information helps clarify the ingroup prototype, ingroup prototype clarity itself is probably the most direct source of self-deﬁnitional information (Yzerbyt, Castano, Leyens, & Paladrino, 2000). Knowing what we are like is usually of more immediate self-deﬁnitional use than knowing how we diﬀer from them (e.g., Pickett & Brewer, 2001; Smith & Henry, 1996). Remodeling may also involve an overexclusion eﬀect (e.g., Leyens & Yzerbyt, 1992; Yzerbyt, Leyens, & Bellour, 1995), in which people, particularly those who are highly identiﬁed with the group, adopt very strict criteria for admission of any new members to the group. As foreshadowed above, entitativity threat does not necessarily lead to intragroup homogenization around a clear consensual prototype—it can also lead to intragroup diﬀerentiation. Often such diﬀerentiation spawns conﬂict between nested or cross-cutting categories—leading to group fragmentation. However, this is not inevitable. Under some circumstances diﬀerentiation in the face of entitativity threat can capture harmonious role relations among subgroups, cross-cutting categories, and other forms of intragroup diversity. The key here is that role or category diversity becomes a deﬁning feature of the group—something that, paradoxically, enhances the group’s entitativity, distinctiveness, and prototype clarity. The group essentially remodels itself around a clear and consensual prototype that explicitly recognizes and celebrates intragroup categorial diﬀerentiation.

DIFFERENTIATION AND DIVERSITY WITHIN GROUPS Almost all groups are internally diﬀerentiated in a variety of ways. In many cases such diﬀerentiation is unproblematic, and is in fact essential to group functioning—for example, role diﬀerentiation and diﬀerent divisions in a company. Subgroup diversity within a group may also be valuable in its own right because it enriches the superordinate group and the self-concepts of its members—for example, at the societal level, successful cultural pluralism. There is also evidence that task-oriented groups that embrace diverse views and diverse subgroups function better than groups that are homogeneous in terms of attitudes, positions, and demographic characteristics. An eﬀective way to combat groupthink is to ensure attitudinal diversity within a decisionmaking group (e.g., Postmes, Spears, & Cihangir, 2001; Stasser, Stewart, & Wittenbaum, 1995), and many organizations beneﬁt from demographic diversity (e.g., Brewer, 1996). Generally, there is evidence that unshared information, perhaps reﬂecting diverse subgroups, can have a distinct advantage for overall group functioning (e.g., Larson, Foster-Fishman, & Keys, 1994; for reviews see Tindale, Kameda, & Hinsz, 2003; Wittenbaum & Stasser, 1996). There is also evidence that a degree of dissent created by minorities within

Hogg and Hornsey

119

groups can be beneﬁcial, largely through its impact on creativity and innovation (e.g., Nemeth & Owens, 1996; Nemeth & Staw, 1989; Nemeth & Wachtler, 1983). Finally, internal criticism of a group’s culture or operations can play a crucial role in reinvigorating groups and laying the groundwork for positive change (e.g., Hornsey & Imani, 2004; Hornsey, Oppes, & Svensson, 2002). However, there is no doubt that some diﬀerentiation, for example reﬂecting profound intergroup hostility, is problematic and may detract from group functioning and undermine the quality of group life for some or all members. The motivational analysis above has suggested that social identity threat may arouse diﬀerent motivations, speciﬁcally self-enhancement and uncertainty reduction, which, in turn, produce diﬀerent types and qualities of intragroup diﬀerentiation. In the following section we discuss a number of instances of such intragroup diﬀerentiation. The classic case is where a group contains subgroups or cross-cutting categories—this occurs in multiethnic societies and almost all organizations. We then discuss schisms and splinter groups—situations where an existing group breaks down along ideological or attitudinal lines into warring factions. Next we discuss situations where groups contain sets of individuals who are treated as deviants or marginal members— people who are often viliﬁed and rejected. The ﬂip side of the marginal member coin is leadership—the structural diﬀerentiation of groups into leaders and followers. We discuss how leader–follower relations can become destructively hierarchical and characteristic of dysfunctional totalist groups. A key question throughout is how can we protect against destructive intragroup diﬀerentiation, and encourage constructive diﬀerentiation within groups.

Subgroups and cross-cutting categories Groups often have, wholly or partially, nested within them a number of other groups that members could or do identify with. For example, most organizations embrace a range of work groups, divisions, professional identities, and ethnic, gender, socioeconomic and racial groups, and most nations encompass diﬀerent regional groups, diﬀerent ethnic groups, diﬀerent religions, and so forth. A critical question concerns the psychological prominence of groups that are nested within a superordinate identity. To what extent do members identify with the superordinate group or with one of the other groups, and what is the quality of the relations between groups within the superordinate group? The social-psychological literature dealing with these questions is enormous (this volume is part of that literature). For example, it includes literature on the contact hypothesis (Hewstone, 1996; Miller & Brewer, 1984; Pettigrew, 1998), cultural divides (Prentice & Miller, 1999), crossed categorization eﬀects (Crisp, Ensari, Hewstone, & Miller, 2003; Vescio, Hewstone, Crisp, & Rubin, 1999), decategorization and recategorization (e.g., Gaertner, Dovidio, Anastasio, Bachman, & Rust, 1993; Gaertner, Mann, Murrell, & Dovidio, 1989), organizational mergers and acquisitions (Terry, Carey, & Callan, 2001), and threat-based perceptions of group variability (e.g., Rothgerber, 1997).

120

Threat and multiple categorization

One obvious reaction to intergroup threat is for the group to draw together and dissolve subgroups into the superordinate entity—producing a single homogeneous category with a common ingroup identity (Gaertner et al., 1993). In his boys’ camp studies, Sherif (1958, 1966) tried to defuse intergroup conﬂict by creating a perception of common fate/shared threat to focus the boys’ attention on superordinate group membership. Similarly, political parties can pull together or pull apart depending on the extent of a realistic threat they face from other political forces. For example, in a longitudinal analysis of a union election in Australia, Hornsey, Blackwood, and O’Brien (2002) showed that political parties within the Broad Left frequently questioned each other’s right to claim membership of the superordinate Left identity. These parties only pulled together under the banner of the Broad Left to the extent that the Right was seen to be a realistic threat in an upcoming election. This dynamic can also be witnessed at the societal level. National leaders often create a common foe in order to bind the nation together and defuse internal subgroup conﬂict. More broadly, many societies that feel that the unity of their national identity is under threat pursue a policy of assimilation where ethnic subgroups are encouraged to abandon their ethnic origins and re-identify at the superordinate national level. However, intergroup threat does not always produce this outcome. It can often be easier and more logical for subgroups to circumvent the superordinate group threat by disidentifying with the superordinate group altogether and identifying at the subgroup level only. Where deep and historically enduring divisions marked by a history of mutual atrocities exist between groups (e.g., among ethnic groups in Yugoslavia), no amount of superordinate intergroup threat is going to easily bind the groups together into a single entity (Miller & Prentice, 1999). The key problem is that losing subgroup distinctiveness can itself be an identity threat. For example, Hornsey and Hogg (2000b, 2000c) found that when maths-science and social science students were categorized exclusively at the superordinate university level, there was evidence of a reactive increase in inter-subgroup bias relative to conditions where the subgroups were acknowledged (see also Crisp, this volume). Even under conditions of extreme intergroup threat at the superordinate level, the prospect of resolving that threat by losing subgroup distinctiveness can be problematic. Thus, when engaged in superordinate conﬂict, members of certain subgroups may experience a conﬂict between: (a) needing to homogenize with other subgroup members in order to maximize the inﬂuence and political eﬀectiveness of their message at the superordinate level; (b) needing to defend the position of their subgroup relative to other subgroups. This conﬂict may be especially strong for subgroups that occupy a peripheral

Hogg and Hornsey 121 position with respect to the superordinate prototype, or for subgroups with low status and/or power (Hornsey & Hogg, 2002). If they were to conform to a single superordinate position, members of subgroups such as these would be vulnerable to domination or appropriation by other subgroups. Thus, what they gain in inﬂuence and proﬁle at the superordinate level, they may lose at the subgroup level. Thus, there is an identity dilemma, which more often than not seems to be resolved by retaining subgroup distinctiveness even if that course of action does not reduce the superordinate intergroup threat. For example, research on organizational mergers and acquisitions (e.g., Terry et al., 2001) shows that where one company takes over or merges with another company, the premerger entities can remain distinct and even hostile to one another for many years, even when the superordinate entity is itself under threat. The identity dilemma is further complicated because in most nested subgroup situations, the properties of one subgroup are better (i.e., more fully) represented than the properties of other subgroups in the superordinate entity (Mummendey & Wenzel, 1999). The superordinate group looks more like one subgroup than other subgroups. To some extent, one subgroup is dominant (it occupies the central position in the superordinate group), whereas other subgroups are subordinate (they occupy more marginal positions). Clearly, superordinate identity threat says more about the better represented (dominant) subgroup than the less well-represented (subordinate) subgroups. The dominant subgroup might respond by marginalizing other subgroups as scapegoats—this might occur under conditions of entitativity or valence threat. In summary, attempts to get subgroups to merge to confront a threat to the superordinate group can backﬁre and produce enhanced conﬂict. Harmonious relations among subgroups are more readily achieved by encouraging the subgroups to remain distinct, to recognize and respect their mutual diﬀerences, and to view themselves as contributing valued complementary qualities to the superordinate group (Hewstone, 1996; Hornsey & Hogg, 2000a). The implication of this analysis is that intergroup threat that encourages a reaction framed in terms of diverse and distinct subgroups pulling together, each making its unique contribution to the common cause, can produce quite positive outcomes. There will be cooperative and relatively harmonious relations between subgroups that retain their distinctiveness and social value. We can take this argument further. Subgroup diversity may actually be a valued resource for superordinate groups that are under entitativity or valence threat. Although this value may be relatively instrumental in terms of shared goals and cooperative interdependence in the face of threat, under the right conditions the subgroup prototypes, and thus social identities of subgroup members, may include beliefs about positive relations with other subgroups and about the value of diversity itself (see Niedenthal & Beike, 1997; Roccas & Brewer, 2002; Wright, Aron, & Tropp, 2002). Under these circumstances intergroup threat has created a situation where subgroup structure

122

Threat and multiple categorization

remains intact, and people identify at both the subgroup and the superordinate group level. At the subgroup level the ingroup prototype has a relational component that deﬁnes membership in the ingroup in terms of positive relations with the outgroup. At the superordinate group level the ingroup prototype has a self-deﬁnitional component that places value on positive subgroup relations and on diversity within the superordinate group. However, from our motivational analysis above, we might further speculate that the favorable situation we have just described is more easily achieved where the superordinate intergroup threat is mainly a valence, not an entitativity, threat. Where valence is threatened, a group may readily see how embracing diversity and structural diﬀerentiation enhances the group’s prestige and status, and thus its valence for self. Where entitativity is threatened, the immediate reaction might be to enhance homogeneity, purge the group of dissenters, suppress diversity, and impose orthodoxy. However, as described above, entitativity is not isomorphic with homogeneity. Entitativity can be reinstated by constructing a clear and distinct image of the group as a group that is structured into a variety of subgroups and roles—a group that embraces diversity as a very clear, self-deﬁning attribute. This idea is nicely captured by a statement made by Prime Minister Trudeau in the Canadian Parliament in 1971: “Canadian identity will not be undermined by multiculturalism. Indeed, we believe that cultural pluralism is the very essence of Canadian identity” (Government of Canada, 1971).

Schisms and splinter groups A relatively special case of nested subgroups, which focuses on attitudinal and value diﬀerences, is the emergence of schisms in, for example, ideological groups such as religions, cults, political parties, artistic movements, scientiﬁc doctrines, schools of thought, and so forth (e.g., Liebman, Sutton, & Wuthnow, 1988). In their analysis of schisms, Sani and Reicher (1998, 1999, 2000; also see Reicher & Hopkins, 2001) describe the way that identity threat, selfconceptual uncertainty, and a sense of self-conceptual impermanence and instability can arise in groups whose deﬁning properties are suddenly changed. The change can be brought about by the actions of a subgroup or a leadership clique. Members feel that the group is no longer what it used to be—its normative attitudes, values, perceptions, and behaviors have uncompromisingly changed. The prototype has changed, and thus the group’s identity has changed. Those who are taken oﬀ guard by this change suddenly feel uncertain about how, and whether, they ﬁt into the new group—the threat is acute, and very much one that engages uncertainty reduction motivation. Under these circumstances members can either try to re-establish the group’s original identity through discussion, persuasion, and negotiation, or they can split into a separate subgroup that is in conﬂict with the rest of the group. A split, or schism, is most likely to occur if members consider the group to be

Hogg and Hornsey 123 intolerant of dissent, unable to embrace diverse views, and inclined toward marginalization of dissenting individuals. A schism eﬀectively transforms one group, a single category, into two separate groups that are engaged in often highly charged intergroup conﬂict. The split rests on a profound social identity threat that engages a powerful drive to reduce the acute selfconceptual uncertainty that has been aroused. Not surprisingly, schisms can sometimes be particularly destructive to groups—for example, factional conﬂicts within political ideologies (e.g., Stalinists vs. Trotskyites within the Communist Party) and interpretational diﬀerences within religions (e.g., Sunnis vs. Shi’ites within Islam). Where a schism exists, the subgroup that holds the minority position may paradoxically stand a chance of winning over the rest of the group and reinstating a degree of consensus. This might happen if the minority’s position was novel, the minority could lay some claim to being a bone ﬁde part of the larger ingroup, and the minority adopted a consistent yet ﬂexible style of social inﬂuence and persuasion (e.g., Mugny, 1982; Nemeth, 1986). Indeed, although schisms are often highly destructive, the fact that they may sponsor critical thinking, creativity, and innovation may, if properly managed, enhance the larger group (e.g., Nemeth & Owens, 1996; Nemeth & Staw, 1989; Nemeth & Wachtler, 1983).

Deviance, dissent, and marginalization Schism represents a process whereby group members feel self-conceptually marginalized within the group and coalesce to form a separate subgroup— thus the group is rent asunder into two separate groups. Marginalization can also take a diﬀerent form where speciﬁc individuals are evaluatively marginalized within, and sometimes rejected from, the group. The group does not disintegrate but creates a class of stigmatized individuals within its boundaries. Although the group treats these individuals as a category, the individuals themselves may not identify with the category. This process has most systematically been analyzed by Marques and his colleagues through their work on the “black sheep” eﬀect (e.g., Marques & Páez, 1994), and subsequent work on subjective group dynamics (e.g., Abrams, Marques, Bown, & Henson, 2000; Marques, Abrams, & Serodio, 2001; also see Marques, Abrams, Páez, & Hogg, 2001). Marques and colleagues argue that ingroup members who are considered to be only marginally prototypical of the group are liked less as group members than prototypically central members. This diﬀerential is accentuated under high salience so that marginal members may be strongly disliked and entirely rejected as deviants or “black sheep”. By being a-prototypical, particularly in a direction that leans toward a salient outgroup, a marginal ingrouper is considered deviant and jeopardizes the distinctiveness and prototypical clarity and integrity of the ingroup. This may threaten the valence of the group, but it also introduces the threat of uncertainty. Thus,

124

Threat and multiple categorization

fellow ingroupers, especially those for whom uncertainty is particularly threatening, will reject the deviant in order to consolidate a clear prototype to which they can assimilate themselves through self-categorization. Black sheep studies focus on negative deviants, ingroup members who are inclined toward the outgroup prototype. But, what about positive deviants? These are group members who are a-prototypical but in evaluatively favorable ways; for example, overachievers or highﬂyers. On the one hand overachievers should be socially unattractive (disliked as group members) because they are a-prototypical, but on the other hand they should be socially attractive (liked as group members) because the group can bask in their reﬂected glory (e.g., Cialdini, Borden, Thorne, Walker, Freeman, & Sloan, 1976; Cialdini & de Nicholas, 1989; Sigelman, 1986; Wann, Hamlet, Wilson, & Hodges, 1995). There is some evidence that people are particularly harsh on overachievers who suﬀer a setback or experience a fall (e.g., Feather, 1994), but this research does not diﬀerentiate between overachievers who are members of a salient ingroup and those who are not. Let us consider deviance as a whole, taking negative and positive deviance together. Hogg, Fielding, and Darley (2005; also see Hogg & Terry, 2000) suggest that ingroup responses to a-prototypical or marginally prototypical members are inﬂuenced by at least two dimensions: a functional dimension and a social attribution dimension. Where solidarity and consensual prototypicality are important to the group, perhaps due to uncertainty concerns based on self-concept and group entitativity threat, both positive and negative deviants are dysfunctional for the group; they will both be evaluatively downgraded. Where solidarity and consensual prototypicality are less critical, but self-enhancement is important due to valence threat, positive deviants are functional for the group; they will be upgraded as they contribute to a favorable redeﬁnition of ingroup identity. Negative deviants are dysfunctional and will be downgraded as black sheep. The attribution dimension refers to the extent to which deviant behavior can be attributed to the group as a whole, or internally to the deviant him- or herself. Where positively deviant behavior can be “owned” by the group, the deviant will be favorably evaluated. This would be likely if the deviant modestly attributed the behavior to the support of the group rather than to personal ability, and where the deviant had only a short personal history of overachievement (i.e., was a “new” deviant). Where positively deviant behavior cannot readily be “owned” by the group, the deviant will be unfavorably evaluated. This would be likely where the deviant took full personal credit for the behavior without acknowledging the group’s support (i.e., “boasted”), and where the deviant had a long personal history of overachievement (i.e., was an enduring deviant). The relevance of the length of personal history is that an internal attribution to the person is more easily made where the person has behaved in that way consistently over a long period. As regards negative deviants, the group will always try to attribute their behavior internally—indeed, this is one function of negative deviants for

Hogg and Hornsey

125

groups. Negative deviants would attract particularly strong reactions from the group if they try to oﬄoad responsibility for their deviance onto the group. In discussing group deviants, it is important, however, to recognize that groups often go to enormous lengths to bring deviants back in line in order to retain them in the group (e.g., Schachter, 1951; see Darley, 2001). Often, deviants are rejected only after such eﬀorts have failed. According to staﬃng theory, one factor that is important is group size (e.g., Wicker, 1968). Where groups are small, rejection of members may threaten the group’s very existence and so more eﬀort is made to resocialize deviants. In larger groups there is less necessity to retain members, and indeed group distinctiveness may actually be served by rejecting marginal members. Another factor that may be important is the extent to which deviance, dissent, and internal criticism are considered a valued quality of the group. Where these properties are valued, deviants, dissenters, and critics are less likely to be ejected and may even be celebrated (e.g., Hornsey & Imani, 2004; Hornsey, Oppes, & Svensson, 2002; McAuliﬀe, Jetten, Hornsey, & Hogg, 2003). Deviance and dissent may even be considered to lend the group a distinctive identity. Although deviance may not lead to group dissolution, it may introduce a tense structural diﬀerentiation between members who are central and favored, and those who are marginal and disfavored. Negative a-prototypical members are most likely to be marginalized if group entitativity and valence are threatened, if the group is relatively large, if dissent and deviance are not deﬁning features of the group, and if the deviant tries to blame the group for his or her deviance. Positive a-prototypical members are most likely to be marginalized if entitativity alone is threatened, if the group is relatively large, if dissent and deviance are not deﬁning features of the group, and if the deviant tries to take personal responsibility for his or her deviance.

Hierarchy and totalism The idea that a group can be structurally diﬀerentiated into prototypically central and prototypically marginal members can be taken further if we shift focus from prototypically marginal members to prototypically central members. The social identity analysis of leadership suggests that as group membership becomes increasingly salient, eﬀective leadership increasingly rests on the leader being perceived to possess qualities that are prototypical of the group (e.g., Hogg, 2001b; Hogg, 2005; Hogg, Hains, & Mason, 1998; Hogg & van Knippenberg, 2003). Leadership is, of course, generally a good thing for groups. However, some forms of leadership are not—for example, autocratic, highly hierarchical, power-based leadership structures in totalist groups can be harmful. According to the social identity analysis of leadership, destructive leadership structures are particularly likely to arise under conditions of social identity threat (Hogg, 2001e; Hogg & Reid, 2001). Where the group’s valence or

126

Threat and multiple categorization

entitativity is threatened and members are unable or unwilling to psychologically leave the group, group membership becomes highly salient and self-conceptually important. The group can become tightly cohesive with a very simple, clear, and consensual prototype. Members pay close attention to prototypicality, as described above, and thus leadership processes rest heavily on how prototypical the leader is perceived to be. Being prototypical allows the leader to lead through inﬂuence rather than the exercise of power. Being prototypical also means that there is a common empathic membership bond between leader and follower. Therefore, members comply with the leader’s directions, which are after all the prototypical group position, through selfcategorization-based conformity processes and also because the leader is consensually socially attractive. In addition, by being prototypical, the leader’s actions are seen to be group-serving and fair, and thus the leader is trusted and his or her leadership is considered legitimate (Tyler, 1997). The leader is thus allowed a relatively high degree of latitude to diverge from the prototype, and thus to be innovative. The paradox is that precisely these conditions, if they persist for long, may spawn destructive leadership based on power. Being a highly prototypical leader of a tightly cohesive group with which members identify very strongly endows the leader with the appearance of being very eﬀective and having substantial leadership charisma and power. The fundamental attribution error (Ross, 1977), or correspondence bias (Gilbert & Malone, 1995), then ensures that these attributes are internally attributed to enduring dispositional properties of the leader—the leader is viewed as charismatic, a “great person”. In this way the leader, who was viewed as “one of us”, is gradually categorized as “other”, as diﬀerent from and better than the rest of the group. The common empathic bond between leader and followers is severed and a deep status-based intergroup schism is gradually created between leader (or leadership clique) and followers—the group has to diﬀerentiate into leadership and follower subgroups. There is now great power distance (Mulder, 1977) between leader and followers, and the leader has signiﬁcant reward power over the followers—together these encourage discrimination against the followers (Ng, 1996; Sachdev & Bourhis, 1985), and leadership by coercion. The relationship within the group between leader(s) and followers is now, eﬀectively, an intergroup relationship in which status diﬀerences are vast and consensual. The end result is a totalist group with a rigidly hierarchical power structure and an all-powerful leader who leads largely by coercion. To maintain power, such a leader may collect around him- or herself a group of loyal followers by treating them better than the rest of the group—thus creating an atmosphere of distrust, cronyism, and rivalry in the group and a sense that the leader’s position is no longer legitimate. This is a highly dysfunctional group, which history shows is heading for serious trouble—this is the situation that exists in dictatorships, cults, and other large and small totalist groups.

Hogg and Hornsey 127 BUFFERING THREAT AND PROTECTING AGAINST TOTALISM AND INTRAGROUP MELTDOWN Let us assume that it is desirable for a group to maintain itself as a distinct self-deﬁning entity that one feels relatively good about belonging to. Let us also assume that such a group embraces harmonious relations among diﬀerent subgroups and celebrates a degree of diversity against a background of shared goals and values. What can be done to protect such a group against some of the harmful consequences of identity threat described above? To recap, multiple categorization can be a strategic response to identity threat. The principal harmful consequences of identity threat we discussed are: (1) Dissolution: the group fragments into a rabble of disidentiﬁed individuals. (2) Cleavage: the group breaks down into irreconcilable warring factions that represent pre-existing subgroups or cross-cutting categories, or alternative visions of what the group stands for. (3) Witch hunts: the group marginalizes, persecutes, and rejects peripheral members as deviant individuals or deviant subgroups. (4) Totalism: the group becomes tightly cohesive and monolithically uniform with associated orthodoxy and hierarchy. Dissolution is most likely where the group is only a nebulous entity to begin with. Members are only weakly tied to or invested in the group, the group is not a very signiﬁcant contributor to self-conception (it is very low in one’s self-conceptual hierarchy), the group has little instrumental value or importance for its members, and it is very easy and uncostly to leave the group psychologically. Valence or entitativity threat simply conﬁrms the relative inadequacy of the group and points to “exit” as the best and easiest way to avoid the threat. To prevent threat producing dissolution, the group would need to be a more distinct entity that members are more strongly tied to and invested in, and that contributes more signiﬁcantly to self-conception. The group should also have greater instrumental value for its members, and it should also appear relatively diﬃcult and costly psychologically to leave the group. If a group of this type is relatively homogeneous, and poorly diﬀerentiated in terms of roles, belief systems, and nested or cross-cutting subgroups, threat is likely to engage witch hunts targeting marginal members (under valence threat only negative deviants, under entitativity threat all deviants—e.g., Hogg, Fielding, & Darley, in press). This produces an even ﬂatter and more homogeneous group structure. The threat of descent into varying degrees of totalism is now very real. If a group is relatively ﬂat, as described, but does have some potential diﬀerentiation in terms of belief systems, then, in addition to witch hunts, schism as a form of cleavage is likely. This, of course, protects against totalism,

128

Threat and multiple categorization

and instead introduces intragroup diﬀerentiation where little was apparent before. Schism may rip the group asunder, or it may, to combat entitativity threat, generate an enduring internal dialectic that celebrates criticism and diversity as an explicitly group-deﬁning property. Many groups are not ﬂat—they are relatively diverse insofar as they embrace distinct subgroups that are more or less chronically salient. Where these subgroups reﬂect functional diﬀerentiation for group goal achievement, both valence and entitativity threats are likely to have a beneﬁcial eﬀect in motivating the subgroups to work more closely together to promote the group and deﬂect the threat. Subgroup diﬀerentiation is not threatened as the subgroups are largely deﬁned by their diﬀerent roles within the group. There may, however, be some witch-hunting in which some subgroups are blamed for shortcomings of the group—for example, in an organization, the sales division may be blamed. Where these subgroups reﬂect essential self-conceptual diﬀerences, for example those based on race, ethnicity, ideology, or religion, threat will most readily produce cleavage and witch hunts. These subgroups reﬂect deep cultural divides (Miller & Prentice, 1999) that are very diﬃcult to transcend as people’s self-concepts are profoundly rooted in these kinds of groups. Whether the value or the entitativity, or both, of the superordinate group are threatened, people are most likely to respond by disidentifying from the superordinate group and identifying more with the subgroup. Subgroup relations will now reﬂect the history of such relations—conﬂict, coexistence, or harmony. Conﬂict is most likely because cleavage is likely to be accompanied by witch-hunting in which the dominant subgroup, the one whose properties are best represented in the superordinate entity, blames the plight of the group on the activities of one or more subordinate subgroups. However, harmonious subgroup relations within the superordinate group can arise if, as described above, threat creates a situation where subgroup structure remains intact, people identify at both the subgroup and the superordinate group level, and group prototypes and self-conception rest on the value of speciﬁc inter-subgroup relations and on the value of diversity within the larger group. In practice this is only likely to be successful where subgroup relations are not too hostile or polarized to begin with. A practical ﬁrst step would be an instrumental emphasis on the mutual and superordinate beneﬁts of cooperation between distinct subgroups, with each contributing their own special and valued qualities to the larger group.

CONCLUSION Groups and categories thrive to the extent that they provide their members with a clear and evaluatively positive sense of who they are and how they relate to others. Such groups should typically have high entitativity and positive valence. High entitativity does not equate to homogeneity—a high

Hogg and Hornsey 129 entitativity group can be one that embraces, as part of its distinctive identity, diversity (diﬀerent subgroups, cross-cutting categories, critics, dissenters, and so forth). A perceived threat to the group’s entitativity and/or valence can aﬀect the internal form and structure of a group. It can cause the group to fragment into warring factions that recreate external intergroup conﬂicts or an ideological struggle for the group’s destiny, or the group can simply dissolve and disappear. At the other extreme the group can purge itself of deviants, dissenters, and critics, and suppress diversity as it morphs into a monolithic and orthodox monoculture. These are both undesirable outcomes in terms of group life and social identity. In this chapter we discussed how perceived threat produces diﬀerent intragroup categorizations, structures, and relations, depending on the nature of the group to begin with, and the extent to which the threat primarily aﬀects the group’s valence or the group’s entitativity. The analysis is not intended to be exhaustive, but to be an overview of possible ways that social identityrelated motivational responses mediate the impact of threat on intragroup life. An understanding of such processes can help us to speculate about ways in which the harmful consequences of threat can be muted or avoided entirely.

REFERENCES Abrams, D., & Hogg, M. A. (1988). Comments on the motivational status of selfesteem in social identity and intergroup discrimination. European Journal of Social Psychology, 18, 317–334. Abrams, D., & Hogg, M. A. (2001). Collective identity: Group membership and selfconception. In M. A. Hogg & R. S. Tindale (Eds.), Blackwell handbook of social psychology: Group processes (pp. 425–460). Oxford: Blackwell Publishing. Abrams, D., Marques, J. M., Bown, N. J., & Henson, M. (2000). Pro-norm and antinorm deviance within in-groups and out-groups. Journal of Personality and Social Psychology, 78, 906–912. Baumeister, R. F., & Learly, M. R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497–529. Branscombe, N. R., Wann, D. L., Noel, J. G., & Coleman, J. (1993). In-group or outgroup extremity: Importance of the threatened social identity. Personality and Social Psychology Bulletin, 19, 381–388. Brewer, M. B. (1991). The social self: On being the same and diﬀerent at the same time. Personality and Social Psychology Bulletin, 17, 475–482. Brewer, M. B. (1996). Managing diversity: The role of social identities. In S. Jackson & M. Ruderman (Eds.), Diversity in work teams (pp. 47–68). Washington, DC: American Psychological Association. Brewer, M. B., & Gardner, W. (1996). Who is this we? Levels of collective identity and self representations. Journal of Personality and Social Psychology, 71, 83–93. Brewer, M. B., von Hippel, W., & Gooden, M. P. (1999). Diversity and organizational identity: The problem of entrée after entry. In D. A. Prentice & D. T. Miller (Eds.),

130

Threat and multiple categorization

Cultural divides: Understanding and overcoming group conﬂict (pp. 337–363). New York: Russell Sage Foundation. Campbell, D. T. (1958). Common fate, similarity, and other indices of the status of aggregates of persons as social entities. Behavioral Science, 3, 14–25. Cialdini, R. B., Borden, R. J., Thorne, A., Walker, M. R., Freeman, S., & Sloan, L. R. (1976). Basking in reﬂected glory: Three (football) ﬁeld studies. Journal of Personality and Social Psychology, 34, 366–375. Cialdini, R. B., & de Nicholas, M. E. (1989). Self-presentation by association. Journal of Personality and Social Psychology, 57, 626–631. Crisp, R. J., Ensari, N., Hewstone, M., & Miller, N. (2003). A dual-route model of crossed categorization eﬀects. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 13, pp. 35–74). New York: Psychology Press. Crocker, J., & Major, B. (1994). Reactions to stigma: The moderating role of justiﬁcations. In M. P. Zanna & J. M. Olson (Eds.), The psychology of prejudice: The Ontario symposium (Vol. 7, pp. 289–314). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Crocker, J., Major, B., & Steele, C. (1998). Social stigma. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (4th ed., Vol. 2, pp. 504–553). New York: McGraw-Hill. Crosby, F. (1984). The denial of personal discrimination. American Behavioral Scientist, 27, 371–386. Darley, J. (2001). Social comparison motives in ongoing groups. In M. A. Hogg & R. S. Tindale (Eds.), Blackwell handbook of social psychology (pp. 334–351). Oxford: Blackwell Publishing. Ellemers, N. (1993). The inﬂuence of socio-structural variables on identity management strategies. European Review of Social Psychology, 4, 27–57. Feather, N. T. (1994). Attitudes towards high achievers and reactions to their fall: Theory and research concerning tall poppies. Advances in Experimental Social Psychology, 26, 1–73. Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 117–140. Gaertner, S. L., Dovidio, J. F., Anastasio, P. A., Bachman, B. A., & Rust, M. C. (1993). Reducing intergroup bias: The common ingroup identity model. European Review of Social Psychology, 4, 1–26. Gaertner, S. L., Mann, J., Murrell, A., & Dovidio, J. F. (1989). Reducing intergroup bias: The beneﬁts of recategorization. Journal of Personality and Social Psychology, 57, 239–249. Gilbert, D. T., & Malone, P. S. (1995). The correspondence bias. Psychological Bulletin, 117, 21–38. Government of Canada (1971). Multicultural policy: Statement to House of Commons. Ontario, Canada: Hansard. Greenberg, J., Solomon, S., & Pyszczynski, T. (1997). Terror management theory of self-esteem and cultural worldviews: Empirical assessments and conceptual reﬁnements. In M. Zanna (Ed.), Advances in experimental social psychology (Vol. 29, pp. 61–139). Orlando, FL: Academic Press. Hamilton, D. L., & Sherman, S. J. (1996). Perceiving persons and groups. Psychological Review, 103, 336–355. Hamilton, D. L., Sherman, S. J., & Castelli, L. (2002). A group by any other name: The role of entitativity in group perception. European Review of Social Psychology, 12, 139–166.

Hogg and Hornsey 131 Hewstone, M. (1996). Contact and categorization: Social-psychological interventions to change intergroup relations. In C. N. Macrae, C. Stangor, & M. Hewstone (Eds.), Stereotypes and stereotyping (pp. 323–368). New York: The Guilford Press. Hogg, M. A. (2000). Subjective uncertainty reduction through self-categorization: A motivational theory of social identity processes. European Review of Social Psychology, 11, 223–255. Hogg, M. A. (2001a). Social categorization, depersonalization, and group behavior. In M. A. Hogg & R. S. Tindale (Eds.), Blackwell handbook of social psychology: Group processes (pp. 56–85). Oxford: Blackwell Publishing. Hogg, M. A. (2001b). A social identity theory of leadership. Personality and Social Psychology Review, 5, 184–200. Hogg, M. A. (2001c). Social identity and the sovereignty of the group: A psychology of belonging. In C. Sedikides & M. B. Brewer (Eds.), Individual self, relational self, and collective self: Partners, opponents, or strangers (pp. 125–146). Philadelphia, PA: Psychology Press. Hogg, M. A. (2001d). Self-categorization and subjective uncertainty resolution: Cognitive and motivational facets of social identity and group membership. In J. P. Forgas, K. D. Williams, & L. Wheeler (Eds.), The social mind: Cognitive and motivational aspects of interpersonal behavior (pp. 323–349). New York: Cambridge University Press. Hogg, M. A. (2001e). From prototypicality to power: A social identity analysis of leadership. In S. R. Thye, E. J. Lawler, M. W. Macy, & H. A. Walker (Eds.), Advances in group processes (Vol. 18, pp. 1–30). Oxford: Elsevier. Hogg, M. A. (2003). Social identity. In M. R. Learly & J. P. Tangney (Eds.), Handbook of self and identity (pp. 462–479). New York: The Guilford Press. Hogg, M. A. (2004). Uncertainty and extremism: Identiﬁcation with high entitativity groups under conditions of uncertainty. In V. Yzerbyt, C. M. Judd, & O. Corneille (Eds.), The psychology of group perception: Perceived variability, entitativity, and essentialism (pp. 401–418). New York: Psychology Press. Hogg, M. A. (2005). Social identity and leadership. In D. M. Messick & R. Kramer (Eds.), The psychology of leadership: Some new approaches (pp. 53–80). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Hogg, M. A., & Abrams, D. (1988). Social identiﬁcations: A social psychology of intergroup relations and group processes. London: Routledge. Hogg, M. A., Fielding, K. S., & Darley, J. (2005). Deviance and marginalization. In D. Abrams, J. Marques, & M. A. Hogg (Eds.), The social psychology of inclusion and exclusion (pp. 191–210). New York: Psychology Press. Hogg, M. A., Hains, S. C., & Mason, I. (1998). Identiﬁcation and leadership in small groups: Salience, frame of reference, and leader stereotypicality eﬀects on leader evaluations. Journal of Personality and Social Psychology, 75, 1248–1263. Hogg, M. A., & Reid, S. A. (2001). Social identity, leadership, and power. In A. Y. Lee-Chai & J. A. Bargh (Eds.), The use and abuse of power: Multiple perspectives on the causes of corruption (pp. 159–180). Philadelphia, PA: Psychology Press. Hogg, M. A., & Terry, D. J. (2000). Social identity and self-categorization processes in organizational contexts. Academy of Management Review, 25, 121–140. Hogg, M. A., & van Knippenberg, D. (2003). Social identity and leadership processes in groups. In M. P. Zanna (Ed.), Advances in experimental social psychology (Vol. 35, pp. 1–52). San Diego, CA: Academic Press.

132

Threat and multiple categorization

Hogg, M. A., & Williams, K. D. (2000). From I to we: Social identity and the collective self. Group Dynamics: Theory, Research, and Practice, 4, 81–97. Hornsey, M. J., Blackwood, L., & O’Brien, A. (2002). Subgroup relations within the Broad Left: An analysis of an election. Manuscript submitted for publication, University of Queensland. Hornsey, M. J., & Hogg, M. A. (2000a). Assimilation and diversity: An integrative model of subgroup relations. Personality and Social Psychology Review, 4, 143–156. Hornsey, M. J., & Hogg, M. A. (2000b). Subgroup relations: A comparison of the mutual intergroup diﬀerentiation and common ingroup identity models of prejudice reduction. Personality and Social Psychology Bulletin, 26, 242–256. Hornsey, M. J., & Hogg, M. A. (2000c). Intergroup similarity and subgroup relations: Some implications for assimilation. Personality and Social Psychology Bulletin, 26, 948–958. Hornsey, M. J., & Hogg, M. A. (2002). The eﬀects of status on subgroup relations. British Journal of Social Psychology, 41, 203–218. Hornsey, M. J., & Imani, A. (2004). Criticising groups from the inside and the outside: An identity perspective on the intergroup sensitivity eﬀect. Personality and Social Psychology Bulletin, 30, 365–383. Hornsey, M. J., Oppes, T., & Svensson, A. (2002). “It’s OK if we say it, but you can’t”: Responses to intergroup and intragroup criticism. European Journal of Social Psychology, 32, 293–307. Jetten, J., Spears, R., & Manstead, A. S. R. (1997). Identity threat and prototypicality: Combined eﬀects on intergroup discrimination and collective self-esteem. European Journal of Social Psychology, 27, 635–657. Jost, J. T., & Kramer, R. M. (2003). The system justiﬁcation motive in intergroup relations. In D. M. Mackie & E. R. Smith (Eds.), From prejudice to intergroup emotions: Diﬀerentiated reactions to social groups (pp. 227–246). New York: Psychology Press. Larson, J. R., Jr., Foster-Fishman, P. G., & Keys, C. B. (1994). Discussion of shared and unshared information in decision-making groups. Journal of Personality and Social Psychology, 67, 446–461. Leyens, J.-P., & Yzerbyt, V. Y. (1992). The ingroup overexclusion eﬀect: Impact of valence and conﬁrmation on stereotypical information search. European Journal of Social Psychology, 22, 549–569. Liebman, R. C., Sutton, J. R., & Wuthnow, R. (1988). Exploring the social sources of denominationalism: Schisms in American Protestant denominations, 1890–1980. American Sociological Review, 53, 343–52. Marques, J. M., Abrams, D., Páez, D., & Hogg, M. A. (2001). Social categorization, social identiﬁcation, and rejection of deviant group members. In M. A. Hogg & R. S. Tindale (Eds.), Blackwell handbook of social psychology: Group processes (pp. 400–424). Oxford: Blackwell Publishing. Marques, J. M, Abrams, D., & Serodio, R. (2001). Being better by being right: Subjective group dynamics and derogation of in-group deviants when generic norms are undermined. Journal of Personality and Social Psychology, 81, 436–447. Marques, J. M., & Páez, D. (1994). The black sheep eﬀect: Social categorization, rejection of ingroup deviates, and perception of group variability. European Review of Social Psychology, 5, 37–68. McAuliﬀe, B. J., Jetten, J., Hornsey, M. J., & Hogg, M. A. (2003). Individualist and

Hogg and Hornsey 133 collectivist group norms: When it’s OK to go your own way. European Journal of Social Psychology, 33, 57–70. Miller, D. T., & Prentice, D. A. (1999). Some consequences of a belief in group essence: The category divide hypothesis. In D. A. Prentice & D. T. Miller (Eds.), Cultural divides: Understanding and overcoming group conﬂict (pp. 213–238). New York: Russell Sage Foundation. Miller, N., & Brewer, M. B. (Eds.) (1984). Groups in contact: The psychology of desegregation. New York: Academic Press. Mugny, G. (1982). The power of minorities. London: Academic Press. Mulder, M. (1977). The daily power game. Leiden, The Netherlands: Martinus Nijhoﬀ Social Sciences Division. Mummendey, A., & Wenzel, M. (1999). Social discrimination and tolerance in intergroup relations: Reactions to intergroup diﬀerence. Personality and Social Psychology Review, 3, 158–174. Nemeth, C. (1986). Diﬀerential contributions of majority and minority inﬂuence. Psychological Review, 93, 23–32. Nemeth, C., & Owens, P. (1996). Making work groups more eﬀective: The value of minority dissent. In M. A. West (Ed.), The handbook of workgroup psychology (pp. 125–141). Chichester, UK: Wiley International. Nemeth, C., & Staw, B. M. (1989). The tradeoﬀs of social control and innovation in groups and organizations. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 22, pp. 175–210). San Diego, CA: Academic Press. Nemeth, C., & Wachtler, J. (1983). Creative problem solving as a result of majority vs. minority inﬂuence. European Journal of Social Psychology, 13, 45–55. Ng, S. H. (1996). Power: An essay in honour of Henri Tajfel. In W. P. Robinson (Ed.), Social groups and identities: Developing the legacy of Henri Tajfel (pp. 191–214). Oxford: Butterworth-Heinemann. Niedenthal, P. M., & Beike, D. R. (1997). Interrelated and isolated self-concepts. Personality and Social Psychology Review, 1, 106–128. Pettigrew, T. F. (1998). Intergroup contact theory. Annual Review of Psychology, 49, 65–85. Pickett, C. L., & Brewer, M. B. (2001). Assimilation and diﬀerentiation needs as motivational determinants of perceived in-group and out-group homogeneity. Journal of Experimental Social Psychology, 37, 341–348. Postmes, T., Spears, R., & Cihangir, S. (2001). Quality of decision making and group norms. Journal of Personality and Social Psychology, 80, 918–930. Prentice, D. A., & Miller, D. T. (Eds.) (1999). Cultural divides: Understanding and overcoming group conﬂict. New York: Russell Sage Foundation. Reicher, S. D., & Hopkins, N. (2001). Self and nation. London: Sage Publications. Reicher, S. D., Spears, R., & Postmes, T. (1995). A social identity model of deindividuation phenomena. European Review of Social Psychology, 6, 161–198. Roccas, S., & Brewer, M. B. (2002). Social identity complexity. Personality and Social Psychology Review, 6, 88–109. Ross, L. (1977). The intuitive psychologist and his shortcomings. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 10, pp. 174–220). New York: Academic Press. Rothgerber, H. (1997). External intergroup threat as an antecedent to perceptions of in-group and out-group homogeneity. Journal of Personality and Social Psychology, 73, 1206–1212.

134

Threat and multiple categorization

Rubin, M., & Hewstone, M. (1998). Social identity theory’s self-esteem hypothesis: A review and some suggestions for clariﬁcation. Personality and Social Psychology Review, 2, 40–62. Sachdev, I., & Bourhis, R. Y. (1985). Social categorization and power diﬀerentials in group relations. European Journal of Social Psychology, 15, 415–434. Sani, F., & Reicher, S. D. (1998). When consensus fails: An analysis of the schism within the Italian Communist Party (1991). European Journal of Social Psychology, 28, 623–645. Sani, F., & Reicher, S. D. (1999). Identity, argument and schisms: Two longitudinal studies of the split in the Church of England over the ordination of women to the priesthood. Group Processes and Intergroup Relations, 2, 279–300. Sani, F., & Reicher, S. D. (2000). Contested identities and schisms in groups: Opposing the ordination of women as priests in the Church of England. British Journal of Social Psychology, 39, 95–112. Schachter, S. (1951). Deviation, rejection, and communication. Journal of Abnormal and Social Psychology, 46, 190–207. 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. 209–296). New York: Academic Press. Sherif, M. (1958). Superordinate goals in the reduction of intergroup conﬂicts. American Journal of Sociology, 63, 349–356. Sherif, M. (1966). In common predicament: Social psychology of intergroup conﬂict and cooperation. Boston, MA: Houghton-Miﬄin. Sigelman, L. (1986). Basking in reﬂected glory revisited: An attempt at replication. Social Psychology Quarterly, 49, 90–92. Smith, E. R., & Henry, S. (1996). An in-group becomes part of the self: Response time evidence. Personality and Social Psychology Bulletin, 22, 635–642. Stasser, G., Stewart, D. D., & Wittenbaum (1995). Expert roles and information exchange during discussion: The importance of knowing who knows what. Journal of Experimental Social Psychology, 31, 244–265. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S. Worchel & W. Austin (Eds.), Psychology of intergroup relations (pp. 7–24). Chicago: Nelson-Hall. Terry, D. J., Carey, C. J., & Callan, V. J. (2001). Employee adjustment to an organizational merger: An intergroup perspective. Personality and Social Psychology Bulletin, 27, 267–280. Tindale, R. S., Kameda, T., & Hinsz, V. B. (2003). Group decision-making. In M. A. Hogg & J. Cooper (Eds.), The Sage handbook of social psychology (pp. 381–403). London: Sage Publications. 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 (pp. 6–34). Oxford: Blackwell. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. Oxford: Blackwell Publishing. Tyler, T. R. (1997). The psychology of legitimacy: A relational perspective on voluntary deference to authorities. Personality and Social Psychology Review, 1, 323–345.

Hogg and Hornsey 135 Vescio, T. K., Hewstone, M., Crisp, R. J., & Rubin, J. M. (1999). Perceiving and responding to multiple categorizable individuals: Cognitive processes and aﬀective intergroup bias. In D. Abrams & M. A. Hogg (Eds.), Social identity and social cognition (pp. 111–140). Oxford: Blackwell Publishing. Wann, D. L., Hamlet, M. A., Wilson, T. M., & Hodges, J. A. (1995). Basking in reﬂected glory, cutting oﬀ reﬂected failure, and cutting oﬀ future failure: The importance of group identiﬁcation. Social Behavior and Personality, 23, 377–388. Wicker, A. W. (1968). Undermanning, performances, and students’ subjective experiences in behavior settings of large and small high schools. Journal of Personality and Social Psychology, 10, 225–261. Wittenbaum, G. M., & Stasser, G. (1996). Management of information in small groups. In J. L. Nye & A. M. Brower (Eds.), What’s social about social cognition (pp. 3–28). Thousand Oaks, CA: Sage Publications. Wright, S. C., Aron, A., & Tropp, L. R. (2002). Including others (and groups) in the self: Self-expansion and intergroup relations. In J. P. Forgas & K. D. Williams (Eds.), The social self: Cognitive, interpersonal, and intergroup perspectives (pp. 343–363). New York: Psychology Press. Yzerbyt, V., Castano, E., Leyens, J.-P., & Paladrino, M.-P. (2000). The primacy of the ingroup: The interplay of entitativity and identiﬁcation. European Review of Social Psychology, 11, 257–295. Yzerbyt, V. Y., Leyens, J.-P., & Bellour, F. (1995). The ingroup overexclusion eﬀect: Identity concerns in decisions about group membership. European Journal of Social Psychology, 25, 1–16.

Part IV

Cross-cutting categorization and evaluation

7

The crossed categorization hypothesis Cognitive mechanisms and patterns of intergroup bias Theresa K. Vescio, Charles M. Judd, and Poh-Pheng Chua

The social categorization of people as ingroup and outgroup members has been shown to lead to two robust eﬀects. First, categorization (e.g., according to gender) leads to the perceptual accentuation of similarities within groups (e.g., women perceived as more similar to other women and men to other men) and diﬀerences between groups (e.g., women and men perceived as increasingly diﬀerent from one another; see Doise, 1978; Doise, Deschamps, & Meyer, 1978; Tajfel & Wilkes, 1963). Second, bias and discrimination follow from social categorization when individuals belong to one of the groups (for reviews see Brewer, 1979; Doise, 1978; Messick & Mackie, 1989; Tajfel, 1978; Wilder, 1986). In fact, the mere classiﬁcation of people into ingroups and outgroups on the basis of trivial distinctions has been shown to result in biased evaluations (e.g., Brewer & Silver, 1978; Doise et al., 1978; Kahn & Ryen, 1972) and the unequal distribution of rewards (e.g., Tajfel, Flament, Billig, & Bundy, 1971; Turner, 1975). Deschamps and Doise (1978) assumed that intergroup bias resulted from the accentuation processes underlying social categorization. In simple categorization conditions, where individuals can be classiﬁed into ingroups or outgroups according to only one dimension (e.g., race ingroup or outgroup), the cognitive accentuation of similarities within groups and diﬀerences between groups should motivate intergroup bias. In most situations, however, individuals can be classiﬁed as ingroup or outgroup members along multiple dimensions. Not surprisingly, when multiple dimensions converge to diﬀerentiate ingroups from outgroups (e.g., all ingroup members are Latino and Democrat, whereas all outgroup members are white and Republican), category diﬀerentiation is stronger and intergroup bias is exacerbated (e.g., Biernat & Vescio, 1993; see also Deschamps, 1977; Oakes, Turner, & Haslam, 1991). In other crossed categorization situations, categorization dimensions cross such that others belong to one of four subgroups created by the conjunction of the dimensions. When race (Asian or white) and gender (women and men) are crossed, for example, individuals are Asian women, Asian men, white women, or white men. With respect to a given perceiver (e.g., an Asian man), others are then either ingroup members according to both dimensions (or

140

Categorization and intergroup bias

double ingroups, e.g., other Asian men), outgroup members according to both dimensions (or double outgroups, e.g., white women), or ingroup members according to one dimension but outgroup members according to the second dimension (or mixed groups, e.g., Asian women and white men). In crossed categorization contexts, Deschamps and Doise (1978) suggested that the category accentuation processes underlying social categorization along each individual dimension conﬂict and cancel. The accentuation of similarities according to one dimension (e.g., men similar to one another) should be counteracted by the simultaneous tendency to accentuate diﬀerences along the second dimension (e.g., Asian men and white men diﬀerent). Similarly, the accentuation of diﬀerences according to the ﬁrst dimension (e.g., women and men diﬀerent) should be counteracted by the tendency to accentuate similarities along the second dimension (e.g., Asian women and Asian men similar to one another). Because Deschamps and Doise assumed that categorization was the mechanism underlying bias, the cancellation of category diﬀerentiation processes within crossed conditions was predicted to result in the elimination of intergroup bias (see also Crisp & Hewstone, this volume). To examine this possibility, Deschamps and Doise (1978) crossed an experimentally created dimension with gender. They found that bias was eliminated in crossed conditions; double ingroups, mixed groups, and double outgroups were evaluated similarly. Although there was no direct examination of the category diﬀerentiation processes underlying the perception of social stimuli in crossed contexts, Deschamps (1977) demonstrated that the perceptual diﬀerentiation of squares (e.g., accentuation of similarities and diﬀerences between squares belonging to small or large groups) was reduced when a second dimension was crossed with size (i.e., size × color crossed conditions) compared to when squares diﬀered only in size (e.g., small or large group; see also Crisp & Hewstone, 1999). Deschamps and Doise’s (1978) original theorizing and ﬁndings have inspired a great deal of research, most of which has examined patterns of evaluative judgment (but not underlying cognitive processes) in crossed contexts. However, because few studies have precisely replicated Deschamps and Doise’s ﬁndings that bias is eliminated in crossed contexts, subsequent researchers have modiﬁed this original hypothesis. Some researchers have addressed a weaker version of Deschamps and Doise’s prediction, questioning: is bias reduced (rather than eliminated) in crossed versus simple conditions? (e.g., Brown & Turner, 1979; Commins & Lockwood, 1978; Deschamps & Doise, 1978, Eurich-Fulcer & Schoﬁeld, 1995; Marcus-Newhall, Miller, Holtz, & Brewer, 1993; Migdal, Hewstone, & Mullen, 1998; Rehm, Lilli, & Van Eimeren, 1988; Vanbeselaere, 1987, 1991). Other researchers have considered patterns of bias just within crossed conditions, addressing the question of whether bias toward some subgroups (e.g., mixed groups), but perhaps not all subgroups (e.g., double outgroups), is reduced in crossed contexts (e.g., Brewer, Ho, Lee, & Miller, 1987; Hagendorn & Henke, 1991; Hewstone,

Vescio et al.

141

Islam, & Judd, 1993; Migdal et al., 1998; Singh, Yoeh, Lim, & Kim, 1997; Urban & Miller, 1998; see also Miller et al., this volume). The goal of this chapter is to consider both the intergroup bias and categorization process outcomes associated with each of these two questions. Although many studies have examined patterns of intergroup bias in crossed categorization conditions, very few studies have examined the category differentiation processes underlying the perception of social stimuli in crossed conditions. Therefore the questions of whether the category diﬀerentiation of social stimuli is reduced in crossed contexts, and whether reductions in category diﬀerentiation are responsible for any reduction in intergroup bias, have not been comprehensively considered. To provide a foundation for a consideration of these questions, we do three things. First, we articulate the two research questions that have motivated prior crossed categorization research, noting the eﬀects that each hypothesis would predict on both categorization and intergroup bias outcomes. Second, we present a research design that permits a simultaneous examination of the intergroup bias and social categorization outcomes predicted by these hypotheses, which we call a full crossed categorization design. Third, we discuss how the categorization and bias eﬀects predicted by each hypothesis could be examined within such a design. We then review some ﬁndings from our lab and discuss the implications of these ﬁndings for future crossed categorization research and theorizing.

THE TWO CROSSED CATEGORIZATION HYPOTHESES As we have noted, two questions have motivated most crossed categorization research.

Question 1: The classic hypothesis Paralleling the classic theorizing of Deschamps and Doise (1978; Deschamps, 1977), some crossed categorization researchers have designed studies to examine the question of whether categorization and intergroup bias is reduced in crossed compared to simple contexts. If the accentuation processes associated with categorization cancel each other out in crossed contexts (Deschamps & Doise, 1978), then the strength of categorization along each dimension should be reduced in crossed compared to simple contexts. If categorization aﬀects intergroup bias, then bias should also be reduced in crossed as compared to simple contexts. If race and gender are crossed, for example, then categorization and bias according to both race and gender should be weaker in crossed contexts (where targets are Asian women, Asian men, white women, and white men) than in appropriate simple comparison contexts (i.e., targets are Asian and white or women and men targets). Testing these predictions involves comparing the strength of the ingroup–outgroup main eﬀect

142

Categorization and intergroup bias

along each dimension in the crossed versus simple contexts, on both categorization and intergroup bias variables.

Question 2: The moderation hypothesis Some contemporary crossed categorization research has examined patterns of intergroup bias just within crossed conditions, where individuals belong to one of four subgroups created by the conjunction of two crossed dimensions (e.g., in a race × gender crossed design people are Asian women, Asian men, white women, and white men). These inquiries have been based on the logic that bias toward people who are outgroup members according to one dimension might be reduced if those people are simultaneously ingroup members rather than outgroup members according to the second cross-cutting dimension (or belong to mixed groups rather than double outgroups). The notion that bias toward mixed groups may be reduced in crossed contexts suggests a moderation eﬀect. For example, if race and gender are crossed, gender bias should be less when others are racial ingroups rather than racial outgroups. Equivalently, racial bias should be reduced when others are gender ingroup members rather than gender outgroup members. To illustrate how such a relationship reﬂects moderation, consider the subgroups of a race × gender crossed condition. If bias toward mixed groups is reduced, this would be clearly demonstrated if ratings (e.g., on a ﬁve-point scale with higher numbers more favorable) of mixed groups were more similar to ratings of double ingroups than double outgroups, as is the case depicted in Outcome 1 of Table 7.1. In this situation, race bias (or the diﬀerence in race ingroup versus race outgroup ratings) is less when targets are gender ingroup (e.g., 5 − 4 = 1) than gender outgroup (e.g., 4 − 1 = 3) and gender bias is less when targets are race ingroup (e.g., 5 − 4 = 1) than race outgroup (e.g., 4 − 1 = 3), reﬂecting moderation of bias along one dimension by the ingroup–outgroup status of the targets on the second dimension. Evidence of such moderation would be revealed by a signiﬁcant interaction of the two dimensions on bias measures. If ratings of mixed groups are more similar to ratings of double outgroups than ratings of double ingroups, there would also be evidence of moderation, but in the opposite direction from that predicted by the hypothesis. As Outcome 2 of Table 7.1 shows, in such a situation, race bias is greater when targets are gender ingroup (e.g., 5 − 2 = 3) than gender outgroup (e.g., 2 − 1 = 1), and gender bias is greater when targets are race ingroup (e.g., 5 − 2 = 3) than race outgroup (e.g., 2 − 1 = 1). Finally, there is a third possible pattern of bias that may emerge. Ratings of mixed groups may be between (i.e., approximately equal distance from) double ingroup and double outgroup ratings. This situation is depicted in Outcome 3 of Table 7.1. When such patterns of bias emerge, conclusions about the relative bias reduction beneﬁts of crossed contexts cannot be determined via an examination of bias solely within crossed conditions. This

Vescio et al.

143

Table 7.1 Illustration of the moderation hypothesis in a race × gender crossed context Outcome 1: Mixed groups more similar to double ingroups than double outgroups Race ingroup

Race outgroup

Race bias ingroup–outgroup

Gender ingroup

5

4

Gender outgroup

4

1

1 Gender ingroup 3 Gender outgroup

Gender bias ingroup– outgroup

1 Race ingroup

3 Race outgroup

Outcome 2: Mixed groups more similar to double outgroups than double ingroups Race ingroup

Race outgroup

Race bias ingroup– outgroup

Gender ingroup

5

2

Gender outgroup

2

1

3 Gender ingroup 1 Gender outgroup

Gender bias ingroup– outgroup

3 Race ingroup

1 Race outgroup

Outcome 3: Mixed groups between double ingroups and double outgroups Race ingroup

Race outgroup

Race bias ingroup– outgroup

Gender ingroup

5

3

Gender outgroup

3

1

2 Gender ingroup 2 Gender outgroup

Gender bias ingroup– outgroup

2 Race ingroup

2 Race outgroup

pattern may emerge if bias is reduced in crossed as compared to simple conditions. Alternatively, this pattern may result from the simple additive combination of two main eﬀects that are equally strong in simple and crossed conditions (see Brown & Turner, 1979). Thus, conclusions regarding whether bias is reduced given this pattern necessitates comparisons across simple and crossed conditions. Researchers who have examined patterns of intergroup bias within crossed conditions have not typically examined category diﬀerentiation eﬀects.

144

Categorization and intergroup bias

However, it seems reasonable that similar results would occur if categorization eﬀects indeed underlie intergroup bias eﬀects, as Deschamps and Doise (1978) claimed. Additionally, it has recently been suggested that theoretical arguments positing category diﬀerentiation processes as the sole mechanisms driving intergroup bias in crossed contexts (compared to perspectives positing both category diﬀerentiation and motivational mechanisms; e.g., Brown & Turner, 1979) provide the best account of the commonly found patterns of intergroup bias (e.g., Migdal et al., 1998). If this is the case, then predictions for categorization eﬀects should parallel the intergroup bias predictions. The strength of category diﬀerentiation along one dimension (e.g., race) should be reduced when targets are ingroup members on the second dimension (i.e., gender ingroup) compared to when they are outgroup members on that second dimension (i.e., gender outgroup).

A full crossed categorization design To comprehensively test the predictions of the classic hypothesis and the moderation hypothesis in a single experimental design, one must use a full crossed categorization design, which has three critical features. First, a full crossed categorization design includes three experimental conditions: a crossed condition and two simple conditions. In the crossed condition, two dimensions are arrayed such that targets belong to each of the four subgroups represented by the category conjunctions. There are also two simple conditions in which each of the categorization dimensions used in the crossed condition is presented in isolation. If race and gender are the dimensions used in the crossed condition, then in a simple race condition people can be classiﬁed as ingroup or outgroup members solely according to race (e.g., people are Asian or white with gender unspeciﬁed). Likewise, in a simple gender condition people can be classiﬁed as ingroup or outgroup only on the basis of gender (e.g., people are men or women with race unspeciﬁed). Second, participants must be recruited on the basis of their membership in the subgroups represented in the crossed condition. For example, if an experiment uses target stimuli in which race and gender are crossed (where targets are Asian women, Asian men, white women, and white men), then these dimensions must also be crossed in terms of participants’ own group membership (participants must include Asian women, Asian men, white women, and white men). If this is not the case, then ingroup–outgroup diﬀerences will be confounded with one or both of the categorization dimensions (race and/or gender). Finally, critical tests of the two hypotheses require the inclusion of measures that examine both category diﬀerentiation and intergroup bias outcomes.

Vescio et al.

145

Estimating eﬀects in the full crossed categorization design Given this design, we now turn to procedures that might be used to examine both hypotheses (the classic one and the moderation one) on both bias and categorization outcomes. In the case of bias, typically participants have been asked to evaluate members of the various groups or to choose members for future interactions. Given these measures, the analytic approaches to the classic and moderation hypotheses in the full crossed design are relatively straightforward. To assess the classic hypothesis (that intergroup bias along one dimension is less strong in crossed than in simple conditions), one wants to compare the magnitude of the ingroup–outgroup diﬀerence in evaluation (or choice probability) along a given dimension in the crossed condition to the same diﬀerence in the appropriate simple condition. From the crossed condition, in order not to confound the classic hypothesis with the moderation one, the estimation of the ingroup–outgroup diﬀerence in evaluation should collapse across both levels of the second crossed dimension. Thus the appropriate comparison is the main eﬀect of one dimension in the crossed condition with the eﬀect of that same dimension in the appropriate simple condition. The moderation hypothesis focuses only on the crossed condition. It is tested by examining the interaction of the two dimensions: the ingroup– outgroup evaluation (or choice probability) diﬀerence should be greater along one dimension when the groups are ingroup on the second than when they are outgroup on the second. Of course, given that interactions can be interpreted equivalently by focusing on comparisons of simple eﬀects of either one or the other dimension, if the ingroup–outgroup status on a second dimension aﬀects the magnitude of the ingroup–outgroup diﬀerence on the ﬁrst, then the same holds when the two dimensions are switched. In other words, if one dimensional eﬀect is moderated, then the other is as well. Turning to categorization outcomes, the most common method that has been used to assess categorization in crossed designs is the “who said what?” paradigm (originally developed by Taylor, Fiske, Etcoﬀ, & Ruderman, 1978; and used in crossed designs by Arcuri, 1982; Biernat & Vescio, 1993; Stangor, Lynch, Duan, & Glass, 1992; van Knippenberg, van Twuyver, & Pepels, 1994; van Twuyver & van Knippenberg, 1995). This procedure involves participants attempting to recognize and correctly attribute to speakers comments that they previously heard from those speakers. The crucial analysis focuses on recognition errors, when statements are misattributed to a person other than the correct speaker. Stemming from the operation of underlying accentuation processes (i.e., accentuation of similarities within and diﬀerences between categories), the categorization prediction is that more withincategory recognition errors (e.g., falsely attributing a statement said by a female to another female) should occur than between-category errors (e.g., falsely attributing a statement said by a female to a male). Examining the categorization eﬀects predicted by the classic hypothesis in

146

Categorization and intergroup bias

the context of the “who said what?” paradigm requires the computation of within- and between-category errors along each dimension of the crossed conditions. If race and gender are crossed, within- and between-race errors are calculated in the crossed conditions ignoring target gender. Similarly, within- and between-gender errors are calculated in crossed condition ignoring target race. Parallel within- and between-category errors are also computed in the simple conditions (e.g., the simple race and simple gender conditions). Then the magnitude of the within versus between error diﬀerence along each dimension is compared in crossed and appropriate simple conditions. If category diﬀerentiation processes conﬂict and cancel one another in crossed contexts, then there should be evidence of weaker categorization along each dimension in the crossed condition rather than the simple conditions. In the context of the “who said what?” paradigm, the moderation hypothesis is assessed by examining recognition errors along each dimension, considering the ingroup versus outgroup status of the person who actually made a given comment and the ingroup versus outgroup status of the person to whom the comment was erroneously attributed. Consider, for example, a race × gender crossed design. There are four features of the comments, each with two levels (ingroup or outgroup), that when crossed depict the possible errors in the “who said what?” paradigm. These include: (1) Speaker Race (was the speaker race ingroup or race outgroup?); (2) Speaker Gender (was the speaker gender ingroup or gender outgroup?); (3) Race Misattributed (was the comment misattributed to a race ingroup or race outgroup person?), and (4) Gender Misattributed (was the comment misattributed to a gender ingroup or gender outgroup person?). The possible errors are depicted in Table 7.2, with these four factors deﬁning the table’s rows and columns. To test whether gender categorization is moderated by the ingroup– Table 7.2 Possible memory errors in a race × gender crossed categorization condition Actual speaker Ingroup race

Outgroup race

Ingroup gender

Outgroup gender

Ingroup gender

Outgroup gender

Ingroup race Ingroup gender Outgroup gender

WR-WG WR-BG

WR-BG WR-WG

BR-WG BR-BG

BR-BG BR-WG

Outgroup race Ingroup gender Outgroup gender

BR-WG BR-BG

BR-BG BR-WG

WR-WG WR-BG

WR-BG WR-WG

Misattributed to:

Note: WR = within-race error; BR = between-race error; WG = within-gender error; BG = between-gender error.

Vescio et al.

147

outgroup race status of the speaker (i.e., is gender categorization weaker when targets are race ingroup than race outgroup?), the diﬀerence between within-gender errors (WG = WR − WG + BR − WG) and between-gender errors (BG = WR − BG + BR − BG) should be compared across race ingroup targets (the errors in columns 1 and 2) and race outgroup targets (the errors in columns 3 and 4). To examine whether race categorization is moderated by the ingroup–outgroup gender status of speakers, the diﬀerence between within-race errors (WR = WR − WG + WR − BG) and between-race errors (BR = BR − WG + BR − BG) should be compared across gender ingroup targets (the errors in columns 1 and 3) and gender outgroup targets (the errors in columns 2 and 4). Insofar as the within- versus between-category diﬀerence in errors depends on the ingroup–outgroup status of speakers along the second dimension, the moderation hypothesis would be supported.1

RESEARCH FROM OUR LAB We used a full crossed categorization design to examine both the categorization and intergroup bias predictions from the classic and the moderation hypotheses (Vescio, Judd, & Kwan, 2004). To our knowledge, this work is the ﬁrst to examine both hypotheses and to examine them for both categorization (using the “who said what?” paradigm) and intergroup bias. Under the guise of an investigation of college adjustment experiences, we asked participants in two diﬀerent experiments to read a transcript of a group discussion in which eight students talked about the problems they faced adjusting to college. Each of the eight targets contributed to the discussion twice and the discussion contributions were presented on 16 consecutive pages of an experimental booklet. On each page, participants saw a photograph of a target individual, beneath which his or her discussion contributions were presented. Across experiments, discussion contributions were unrelated to categorization dimensions and the speciﬁc discussion contributions were counterbalanced with the category membership of the target who made them (for details see Vescio et al., 2004).2 After reading a discussion, participants in both experiments rated each of the eight targets along several 1

2

Alternatively, the moderation hypothesis could be tested by submitting each of the 16 kinds of errors in the full matrix (Table 7.2) to an ANOVA in which each of the four factors that deﬁne the matrix are treated as within-participant factors. Two three-way interactions would provide evidence of moderation. A speaker gender × gender misattributed × speaker race interaction would reveal that the gender categorization varies depending on the race of the speaker, whereas a speaker race × race misattributed × speaker gender interaction would reveal that race categorization varies depending on the gender of the speaker. In both studies we made sure that discussion contribution-target photo pairings were fully counterbalanced and that the content of the stimulus materials was unrelated to the available categorization dimensions. This prohibited the use of category-based guessing strategies that can inﬂate the ratio of within-to between-category errors (see Klauer & Wegener, 1998).

148

Categorization and intergroup bias

evaluative and behavioral dimensions (e.g., participants rated how much they would like to have this person as a friend, indicated how comfortable they would be interacting with this person, and selected four targets with whom they would like to interact). These ratings provided measures of intergroup bias. Participants also completed a recall task, which required that they match comments made during the discussion to the speaker, who made the comment. This permitted an estimation of within-group errors versus between-group errors, which provided a measure of categorization strength.

Experiment 1: Race × gender full crossed categorization design In Experiment 1, the racial and gender composition of the discussion group was manipulated to create crossed and simple categorization conditions. In the crossed conditions, race and gender were crossed such that the discussion group involved two Asian women, two Asian men, two white women, and two white men. In the simple race condition discussants were four Asian and four white targets (all of the same gender as participants). In the simple gender condition discussants were four women and four men (all of the same race as participants). Participants were sampled from each of the subgroups represented in the crossed conditions and included 99 Asian women, 72 Asian men, 67 white women, and 43 white men. After reading the transcript of the group discussion, participants indicated their favorability toward each target on a variety of ratings. We averaged across ratings of appropriate targets to create ingroup and outgroup liking measures, which provided measures of intergroup bias.3 At the conclusion of the experiment, participants completed a surprise recognition task, where they matched quotes taken from the discussion with the speaker who made the comment. Recognition errors were coded and within- and betweencategory errors were computed as described above, which provided a measure of categorization strength. Categorization ﬁndings The classic hypothesis asks if the strength of categorization is reduced in crossed compared to simple conditions. Figure 7.1 (four leftmost bars) presents a summary of results stemming from our comparison of the magnitude of the within-group versus between-group error diﬀerence in simple versus crossed conditions on each dimension. Higher bars reﬂect stronger categorization eﬀects (within- minus between-group errors). Strongly supporting the classic hypothesis, categorization on the basis of each dimension (race and 3

Participants also selected four of the eight participants with whom they would like to interact. These ratings were intended to provide a measure of behavioral choice (or a more behavior-like measure of intergroup bias). Findings on this variable parallel the ﬁndings of liking (for a full report see Vescio et al., 2004).

Vescio et al.

149

Figure 7.1 Categorization (within-group minus between-group errors) in simple versus crossed conditions, Experiments 1 and 2.

gender) was reduced in crossed conditions compared to relevant simple conditions. In both the crossed and the simple conditions, there was evidence of signiﬁcant categorization according to both race and gender; the withinminus between-group error diﬀerence signiﬁcantly diﬀered from zero on each dimension (race and gender) in each condition (crossed and simple conditions). However, the magnitude of both race-based and gender-based categorization eﬀects was signiﬁcantly weaker in crossed than simple conditions, strongly supporting the classic hypothesis. Consistent with the moderation hypothesis, there was also evidence that race-based categorization was moderated by the gender ingroup versus outgroup status of the targets. As the left panel of Figure 7.2 shows, categorization on the basis of race was weaker when targets were gender ingroup members than gender outgroup members. There was, however, no evidence that the strength of gender-based categorization varied depending on the racial ingroup versus outgroup status of the targets. Taken together, the categorization ﬁndings were strongly consistent with the classic hypothesis and partially consistent with the moderation hypothesis (i.e., categorization along one dimension, but not the other, was moderated by the targets’ ingroup versus outgroup status on the cross-cutting second dimension). Intergroup bias ﬁndings As we have noted, the classic hypothesis of Deschamps and Doise (1978) states that category diﬀerentiation processes operating across the two dimensions in

150

Categorization and intergroup bias

Figure 7.2 Categorization (within-group minus between-group errors) within crossed conditions, Experiments 1 and 2. IG: ingroup; OG: outgroup.

crossed conditions conﬂict and cancel one another. Furthermore, because categorization is presumed to drive patterns of intergroup bias, reductions in the strength of intergroup bias should follow. Bias should be weaker in crossed than simple conditions. To examine this prediction we compared the strength of the (a) race ingroup–outgroup bias main eﬀect in the crossed versus the simple race condition, and (b) gender ingroup–outgroup bias main eﬀect in the crossed versus the simple gender condition. Results revealed evidence of bias along each dimension. Participants liked race ingroup members more than race outgroup members, and they liked gender ingroup members more than gender outgroup members. However, there was no evidence that the magnitude of bias along either dimension was less strong in the crossed than in the simple conditions. Thus, on bias, there was no evidence in support of the classic hypothesis. Within crossed conditions, there was also no evidence consistent with the moderation hypothesis. There was evidence of signiﬁcant race and gender bias within the crossed conditions (the race bias main eﬀect was the stronger of the two bias main eﬀects). However, the race bias × gender bias interaction did not approach signiﬁcance. Instead, patterns of intergroup bias resembled those depicted in Outcome 3 of Table 7.1. Replicating the most common patterns of bias found by others, as Table 7.3 shows, double ingroups were liked more than mixed groups, who were liked more than double outgroups. This pattern is commonly known as an additive pattern (i.e., no interaction): bias toward groups in the crossed condition is an additive function of bias on

Vescio et al.

151

Table 7.3 Bias toward subgroups of crossed conditions compared to ingroups and outgroups of simple conditions Target group membership Experiment

Condition

Double ingroup

Mixed group

Double outgroup

Simple Crossed

5.96a 6.00a

— 5.67b

5.66b 5.37c

Simple Crossed

5.73a 5.96a

— 5.68b

5.47c 5.44c

Experiment 1

Experiment 2

Note: Within experiments, means with diﬀerent superscripts diﬀer signiﬁcantly at p < 0.05.

each dimension. Importantly, in the present context where there is no evidence that bias along each dimension in the crossed conditions is reduced compared to the simple condition, this additivity pattern is entirely consistent with the simple prediction that crossed categories do not reduce intergroup bias. The relationship between categorization and intergroup bias To examine the relationship between categorization and intergroup bias, we created a series of variables mapping onto the eﬀects predicted by each hypothesis (and reported above). The categorization variables included indices of race categorization (i.e., within- minus between-race errors), gender categorization (i.e., within- minus between-gender errors), race categorization moderated by target’s gender ingroup versus outgroup status (or race categorization of gender ingroup targets minus gender outgroup targets), and gender categorization moderated by target’s race ingroup versus outgroup status (or gender categorization of race ingroup targets minus race outgroup targets). We created parallel intergroup bias variables. Then we examined correlations between the categorization and bias variables within each condition, collapsing across the two simple conditions, just within the crossed conditions, and collapsing across all conditions. Not surprisingly, the various intergroup bias variables were highly related. Bias in liking along one dimension was highly related to bias along the other dimension. The principal categorization variables also hung together. However, there were virtually no signiﬁcant categorization-bias correlations. The only one of these correlations to emerge as signiﬁcant was in the direction opposite to what one would expect, such that more categorization was associated with less bias. The absence of categorization–bias correlations in these data is quite surprising given the prominence of such assumed links and the

152

Categorization and intergroup bias

conclusions that have been reached by crossed categorization researchers regarding the viability of models positing category diﬀerentiation processes as the primary mechanisms motivating intergroup bias (e.g., Migdal et al., 1998). Why was there no reduction in bias in the crossed conditions, or any of the expected categorization–bias correlations, in Experiment 1? There are two possible explanations. First, race and gender are salient and meaningful social categories that people apply immediately, almost automatically, to others (see Brewer, 1988; Fiske & Neuberg, 1990). So, people may have extensive histories of biased responding according to both race and gender. In such instances, biased responding may become a sort of well-rehearsed “bad habit” that is not easily broken. If this is the case, then biased responding could have persisted out of habit in Experiment 1, despite evidence that the category diﬀerentiation processes had begun to break down in the crossed as compared to simple conditions. Second, crossed categorization researchers have suggested that crossed contexts should reduce bias only if the categorization dimensions are of equal psychological signiﬁcance (Diehl, 1989). Although race and gender are both important social categories, in Experiment 1 the race bias main eﬀect was a stronger eﬀect than was the gender bias main eﬀect. If race is a psychologically more signiﬁcant categorization dimension than gender, then crossing these dimensions should not reduce bias. To rule out these possibilities, we conducted a second experiment that used less socially charged categorization dimensions. The dimensions used in Experiment 2 were relationship status (single or partner/signiﬁcant other) and hometown size (big or little town), which are dimensions that split our participants into roughly equal subgroups.

Experiment 2: Relationship status × hometown size full crossed categorization design The procedure, counterbalancing, instructions, stimulus materials, and dependent variables were the same as in Experiment 1 with two exceptions. First, the target photographs were altered to ﬁt the new context. In the simple relationship status condition, only information about the targets’ relationship status was provided. This was indicated by the shape of the background on which the photographs were mounted. Photographs of single targets (not in romantic relationships) were mounted on oval backgrounds, whereas targets with romantic partners were mounted on heart-shaped backgrounds. In the simple hometown condition, only information about the targets’ hometown size was provided. This was indicated by the size of the background on which photographs were mounted. Big and small backgrounds indicated that targets were from big and little towns, respectively. In the crossed conditions, target relationship status and hometown size were crossed—there were two single/big town targets (on big ovals), two single/little town targets (on little

Vescio et al.

153

ovals), two partner/big town targets (on big hearts), and two partner/little town targets (on little hearts). All participants saw same-gender photographs, which were pilot tested for equivalence in attractiveness and randomly assigned to subgroup. Second, because we used novel visual cues to convey group membership, attention to multiple cues in the crossed conditions (mounting size and shape) may have been more taxing than attention to a single cue in the simple conditions (mounting size or shape). If so, then discussion comments might have been attended to less carefully or processed less deeply in the crossed conditions than in the simple conditions. If participants in the crossed conditions did not attend to or process speciﬁc comments, then random guessing could occur during the recognition task. Random guessing could lead to a smaller diﬀerence in within- versus between-category errors, creating the false appearance that categorization was weaker, when in fact information was simply not encoded. To rule out this possibility, the recognition task was expanded to include 16 foils (items not in the original discussion), as well as 16 quotes taken from the discussion. Participants were ﬁrst asked whether each comment actually occurred during the discussion. For items identiﬁed as discussion contributions, participants further indicated which speaker made each comment. The inclusion of foils allowed us to test how well participants could distinguish between comments that were made during the group discussion from comments that were not made. If the crossed conditions were indeed more taxing than the simple conditions, then participants in the crossed (compared to simple) conditions should be less able to distinguish actual contributions from the foils. There was no evidence that this was the case in the present data. Therefore, we turned attention to tests of the categorization and bias eﬀects predicted by the two hypotheses. Categorization ﬁndings The categorization ﬁndings of Experiment 2 replicated the Experiment 1 ﬁndings. As the right four bars of Figure 7.1 show, categorization according to both relationship status and hometown size was stronger in the simple condition than in the crossed condition. In fact, there was evidence of signiﬁcant categorization along each dimension in the simple conditions, but no evidence of categorization along either dimension in the crossed conditions. These ﬁndings are strongly supportive of the predictions of the classic hypothesis. As in Experiment 1, tests of the extent to which categorization along one dimension was moderated by the targets’ ingroup versus outgroup status on the second dimension were partially consistent with the predictions of the moderation hypothesis. Categorization according to relationship status (but not hometown size) was moderated by targets’ ingroup versus outgroup status on the second dimension. As the right panel of Figure 7.2 shows,

154

Categorization and intergroup bias

categorization by relationship status was signiﬁcant when targets were hometown outgroup members, but there was no evidence of reliable categorization on the basis of relationship status when targets were hometown ingroup members. Intergroup bias ﬁndings To test the bias predictions of the classic hypothesis, we examined the strength of the bias main eﬀects along each dimension. There was signiﬁcant bias related to relationship status and hometown size. Participants liked relationship ingroup members more than relationship outgroup members, and they liked hometown ingroup members more than hometown outgroup members. In contrast to the predictions of the classic hypothesis, however, these bias main eﬀects were as strong in the crossed as in the simple conditions. To test the predictions of the moderation hypothesis on intergroup bias, we examined patterns of liking just within the crossed conditions. Again, there was evidence of bias according to both relationship status and hometown size, but the interaction between these two eﬀects did not approach signiﬁcance on liking. Therefore, there was no evidence supportive of the predictions of the moderation hypothesis. As in Experiment 1, the patterns of bias that emerged within the crossed conditions resembled the patterns depicted by Outcome 3 of Table 7.1—double ingroups were liked more than were mixed groups, who were liked more than double outgroups (see Table 7.3). However, this pattern resulted from the additive combination of the bias main eﬀects along the two dimensions that were of equivalent magnitude in crossed as in the simple conditions. The relationship between categorization and intergroup bias As in Experiment 1, we computed categorization-bias correlations within crossed conditions, within the simple conditions, and collapsing across conditions. Not a single categorization–bias correlation was signiﬁcant in any of these matrices. Summary Across the two experiments, our categorization ﬁndings were strongly supportive of the classic hypothesis. Categorization on each dimension of both studies was weaker in the crossed than in the simple conditions. Examination of categorization eﬀects just within the crossed conditions also revealed ﬁndings partially consistent with the moderation hypothesis. In each experiment categorization along one dimension (but not the other) was weaker when targets were ingroup members rather than outgroup members according to the second dimension. However, we found no evidence in either experiment that bias was reduced in crossed conditions. While there was evidence of bias

Vescio et al.

155

along each dimension of both studies, the magnitude of that bias did not vary across crossed versus simple conditions, contrary to the classic hypothesis. And contrary to the moderation hypothesis, within the crossed conditions, bias on one dimension did not vary as a function of targets’ ingroup versus outgroup status on the second dimension. Therefore, the additivity pattern that we found on bias in the crossed conditions of each experiment (i.e., double ingroups liked more than mixed groups, who were liked more than double outgroups) resulted from the additive combination of the bias main eﬀects along the two dimensions that were equivalently strong in crossed as in the simple conditions. These ﬁndings raise an important methodological point that past studies of bias in crossed designs have tended to ignore. If researchers include only crossed conditions in their design and ﬁnd evidence of an additivity pattern on bias, there is no way to assess whether bias is reduced in crossed conditions (compared to simple ones). Tests of the two questions that have motivated most prior crossed categorization research require both (a) the use of a full crossed categorization design, and (b) examination of the bias main eﬀects in crossed versus simple conditions and the interaction within crossed conditions. An examination of the relative evaluations of the subgroups of crossed contexts in the absence of tests of the interaction and comparisons to simple conditions does not address the question of whether bias along the individual dimensions is reduced in crossed conditions.

THEORETICAL IMPLICATIONS The additivity pattern of intergroup bias has been the pattern most frequently found by crossed categorization researchers (i.e., double ingroups rated more favorably than mixed groups, who are rated more favorably than double outgroups). Interestingly, two models predict such patterns of bias in crossed contexts: the category diﬀerentiation model and the social identity model. Although a detailed consideration of the diﬀerences of these models is beyond the scope of the present work, one critical diﬀerence is important. The models diﬀer in their focus on underlying mechanisms. The category diﬀerentiation model posits that intergroup bias follows from the category diﬀerentiation processes underlying social categorization. In contrast, the social identity model suggests social identity motives are critical contributors to intergroup bias. It is not surprising, therefore, that considerable attention has focused on the question of the relative viability of the category diﬀerentiation model versus the social identity model. Quite surprisingly, however, is the fact that the models diﬀer in terms of their purely cognitive mechanisms (i.e., category diﬀerentiation model) versus mutual cognitive and motivational mechanisms (i.e., social identity model), yet very little prior research has simultaneously examined categorization eﬀects and patterns of intergroup bias in a single experimental context.

156

Categorization and intergroup bias

Instead, researchers have conducted post hoc inquiries in attempts to identify the potential mechanisms driving intergroup bias. For example, independent raters have been presented with descriptions of experimental contexts and asked to report their perceptions of the extent to which it seemed various factors would have inﬂuenced target evaluations (e.g., ingroup identity, identity threat, category salience; see Migdal et al., 1998). And based on these post hoc appraisals, recent conclusions have suggested that a cognitive model (i.e., one that posits category diﬀerentiation processes as the sole mechanisms motivating intergroup bias) best accounts for emergent patterns of intergroup bias in crossed contexts (see Migdal et al., 1998). The results from our research, however, suggest a strikingly diﬀerent conclusion. We found that the strength of categorization was reduced in crossed conditions and we documented an additivity pattern of intergroup bias. These ﬁndings are consistent with the logic underlying both the category diﬀerentiation model and the social identity model. However, the models then diverge. The category diﬀerentiation model predicts (a) a relationship between the two eﬀects such that categorization motivates intergroup bias, and (b) that the additivity pattern of intergroup bias stems from reductions in the strength of intergroup bias in crossed as compared to simple conditions. In contrast, the social identity model suggests that while categorization strength may be diminished, social identity motives remain intact. In both crossed and simple conditions, it is presumed that people want to distinguish ingroups from outgroups along each dimension (which is of equal magnitude in crossed and simple conditions). In crossed conditions, these two eﬀects have been suggested to combine additively, resulting in the most favorable evaluation of double ingroups then mixed groups then double outgroups. While we did not assess the critical mechanisms of social identity theory (e.g., self-esteem, ingroup identiﬁcation), we did ﬁnd two patterns in our data that are at once consistent with the social identity model and inconsistent with the category diﬀerentiation model. Namely, there was a complete lack of association between our categorization measures on the one hand and intergroup bias eﬀects on the other hand. We also found evidence of equally strong bias in crossed and simple conditions. Consistent with the contentions of the social identity model, categorization alone did not drive intergroup bias in our data and the magnitude of intergroup bias was not reduced. Instead, emergent additivity patterns of intergroup bias stemmed from the mere additive combination of the two bias main eﬀects (which were equally strong in crossed conditions and simple conditions), as suggested by Brown and Turner (1979).

CONCLUSION Social scientists have long assumed that cross-cutting social categories have the potential for reduced intergroup conﬂict and prejudice. Since Deschamps

Vescio et al.

157

and Doise (1978), the argument has been that when one dimension is crossed with another, the tendency to categorize individuals along one dimension will be reduced by the cross-cutting second dimension. Since the degree to which target individuals are categorized along a dimension was presumed to aﬀect the degree to which intergroup bias is manifested along that dimension, the argument went that this reduction in categorization in crossed contexts should be accompanied by a reduction in intergroup bias. We believe that the present data represent the ﬁrst complete analyses of the hypotheses contained in this argument. This is because of the characteristics of what we have referred to as the full cross categorization design, which has never before been implemented. The results of both studies suggest that the theoretical story is decidedly more mixed about the beneﬁts of crossed categories than the previous literature suggests. On the one hand, we have consistently found that categorization strength is reduced in crossed category conditions, either when we compare that condition with single conditions or when we ask whether categorization along one dimension is reduced in strength when targets are ingroup members on a second dimension. On the other hand, when we tested the intergroup bias hypotheses, we found no evidence that intergroup bias was reduced as a function of crossed categorization. Finally, the inconsistency of this very mixed message about the beneﬁts of crossed classiﬁcations is reﬂected by the absence of correlations between our categorization and bias measures. The absence of such correlations suggests to us that perhaps current theoretical models of complex intergroup relations need to be modiﬁed. Categorization and bias do not seem to be linked in the way we expected they would be if category diﬀerentiation processes drive, or contribute to, intergroup bias in crossed categorization contexts.

REFERENCES Arcuri, L. (1982). Three patterns of social categorization in attribution memory. European Journal of Social Psychology, 12, 271–282. 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. Brewer, M. B. (1979). Ingroup bias in the minimal intergroup situation: A cognitivemotivational analysis. Psychological Bulletin, 86, 307–324. Brewer, M. B. (1988). A dual process model of impression formation. In R. Wyer & T. Srull (Eds.), Advances in social cognition (Vol. 1, pp. 1–36). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Brewer, M. B., Ho, H., Lee, J., & Miller, N. (1987). Social identity and social distance among Hong Kong schoolchildren. Personality and Social Psychology Bulletin, 13, 156–165. Brewer, M. B., & Silver, M. (1978). Ingroup bias as a function of task characteristics. Journal of Social Psychology, 8, 393–400.

158

Categorization and intergroup bias

Brown, R. J., & Turner, J. C. (1979). The criss-cross categorization eﬀect in intergroup discrimination. British Journal of Social and Clinical Psychology, 18, 371–383. Commins, B., & Lockwood, J. (1978). The eﬀects on intergroup relations of mixing Roman Catholics and Protestants: An experimental investigation. European Journal of Social Psychology, 8, 383–386. Crisp, R. J., & Hewstone, M. (1999). Subcategorization of physical stimuli: Category diﬀerentiation and decategorization processes. European Journal of Social Psychology, 29, 665–671. Deschamps, J. C. (1977). Eﬀect of crossing category memberships on quantitative judgement. European Journal of Social Psychology, 7, 517–522. Deschamps, J. C., & Doise, W. (1978). Crossed category membership in intergroup relations. In H. Tajfel (Ed.), Diﬀerentiation between social groups (pp. 141–158) London: Academic Press. Diehl, M. (1989). Dichotomie und Diskriminierung: Die Auswirkungen von Kreuzkategorisierungen auf die Diskriminierung im Paradigma der minimalen Gruppen. Zeitschrift für Soziolpsychologie, 20, 92–102. Doise, W. (1978). Groups and individuals: Explanations in social psychology. Cambridge: Cambridge University Press. Doise, W., Deschamps, J.-C., & Meyer, G. (1978). The accentuation of intracategory similarities. In H. Tajfel (Ed.), Diﬀerentiation between social groups (pp. 141–158). London: Academic Press. Eurich-Fulcer, R., & Schoﬁeld, J. W. (1995). Correlated vs. uncorrelated social categorizations: The eﬀect on intergroup bias. Personality and Social Psychology Bulletin, 21, 149–159. Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression formation from category based to individuating process: Inﬂuences of information and motivation on attention and interpretation. Advances in Experimental Social Psychology, 23, 1–74. Hagendoorn, L., & Henke, R. (1991). The eﬀect of multiple category membership on intergroup evaluations in a North-Indian context: Class, caste, and religion. British Journal of Social Psychology, 30, 247–260. Hewstone, M., Islam, M. R., & Judd, C. M. (1993). Models of crossed categorization and intergroup relations. Journal of Personality and Social Psychology, 64, 779–793. Kahn, A., & Ryen, A. H. (1972). Factors inﬂuencing the bias towards one’s own group. International Journal of Group Tensions, 2, 33–50. Klauer, K. C., & Wegener, I. (1998). Unraveling social categorization in the “Who said what?” paradigm. Journal of Personality and Social Psychology, 75, 1155–1178. Marcus-Newhall, A., Miller, N., Holtz, R., & Brewer, M. B. (1993). Cross-cutting category membership with role assignment: A means of reducing intergroup bias. European Journal of Social Psychology, 32, 125–146. Messick, D. M., & Mackie, D. M. (1989). Intergroup relations. Annual Review of Psychology, 40, 45–81. Migdal, M. J., Hewstone, M., & Mullen, B. (1998). The eﬀects of crossed categorization on intergroup evaluations: A meta-analysis. British Journal of Social Psychology, 37, 303–324. Oakes, P. J., Turner, J. C., & Haslam, S. A. (1991). Perceiving people as group members: The role of ﬁt in the salience of social categorizations. British Journal of Social Psychology, 30, 125–144. Rehm, J., Lilli, W., & Van Eimeren, B. (1988). Reduced intergroup diﬀerentiation as a

Vescio et al.

159

result of self-categorization in overlapping categories: A quasi-experiment. European Journal of Social Psychology, 18, 375–379. Singh, R., Yeoh, B. S. E., Lim, D. I., & Kim, K. K. (1997). Crossed categorization eﬀects in intergroup discrimination: Adding vs. averaging. British Journal of Social Psychology, 36, 121–138. 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, 207–218. Tajfel, H. (1978). Diﬀerentiation between social groups: Studies in the social psychology of intergroup behavior. London: Academic Press. Tajfel, H., & Wilkes, A. L. (1963). Classiﬁcation and quantitative judgment. British Journal of Psychology, 54, 79–97. Tajfel, H., Flament, C., Billig, M., & Bundy, R. F. (1971) Social categorization and intergroup behaviour. Enterprise Journal of Social Psychology, 1, 149–178. Taylor, S. E., Fiske, S. T., Etcoﬀ, N., & Ruderman, A. (1978). The categorical and contextual bases of person memory and stereotyping. Journal of Personality and Social Psychology, 36, 778–793. Turner, J. W. (1975). Social comparison and social identity: Some prospects for intergroup behaviour. European Journal of Social Psychology, 5, 5–34. Urban, L. M., & Miller, N. (1998). A theoretical analysis of crossed categorization eﬀects: A meta-analysis. Journal of Personality and Social Psychology, 74, 894–908. Vanbeselaere, N. (1987). The eﬀect of dichotomous and crossed social categorizations upon intergroup discrimination. European Journal of Social Psychology, 17, 143–156. Vanbeselaere, N. (1991). The diﬀerent eﬀects of simple and crossed categorizations: A result of the category diﬀerentiation process or of diﬀerential category salience? In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 2, pp. 560–578). Chichester, UK: Wiley International. van Knippenberg, A., van Twuyver, M., & Pepels, J. (1994). Factors aﬀecting social categorization processes in memory. British Journal of Social Psychology, 33, 419–431. van Twuyver, M., & van Knippenberg, A. (1995). Social categorization as a function of priming. European Journal of Social Psychology, 25, 695–701. Vescio, T. K., Judd, C. M., & Kwan, V. S. Y. (2004). The crossed-categorization hypothesis: Evidence of reductions in the strength of categorization, but not intergroup bias. Journal of Experimental Social Psychology, 40, 478–496. Wilder, D. A. (1986). Social categorizations: Implications for creation and reduction of intergroup bias. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 19, pp. 291–345). New York: Academic Press.

8

Explaining the eﬀects of crossed categorization on ethnocentric bias Norman Miller, Jared B. Kenworthy, Carrie J. Canales, and Douglas M. Stenstrom

One of Northern Ireland’s most repeated jokes perfectly illustrates the tendency for social group memberships to determine our actions toward one another. A guy walks out of a Belfast bar into a dark alley and is soon backed up against the wall by a threatening man, who asks, “Are you Catholic or Protestant?” The guy replies, disconcertedly, “Neither. I’m Jewish.” The attacker pauses brieﬂy, then presses his potential victim again, “Well . . . are you a Catholic Jew or a Protestant Jew?” This joke also shows that some group memberships can be more important than others. This chapter discusses crossed and multiple group membership. We delineate factors that moderate the degree to which ethnocentric bias is reduced when a target person or group exhibits conﬂicting cues regarding ingroup and outgroup identity. We then discuss the relative importance of dimensions of categorization, including their relevance both to anticipated task activities and to other social categories. We next consider the eﬀects of the aﬀective state of the respondent and individuated or personalized interaction with outgroup members. We examine the eﬀects of the former variable separately and comparatively with respect to positive and negative aﬀect. We also consider the source of aﬀect, whether integral or incidental, as well as diﬀerential eﬀects of speciﬁc types of emotion and the degree to which they elicit approach or avoidance tendencies.

CROSSED CATEGORIZATION EFFECTS As with most social-psychological phenomena, ethnocentric bias can be moderated. Although its incidence in intergroup contexts is ubiquitous, its degree is less predictable. Recent research programs, however, have identiﬁed factors that reduce the bias displayed toward outgroup members. One such program is the crossed categorization paradigm. The crossed categorization paradigm (Deschamps & Doise, 1978) examines the inﬂuence of multiple group memberships on behavior. Speciﬁcally, it attempts to answer the question of whether bias toward outgroup individuals

Miller et al.

161

can be mitigated by the simultaneous presence of one or more ingroup cues. In instances of crossed categorization, a person is perceived as an ingroup member on one dimension of social categorization, but as an outgroup member on a second dimension. For example, a Spanish Catholic and a Corsican Catholic belong to the same religious group but to diﬀerent national groups. In typical crossed categorization experiments, a perceiver does not just evaluate other people who have conﬂicting category identities, such as ingroup– outgroup (io) or outgroup–ingroup (oi) people, but also people with simpler, convergent ingroup (ii) or outgroup (oo) identities, who can act as comparative baseline targets. Though this paradigm clearly remains a drastic oversimpliﬁcation of the social world, it is important because—like real-world settings— it provides conﬂicting cues about the ingroup and outgroup identities of some of the persons to be evaluated. In this respect its ecological validity exceeds that of many paradigms used to study prejudice. Evaluations of the four types of target persons deﬁned by the crossed categorization paradigm (ii, io, oi, oo) yield a variety of patterns (Migdal, Hewstone, & Mullen, 1998; Miller, Urban, & Vanman, 1998; Urban & Miller, 1998). In the additivity pattern (e.g., Hagendoorn & Henke, 1991), which is the most fundamental pattern (Migdal et al., 1998), evaluations of the double ingroup target (ii) are most positive, the mixed targets (io and oi) are equally intermediate, and the double outgroup target (oo) is most negative: (ii > io = oi > oo). Theorists have viewed the overlapping group memberships implicit in crossed categorizations as a structural feature of societies that reduces bias and discrimination (Coser, 1956; LeVine & Campbell, 1972; Simmel, 1908; see also Crisp & Hewstone, this volume). In experimental research, however, this hopeful expectation has not been consistently realized, suggesting that additional moderators are at work. At least ﬁve other patterns of evaluation have been obtained, with corresponding theoretical considerations vying for their explanation (see Migdal et al., 1998; Urban & Miller, 1998) (see Crisp & Hewstone, Vescio et al., this volume). Instead of focusing on the cognitive processes that underlie the general crossed categorization eﬀect (see Crisp, Ensari, Hewstone, & Miller, 2002, for an examination of these issues), we focus on factors that lead to deviation from the basic additive ﬁnding. A meta-analysis of the crossed categorization literature (Urban & Miller, 1998) revealed a number of theoretically derived moderators of the basic additive model: (a) the diﬀerential importance of the two social category dimensions; (b) the aﬀective valence of the experimental setting; (c) the opportunity for more individuated contact; and (d) cognitive overload. Because no crossed categorization research has yet to address cognitive overload eﬀects, this chapter deals only with the former three variables.

162

Moderating ethnocentric bias

IMPORTANCE OF CATEGORIES Within the crossed categorization paradigm, discrepant category importance refers to a situation in which one of the two categories that comprise a target person’s description has more cognitive impact than the other. For example, if two known category dimensions of a target are political aﬃliation and sports team aﬃliation, ingroup/outgroup diﬀerences on the latter will be more important at a hockey game whereas such diﬀerences on the former will be more important during an election campaign. In this example, importance is situationally moderated. It also can vary as a function of stable person factors (Urada & Miller, 2000). Across time and situations certain categories are more important to perceivers than others (see also Haslam, Turner, Oakes, McGarty, & Hayes, 1992; Oakes, 1987; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987; van Rijswijk & Ellemers, 2002). Although crossed categorization research has only examined stable diﬀerences in category importance, parallel eﬀects are expected from temporal/situational manipulations of importance (see McGarty, this volume, for a discussion of how context has implications for multiple categorization). This situational/chronic distinction aside, note that it is not category importance per se that moderates (as yet, only correlationally) crossed categorization eﬀects. Rather, it is the diﬀerential importance between two or more of the available dimensions that produces deviations from the additivity pattern. What is the eﬀect of diﬀerential importance? Typically, it yields a category dominance pattern. Diﬀerences on the important category dimension exert more inﬂuence on ratings of target persons than do diﬀerences on the nonimportant dimension. In an idealized form of the dominance pattern, favorable evaluation of the Io1 target is raised to the level of the Ii target, and evaluation of the Oi target is lowered to the level of the Oo target (i.e., Ii = Io > Oi = Oo). Few crossed categorization studies experimentally examine importance, yet those that do shed light on its theoretical implications; for example, in Urada and Miller’s (2000) study of category importance, positive aﬀect (a topic that we will revisit) improved the evaluations of mixed targets with a dominant, or important, ingroup category, but had no eﬀect on targets with less important ingroups. To demonstrate, however, that diﬀerential importance of a category dimension produces a favorable bias toward one’s ingroup on the more important dimension is simply an observation, but not an explanation. Indeed, an explanation requires that we consider what psychological eﬀects underlie diﬀerential importance. An obvious ﬁrst answer is heightened

1

Throughout this chapter, the notation of a capital letter “I” or “O” denotes that it is either a dominant category or a category along a dominant dimension.

Miller et al.

163

salience of the important dimension, which then makes diﬀerences along it more likely to produce more evaluative bias than diﬀerences on the less salient dimension (Ensari & Miller, 1998). This occurs because greater salience allows the inherent aﬀective valence of ingroup and outgroup characteristics to more strongly inﬂuence judgments (Crisp et al., 2002). Not only will the salient dimension dominate one’s current cognitive processes, but diﬀerences along it, by contrast with the less salient dimension, will seem more pronounced, leading to greater evaluative bias. When the conﬂicting ingroup and outgroup characteristics of the two mixed targets are of equal importance (i.e., io and oi), they balance each other. By contrast, with diﬀerentiated importance the less salient dimension is virtually ignored, making an otherwise conﬂicting description of a target person into a less conﬂicting, predominantly ingroup-like (Io) or predominantly outgroup-like (Oi) individual. Consequently, perceivers are likely to assimilate evaluations of these mixed targets toward their ingroup (Ii) and outgroup (Oo) counterparts.

Relevance of categories Relevancy can be viewed as a subset, or a speciﬁc type, of category importance. According to social identity theory, individuals achieve a positive social identity by evaluating their ingroup in a manner that makes it positively distinctive by comparison with relevant outgroups (Tajfel, 1978; Tajfel & Turner, 1979; Turner, 1982). Some researchers measure category relevancy as the degree to which perceivers attribute importance to a category (Devos, 1998; Spears & Manstead, 1989), arguing that the more important the category dimension, the more relevant it is to the perceiver. Therefore, it is more likely to produce the ethnocentric bias that occurs with relevant outgroups in social identity research (Mullen, Brown, & Smith, 1992; Tajfel, 1982).2 Research on both types of relevancy (see footnote 2) has typically used single categorization paradigms that involve only two mutually exclusive groups—whether based on either real-world categories or experimentally induced trivial criteria (Mullen et al., 1992; Spears & Manstead, 1989; Vanbeselaere, 2000). Relevancy has only rarely been manipulated in the crossed categorization literature, and with mixed ﬁndings. Puzzlingly, in

2

Other researchers measure relevancy as the degree to which social category dimensions correlate or predict other comparison dimensions (Kelly, 1989; Simon & Brown, 1987; Wilder 1984). Irrespective of dimensional importance, comparison dimensions in these studies that were strongly correlated with, or logically connected to, the dimension of social categorization evidenced stronger intergroup bias eﬀects because they were argued to be more relevant. Thus, there are two distinct lines of relevancy research that operationalize relevancy as either based on the logical connection to the category dimension or the importance of the category dimension to the participant. Even though these two lines of research are conceptually distinct, they both produce the same eﬀect of creating diﬀerential levels of dominance based on the degree of relevance of social categories.

164

Moderating ethnocentric bias

Vanbeselaere (2000), the relevancy of category dimensions did not aﬀect intergroup discrimination. In his experiment participants evaluated both actual and hypothetical (i.e., nonpresent) outgroup targets. Because Vanbeselaere considered hypothetical groups to be irrelevant, he argued that evaluating them would not impact participants’ self-esteem or the positivity of their social identity. Vanbeselaere’s (2000) unexpected lack of eﬀect may have occurred because the hypothetical groups inadvertently did possess category importance. The participants, 13-year-old boys from the same high school, ﬁrst completed a diﬃcult perceptual decision-making task. Their next task allegedly assessed whether people of their age could accurately evaluate the decision-making performances (on that same task) by the diﬀerent groups reﬂected in the categorization conditions. Thus, the boys were competing against friends and schoolmates on a task acknowledged as diﬃcult. Consequently, even though one of the target groups was hypothetical, it may have had the same level of importance to the boys as did real groups. When they were in competition with their schoolmates, negatively rating the hypothetical groups might have served to bolster self-esteem and social identity as readily as did rating of real groups. Thus, contrary to Vanbeselaere’s (2000) conclusions, it may have been that real and hypothetical social comparison targets evoked diﬀerent processes. Ensari and Miller (1998, Study 2) also found no relevancy-related eﬀects. Participants indicated their preferences among the four crossed categorization targets after several members of an outgroup category had insulted one of their ingroup category memberships (via a bogus newspaper article). In the high-relevancy condition one of the category memberships for each of the four targets was from the same category dimension that contained the insulted category. In the irrelevant category condition, none of the targets (except the Oi target) possessed a category membership from the insultsource dimension. Instead, all (except the Oi target) belonged to unrelated dimensions. As in Vanbeselaere (2000), no diﬀerences in partner evaluations were found between the relevant and irrelevant groups.3 Unlike Vanbeselaere (2000), however, wherein the irrelevant group may have inadvertently been made too important and therefore not irrelevant, Ensari and Miller (1998) argued that the relevant group failed to yield predicted eﬀects because of their weak relevancy manipulation. Thus, in both studies, the diﬀerential importance of the relevant and irrelevant conditions was arguably not discrepant enough to create a diﬀerence in target ratings.

3

Although this lack of diﬀerence is technically true given the two conditions did not interact, we point out here that simple contrast analysis revealed that the relevant condition was slightly diﬀerent from the irrelevant condition. In the latter, a hierarchical pattern emerged; in the former, a dominance pattern was obtained (see Ensari & Miller, 1998, Table 2). Because the relevancy condition confounded a dominant dimension with strong aﬀect, it is unclear what the implications are for relevancy per se. We discuss this issue further in the section, below, entitled crossed categorization patterns versus processes.

Miller et al.

165

In one instance, however, crossed categorization has yielded a relevancy eﬀect (van Knippenburg, van Twuyver, & Pepels, 1994). Target characteristics fell along the two dimensions of male/female and student/teacher. Recall errors in the “who said what?” paradigm (Acuri, 1982; Taylor, Fiske, Etcoﬀ, & Ruderman, 1978) served as the dependent measure. Relevancy was manipulated by having the verbal discussion statements relate either to the gender dimension or the university (i.e., student/teacher) dimension. As expected, the socially dominant dimension of male/female, as compared to student/teacher, produced more within-group errors than between-group errors. Making the verbal topics relevant to a particular dimension of social categorization, however, was expected either to exacerbate or reverse the tendency for the cognitive dominance of gender to evidence more bias. In conﬁrmation, when the topic was gender-relevant, within-group errors exceeded between-group errors to a larger degree along the gender dimension than along the student/teacher dimension. Moreover, this tendency was reversed when the topic was related instead to the student/teacher dimension, thereby showing a strong eﬀect of relevancy on within versus between category errors. Aﬀective arousal may provide another source of diﬀerential relevancy by creating dominant categories and thereby instigating diﬀerential impact on bias toward target category memberships. Relevancy may co-vary with aﬀective arousal because the latter directs attention to the category dimensions or groups associated with that aﬀective arousal. For example, an insult from an outgroup member may increase the salience of that outgroup member’s social category. The category that was the source of insult is likely to incur the negative aﬀect generated by the comment and, at least temporarily, become a dominant category within the stimulus array. In this way, aﬀect-induced relevancy acts to trigger the priming function of the insult and direct it toward other persons who belong to the same outgroup category as that of the source of the insult. An insult from an outgroup member may also emphasize one’s self-identity in the ingroup category of the relevant dimension and accentuate preference for this ingroup. For example, a liberal who is insulted by a conservative may not only evidence bias against other conservatives, but also evidence preference for other liberals. We argue that this eﬀect may occur not just with respect to insults, but also in any circumstance that induces speciﬁc negative aﬀect that is relevant to an intergroup dimension. For example, an outgroup member who is seen to be generous or helpful toward the ingroup may actually serve to threaten the ingroup’s competence and self-reliance (DePaulo & Fisher, 1980; Newman & Goldwin, 1990; Wagner, Lampen, & Syllwasschy, 1986), eliciting negative integral aﬀect with respect to the relevant outgroup category, as well as increased aﬃliation toward other ingroup members. Likewise, in terms of positive aﬀect, a compliment from an outgroup member is likely to generate positive aﬀective arousal, which may then generalize to evaluations of new targets that are relevant to the category dimension

166

Moderating ethnocentric bias

that provided the compliment. Relevancy acts to trigger the guiding function of the positive aﬀect and direct it toward individuals whose outgroup status is relevant to the source of the aﬀect.

THEORIES OF AFFECT AND SOCIAL JUDGMENT Before discussing the research on mood eﬀects in crossed categorization research, we ﬁrst present models of how mood impacts social judgment. These models propose various mechanisms for how mood aﬀects cognitive processing. Hence, they oﬀer potentially distinct perspectives and predictions regarding the eﬀects of mood on crossed categorization. In his associative network theory, Bower (1981) proposed that feelings are nodes in a network of associations in memory. When one experiences a particular mood, a memory node associated with it evokes prior experiences that are associated with that feeling. Thus, a mood state primes feeling-congruent memories. Within the crossed categorization paradigm, this model might predict that evaluations of mixed targets will be elevated under positive incidental mood and lowered under negative incidental mood. In a positive mood, the ingroup attribute of the mixed targets may become more salient and their other positive attributes may be exaggerated via their relationship with positive thoughts and memories, yielding a more positive judgment. Similarly, in a negative mood, the outgroup characteristics of the mixed targets may become more salient and negative via their link with negative thoughts and memories, yielding a more negative judgment. Schwarz and Clore (1983) explain the impact of mood on evaluative judgments by proposing that feelings function as information. They proposed that if people are in a particular mood state and are then asked to make a social judgment, they may misinterpret their mood as being due to their feelings about the person who is to be evaluated and in turn judge that person in a mood-consistent manner (provided that they are not too aware of their mood state; see Schwarz, 2001). Schwarz (1990) also proposed that the relationship between the state of one’s environment and one’s mood is bidirectional, one’s moods reﬂecting the environment and vice versa. This hypothesis stems from Wyer and Carlston’s (1979) idea that moods inform one about the environment (see also the mood and general knowledge structures model; e.g., Bless, Clore, Schwarz, Golisano, Rabe, & Wölk, 1996). Experiencing a bad mood signals a problem in the environment; similarly, experiencing a good mood signals a nonproblematic environment (Schwarz, 1990). Further, the presence of a negative mood can cue a need to change the environment, requiring that one ﬁrst thoroughly assesses and pays close attention to the situation. Conversely, a benign environment does not signal a need for change, precluding the need for careful assessment. Therefore, a negative mood leads to more systematic cognitive processing than a neutral or a positive mood. Within crossed categorization this might lead to mood-congruent (negative)

Miller et al.

167

judgments of mixed and double outgroup targets because their descriptions contain negative information. As long as one’s mood is not made a salient feature of the stimulus array, one is likely to attribute one’s negative mood to the negative targets and evaluate them negatively or choose to avoid aﬃliating with them. Having projected one’s negative mood onto them, their potential presence signals danger. By contrast, positive mood will have a weaker, or no eﬀect, on evaluations. The heuristic-like processing that is expected under positive mood may have the indirect eﬀect of reducing diﬀerentiation among the crossed categorization targets. Isen’s (1982; Isen, Niedenthal, & Cantor, 1992) mood model stands in contrast with the preceding models, which emphasize a relative deﬁciency in processing under positive, by comparison with negative, mood. Instead, Isen emphasizes the processing advantages of positive mood. Fundamental is the notion that positive aﬀect induces broader, more inclusive categorization. There is an asymmetry to this eﬀect, however, in that positive mood does not lead perceivers to more broadly categorize predominantly negative people into negative social categories (Isen et al., 1992). Isen et al. (1992) point to increased complexity of the cognitive context and the priming of aﬀective features as mechanisms by which positive mood may aﬀect cognitive processes. Speciﬁcally, positive aﬀect is seen as cueing positive material in memory (Isen, Shalker, Clark, & Karp, 1978; Teasdale & Russell, 1983). As a result of the large amount of positive material available, it elicits a wider range of associations than does neutral aﬀect. Thus, by concurrently thinking about more ideas, people will see relationships that otherwise remain unnoticed (Isen & Daubman, 1984). This leads to greater integration of ideas and results in more inclusive, broader categorization. By contrast, because positive aﬀect’s inﬂuence relies on increased accessibility of predominantly positive material, asymmetrical inclusivity occurs; positive aﬀect fails to promote a broader range of associates to negative stimuli. Consequently, under a positive mood, crossed categorization should not induce an assimilation of targets with an undesirable category membership into a more undesirable group (i.e., io or oi = oo). Abele and Gendolla (1999), Abele, Gendolla, and Petzold (1998), and Abele and Petzold (1994) propose that positive mood acts on categorization information via two mechanisms. First, although one’s good mood is one of many pieces of information to be integrated into a judgment, it serves to average a judgment in the positive direction, typically yielding mood-congruent outcomes. Second, positive mood enhances the subjective weight of category information. Within crossed categorization, this suggests more positive evaluations of ii targets, and less positive evaluations of oo targets. This latter eﬀect, however, is likely to be oﬀset by the upward averaging eﬀect of positive mood, yielding a net positive eﬀect only for positive targets. Any positive eﬀect on evaluations of mixed targets only will occur under conditions of diﬀerential weighting or salience of the ingroup or outgroup dimension. Lastly, Forgas’s (1995, 2001) aﬀect infusion model asserts that our tendency

168

Moderating ethnocentric bias

to adopt diﬀerent processing strategies reﬂects diﬀerent contextual requirements. A positive mood generates schematic and heuristic processing whereas a negative mood generates a vigilant processing style (Forgas, 2000). As in other models, situational cues are a factor in determining the eﬀect of mood on cognitive processing. However, contextual requirements include both environmental cues and the nature of the task (e.g., simple versus complex). This latter feature distinguishes his model. Moreover, the amount of cognitive processing required by a task aﬀects the degree to which a mood will infuse. Infusion refers to the process whereby aﬀectively loaded information becomes incorporated into and exerts a congruent inﬂuence on cognitive, judgmental, and behavioral processes. It is most likely to occur when a task requires elaborate processing strategies, as opposed to tasks that require less cognitive activity, such as one for which the solution rests on schema-based knowledge. Elaborate processing allows existing aﬀect more opportunity to intrude and inﬂuence cognitive processing, thereby leading to moodconsistent evaluations. In short, the more diﬃcult or complex the task, the more likely is a mood-congruent judgment. The various constraints that this model imposes on mood processing make it diﬃcult to generate hypotheses for crossed categorization in the absence of knowledge about speciﬁc features of an experimental setting. One simple prediction, however, is that more novel or complex targets will yield more mood-congruent judgments. This is likely because the lack of prior associations with such stimuli, as well as their perceptual salience, makes them more susceptible to the infusion of aﬀect. But does this imply that mixed targets will be aﬀected more strongly by moods than will double-ingroup or double-outgroup targets? Perhaps, but most crossed categorization studies use barren descriptions of the target persons that are constrained largely by a presentation of their category memberships. Hence, it is hard to imagine that the diﬀerential complexity between crossed versus convergent targets is suﬃcient to cause diﬀerential infusion of mood. Although not an exhaustive presentation, this handful of models has bearing on the processes underlying the outcomes of crossed categorization research. They have rarely been invoked, however, much less pitted against one another in terms of their diﬀerential positions on underlying process. We discuss them here and below to help integrate crossed categorization ﬁndings with them, and to promote further research that explores new theoretical and practical avenues with respect to social categorization and intergroup bias.

MOOD AND CROSSED CATEGORIZATION We and others (see Crisp et al., 2002; Kenworthy, Canales, Weaver, & Miller, 2003; Miller et al., 1998; Vanman, Paul, Ito, & Miller, 1997) have proposed that ingroup members are preferred over outgroup members due to their relatively automatic activation of positive and negative aﬀect, respectively.

Miller et al.

169

Indeed, Urban and Miller (1998) showed that the degree of judged positive aﬀect associated with experimental settings was meta-analytically associated with decreased diﬀerential bias toward the four crossed categorization targets (ii, io, oi, and oo). Several parameters, such as intensity or valence, delimit the nature of a given mood and aﬀect the way it inﬂuences judgment. In addition, much recent research has focused on whether the source of the aﬀect is incidental or integral (e.g., Bodenhausen, 1993; Moreno & Bodenhausen, 2001; Wilder & Simon, 1996). Incidental aﬀect is elicited by characteristics of the experimental setting, but not by targets of judgment. Integral aﬀect, by contrast, is aﬀect that is elicited by the targets themselves or their respective categories.

Positive incidental aﬀect Expectations for the eﬀects of positive incidental aﬀect diﬀer in a number of meaningful ways from those for positive integral aﬀect. Speciﬁcally, because it is aﬀect that is not linked directly with a group or group member, no single target in the crossed categorization paradigm is expected to be inﬂuenced in a unique way. A positive aﬀective state is alleged to reduce careful, systematic information processing (e.g., Bodenhausen, Kramer, & Susser, 1994; Schwarz, 1990; Worth & Mackie, 1987), and to increase the degree to which people use superordinate group representations (Dovidio, Gaertner, Isen, & Lowrance, 1995). Relatedly, a positive incidental mood state can increase inclusiveness in the use of categories (Isen & Daubman, 1984) and can augment the positive evaluation of neutral targets (Isen et al., 1992). Following meta-analytic work showing aﬀect to be a correlational moderator of crossed categorization eﬀects (Urban & Miller, 1998), two studies examined the experimental eﬀects of positive incidental aﬀect (Crisp & Hewstone, 2000a; Urada & Miller, 2000). Crisp and Hewstone (2000a) induced positive incidental mood via false performance feedback. Relative to the neutral mood condition, positive mood eliminated the typical additivity pattern and instead produced an equivalence pattern (i.e., all targets are evaluated equally). Further, the participants’ self-reported mood scores mediationally explained the mitigating function of false-positive feedback on evaluations. Additional analyses showed that positive mood did not alter the perceptions of the groups themselves, or diﬀerences between them. It only led to an elimination of the use of such memberships in forming judgments about targets. Although the ﬁndings with respect to the mixed targets (who are, on balance, neutral targets) support the notion that positive and neutral targets are perceived as more positive under positive incidental aﬀect (Isen et al., 1992), this theorizing fails to account for the equivalence of the double outgroup to the other targets; in Isen et al. (1992) evaluations of negative targets were not aﬀected by positive mood. To argue instead, that this latter aspect of Crisp and Hewstone’s (2000a) results parallels those of

170

Moderating ethnocentric bias

Dovidio et al. (1995), in that they reﬂect a greater inclusiveness with respect to a superordinate category, requires data identifying the superordinate category invoked by their participants—perhaps students at the same school. Urada and Miller (2000) manipulated positive incidental aﬀect via autobiographical recollections (see also Bodenhausen et al., 1994; Brewer, Doughtie, & Lubin, 1980; Strack, Schwarz, & Gschneidinger, 1985), and then assessed partner preferences for the four crossed categorization targets. Building on Isen’s model (Isen et al., 1992), they emphasized the guiding and broader categorization functions of positive mood in asymmetrically inﬂuencing judgments of positive and negative stimuli. Speciﬁcally, they expected positive mood to lead to an assimilation of positive targets, with no eﬀect on perceptions of negative targets. Such a prediction can also be derived from Bower’s (1981) associative network theory, in which positive incidental mood primes positive elements in memory, making positive aspects (i.e., ingroup memberships) of mixed targets more salient and thereby rendering them more ingroup-like. A key question, however, is whether mixed targets, who possess both positive (ingroup) and negative (outgroup) characteristics, are perceived as fundamentally neutral, or, instead, are perceived as being more positive (ingroup-like) or more negative (outgroup-like). From the perspective of Isen et al. (1992), this is a crucial point; only positive and neutral, but not negative, targets will be assimilated into the broader categorization scheme produced by positive mood. Urada and Miller (2000) reasoned that whereas mixed targets may seemingly be neutral (an average of positive and negative), in actuality they represent exactly what they are, namely, persons of mixed characteristics: both positive and negative. Moreover, if the importance of the two categories is equal, they lack either a strong positive cue or a strong negative cue that respectively, under positive mood, can promote (or fail to promote) their assimilation into a broadened positive categorization scheme. Thus, Urada and Miller (2000) expected the guiding and broadened categorization eﬀects of positive aﬀect to inﬂuence judgments of mixed targets only when a dominant category dimension exists (i.e., Ii, Io, Oi, and Oo, not ii, io, oi, and oo). Then, the ingroup category of the Io target is salient, whereas the outgroup category of the Oi target is salient, thereby making the two mixed targets seem more ingroup- and outgroup-like, respectively. Supporting their reasoning, overlay of positive mood produced an evaluative assimilation of the Io target (now more inherently positive) to the level of the Ii target, leaving evaluations of the Oi and Oo targets unchanged (see Figure 8.1). The broader inclusiveness explanation was not solely supported by default, however. When the two category dimensions were made equally low in importance, positive mood failed to inﬂuence evaluations of mixed targets (Urada & Miller, 2000, Studies 2 and 4). These ﬁndings compellingly support the guiding and broadened categorization roles of positive incidental aﬀect in crossed categorization. Note, however, that, unlike positive integral aﬀect (see below), incidental aﬀect per se does not cause dominance of one

Miller et al.

171

Figure 8.1 Preference ratings for targets (± SE) as a function of mood condition and target type; Urada and Miller (2000, Study 1). Reprinted with permission by APA.

category over another; it had its eﬀect only when dominance was already present. The ﬁndings discussed in this section generally support models of mood and social judgment, and, more speciﬁcally, lend support to Bower’s (1981) associative network model, which contains features that underlie the subsequently developed models. They show a form of mood-congruent judgments under positive mood. Other theoretical models are also implicated. Urada and Miller’s (2000) and Crisp and Hewstone’s (2000a) results support Isen’s model. One important methodological divergence between the two studies concerns the motivational aspects of the respective judgment processes. In Urada and Miller (2000), participants believed they were going to interact with actual, present individuals that belonged to subjectively important social categories. With meaningful target diﬀerences, the asymmetrical predictions of Isen’s model found support. In Crisp and Hewstone (2000a), targets belonged to minimal groups and participants expected no interaction. The resulting equivalence pattern may reﬂect a superordinate categorization, again supporting Isen’s model. Perhaps, too, this disjunction in results ﬁts with Forgas’s (1995) aﬀect infusion model, which proposes less infusion of aﬀect into judgments when a motivational goal is invoked by the judgment process. With Urada and Miller’s motivationally laden targets,

172

Moderating ethnocentric bias

positive mood only aﬀected the Io target. With targets lacking importance, Crisp and Hewstone’s (2000a) obtained equivalence pattern reﬂects an inﬂuence on all four targets.

Negative incidental aﬀect Induction of negative aﬀect that is independent of the intergroup context has been labeled incidental aﬀect. Its induction is expected to augment the negativity of targets possessing outgroup cues (e.g., Bower, 1981; Forgas, 1995; Miller et al., 1998) because negative mood simultaneously activates negative elements in memory and acts as a ﬁlter for selective processing. Speciﬁcally, negative mood may increase the attention paid to mood-congruent (i.e., negative) stimuli, thus allowing greater scope for primed negative memories, ideas, thoughts, and behavior tendencies to inﬂuence judgments of external events or persons. Thus, crossed categorization targets who possess outgroup characteristics (i.e., io, oi, and oo) will be evaluated more negatively under negative mood than under neutral mood, leading to a social exclusion pattern (ii > crossed = oo, or in less idealized form, ii Ⰷ crossed > oo). Kenworthy et al. (2003) tested this hypothesis by manipulating distinct negative incidental mood states. In Kenworthy et al. (2003, Study 1), we manipulated negative mood by giving participants a very diﬃcult anagram task during which they were rudely repeatedly interrupted and told to speak louder (see Stemmler, 1989). Ostensibly as part of a separate study, participants then evaluated the four crossed categorization targets. Contrast analysis showed that the pattern of target ratings by the negative mood participants matched a social exclusion pattern (contrast weights = 3, –1, –2), whereas additivity (1, 0, –1) was found in the neutral mood condition. Covariance analysis showed that self-reported negative mood, particularly anger, partially accounted for the eﬀects. In Study 2, we manipulated sadness (via a ﬁlm), and found a social exclusion pattern. Studies 3 and 4 manipulated sadness and anger within the same designs. We also induced the two negative moods in a similar manner, namely, via autobiographical recall. In all studies a common ﬁnding emerged. Both anger and sadness produced evaluations of crossed categorization targets that corresponded to a social exclusion pattern (see Figure 8.2). Further, covariance analysis consistently showed that participants’ self-reported negative mood scores were partially responsible for the reliable deviation from the additivity pattern.

Sadness and disgust diﬀer from anger in approach/avoidance tendencies Mood states diﬀer in the degree to which they elicit approach versus avoidance tendencies (see Frijda, Kuipers, & ter Schure, 1989; Mackie, Devos, & Smith, 2000; Roseman, Wiest, & Swartz, 1994). Although no disparity had

Miller et al.

173

Figure 8.2 Mean aﬃliative target evaluations (± SE) on composite index as a function of induced aﬀective state. From Kenworthy et al. (2003, Study 4). Reprinted with permission from Elsevier.

emerged in any of our individual studies (Kenworthy et al., 2003), we collapsed the four studies to meta-analytically examine diﬀerential eﬀects of sadness and anger on the major dependent measures. A main eﬀect emerged, showing that angered participants more strongly avoided mixed and oo targets than did saddened participants. Thus, although the general patterns on the aﬃliation measures were parallel for sadness and anger, anger produced more avoidance than did sadness. Likewise, in Study 4, wherein a measure was introduced to examine the potential diﬀerence in approach behavior between the two moods, anger again diﬀered from sadness. Speciﬁcally, on a hypothetical measure of approach aggression toward the crossed categorization targets, anger produced a greater overall level of aggressive approach by comparison with the control and sadness conditions, which did not diﬀer from each other. Moreover, whereas sadness and anger produced similar patterns of avoidance on the aﬃliation measures (albeit ones that diﬀered in magnitude), they produced strikingly diﬀerent patterns of responding on the aggression measure. In the sadness condition participants aggressed even less (directionally) than those in the neutral mood condition, whereas those in the anger condition aggressed more than those in both the neutral and sadness conditions. Canales (1999) corroborated this latter ﬁnding. Participants read a 350word description of a very obese person overeating at a fast-food restaurant. It included details about disgusting sounds, smells, and visual attributes of the person. The aﬀective state of disgust was incidental in one condition because the descriptions of the four crossed categorization targets, ostensibly

174

Moderating ethnocentric bias

applicants for a tutorial position in the psychology department, contained no information about their weight category (i.e., normal or overweight). Thus, the elicited disgust was unrelated to the four targets. In that condition, job recommendations for the targets did not diﬀer from the neutral condition. Although conclusions drawn from null eﬀects must be tentative, we see a parallel between this ﬁnding and the lack of aggression among sad participants in Kenworthy et al. (2003, Study 4). Speciﬁcally, in both instances the dependent measure was an aggression measure. In Kenworthy et al. (2003), although the measure applied to a hypothetical situation, it was explicitly aggressive. In Canales (1999) it was aggressive to the degree that the withholding of a recommendation for an applicant’s desired job reﬂects one’s intent to do harm (see Pedersen, Gonzales, & Miller, 2000). Moreover, it cannot be argued that these dependent measures represented an avoidance response. Because sadness and disgust are aﬀective states characterized by withdrawal and avoidance (see below), their failure to increase aggressive action makes sense. These ﬁndings support a conceptual distinction between sadness (and likewise, disgust) and anger. As mentioned, diﬀerent emotions are characterized by diﬀerent behavioral tendencies (e.g., Mackie et al., 2000). Some, such as anger, characteristically produce an external focus or a focus on others (Green & Sedikides, 1999; Roseman et al., 1994; Sedikides, 1992), which may be coordinated with either approach or avoidance tendencies. By contrast, an emotional state, such as sadness, can produce an inward focus (e.g., Carlson & Miller, 1987; Roseman et al., 1994; Sedikides, 1992). Yet this inwardfocusing eﬀect of sadness varies as a function of task. For example, a direct request for help (see Carlson & Miller, 1987) will elicit an external focus, even in sad participants. This moderation is evocative of Forgas’s (1995, 2001) arguments concerning the interaction of mood with task type (see above). Nevertheless, under normal circumstances, one can expect sadness to produce a less extreme bias than anger for both approach and avoidance behaviors. Although the evaluative outcomes with respect to negative incidental aﬀect generally support models of mood and social judgment (e.g., Bower, 1981; Forgas, 1995), nevertheless, like the outcomes of positive incidental aﬀect, these outcomes do not necessarily illuminate the processes by which these moods produce their eﬀects. Indeed, we argue that when crossed categorization researchers studied mood, it was not their primary aim to show how mood aﬀects judgments in the crossed categorization paradigm. Instead, the work was pioneering in the sense that it was investigating what kinds of diﬀerences would result from the overlay of mood onto the study of crossed categorization eﬀects. Researchers now have the happy burden of going further in exploring both the consequences of moods as well as the processes that underlie the eﬀects. Before further discussing this topic, we turn ﬁrst to positive and negative integral aﬀect. Note, however, that the models of mood and social judgment

Miller et al.

175

discussed above have little bearing here; they primarily concern the inﬂuence of incidental aﬀect on judgments of social others.

Positive integral aﬀect With positive integral aﬀect, by contrast with positive incidental aﬀect, its source is an individual or group. When elicited, the expectation is that any target possessing a salient characteristic identical to that of the source of positive aﬀect will be positively aﬀected. Ensari and Miller (1998, Study 1) tested this idea. In a newspaper article, alleged outgroup members from a single category praised the participant’s ingroup category (e.g., participants who were young adults read interview statements by older adults that praised young adults). In the neutral comparison condition, participants read an irrelevant, neutrally valenced article. Participants were then presented with four (bogus) potential discussion partners who corresponded to the four crossed categorization targets: ii, io, oi, and oo. All targets were new individuals, not those who had given compliments in the newspaper article. In the positive integral aﬀect condition, the outgroup category for the Oi target was made identical to the category membership shared by the complimenting newspaper interviewees. In the neutral condition, the outgroup category for the oi target, as well as the io and oo targets, diﬀered from the outgroup category of those who had provided compliments in the news article. For the positive integral aﬀect condition, Ensari and Miller (1998) expected a hierarchical rejection pattern in which the evaluation for the Oi target is raised to the level of the ii target (i.e., ii = Oi > io > oo), leaving the io and oo targets unaﬀected relative to the neutral condition. Why expect this pattern? As mentioned, a positive aﬀective state can increase the degree to which people use superordinate group representations (Dovidio et al., 1995). This may result in improved outgroup evaluations, thus mitigating intergroup bias. When positive aﬀective arousal is linked directly to an outgroup individual or category, it increases generalized positive responses toward that particular outgroup as a whole (Hewstone & Lord, 1998). The salient general positive aﬀect toward the outgroup that was made the source of positive aﬀect should result in an evaluation of a new mixed target (Oi) member of that outgroup category that is as positive as that of a double ingroup target (ii). Further, Ensari and Miller’s implementation of the crossed categorization paradigm precluded this generalized aﬀect from equally inﬂuencing judgments of outgroup members of other categories because they made the distinctive salience of all categories maximal. In Ensari and Miller (1998, Study 1) the hierarchical rejection pattern was, in fact, obtained. In the neutral condition preferences for discussion partners matched the additivity pattern, whereas positive integral aﬀect caused a new Oi target to be preferred as much as the ii target. The oo target was preferred the least, and the io target preferred intermediately. This study was the ﬁrst to

176

Moderating ethnocentric bias

demonstrate that positive integral aﬀect acts to mitigate bias toward new relevant outgroup targets. It was also the ﬁrst to produce a hierarchical rejection pattern within the crossed categorization paradigm. Of equal importance, mediation analyses showed that variation in self-reported mood states caused these eﬀects.

Negative integral aﬀect As with positive aﬀect, negative aﬀect can be either integral or incidental. Negative integral aﬀect is an induction of negative aﬀect that has as its source an individual or group. When negative integral aﬀect is elicited, the expectation is that the evaluation of any target possessing a characteristic identical to that of the source of negative aﬀect will be negatively inﬂuenced. That is, similar to positive integral aﬀect, negative integral aﬀect should promote dominance of the category primed by the negative mood manipulation. Ensari and Miller (1998) tested the hypothesis that negative integral aﬀect will produce a hierarchical acceptance pattern (ii > io > Oi = oo). They manipulated negative integral aﬀect by having participants read a news article containing several interviews in which outgroup members from a single category disparaged the participant’s ingroup. They then presented the four crossed categorization targets to participants, with the ostensible goal of having them pick a discussion partner from among the four (bogus) “others” in the study. One of these potential partners, the Oi target, belonged to the same outgroup that had insulted the ingroup. Conﬁrming expectations, the Oi target (whose outgroup category membership matched the source of negative aﬀect) was evaluated as negatively as was the oo target. The ii and io targets were unaﬀected, relative to the neutral control condition, yielding a hierarchical acceptance pattern (ii > io > Oi = oo). Further, mediational analyses showed that subjective ratings of negative mood mediated the relationship between manipulations of mood and ratings of the Oi target. From the design and results of this study, however, it is not clear whether the increase in salience of the source category is also partly responsible for the eﬀect of negative integral aﬀect on ratings of the Oi target. That is, inherent in a manipulation of negative integral aﬀect is the cognitive dominance of that source category over others. A true test of whether dominance of the integral category inﬂuenced the negativity of the Oi target requires a neutral mood condition in which negative aﬀect is absent, and yet the same category is in some way made to be dominant (cf. Urada & Miller, 2000). Unfortunately, the control condition in Ensari and Miller (1998, Study 2) had no such explicit dominance, leaving aﬀect-induced dominance as well as aﬀect per se as viable alternative explanations of their outcome. Thus, the issue remains empirically unresolved. (Parenthetically, although not considered in our discussion of integral positive aﬀect, this same issue applies there too.) These latter considerations notwithstanding, negative integral aﬀect was a unique predictor in the ratings of the Oi target.

Miller et al.

177

Another test of negative integral aﬀect (Canales, 1999; see above) manipulated the emotion of disgust. The aﬀective state was integral in one condition because the descriptions of the four crossed categorization targets, ostensibly applicants for a tutorial position in the psychology department (see above), all contained information about the individual’s weight category (i.e., normal weight or overweight). Canales expected targets whose weight category matched the source of disgust (i.e., obese) to receive less favorable evaluations and job recommendations than neutral mood controls. Conﬁrming this hypothesis, pair-wise comparisons revealed a reliable diﬀerence between the Oi and Oo targets in the integral disgust condition, versus their respective neutral mood counterparts. The disgust elicited by the description of the obese person generalized to other targets who belonged to the same category with respect to relative body weight.4 With respect to both positive and negative integral aﬀect within crossed categorization, we repeat the observation that all relevant psychological processes were not always assessed in these studies, thereby precluding a full understanding of the underlying mediators of the obtained eﬀects.

Mood and crossed categorization: current juncture The mood-related crossed categorization studies discussed in this chapter are the ﬁrst steps toward discovering the consequences of incidental and integral aﬀect on evaluative bias toward ingroup and outgroup members. They also highlight underlying mechanisms that may cause the diﬀerential evaluations that result from valence of aﬀect. As mentioned previously, the outcomes of these studies can be seen as superﬁcially supportive of models of mood and social judgment. Speciﬁcally, several models—for example, the associative network model (Bower, 1981), the feelings as information model (Schwarz, 1990), and Isen’s model of broadened categorization under positive mood (Isen et al., 1992)—predict that under certain circumstances judgments will be mood-congruent under positive and negative moods. To varying degrees, evaluative target ratings were congruent with valence of mood within each of the studies that manipulated either incidental or integral aﬀect. When the source of the aﬀect was either incidental or integral, mixed targets received more favorable evaluations following manipulations of positive mood and less favorable ones after induced negative moods. For example, Ensari and Miller (1998) found that when positive aﬀect was linked directly to an

4

The reader will notice that this eﬀect seems to undermine the discussion above concerning the distinction between approach/avoid measures for moods. We argue, however, that integral aﬀect makes certain cues more salient, and this augmented importance of relevant cues is likely to interact with approach/avoid tendencies. For example, the salient semantic connection between the induction of disgust and the targets in the integral condition may produce a generalized negative responding toward relevant targets on any available measure, including a more aggressive one.

178

Moderating ethnocentric bias

outgroup target category, it raised a mixed target possessing that outgroup membership to the level of an ii target. Conversely, negative integral aﬀect lowered an equivalent mixed target to the level of an oo target. The broader inclusiveness produced by positive incidental aﬀect (Crisp & Hewstone, 2000a; Urada & Miller, 2000) was thought to be a result of the priming of positive material in memory. In turn, it focused participants’ attention on targets’ positive features. Conversely, the social exclusion pattern (Kenworthy et al., 2003) was interpreted as due to the priming of negative thoughts and feelings, which made the more negative aspects of the mixed targets more salient. In this way, the increased salience of the moodcongruent category memberships of the targets produced opposing results that were based upon the valence of the aﬀect. These mood-congruent eﬀects produced from incidental aﬀect can be explained by the activation of the mood-congruent cognitive priming that is central to some of the models of mood and social judgment. Alternatively, the diﬀerential results between studies of positive and negative incidental aﬀect may seemingly be explained by theories of mood and social judgment that concern attentional processes. Some theories, for example, the feeling as information and aﬀect infusion models, argue that positive aﬀect leads to less careful attention to stimuli, whereas negative aﬀect increases systematic attention to information. Extending this reasoning, the more thorough attention elicited by negative mood may make the negative aspect of the mixed targets more noticeable, thereby accounting for the lowered ratings for targets with outgroup memberships in Kenworthy et al. (2003). Conversely, positive incidental aﬀect may have led to broader inclusiveness (Crisp & Hewstone, 2000a; Urada & Miller, 2000) by reducing attention to target diﬀerences. However, because the target stimuli in the crossed categorization studies discussed above are perceptually barren, typically containing little more than the relevant category membership information, we do not see this reasoning as readily applicable. That is, given the characteristic simplicity of the stimuli, there is little reason to expect diﬀerential attention eﬀects. A more plausible conceptualization of positive aﬀect argues that broader inclusiveness is a result of increased processing, not decreased attention. According to the model of positive mood and social judgment advocated by Isen and colleagues (Isen et al., 1992), broader inclusiveness is a result of calling to mind the wider range of thoughts and ideas that are cued by positive mood. Again, however, we note that although self-reported mood scores have been shown to partially mediate the relationship between manipulated mood and the eventual evaluative outcomes, there are no data to suggest by what means mood accomplishes these eﬀects. In other words, even though priming, attention, or associative processing may be underlying mechanisms in these studies, we remain relatively ignorant about the details of the psychological processes that are directly triggered by the onset of aﬀect in the context of crossed categorization.

Miller et al.

179

What might be mediating these eﬀects? To what degree is it similar activated memories, as suggested by Bower (1981)? To what extent does mood trigger more heuristic styles of processing social information? When is mood misattributed from its original source to the targets themselves (Schwarz, 1990), and does the process of misattribution diﬀer between integral and incidental aﬀect? We ask these questions not to suggest mistakes in prior research on mood and crossed categorization, but rather to point future research in potentially fruitful directions. If we are interested in knowing how moods, both integral and incidental, interact with situations of crossed categorization and multiple categorization, then it is incumbent upon us to tackle these empirical questions using experimental methodologies.

Opportunity for personalized contact Before embarking on the discussion of personalization as a moderator of crossed categorization eﬀects, we oﬀer a theoretical caveat. In their metaanalysis of crossed categorization eﬀects Urban and Miller (1998) had judges rate: (a) the level of judgment (group or individual); (b) whether the targets had names; (c) whether any visual contact occurred between participants; and (d) the dichotomized (none versus some) amount of cooperative interaction between participants in the experiment. This aggregate, which they termed personalization, was shown to moderate reduced bias among the four crossed categorization targets. One of us has recently argued (Miller, 2002) that there is an important theoretical distinction between the personalization and the individuation of social targets. Personalization involves: (a) the selfdisclosure of unique attributes that diﬀerentiate the self from the social group; and/or (b) self/other comparisons with respect to such personal information. By contrast, individuation is merely the acquisition of knowledge about target persons’ unique attributes. Note, however, that such knowledge need not come from the target person, but instead can be obtained from a third party (i.e., not from the individual him/herself). Thus, although personalization necessarily includes individuation, individuation does not require self-disclosure or self/other comparisons. In support of the need for this distinction, evidence shows that individuating information per se reduces group-based (e.g., Locksley, Borgida, Brekke, & Hepburn, 1980) and individual bias (Alicke, Klotz, Breitenbecher, Yurak, & Vredenburg, 1995). Given the aggregate of constructs implicated by their operational deﬁnition, in retrospect it appears that Urban and Miller (1998) assessed the moderating eﬀect of individuation rather than that of personalization. The increased level of intimate contact induced by personalization, by comparison with that of mere individuation, suggests that the former may even more strongly moderate reduced bias in crossed categorization situations as well as other intergroup contexts. In the single published experimental examination of personalized contact within crossed categorization (Ensari & Miller, 2001), intergroup bias was

180

Moderating ethnocentric bias

reduced when participants were told to pay attention to unique attributes rather than to category distinctions. Speciﬁcally, after assignment to minimal groups and a personalized interaction with outgroup individuals (i.e., the self-disclosure of unique personal attributes in a contact setting), participants did not discriminate between ingroup and outgroup members on team composition preferences in the condition wherein attention to unique, individual attributes had been emphasized. By contrast, when category memberships were emphasized as the decision rule for team compositions, the unique attributes self-disclosed during the previous personalized contact were ignored, leading instead to bias between groups. This suggests that when the salience of outgroup cues is augmented by a group formation rule, the beneﬁcial eﬀect of personalized interaction can be vitiated or eliminated. Although not a crossed categorization study, Ensari and Miller (2002) corroborated these eﬀects when they showed that the combination of personalized contact and group typicality reduced bias toward new outgroup members. The implications of this ﬁnding bear on a current theoretical and empirical debate among categorization researchers. Based on recent ﬁndings (e.g., Crisp & Hewstone, 2000b; Crisp, Hewstone, & Rubin, 2001; Ensari & Miller, 2002; Hornsey & Hogg, 2000), the beneﬁcial eﬀects of personalized interaction may be reduced (a) when one’s ingroup identity is threatened by a perceived loss of diﬀerentiation, and, relatedly, (b) when, after a superordinate category has been made salient, the perceiver’s ingroup identity has not also been made salient. In further support of this, Marcus-Newhall, Miller, Holtz, and Brewer (1993) either cross-cut or converged minimal group assignments not with a second social category, but rather with role assignments on a cooperative task. Their aim was to compare the predictions of a model based on social identity theory (e.g., Brown & Wade, 1987) with those of a model based on decategorized personalization (e.g., Brewer & Miller, 1984). When social categories were crossed with role assignments (i.e., when participants worked with outgroup members on a similar, as compared to a dissimilar, task), personalized contact reduced intergroup bias, provided that task interaction attitudes were positive (Marcus-Newhall et al., 1993). By contrast, personalization did not reduce intergroup bias when task interaction attitudes were negative. Presumably, such negative attitudes were a marker of subjective feelings of threat of one kind or another.

CROSSED CATEGORIZATION PATTERNS VERSUS PROCESSES Throughout this chapter, and indeed throughout the crossed categorization literature, the discussion often centers on patterns of evaluation as the evidence for either a reduction in, or an exacerbation of, intergroup bias. For example, the additivity pattern (ii > io = oi > oo) shows that by crossing an

Miller et al.

181

ingroup category with an outgroup category (mixed targets), ingroup bias can be reduced. Likewise, the hierarchical acceptance pattern (ii > io > oi = oo; Brewer, Ho, Lee, & Miller, 1987; Ensari & Miller, 1998; Park & Rothbart, 1982) indicates that when others are ingroup members along a ﬁrst dimension, their ingroup and outgroup memberships along a second dimension are more diﬀerentially evaluated than when others are outgroup members on the ﬁrst dimension. There are other patterns as well (see Migdal et al., 1998; Urban & Miller, 1998; see also Crisp & Hewstone; Vescio et al., this volume), which, when discussed, are typically accompanied by an analysis of how bias has been either mitigated or augmented. Such labels for patterns of evaluation, however, should not be mistaken for information about the processes that underlie them. Above, in our discussion of negative integral aﬀect, we mentioned that Ensari and Miller (1998) predicted the hierarchical acceptance pattern. Previous papers (e.g., Brewer et al., 1987; Hewstone, Islam, & Judd, 1993; Park & Rothbart, 1982) had also found and discussed this pattern. Note, however, that Ensari and Miller’s (1998) prediction of it was based on a very diﬀerent process than that proposed by earlier researchers. Speciﬁcally, Brewer et al. (1987) predicted hierarchical ordering of targets when one dimension of categorization was made dominant. In their study, schoolchildren evaluated other nationalities diﬀerently only when the targets were ingroups with respect to sex; when they were outgroups on the sex dimension, nationality made little diﬀerence to them. Likewise, Park and Rothbart (1982) showed that memory for characteristics of same-sex ingroup members was better than that for members of the opposite sex. This was seen as evidence that a dominant dimension was required for hierarchical ordering of targets. By contrast, Ensari and Miller (1998) reformulated this process hypothesis and argued that a dominant category with a strong aﬀective link (rather than dominance of the entire category dimension) was suﬃcient to produce hierarchical ordering. In this comparison between studies, the simple descriptive name for the pattern of evaluation is not helpful for understanding the conditions that produce it. The methodological diﬀerences between the previous work (e.g., Brewer et al., 1987; Hewstone et al., 1993; Park & Rothbart, 1982) and the latter (Ensari & Miller, 1998) can reveal important parameters regarding the theorizing about underlying processes. In work promoting the importance of a single, dominant dimension as necessary for hierarchical ordering (e.g., Brewer et al., 1987; Hewstone et al., 1993; Park & Rothbart, 1982), the participants were presented with targets who varied only along two predetermined dimensions of categorizations (e.g., race and sex). By contrast, when researchers have instead argued for the importance of a single, aﬀectively laden category for hierarchical ordering (e.g., Ensari & Miller, 1998), the participants evaluated targets who varied along multiple, ideographically selected dimensions of equal importance (e.g., sex, year in school, ethnicity, sports team, political aﬃliation, etc.). Could it be that the diﬀering respective methodological constraints

182

Moderating ethnocentric bias

eliminate the need for either/or thinking with respect to the processes underlying hierarchical ordering? Speciﬁcally, it appears that because the two predictions (i.e., dominant dimension versus dominant category) are made under diﬀerent methodologies, namely, two dimensions versus multiple dimensions, it is therefore not necessary to debate whether the cognitive dominance of a dimension versus the aﬀective dominance of a category produces the pattern. Rather, both may be true, and theorists must specify the conditions in which cognitive versus aﬀective variables will be more important in producing hierarchical ordering of targets. Thus, we see that the same pattern, whether it be additivity, dominance, conjunction dissimilarity (also known as social exclusion; see Crisp & Hewstone, 1999), or hierarchical acceptance, can result from very diﬀerent means. In some cases, aﬀect may underlie a pattern; in others, cognitive factors may be primarily responsible. Often, however, both processes are likely to be dynamically involved in what eventually becomes labeled a pattern of evaluation. Thus, the names for patterns are descriptive shortcuts, and the preceding considerations underscore the necessity to avoid assuming a single causal process when a pattern label is invoked.

SUMMARY AND CONCLUSION The goal of experimental social psychology is to explain social phenomena and generate predictive hypotheses. It also seeks to answer questions about psychological processes underlying aﬀective, cognitive, and behavioral eﬀects. In this chapter we examine processes that moderate a more general ﬁnding characteristically obtained within the crossed categorization paradigm. Speciﬁcally, outgroup members who share ingroup characteristics with the perceiver are evaluated more positively than those who do not. We refer to these persons as mixed targets because despite their salient membership in an outgroup, they share a salient ingroup membership with the perceiver. Diﬀerential category importance was found to be an important moderator of crossed categorization eﬀects in that by heightening the salience or relevancy of a category or its entire dimension, the eﬀects of mood can more readily act to accentuate or diminish bias between groups. We have seen that, as intuitively expected, positive mood tends to increase the favorability of evaluations of mixed targets, whereas negative mood tends to lower them. However, the story is not as simple as that. There are several important qualiﬁcations. First, with respect to positive aﬀect, apparently, an incidental good mood per se may improve outgroup evaluations provided the ingroup–outgroup categorization dimension is of low importance (see Crisp & Hewstone, 2000a). For more important dimensions, category dominance was necessary to produce more favorable judgments. Perhaps this occurs because the true ambivalence toward targets who possess important, yet conﬂicting, group

Miller et al.

183

cues must be overcome by the augmented salience of one dimension (Urada & Miller, 2000). Positive integral aﬀect reduced bias toward mixed targets who possessed an outgroup membership on the dimension that was the source of aﬀect, but had no eﬀect on target persons whose outgroup memberships were irrelevant (i.e., not a source of positive integral aﬀect; Ensari & Miller, 1998). These ﬁndings imply that when categories are saliently crossed, individuals experiencing a positive mood are, at best, less likely to exhibit bias and, at worst, will not diﬀer from those in a neutral mood. Overall, no evidence suggests that positive mood augments bias within crossed categorization. Next, we have seen that incidental sadness and anger (Kenworthy et al., 2003) increased negativity toward outgroup members. This was only true, however, for avoidance measures; when instead approach/aggression measures were used, only anger (but not sadness) increased negativity. Likewise, incidental disgust failed to produce an eﬀect for a measure that was aggressive in nature. By contrast, negative integral aﬀect, via insults or disgust (Canales, 1999; Ensari & Miller, 1998), produced more negative evaluations only when targets shared a category membership with the source of negative aﬀect. The implications of these ﬁndings are less clear than those for positive aﬀect. Incidental negative moods generally produce more avoidance of outgroup targets, provided that avoidance, and not aggression, is indeed measured. Not surprisingly, when an outgroup is the source of negative aﬀect, the unfortunate eﬀect of generalization of negativity to innocent outgroup members has been consistently found. Such an eﬀect qualiﬁes as an instance of displaced aggression (e.g., Marcus-Newhall, Pedersen, Carlson, & Miller, 2000). The opportunity for personalized contact has recently been examined experimentally (Ensari & Miller, 2001). Personalization of outgroup targets had a beneﬁcial eﬀect on later evaluations only insofar as category diﬀerences were not then emphasized, thus undermining the decategorizing eﬀect of personalization. We have explored the ﬁndings and implications of several moderators of bias within crossed categorization: category importance, positive and negative moods, and personalized interaction. Although cognitive overload was meta-analytically revealed to be a moderator that reduces diﬀerential bias across the four crossed categorization targets (Urban & Miller, 1998), no research has examined its eﬀects within crossed categorization. Moreover, the variables discussed herein have yet to be examined exhaustively. We hope the cumulative outcomes of these and future experiments will lead to a broad, yet precise, understanding of the bias that accompanies categorization processes. Equally important, however, is the development of practical tools that can serve to mitigate such bias.

184

Moderating ethnocentric bias

REFERENCES Abele, A. E., & Gendolla, G. H. E. (1999). Satisfaction judgments in positive and negative moods: Eﬀects of concurrent assimilation and contrast producing processes. Personality and Social Psychology Bulletin, 25, 883–895. Abele, A. E., Gendolla, G. H. E., & Petzold, P. (1998). Positive mood and ingroup– outgroup diﬀerentiation in a minimal group setting. Personality and Social Psychology Bulletin, 24, 1343–1357. Abele, A. E., & Petzold, P. (1994). How does mood operate in an impression formation task? An information integration approach. European Journal of Social Psychology, 24, 173–187. Alicke, M. D., Klotz, M. L., Breitenbecher, D. L., Yurak, T. J., & Vredenburg, D. S. (1995). Personal contact, individuation, and the better-than-average eﬀect. Journal of Personality and Social Psychology, 68, 804–825. Arcuri, L. (1982). Three patterns of social categorization in attribution memory. European Journal of Social Psychology, 12, 271–282. Bless, H., Clore, G., Schwarz, N., Golisano, V., Rabe, C., & Wölk, M. (1996). Mood and the use of scripts: Does a happy mood really lead to mindlessness? Journal of Personality and Social Psychology, 71, 665–679. Bodenhausen, G. V. (1993). Emotions, arousal, and stereotypic judgments: A heuristic model of aﬀect and stereotyping. In D. M. Mackie & D. L. Hamilton (Eds.), Aﬀect, cognition, and stereotyping: Interactive processes in group perception (pp. 13–37). San Diego, CA: Academic Press. Bodenhausen, G. V., Kramer, G. P., & Susser, K. (1994). Happiness and stereotypic thinking in social judgment. Journal of Personality and Social Psychology, 18, 335–345. Bower, G. H. (1981). Mood and memory. American Psychologist, 36, 129–136. Brewer, D., Doughtie, E. B., & Lubin, B. (1980). Induction of mood and mood shift. Journal of Clinical Psychology, 36, 215–226. Brewer, M. B., Ho, H., Lee, J., & Miller, N. (1987). Social identity and social distance among Hong Kong schoolchildren. Personality and Social Psychology Bulletin, 13, 156–165. Brewer, M. B., & Miller, N. (1984). Beyond the contact hypothesis: Theoretical perspectives on desegregation. In N. Miller & M. B. Brewer (Eds.), Groups in contact: The psychology of desegregation (pp. 281–302). San Diego, CA: Academic Press. Brown, R. J., & Wade, G. (1987). Superordinate goals and intergroup behavior: The eﬀect of role ambiguity and status on intergroup attitudes and task performance. European Journal of Social Psychology, 17, 131–142. Canales, C. J. (1999). Disgust as negative aﬀect: A crossed categorization study. Unpublished master’s thesis, University of Southern California. Carlson, M., & Miller, N. (1987). Explanation of the relation between negative mood and helping. Psychological Bulletin, 102, 91–108. Coser, L. A. (1956). The functions of social conﬂict. New York: Free Press. Crisp, R. J., Ensari, N., Hewstone, M., & Miller, N. (2002). A dual-route model of crossed categorization eﬀects. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 13, pp. 35–74). Hove, UK & Philadelphia, PA: Psychology Press. Crisp, R. J., & Hewstone, M. (1999). Diﬀerential evaluation of crossed category

Miller et al.

185

groups: Patterns, processes, and reducing intergroup bias. Group Processes and Intergroup Relations, 2, 307–333. Crisp, R. J., & Hewstone, M. (2000a). Crossed categorization and intergroup bias: The moderating roles of intergroup and aﬀective context. Journal of Experimental Social Psychology, 36, 357–383. Crisp, R. J., & Hewstone, M. (2000b). Multiple categorization and social identity. In D. Capozza & R. Brown (Eds.), Social identity processes: Trends in theory and research (pp. 149–166). London: Sage Publications. Crisp, R. J., Hewstone, M., & Rubin, M. (2001). Does multiple categorization reduce intergroup bias? Personality and Social Psychology Bulletin, 27, 76–89. DePaulo, B., & Fisher, J. (1980). The costs of asking for help. Basic and Applied Social Psychology, 7, 23–35. Deschamps, J.-C., & Doise, W. (1978). Crossed category memberships in intergroup relations. In H. Tajfel (Ed.), Diﬀerentiation between social groups (pp. 141–158). Cambridge: Cambridge University Press. Devos, T. (1998). Ingroup bias and dimensional relevance. Swiss Journal of Psychology, 57, 153–162. Dovidio, J. F., Gaertner, S. L., Isen, A. M., & Lowrance, R. (1995). Group representations and intergroup bias: Positive aﬀect, similarity, and group size. Personality and Social Psychology Bulletin, 21, 856–865. Ensari, N., & Miller, N. (1998). Eﬀect of aﬀective reactions by an outgroup on preferences for crossed categorization discussion partners. Journal of Personality and Social Psychology, 75, 1503–1527. Ensari, N., & Miller, N. (2001). Decategorization and the reduction of bias in the crossed categorization paradigm. European Journal of Social Psychology, 31, 193–216. Ensari, N., & Miller, N. (2002). The outgroup must not be so bad after all: The eﬀects of disclosure, typicality, and salience on intergroup bias. Journal of Personality and Social Psychology, 83, 313–329. Forgas, J. P. (1995). Strange couples: Mood eﬀects on judgments and memory about prototypical and atypical targets. Personality and Social Psychology Bulletin, 21, 747–765. Forgas, J. P. (2000). The role of aﬀect in social cognition. In J. P. Forgas (Ed.), Feeling and thinking: The role of aﬀect in social cognition (pp. 1–30). Cambridge: Cambridge University Press. Forgas, J. P. (2001). The aﬀect infusion model (AIM): An integrative theory of mood eﬀects on cognition and judgments. In L. L. Martin & G. L. Clore (Eds.), Theories of mood and cognition: A user’s guidebook (pp. 99–134). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Frijda, N. H., Kuipers, P., & ter Schure, E. (1989). Relations among emotion, appraisal, and emotional action readiness. Journal of Personality and Social Psychology, 57, 212–228. Green, J. D., & Sedikides, C. (1999). Aﬀect and self-focused attention revisited: The role of aﬀect orientation. Personality and Social Psychology Bulletin, 25, 104–119. Hagendoorn, L., & Henke, R. (1991). The eﬀect of multiple category membership on intergroup evaluations in a north Indian context: Class, caste, and religion. British Journal of Social Psychology, 30, 247–260. Haslam, S. A., Turner, J. C., Oakes, P. J., McGarty, C., & Hayes, B. K. (1992). Contextdependent variation in social stereotyping: The eﬀects of intergroup relations as

186

Moderating ethnocentric bias

mediated by social change and frame of reference. European Journal of Social Psychology, 22, 3–20. Hewstone, M., Islam, M. R., & Judd, C. M. (1993). Models of crossed categorization and intergroup relations. Journal of Personality and Social Psychology, 64, 779–793. Hewstone, M., & Lord, C. G. (1998). Changing intergroup cognitions and intergroup cognitions and intergroup behavior: The role of typicality. In C. Sedikides, J. Schopler, & C. A. Insko (Eds.), Intergroup cognition and intergroup behavior (pp. 367–392). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Hornsey, M. J., & Hogg, M. A. (2000). Subgroup relations: A comparison of mutual intergroup diﬀerentiation and common ingroup identity models of prejudice reduction. Personality and Social Psychology Bulletin, 26, 242–256. Isen, A. M., & Daubman, K. (1984). The inﬂuence of aﬀect on categorization. Journal of Personality and Social Psychology, 47, 1206–1217. Isen, A. M., Niedenthal, P. M., & Cantor, N. (1992). An inﬂuence of positive aﬀect on social categorization. Motivation and Emotion, 16, 65–78. Isen, A. M., Shalker, T., Clark, M. S., & Karp, L. (1978). Aﬀect, accessibility of material and behaviour: A cognitive loop? Journal of Personality and Social Psychology, 36, 1–12. Kelly, C. (1989). Political identity and perceived intragroup homogeneity. British Journal of Social Psychology, 28, 239–250. Kenworthy, J. B., Canales, C. J., Weaver, K. D., & Miller, N. (2003). Negative incidental aﬀect and mood congruency in crossed categorization. Journal of Experimental Social Psychology, 39, 195–219. LeVine, R. A., & Campbell, D. T. (1972). Ethnocentrism: Theories of conﬂict, ethnic attitudes, and group behavior. New York: Wiley International. Locksley, A., Borgida, E., Brekke, N., & Hepburn, C. (1980). Sex stereotypes and social judgment. Journal of Personality and Social Psychology, 39, 821–831. Mackie, D. M., Devos, T., & Smith, E. R. (2000). Intergroup emotions: Explaining oﬀensive action tendencies in an intergroup context. Journal of Personality and Social Psychology, 79, 602–616. Marcus-Newhall, A., Miller, N., Holtz, R., & Brewer, M. B. (1993). Cross-cutting category memberships with role assignment: A means of reducing intergroup bias. British Journal of Social Psychology, 32, 125–146. Marcus-Newhall, A., Pedersen, W. C., Carlson, M., & Miller, N. (2000). Displaced aggression is alive and well: A meta-analytic review. Journal of Personality and Social Psychology, 78, 670–689. Migdal, M., Hewstone, M., & Mullen, B. (1998). The eﬀects of crossed categorization in intergroup evaluations: A meta-analysis. British Journal of Social Psychology, 37, 303–324. Miller, N. (2002). Personalization and the promise of contact theory. Journal of Social Issues, 58, 387–410. Miller, N., Urban, L. M., & Vanman, E. J. (1998). A theoretical analysis of crossed social categorization eﬀects. In C. Sedikides, J. Schopler, & C. A. Insko (Eds.), Intergroup cognition and intergroup behavior (pp. 313–419). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Moreno, K. N., & Bodenhausen, G. V. (2001). Intergroup aﬀect and social judgement: Feelings as inadmissible information. Group Processes and Intergroup Relations, 4, 21–29. Mullen, B., Brown, R., & Smith, C. (1992). Ingroup bias as a function of salience,

Miller et al.

187

relevance, and status: An integration. European Journal of Social Psychology, 22, 103–122. Newman, R., & Goldwin, L. (1990). Children’s reluctance to seek help with schoolwork. Journal of Educational Psychology, 82, 71–80. Oakes, P. J. (1987). Salience and categorization. In J. C. Turner, M. A. Hogg, P. J. Oakes, S. D. Reicher, & M. S. Wetherell (Eds.), Rediscovering the social group: A self-categorization theory. Oxford: Blackwell Publishing. Park, B., & Rothbart, M. (1982). Perception of outgroup homogeneity and levels of social categorization: Memory for the subordinate attributes of ingroup and outgroup members. Journal of Personality and Social Psychology, 42, 1051–1068. Pedersen, W. C., Gonzales, C., & Miller, N. (2000). The moderating eﬀect of trivial triggering provocation on displaced aggression. Journal of Personality and Social Psychology, 78, 913–927. Roseman, I. J., Wiest, C., & Swartz, T. S. (1994). Phenomenology, behaviors, and goals diﬀerentiate discrete emotions. Journal of Personality and Social Psychology, 67, 206–221. Schwarz, N. (1990). Feelings as information: Informational and motivational functions of aﬀective states. In E. T. Higgins & R. M. Sorrentino (Eds.), Handbook of motivation and cognition: Foundations of social behavior (Vol. 2, pp. 527–561). New York: The Guilford Press. Schwarz, N. (2001). Feelings as information: Implications for aﬀective inﬂuences on information processing. In L. Martin & G. Clore (Eds.), Theories of mood and cognition (pp. 159–176). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Schwarz, N., & Clore, G. L. (1983). Mood, misattribution, and judgments of wellbeing: Informative and directive functions of aﬀective states. Journal of Personality and Social Psychology, 45, 513–523. Sedikides, C. (1992). Mood as a determinant of attentional focus. Cognition and Emotion, 6, 129–148. Simmel, G. (1908). The sociology of Georg Simmel. Glencoe, IL: Free Press. Simon, B., & Brown, R. (1987). Perceived intragroup homogeneity in minoritymajority context. Journal of Personality and Social Psychology, 53, 703–711. Spears, R., & Manstead, A. S. R. (1989). The social context of stereotyping and diﬀerentiation. European Journal of Social Psychology, 19, 101–121. Stemmler, G. (1989). The autonomic diﬀerentiation of emotions revisited: Convergent and discriminant validation. Psychophysiology, 26, 617–632. Strack, F., Schwarz, N., & Gschneidinger, E. (1985). Happiness and reminiscing: The role of time perspective, aﬀect, and mode of thinking. Journal of Personality and Social Psychology, 49, 1460–1469. Tajfel, H. (1978). Diﬀerentiation between social groups: Studies in the social psychology of intergroup relations. London: Academic Press. Tajfel, H. (1982). Social identity and intergroup relations. Cambridge: Cambridge University Press. 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. 33–47). Monterey, CA: Brooks/Cole. 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, 778–793. Teasdale, J. D., & Russell, M. L. (1983). Diﬀerential eﬀects of induced mood on

188

Moderating ethnocentric bias

the recall of positive, negative, and neutral words. British Journal of Clinical Psychology, 22, 163–171. Turner, J. C. (1982). Towards a cognitive redeﬁnition of the social group. In H. Tajfel (Ed.), Social identity and intergroup relations. Cambridge: Cambridge University Press. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. Cambridge, MA: Basil Blackwell, Inc. Urada, D. I., & Miller, N. (2000). The impact of positive mood and category importance on crossed categorization eﬀects. Journal of Personality and Social Psychology, 78, 417–433. Urban, L. M., & Miller, N. (1998). A theoretical analysis of crossed categorization eﬀects: A meta-analysis. Journal of Personality and Social Psychology, 74, 894–908. van Knippenberg, A., van Twuyver, M., & Pepels, J. (1994). Factors aﬀecting social categorization processes in memory. British Journal of Social Psychology, 33, 419–431. van Rijswijk, W., & Ellemers, N. (2002). Context eﬀects on the application of stereotype content to multiple categorizable targets. Personality and Social Psychology Bulletin, 28, 90–101. Vanbeselaere, N. (2000). The treatment of relevant and irrelevant outgroups in minimal group situations with crossed categorizations. Journal of Social Psychology, 140, 515–526. Vanman, E. J., Paul, B. Y., Ito, T., & Miller, N. (1997). The modern face of prejudice and structural features that moderate the eﬀect of cooperation on aﬀect. Journal of Personality and Social Psychology, 73, 941–959. Wagner, U., Lampen, L., & Syllwasschy, J. (1986). Ingroup inferiority, social identity and outgroup devaluation in a modiﬁed minimal group study. British Journal of Social Psychology, 25, 15–23. Wilder, D. A. (1984). Predictions of belief homogeneity and similarity following social categorization. British Journal of Social Psychology, 23, 323–333. Wilder, D. A., & Simon, A. F. (1996). Incidental and integral aﬀect as triggers of stereotyping. In R. M. Sorrentino & E. T. Higgins (Eds.), Handbook of motivation and cognition: The interpersonal context (Vol. 3, pp. 397–419). New York: The Guilford Press. Worth, L. T., & Mackie, D. M. (1987). Cognitive mediation of positive mood in persuasion. Social Cognition, 5, 76–94. Wyer, R. S., & Carlston, D. E. (1979). Social cognition, inference, and attribution. Mahwah, NJ: Lawrence Erlbaum Associates, Inc.

9

Gender among multiple social categories Social attraction in women but interpersonal attraction in men Ramadhar Singh

Every person can be categorized in multiple ways. The same Singaporean person, for example, can also be categorized as Buddhist, Chinese, doctor, rich, and woman. Among such multiple social categorizations, gender is the ﬁrst to be acquired by children around two years of age (see, e.g., Ruble & Martin, 1998, for a review and discussion), and it is the only social category that divides the world into nearly equal halves (United Nations, 2001). Given this apparently high level of social signiﬁcance, one might reasonably ask the question as to whether gender, in relation to all other categories, might have some special status in terms of how it is represented and used in social judgment. In this chapter, we present ﬁndings from a program of work that has attempted to provide some deﬁnitive answers to this question. This chapter has four goals. First, to describe brieﬂy the two approaches that have so far guided the social psychological study of the gender categorization. Second, to show that females and males process gender categories at intergroup and interpersonal levels respectively (see, e.g., Turner & Oakes, 1989, for a discussion of levels of processing in social perception). Third, to argue that crossing gender with another social category, as it is done within the crossed categorization paradigm (Deschamps & Doise, 1978), has the potential to distinguish social attraction from interpersonal attraction (e.g., Hogg & Hains, 1996), corresponding to the gender diﬀerence in intergroup and interpersonal processing tendencies, particularly when measures of attraction and intelligence are simultaneously taken. Fourth, to highlight the utility of examining patterns in factorial plots of crossed categorization eﬀects for testing between diﬀerent models of crossed categorization (Singh, Yeoh, Lim, & Lim, 1997).

TWO APPROACHES TO GENDER CATEGORIZATION Social-psychological studies of gender categorization have been carried out within either the social cognition approach (Devine, Hamilton, & Ostrom, 1994) or the social identity approach (Tajfel & Turner, 1986). In the former view, every natural category is believed to be associated with some unique

190

Gender among multiple categories

socially shared information. Perception and evaluation of people belonging to that category entails the mere activation of socially shared information from long-term memory. In the case of gender groups, for example, a warmthexpressive cluster of traits (e.g., sociable, warm, pleasant) is associated with women, but a competence cluster of traits (e.g., competent, determined, practical ) is associated with men (Broverman, Vogel, Broverman, Clarkson, & Rosenkrantz, 1972). In their review of the literature, Eagly and Mladinic (1994) characterized the foregoing warmth-expressive and competence clusters of traits as representing the communal and agentic dimensions of stereotypes, respectively. In addition, they reported men to be more agentic than women, but women to be more communal than men (see, for example, the “no occupation” condition of Figure 1.5, p. 24; Figure 1.6, p. 25). In a recent study men were accorded a higher rating than women on the positive masculine dimension of personality, whereas women were accorded a higher rating than men on the positive feminine dimension of personality (Diekman & Eagly, 2000). Accordingly, agency and communality are widely regarded as the respective stereotypes of men and women (see, e.g., Eagly & Wood, 1999, for a review). From a social cognitive perspective, then, gender groups are perceived and judged in accord with the stereotypes of men and women independent of the perceivers’ own gender. From the social identity perspective, gender categorization is believed to trigger an ingroup versus outgroup social comparison process, as a function of self-inclusion or exclusion. To aﬃrm or promote a positive social identity, moreover, the ingroup is perceived and evaluated more positively than the outgroup (see Hewstone, Rubin, & Willis, 2002, for a review). From this perceptive, men consider men and woman as the ingroup and outgroup, respectively, but women do the opposite. Consequently, respondents rate their respective ingroup more positively than the outgroup based on the gender (see, e.g., Deschamps & Doise, 1978). Does the presence of an intergroup comparison (ingroup vs. outgroup) always lead to ingroup bias? Eagly and Mladinic (1994) noted an absence of ingroup bias in their program of research on gender stereotypes, and a tendency for perceivers to consistently rate women as positive, irrespective of participant gender. They also claimed that a “women-are-wonderful eﬀect” held with men as well as women at least in the United States. In a 30-nation study of gender stereotypes, Williams and Best (1982) had found that women were evaluated more positively than men in some countries (e.g., Australia, Italy, the USA), but not in other countries (e.g., Israel, Japan, Malaysia). Such national diﬀerences are consistent with the social cognition approach that regards stereotypes as indices of societal beliefs. Given a broadly positive stereotype of women, in gender intergroup contexts the ﬁnding that women rate women more positively than men could be the generalized women-are-wonderful eﬀect (Eagly & Mladinic, 1994); that is, a reporting of a positive stereotype instead of ingroup bias. From the

Singh

191

perspective of theoretical parsimony, this social cognitive explanation in terms of knowledge activation (i.e., stereotypes) appears a more appealing explanation than that of identity enhancement. The two explanations, however, remain confounded, and in order to deal with this it may be necessary to consider other aspects of groups that can exert a moderating inﬂuence. One reason for the contemporary stereotypic preference for women pointed out by Eagly and Mladinic (1994) is political: “. . . it is no longer politically acceptable to derogate women” (p. 13). Consistent with this interpretation, Fiske, Cuddy, Glick, and Xu (2002) showed that most high-status groups, relative to low-status ones, are perceived as more competent but less warm, but the reverse holds with low-status groups. Since “. . . women and activities associated with women are considered less valuable and less prestigious than men and the activities associated with them” (Deaux & LaFrance, 1998, p. 791), it is likely that women and men perceive the social category of gender from their respective high- and low-status perspectives. Consideration of other variables such as status and status-related motives (Deaux, Reid, Mizrahi, & Cotting, 1999; Gaertner & Insko, 2000, pp. 90–91) in gender categorization is, therefore, important from both conceptual and applied angles.

INTERGROUP VERSUS INTERPERSONAL ORIENTATIONS The status of people in a social group aﬀects not only how others perceive them but also how they perceive themselves and behave with others (Fiske et al., 2002; Jackman, 1994), and if women perceive their group as of lower status than men (Deaux & LaFrance, 1998), women might be expected to be concerned more by their gender than other categories. If so, they should engage themselves in a number of identity-enhancement strategies (Singh, Choo, & Poh, 1998; Singh, Sharmini, & Choo, 2004). The situation with men should, however, be diﬀerent. Men should show greater attraction toward women than men for three reasons. First, the higher status of men may predispose them toward preserving the existing system of inequality through a paternalism in which women, like any other lower-status group, are simply stereotyped as communal (Fiske et al., 2002; Jackman, 1994). Second, stereotypes of gender groups (Eagly & Wood, 1999) make the female stereotype more positive than the male stereotype. Finally, men often desire to form new relationships (see, e.g., Baumeister & Learly, 1995, for a review), and women are seen to be more eﬀective than men in making any interaction intimate (see, e.g., Reis, 1998, for a review). Notably, the ﬁrst reason conceptualizes women as prototypical of lower-status groups; the remaining two reasons recognize woman as a unique social category; and the third one implies men’s dependence on women (Glick & Fiske, 1999). These conﬂicting motives should lead to the prediction that gender categorization will evoke intergroup responding (ingroup favoritism) in women but interpersonal responding (outgroup favoritism) in men, and that such a diﬀerence

192

Gender among multiple categories

in level of responding can be identiﬁed by using appropriate response measures of attraction and intelligence. Some initial evidence for this hypothesis was provided by Singh and Dalal (2002). Participants were asked to indicate whether they considered individual male and female targets as male or female, or as ingroup or outgroup according to their own gender. Agreement with each statement was obtained along a 9-point scale (1 = strong disagreement; 9 = strong agreement). Figure 9.1 presents the mean agreement of men and women with the two types of categorization. Evidently, men agreed with both types of categorization equally. In contrast, women described men and women more as an ingroup versus an outgroup than as males and females. These ﬁndings suggest that women, compared to men, were more inclined to process the gender category at the intergroup level.

Dimensions of gender perception One of the motives used to explain ingroup favoritism within social identity theory is competition (Brown, 2000; Tajfel & Turner, 1986). However, Deaux et al. (1999) identiﬁed as many as seven motives behind belonging to a group. Those motives ranged from comparison and competition to romantic involvement. Singh (2001) hypothesized that women might be driven more by competition with men, but men more by social interaction with women. This prediction of competitiveness in women was derived from the ﬁnding described above that women categorize gender groups more as “ingroup”

Figure 9.1 Mean agreement with man versus woman and ingroup–outgroup categorizations of gender groups by men and women in India. Adapted from Singh and Dalal (2002, Experiment 3). Reproduced with permission.

Singh

193

versus “outgroup” than as “women” versus “men” (Singh & Dalal, 2002). The prediction of social interaction in men was based on the ﬁndings that men, compared to women, have a stronger tendency to assume a sexual component in any heterosexual encounter (Abbey, 1982), and that men are dependent on women for sex, sexual reproduction, homemaking, and child care (Glick & Fiske, 1999). Following the work of Deaux et al. (1999), Singh (2001) developed a 28-item survey for perception of gender groups by men and by women. Both versions included items pertaining to identiﬁcation with one’s gender group (e.g., I am proud to be a man/woman) as well as how one responds to people of the opposite gender (e.g., As a man/woman, I actively compete with women/men; I feel comfortable in the company of men/women). Factor analysis, after removal of the ambiguous items, yielded six factors: comparison, competition, identiﬁcation, social interaction, romance, and respect. Figure 9.2 presents the mean factor scores of female and male participants. The three dimensions showing gender diﬀerence are listed on the left side and the three showing no such diﬀerence are shown on the right side of Figure 9.2. As expected, women scored higher than men along comparison and competition dimensions, whereas men scored higher than women in social interaction. Failure to ﬁnd diﬀerences on the dimensions of romance, identiﬁcation, and respect further indicated that the crucial diﬀerences between men and women lay in the motives of comparison, competition, and social interaction. These diﬀerences are in line with the past literature showing that

Figure 9.2 Mean factor scores of men and women in Singapore. Adapted from Singh (2001). Reproduced with permission.

194

Gender among multiple categories

people of lower status compare and contrast themselves with those of higher status (e.g., Martin, 1994), and that men depend heavily on women (Abbey, 1982; Baumeister & Learly, 1995; Glick & Fiske, 1999), and they add further support to the hypothesis of a qualitative diﬀerence in how women and men process gender categories. Taken as a whole, these ﬁndings suggest that both males and females identify highly with their respective gender group. Nevertheless, males were more motivated by social interaction with females, but females were more motivated by competition with males. This supports the argument that gender categories were processed by women and men at the respective intergroup and interpersonal levels. If this is indeed the case, there should be marked gender diﬀerences in the processing of gender in multiple categorization contexts.

SOCIAL ATTRACTION IN WOMEN BUT INTERPERSONAL ATTRACTION IN MEN Social attraction refers to people’s attachment to a collective (Hogg & Turner, 1985). When people categorize themselves as members of a group, they feel attracted toward each other merely because of the common membership. This attraction is depersonalized (Hogg & Hains, 1996), because each member is viewed as a prototype of the group. Such social attraction diﬀers from interpersonal attraction (Byrne, 1971), in which proximity, reciprocity, and similarity between the two persons and/or their other idiosyncratic characteristics are the critical factors. Recent studies (Singh & Ho, 2000; Singh & Teoh, 1999, 2000) have shown that interpersonal attraction is comprised of both social and intellectual evaluations. The former refers to liking a person; the latter refers to acknowledgement of or admiration for his or her capability. The two evaluations converge or diverge contingent upon how threatening the person is for the well-being of the interacting partner (Herbst, Gaertner, & Insko, 2003; Montoya & Horton, 2004; Singh & Teoh, 2000). If, as suggested above, women process gender on an intergroup level, while men process gender on an interpersonal level, this should lead to social attraction-related phenomena for women, and interpersonal attraction-related phenomena for men. The crossed categorization paradigm (see, e.g., Crisp & Hewstone, 1999, for a narrative review; see also Crisp & Hewstone, Miller et al., and Vescio et al., this volume) is ideally suited for separating the hypothesized tendencies toward social attraction of women from interpersonal attraction of men with respect to gender categorization. Since, when using this paradigm, the gender category would be crossed with other social categories, such as race or nationality, women and men should display a similar social attraction with regard to race and nationality categories. They would uniformly judge the ingroup by race or nationality more favorably than the outgroup, the

Singh

195

well-known phenomenon of ingroup bias (Brown, 2000). In the case of gender, however, women should show social attraction (i.e., the same category preference or ingroup bias), but men should show an outgroup preference or greater interpersonal attraction toward women than men. This paradigm would thus enable conclusions to be drawn with respect to the diﬀerent social cognitive and social identity processes operating at the gender categorization level, in comparison with responses to other, equally important, categorizations.

Patterns in crossed categorization eﬀects In the following studies, on the basis of the research reviewed, the social versus interpersonal attraction hypothesis predicted a moderating eﬀect of participant gender. Figure 9.3 presents predicted mean attraction of men toward the ingroup (I) and outgroup (O) by race (separate lines) and men (M) and women (W) targets (marked on the horizontal axis). The four means for the lower and upper lines of each graph represent responses to the outgroup or ingroup as a function of race and gender of target. The graphs in Panels A through D of Figure 9.3 show ingroup bias by race but outgroup preference by gender (i.e., the women-are-wonderful eﬀect). However, they illustrate four diﬀerent mechanisms underlying the processing of the gender category alongside race. Consider, for example, the graph in Panel A. The eﬀects of race and gender categories are nearly the same: Men are more attracted toward the ingroup than outgroup by race but are

Figure 9.3 Mean attraction of men as a function of race and gender of the targets. The patterns of Panels A through F display six ways of processing gender category with race.

196

Gender among multiple categories

attracted more toward a female than male target. This pattern of parallelism supports the additive model of crossed categorization eﬀects (e.g., Hewstone, Islam, & Judd, 1993; Singh et al., 1997; for a list of these models see Crisp & Hewstone, this volume, and also discussions in Miller et al. and Vescio et al., this volume). The graph in Panel B also has the pattern of parallelism, but with a diﬀerence: The eﬀect of social categorization by race (i.e., the distance between the lines) is stronger than that of gender of the target (i.e., the steepness of the slopes). A stronger eﬀect of race than gender supports the model of category dominance (Hagendoorn & Henke, 1991). The patterns in Panels A and B of Figure 9.3 cannot distinguish the women-are-wonderful eﬀect from the eﬀect arising out of men’s interpersonal attraction toward women (i.e., category-based responding in line with a positive stereotype, and interpersonal-based positive attraction). The pattern in Panel C is, therefore, useful in distinguishing the eﬀect of such competing processes. As can be seen in Panel C, men’s preference for women was not just prototype-driven, but was conﬁned to the ingroup by race (i.e., in the upper line). Besides, there is no distinction between men and women of the other race, which is contrary to the women-are-wonderful prediction (Eagly & Wood, 1999; Fiske et al., 2002). The pattern displayed in Panel D of Figure 9.3 retains the two properties (i.e., an ingroup bias by race and a preference for women) of the patterns in Panels A through C. Its uniqueness lies in showing smaller gender discrimination in the ingroup than outgroup by race. If discrimination based on gender is a political taboo (Eagly & Mlandinic, 1994), then the convergent pattern of Panel D may be viewed as a motivated defense of the ingroup but oﬀense against the outgroup by race. Such a pattern might therefore reﬂect an overprotection of the ingroup by race: that is, “Gender discrimination is much less in my race than in your race.” The pattern in Panel E of Figure 9.3 shows ingroup bias by gender; males are rated more positively than females, only with the ingroup by race. With the outgroup by race, however, the same preference for woman as in Panels A, B, and D holds. Although such a pattern shows a double ingroup bias, it does reﬂect the same overprotection of the ingroup by race as in Panel D. The pattern in Panel F of Figure 9.3 illustrates a compromise between the norms of ingroup bias and fair-mindedness; that is, bias according to one category but no discrimination with regard to another category. As can be seen, the line for the ingroup by race has a higher intercept than that for the outgroup by race, an ingroup bias. In contrast, there is no eﬀect of gender category at all (i.e., both lines are ﬂat). This pattern provides the strongest support for the model of category dominance by race. As the null eﬀect of the gender category is independent of the eﬀects of race category, this pattern can also be regarded as a compromise between the norms of ingroup bias by race and fair-mindedness with regard to gender (Singh, Choo, & Poh 1998, 2004). In fact, the weights assigned to the four groups (i.e., OM, OF, IM, and

Singh

197

IF) in ﬁve of the six patterns displayed in Figure 9.3 imply that gender is less predictive of attraction than race. Such dominance of race over gender may be an instance of a compromise between trying to claim superiority in one way but minimize diﬀerence in another way. This interpretation is justiﬁed on the grounds that each category by itself is of the same importance for the participants’ self-deﬁnition (Singh & Dalal, 2002). These predicted patterns were used to guide interpretation of ﬁndings from the program of research presented below. Methodological note From their review and meta-analysis of the literature, Urban and Miller (1998) noted that studies using a between-participants’ design yielded larger eﬀect sizes than did those using a within-participants’ design. They pointed out that the scope of comparison between groups in a repeated-measurements design perhaps “prompts participants to try to appear fair and impartial” (p. 904). In most of the studies within the author’s research program reported in this chapter, a repeated-measures design was used. To make the hypotheses and comparisons less transparent, categories other than race, nationality, and gender were also used. There were some targets described by three or four categories to deﬁne the endpoints of the response measures. The communal and agentic dimensions (Eagly & Wood, 1999) were measured by taking measures of attraction (i.e., How much would you enjoy the company of the person?) and intelligence (i.e., How intelligent is the person?), respectively. Most studies used just one item and a horizontal line along which participants made a vertical mark to indicate their responses. Since results from standard measures containing more items and using between-participants designs (Singh et al., 1997, 1998) converged with those obtained in this research program, attraction and intelligence responses could be regarded as conceptually equivalent to the respective communal or warmth and agentic or intellectual dimensions.

Stereotypes or social versus interpersonal attraction? In their Experiments 1 and 2, Singh and Dalal (2002) presented gender category alone and also crossed it with nationality or race. Participants were Chinese men and women in Singapore. While Experiment 1 took measures of attraction (i.e., liking for the target) only, Experiment 2 took measures of attraction as well as intelligence. There were equal numbers of male and female participants. Figure 9.4 presents mean attraction as a function of nationality or race (separate line) and gender of the target (marked on the horizontal axis). The means of the left, center, and right panels are from the nationality × gender (N × G), race × gender (R × G), and genderonly (G-only) designs, respectively. As the women-are-wonderful eﬀect was

198

Gender among multiple categories

Figure 9.4 Mean attraction as a function of nationality (left graph) or race (center graph) and gender of the targets (listed on the horizontal axis). The line of the rightmost graph is based on gender only. Adapted from Singh and Dalal (2002, Experiment 1). Reproduced with permission.

not moderated by gender of the participants, these ﬁndings did not support the diﬀerential processing hypothesis (i.e., social attraction for women, interpersonal attraction for men). Figure 9.5 presents mean attraction and intelligence (separate lines) of the male and female targets from the three designs of Experiment 2. As in Experiment 1, Singh and Dalal (2002) obtained the women-are-wonderful eﬀect on attraction. Contrary to the social cognitive hypothesis, however, ratings of intelligence of men and women were equal. These results portray women as more communal than agentic. However, they show no diﬀerence on intelligence. Furthermore, evidence for men being more agentic than communal in the left and center panels of Figure 9.5 is contradicted by the evidence for greater warmth than capability in both men and women in the right panel. This discrepancy suggests that the processing of the gender category presented alone was diﬀerent from that when it is crossed with either race or nationality category. Social versus interpersonal attraction The race × gender design of Experiment 2 by Singh and Dalal (2002) had found an interaction between participant gender, race and target gender. Figure 9.6 presents mean attraction in the same format as in Figure 9.4. The left and right graphs are for the responses by men and women, respectively.

Figure 9.5 Mean attraction and intelligence ratings of men and women targets in three designs of Experiment 2. Adapted from Singh and Dalal (2002). Reproduced with permission.

Figure 9.6 Mean attraction of men (left graph) and women (right graph) as a function of race and gender of the targets. Adapted from Singh and Dalal (2002), race × gender design of Experiment 2. Reproduced with permission.

200

Gender among multiple categories

Three results are striking. First, the women-are-wonderful eﬀect held fully with women but only partially with men. Second, Chinese men in Singapore were more attracted toward Chinese women than men. As the eﬀect of gender of the target is conﬁned to the ingroup by race, the obtained pattern supports the hypothesis of interpersonal attraction toward women as envisaged in Panel C of Figure 9.3. Finally, men considered the gender category to be less important than the race category, whereas women considered gender and race to be equally important (i.e., an additive model shown in Panel A of Figure 9.3). Such a gender diﬀerence further indicates that attraction of men and women can be regarded as interpersonal and social, respectively. Results from 15–17-year-old Chinese in Malaysia (Koay, 2002) also showed gender diﬀerences in processing the gender category with race. In particular, attraction of Chinese males in Malaysia was identical to that of Chinese males in Singapore (i.e., as in the left panel of Figure 9.6). However, attraction of females was ingroup-favoring by race but nonbiased with regard to gender, supporting the compromise hypothesis as conceptualized in Panel F of Figure 9.3. Age trends K. J. Tan and Singh (2002) hypothesized that women and men process social categories at the respective intergroup and interpersonal levels in expressing their enjoyment of company, and that such orientations develop with age. In addition, social attraction would be a generalized response to multiple categories, but the interpersonal attraction of men would be conﬁned to enjoyment of company. To test these hypotheses, the authors crossed gender of the target with his or her nationality, and took measures of attraction and intelligence in a counterbalanced order. Participants were 9–10-, 13–14-, 17–18-, and 21–22-year-old Chinese males and females in Singapore. Results from the measures of intelligence and attraction are shown separately in Figures 9.7 and 9.8. Both ﬁgures present mean responses as a function of nationality (separate lines) and gender (listed on the horizontal axis) of the target person. The age diﬀerences are displayed across the four panels; the gender diﬀerences at each age level are shown in the top and bottom panels. Two features of the results are notable. First, there was an ingroup bias by nationality over ages and across responses, conﬁrming the hypothesis of social attraction (Singh & Dalal, 2002). Second, age moderated responses by participants of both gender groups. Of greater importance, processing of the gender category was more variable with age for males than females and also in enjoyment of company. Consider ﬁrst the leftmost top and bottom graphs of Figure 9.7. Whereas the graph of boys has a pattern of convergence, that of the girls has a pattern of divergence. Such discrepant patterns are to be expected when both boys and girls prefer their respective gender categories. Boys as well as girls favored their respective gender group; furthermore the category preference was stronger in

Figure 9.7 Mean intelligence as a function of nationality and gender of the targets. The four panels display results from the four age groups, and the top and bottom panels display results from men and women, respectively. Adapted from K. J. Tan and Singh (2002, Experiment 1). Ke Jia Tan and Ramadhar Singh. Reproduced with permission.

Figure 9.8 Mean attraction as a function of nationality and gender of the targets. The four panels display results from the four age groups, and the top and bottom panels display results from men and women, respectively. Adapted from K. J. Tan and Singh (2002, Experiment 1). Reproduced with permission.

202

Gender among multiple categories

girls than in boys. As Figure 9.8 shows, attraction of the 9–10-year-old boys was moderated by gender only when the target was an outgroup by nationality. Taken together, results from the intelligence and attraction measures argued for a clear gender diﬀerence in processing of the gender category. While girls had an unambiguous ingroup bias by gender, a clearly depersonalized social attraction, boys showed such bias on intelligence only. At the ages of 13–14 years there were also gender diﬀerences. As can be seen in the second graph from the left of Figure 9.7, the female target was judged to be more intelligent than the male by both boys and girls. However, gender was secondary to nationality for boys, but as important as nationality for girls. For attraction, boys showed no gender preference, but girls had a category preference. In fact, nationality and gender were equally important to girls. In the third graph from the left of Figures 9.7 and 9.8, mean attraction followed the additive pattern, which means that the 17–18-year-old males regarded nationality and gender as equally important. This dramatic increase in importance of gender category in attraction of the 17–18-year-old males also reﬂects on their interpersonal attraction. The remaining graphs of Figures 9.7 and 9.8 have a uniform pattern: The 17–18-year-old females and the 21–22-year-olds showed no bias with respect to the gender category. The absence of the women-are-wonderful eﬀect in attraction responses of adults is at odds with the ﬁndings of Koay, Chan, and Singh (2002) and Singh and Dalal (2002), but consistent with those of Goh and Singh (2002, women) and G. L. Tan and Singh (2002, men). Experiment 2 of K. J. Tan and Singh (2002) took measures of attraction with Chinese and Malay women in Singapore. As in the past research, both groups of women showed a category preference, and dominance of nationality over gender. In fact, attraction responses of adults had more cases that were consistent (90%) than inconsistent (10%) with the women-are-wonderful eﬀect in the authors’ program of research, supporting the hypothesis of social attraction in women but interpersonal attraction in men. To summarize, the gender category can be processed at both the intergroup and interpersonal levels. Because judgments of intelligence showed either ingroup bias or no bias, processing of the gender category was always at the intergroup level. It was the enjoyment of company measure that revealed processing of the gender category at the interpersonal level by men. While women indicated a uniform social attraction (ingroup bias), men showed social attraction for nationality but interpersonal attraction in the case of gender.

CONTRIBUTIONS AND IMPLICATIONS One important contribution of the studies reported in this chapter lies in documenting the prevalence of the model of category dominance in Asian

Singh

203

nations (Hewstone et al., 1993). Even an important category like gender (Ruble & Martin, 1998; United Nations, 2001) is often rendered secondary to race, religion, or nationality in South and Southeast Asia. This evidence for the model of category dominance supports Hewstone (1996, p. 346), who noted: Outside the laboratory, one categorization tends to be primary whenever research relies on real social categorizations that have historical, cultural, socio-economic and aﬀective signiﬁcance. Thus, the category dominance model appears to reﬂect reality in multigroup societies (e.g., race in South Africa or the United States, religion in Northern Ireland or Bangladesh, and language in Belgium or Quebec). While the results from laboratory studies of natural social categories also showed the prevalence of the model of category dominance, the present evidence for this model of crossed categorization eﬀects is reﬂective more of the adaptive responses of men and women than social reality prevalent in the settings studied. Less emphasis on gender, relative to other categories, may serve diﬀerent identity needs in men and women. While the former may do so to make a positive self-presentation as fair-minded persons, the latter may do it to aﬃrm a positive social identity: “Women are as intelligent as are men.” If women are perceived as lower in social status than men (Deaux & Lafrance, 1998), then one could argue that equating women with men is a subtle form of ingroup bias (Goh & Singh, 2002). This demonstration, it should be emphasized, has been made possible by crossing the gender category with race, religion, or nationality, as is done within the crossed categorization paradigm (Crisp & Hewstone, 1999; Deschamps & Doise, 1978). By presenting the gender category with no reference to participant aﬃliation, such subtle processing of the gender category by men and women would not have been fully captured. Results also showed support for the additive model (Hewstone et al., 1993; Singh et al., 1997). Such cases were more the case with women than men. In fact, ratings of how much respondents enjoyed outgroup members’ company were only additive for Chinese adolescent males in Singapore. Another contribution lies in documenting that the gender category is processed at both intergroup and interpersonal levels. Consistent with the women-are-wonderful eﬀect, both women and men indicated that they would enjoy company of a female more than a male target. Because there was no such gender diﬀerence in judgments of intelligence, processing of the gender category at the respective intergroup and interpersonal levels by women and men is unique to the dependent variable of enjoyment of company. Hence, what is social attraction in women is possibly interpersonal attraction in men. Could the preceding preference for women over men by men be an outcome of the importance of the comparison dimension of enjoyment of company? Mummendey and Simon (1989), for example, showed that ingroup favoritism

204

Gender among multiple categories

is often higher on dimensions important to the ingroup than on those unimportant to it, but that outgroup favoritism is greater on dimensions which are important to the outgroup but unimportant to the ingroup. Because of the greater importance of the communal dimension for women than men (e.g., Broverman et al., 1972; Eagly & Mladinic, 1994; Fiske et al., 2002), the distinction drawn between social and interpersonal attraction can also be viewed as a subtle means of outgroup discrimination. Had such an importance-driven process operated in the present paradigm, men would have had better ratings than women with regard to intellect. Since there was no apparent discrimination along the measure of intelligence at the adult level, the hypothesis of importance of comparison dimension is not applicable. At the same time, shifts from ingroup bias at a young age to no bias at adolescence reﬂect an importance of personal identity as fair-minded persons in men but of social identity as equals in women. Models of crossed categorization eﬀects have received attention in the literature because of the potential that crossed groups oﬀer for reducing intergroup bias (Deschamps & Doise, 1978; Hewstone et al., 2002; see Migdal, Hewstone, & Mullen, 1998; Mullen, Migdal, & Hewstone, 2001, for metaanalyses). The overwhelming evidence for the model of category dominance reported in this chapter does, however, raise some doubt as to the potential of crossed categorizations as a bias-reducing intervention, and serves to illustrate the importance of continued study in this area.

SUMMARY AND CONCLUSION In this research, gender was processed diﬀerently by men and women, in ways that cannot be fully accounted for by the extant social cognition and social identity approaches in isolation. With respect to a measure of enjoyment of company, gender evoked intergroup responding in women but interpersonal responding in men. Even when gender evoked intergroup responding in judgments of intelligence, the equality between men and women appeared to be guided by the personal identity of fair-mindedness in men but by social identity of being as good as men in women. These processes were only captured because the gender category was crossed with other social categories of race, religion, or nationality. This supports the notion that using crossed categorization paradigms can not only potentially oﬀer the chance of reducing intergroup bias, but also help to shed new light on existing theoretical problems.

ACKNOWLEDGEMENTS This chapter is based on the results of a number of experimental and survey studies supported by Grant RP3950021 and Grant R–107–000–030–112 from

Singh

205

the National University of Singapore to me. I thank Elysia Su-yen Chan, Ajit K. Dalal, Hong Yi Goh, Poay Sun Koay, Ghim Leng Tan, and Ke Jia Tan for permitting me to use the results of their unpublished studies; Susheel Kaur for assisting me to complete some of the survey studies and analyze those data; and Fazlinda Bin Junid for her skillful preparation of the ﬁgures.

REFERENCES Abbey, A. (1982). Sex diﬀerences in attributions for friendly behavior: Do males misperceive females’ friendliness? Journal of Personality and Social Psychology, 42, 830–838. Baumeister, R. F., & Learly, M. R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497–529. Brewer, M. B., Ho, H., Lee, J., & Miller, N. (1987). Social identity and social distance among Hong Kong schoolchildren. Personality and Social Psychology Bulletin, 8, 393–400. Broverman, I. K., Vogel, S. R., Broverman, D. M., Clarkson, F. E., & Rosenkrant, P. S. (1972). Sex-stereotypes: A current appraisal. Journal of Social Issues, 28, 59–78. Brown, R. (2000). Social identity theory: Past achievements, current problems, and future challenges. European Journal of Social Psychology, 30, 745–778. Byrne, D. (1971). The attraction paradigm. New York: Academic Press. Crisp, R. J., & Hewstone, M. (1999). Diﬀerential evaluation of crossed category groups: Patterns, processes, and reducing intergroup bias. Group Processes and Intergroup Relations, 2, 307–333. Deaux, K., & LaFrance, M. (1998). Gender. In D. Gilbert, S. Fiske, & G. Lindzey (Eds.), The handbook of social psychology (4th ed., Vol. II, pp. 788–827). New York: McGraw-Hill. Deaux, K., Reid, A., Mizrahi, K., & Cotting, D. (1999). Connecting the person to the social: The functions of social identiﬁcation. In T. R. Tyler & R. M. Kraemer (Eds.), The psychology of the social self: Applied social research (pp. 91–113). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Deschamps, J. C., & Doise, W. (1978). Crossed category memberships in intergroup relations. In H. Tajfel (Ed.), Diﬀerentiation between social groups (pp. 141–158). Cambridge: Cambridge University Press. Devine, P. G., Hamilton, D. L., & Ostrom, T. M. (Eds.) (1994). Social cognition: Impact on social psychology. San Diego, CA: Academic Press. Diekman, A. B., & Eagly, A. H. (2000). Stereotypes as dynamic constructs: Women and men of past, present, and future. Personality and Social Psychology Bulletin, 26, 1171–1188. Eagly, A. H., & Mladinic, A. (1994). Are people prejudiced against women? Some answers from research on attitudes, gender stereotypes, and judgments of competence. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 5, pp. 1–35). London: Wiley International. Eagly, A. H., & Wood, W. (1999). The origin of sex diﬀerences in human behavior: Evolved disposition versus social roles. American Psychologist, 54, 408–423.

206

Gender among multiple categories

Fiske, S. T., Cuddy, A. J. C., Glick, P., & Xu, J. (2002). A model of (often mixed) stereotype content: Competence and warmth respectively follow from perceived status and competition. Journal of Personality and Social Psychology, 82, 878–902. Gaertner, L, & Insko, C. A. (2000). Intergroup discrimination in the minimal group paradigm: Categorization, reciprocation, or fear? Journal of Personality and Social Psychology, 79, 77–94. Glick, P., & Fiske, S. T. (1999). Sexism and other “isms”: Interdependence, status, and ambivalent content of stereotypes. In W. B. Swan, J. H. Langlois, & L. A. Gilbert (Eds.), Sexism and stereotypes in modern society: The gender science of Janet Taylor Spence (pp. 193–221). Washington, DC: American Psychological Association. Goh, H. Y., & Singh, R. (2002, July). Gender diﬀerences in group and interpersonal attraction. In A. K. Dalal (Chair), Multiple categories and the crossed categorization eﬀects. Symposium conducted at the meeting of the 25th International Congress of Psychology, Singapore. Hagendoorn, L., & Henke, R. (1991). The eﬀect of multiple category membership on intergroup evaluations in a north Indian context: Class, caste, and religion. British Journal of Social Psychology, 30, 247–260. Herbst, K. C., Gaertner, L., & Insko, C. A. (2003). My head says yes but my heart says no: Cognitive and aﬀective attraction as a function of similarity to the ideal self. Journal of Personality and Social Psychology, 84, 1206–1219. Hewstone, M. (1996). Contact and categorization: Social psychological interventions to change intergroup relations. In C. N. Macrae, C. Stangor, & M. Hewstone (Eds.), Stereotypes and stereotyping (pp. 323–368). New York and London: The Guilford Press. Hewstone, M., Islam, M. R., & Judd, C. M. (1993). Models of crossed categorization and intergroup relations. Journal of Personality and Social Psychology, 64, 779–793. Hewstone, M., Rubin, M., & Willis, H. (2002). Intergroup bias. Annual Review of Psychology, 53, 575–604. Hoﬀman, C., & Hurst, N. (1990). Gender stereotypes: Perception or rationalization? Journal of Personality and Social Psychology, 53, 197–208. Hogg, M. A., & Hains, S. C. (1996). Intergroup relations and group solidarity: Eﬀects of group identiﬁcation and social beliefs on depersonalized attraction. Journal of Personality and Social Psychology, 70, 295–309. Hogg, M. A., & Turner, J. C. (1985). When liking begets solidarity: An experiment on the role of interpersonal attraction in psychological group formation. British Journal of Social Psychology, 24, 267–281. Jackman, M. R. (1994). The velvet glove: Paternalism and conﬂict in gender, class, and race relations. Berkeley & Los Angeles: University of California Press. Koay, P. S. (2002). Impression from multiple social categories: Category dominance as a minority eﬀect. Unpublished honors thesis, Department of Social Work and Psychology, National University of Singapore. Koay, P. S., Chan, E. S., & Singh, R. (2002, July). Eﬀects of nationality, race, and gender categories on attraction in Singapore and Malaysia. In A. K. Dalal (Chair), Multiple categories and the crossed categorization eﬀects. Symposium conducted at the meeting of the 25th International Congress of Psychology, Singapore. Martin, J. (1994). Inequality, distributive justice, and organizational illegitimacy. In K. J. Murnigham (Ed.), Social psychology of organizations: Advances in theory and research (pp. 296–321). Englewood Cliﬀs, NJ: Prentice-Hall. Migdal, M. J., Hewstone, M., & Mullen, B. (1998). The eﬀects of crossed categorization

Singh

207

on intergroup evaluations: A meta-analysis. British Journal of Social Psychology, 37, 303–324. Montoya, R. M., & Horton, R. S. (2004). On the importance of cognitive evaluation as a determinant of interpersonal attraction. Journal of Personality and Social Psychology, 86, 696–712. Mullen, B., Migdal, M. J., & Hewstone, M. (2001). Crossed-categorization vs. simple categorization and intergroup evaluations: A meta-analysis. European Journal of Social Psychology, 31, 721–736. Mummendy, A., & Simon, B. (1989). Better or just diﬀerent III: The impact of comparison dimension and relative group size upon intergroup discrimination. British Journal of Social Psychology, 28, 1–16. Reis, H. T. (1998). Gender diﬀerences in intimacy and related behaviors: Context and process. In D. J. Canary & K. Dindia (Eds.), Sex diﬀerences and similarities in communication: Critical essays and empirical investigations of sex and gender interaction (pp. 203–231). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Ruble, D. N., & Martin, C. L. (1998). Gender development. In W. Damon (Series Ed.) & N. Eisenberg (Vol. Ed.), Handbook of child psychology: Social, emotional, and personality development (Vol. 3, 5th ed., pp. 933–1016). New York: Wiley International. Singh, R. (2001). Dimensions of gender perception. Unpublished results, Department of Social Work and Psychology, National University of Singapore. Singh, R., Choo, W. M., & Poh, L. L. (1998). Ingroup bias and fair-mindedness as strategies of self-presentation in intergroup perception. Personality and Social Psychology Bulletin, 24, 147–162. Singh, R., & Dalal, A. K. (2002). Gender among multiple categories: Self-enhancement or stereotypes? In A. K. Dalal (Chair), Multiple categories and the crossed categorization eﬀects. Symposium conducted at the meeting of the 25th International Congress of Psychology, Singapore. Singh, R., & Ho, S. Y. (2000). Attitudes and attraction: A new test of the attraction, repulsion, and similarity-dissimilarity asymmetry. British Journal of Social Psychology, 39, 197–211. Singh, R., & Sharmini Choo, I. (2004). Intergroup perception as a compromise between ingroup bias and fair-mindedness. Journal of Social Psychology, 135, 373–387. Singh, R., & Teoh, B. P. (1999). Attitudes and attraction: A test of two hypotheses for the similarity-dissimilarity asymmetry. British Journal of Social Psychology, 38, 427–443. Singh, R., & Teoh, B. P. (2000). Impression formation from intellectual and social traits: Evidence for behavioral adaptation and cognitive processing. British Journal of Social Psychology, 39, 537–554. Singh, R., & Teo, R. (2002). Attribution of agency to male and female leaders and groups. Unpublished results, Department of Social Work and Psychology, National University of Singapore. Singh, R., Yeoh, B. S. E., Lim, D. I., & Lim, K. K. (1997). Cross categorization eﬀects in intergroup discrimination: Adding versus averaging. British Journal of Social Psychology, 36, 121–138. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S. Worchel & W. A. Austin (Eds.), Psychology of intergroup relations (2nd ed., pp. 7–24). Chicago: Nelson-Hall.

208

Gender among multiple categories

Tan, G. L., & Singh, R. (2002). Ingroup bias and stereotypes in processing of gender category. In A. K. Dalal (Chair), Multiple categories and the crossed categorization eﬀects. Symposium conducted at the meeting of the 25th International Congress of Psychology, Singapore. Tan, K. J., & Singh, R. (2002). Age and gender diﬀerences in processing of nationality and gender categories in Singapore. In A. K. Dalal (Chair), Multiple categories and the crossed categorization eﬀects. Symposium conducted at the meeting of the 25th International Congress of Psychology, Singapore. Turner, J C., & Oakes, P. J. (1989). Self-categorization theory and social inﬂuence. In P. B. Paulus (Ed.), Psychology of group inﬂuence (pp. 233–275). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. United Nations (2001). World population prospects: The 2000 revision. New York: United Nations. Urban, L. M., & Miller, N. M. (1998). A meta-analysis of crossed categorization eﬀects. Journal of Personality and Social Psychology, 74, 898–908. Williams, J. E., & Best, D. L. (1982). Measuring sex stereotypes: A thirty-nation study. Newbury Park, CA: Sage Publications.

Part V

Societal and political perspectives

10 Multiple social categorization and identity among multiracial, multiethnic, and multicultural individuals Processes and implications Jean S. Phinney and Linda L. Alipuria The purpose of this chapter is to examine the nature and meaning of multiple social categorization in the United States in terms of the three “large-scale social cleavages” (Abrams, 1996, p. 144) of race, ethnicity, and culture. We focus on group-related identities and identity formation in relation to these three broad dimensions of society. Most research on multiple categorization within social psychology has focused on simultaneous categorization in terms of two diﬀerent identity domains, such as race and gender; that is, one may be both black and female (Crisp & Hewstone, 2000; see also Crisp & Hewstone, Miller et al., and Vescio et al., this volume). In contrast, we are interested in multiple categorization within a single domain, such as race (being both black and white, based on one’s dual parentage) or culture (being simultaneously Hispanic and American, based on living in both cultures). The issue faced by these individuals is not that of balancing the importance or relevance of two distinct characteristics of the self, but rather that of integrating or otherwise managing an internal complexity involving two potentially conﬂicting, often enriching, parts of one’s ethnic, racial, or cultural self. For multiracial, and multiethnic people, there is not an obvious single ingroup or outgroup. These individuals can claim membership in two or more groups but are sometimes not accepted by others as a member of either. Thus their own sense of self and the way in which society views them require a rethinking of the meaning of groups, particularly the idea of mutually exclusive categories and the distinctions between ingroups and outgroups. The extent of multiple categorization in terms of group-related identity and its importance for the individual and for society have only recently been recognized. Current research examining social categorization among multiracial, multiethnic, and multicultural people in the United States shows that individuals can and often do claim more than one ethnic identity (if multiethnic by birth) (Alipuria, 2002), more than one racial identiﬁcation (if multiracial by birth) (Daniel, 2002; Williams, 1997), and more than one personally meaningful source of cultural competence and values (Alipuria, 2002).

212

Multiple identities

We approach multiple categorization from the perspective of identity development, as conceptualized by Erikson (1968). Developmental psychology is concerned with the processes of identity construction rather than with ethnic, racial, and cultural self-labels. Individuals confront their grouprelated identities as a part of the process of ego identity development, which involves seeking a sense of continuity of self grounded in the social world. Erikson (1968), who introduced the construct of identity to developmental psychology, stressed the role society plays in presenting the individual with issues of importance that need to be resolved. The ways in which individuals who belong to multiple categories construct identities to incorporate their various heritages raise fundamental questions about the deﬁnition of groups, the processes by which people are assigned to groups, and the implications of such group assignment for society. To explore these questions in this chapter, we ﬁrst deﬁne the populations most clearly involved in multiple social categorization in terms of race, ethnicity, and culture. We present brieﬂy the historical and cultural context for understanding these group-related identity issues in the United States, and we deﬁne key concepts for the purposes of the chapter. We discuss identity formation as an approach to understanding the role of multiple categorization in individuals’ lives. We then review literature on identity issues among multiracial, multiethnic and multicultural individuals. We conclude with a consideration of the implications of multiple categories and identities.

POPULATIONS OF INTEREST FOR MULTIPLE SOCIAL CATEGORIZATION The clearest examples of people for whom multiple social categorization is relevant are multiracial or multiethnic individuals whose parents are from diﬀerent ethnic or racial groups, and who thus have a multiple heritage from birth (Alipuria, 2002; Phinney & Alipuria, 1996; Root, 1992b, 1996b; Williams-Leon & Nakashima, 2001). For convenience, the term multiracial is used in the current context to include combinations of conventionally deﬁned races and/or ethnic groups and is used in place of the more inclusive term multiracial/multiethnic, which refers speciﬁcally to both race and ethnicity (later in this chapter we diﬀerentiate between these two concepts). Multiracial people may have parents who are from two diﬀerent groups or parents who are themselves multiracial. Oﬀspring of these parents may be referred to as being of ﬁrst-generation, second-generation, or multigenerationally mixed heritage. These individuals often present to the world an ambiguous phenotype and, as a result, are frequently asked, “What are you?” Multiple social categorization also applies to individuals whose culture of origin diﬀers from the culture in which they reside and who identify to varying degrees with two (or more) cultures. The most obvious examples are immigrants (Phinney, Dupont, Espinosa, Revill, & Sanders, 1994). We refer

Phinney and Alipuria

213

to these people as multicultural, based on their exposure to more than one culture, even though they may not participate equally in these cultures. American-born oﬀspring of immigrants may be able to blend in with the mainstream culture if their appearance and cultural competence allows. Phenotype often prevents full acceptance into white American culture. To the extent that individuals of later immigrant generations maintain identiﬁcation with their distinctive subculture, they continue to be multicultural. American-born ethnic and racial minorities are also multicultural to varying degrees. Any minority group, immigrant or native-born, that does not participate fully in the cultural life of the mainstream dominant group, by choice or exclusion, experiences both cultures (Daniel, 2002). Finally, the discussion of multiple social categorization in terms of culture is not limited to people of multiple heritage, such as immigrants or minority group members. A multiplicity of cultural categories can be experienced by those who have monoethnic and monoracial heritages. Ethnic cultures, as public property, provide insight and content that individuals can use, in part or in whole, to develop their own cultural selves, incorporating an awareness of values, practices, and meanings to enrich and guide their lives. The United States, particularly its large cities, provides “contact zones” (Hermans & Kempen, 1998), where mutual exchange occurs at a deeply penetrating level, leaving everyone changed. Hybridity refers to this ﬂuidity and multiplicity of cultural evolution on the level of ethnic groups as well as individuals. The multiracial and multicultural people deﬁned above participate in this hybridity as well, incorporating varied cultural identiﬁcations. Parental ancestry and ascribed ethnic and racial categories do not dictate cultural orientation (Alipuria, 2002). Multiracial and multicultural people occupy a special position in regard to intergroup relations because of their exposure from early in life to the diﬀering perspectives of the two or more groups of which they are members. As has been shown in experimental study of perspective-taking, the result of a dual perspective is likely to be less bias against other groups (Galinsky & Moskowitz, 2000). The experience of multiracial and multicultural people gives them access to alternative points of view and presumably a broader understanding of other groups than that of the monoracial or monocultural person.

RACE, ETHNICITY, AND CULTURE IN THE UNITED STATES Psychology in the United States has only recently begun to explore the complex and ﬂuid nature of race, ethnicity, and culture and the interrelationships among them. Earlier conceptualizations assumed that persons could be identiﬁed as belonging to discrete groups, and ambiguous cases were ignored or forced into preexisting categories that denied their complex realities. The

214

Multiple identities

1960s Civil Rights movement and the 1965 change in immigration law have changed racial politics and increased the numbers of non-European immigrants. These changes increasingly have required psychologists to acknowledge the complexity of ethnicity, race, and culture and to acknowledge the importance of studying the identity issues and resolutions raised by these socially deﬁned constructs (Phinney, 2000). Although recognition of the complexity is recent in psychology, the United States has always been a land of ethnic and racial complexity. From its beginnings, the United States has encompassed several deep and intermixed racial strands, including indigenous people (Native Americans), European immigrants since the early 1600s, and African slaves in signiﬁcant numbers since the early 1700s. Racial boundaries have persisted, and race is the most highly bounded dimension in North America (Cornell & Hartmann, 1998). In the United States, these racial groups were deﬁned as mutually exclusive for the most part, with little category straddling allowed (Spickard, 1992), even though racial mixing has occurred between the racial populations at all times in the nation’s history (Daniel, 2002; Rockquemore & Brunsma, 2002). In 1848, Mexican territory was annexed into what is now the southwestern United States, incorporating a “Hispanic” population that had a prior history of the multigenerational racial mixing commonly found in Latin America. Instead of challenging or disrupting the existing racial categorization boundaries, this “Hispanic” population was deﬁned in terms of ethnic culture instead of race, and, to this day, it is the only ethnic group speciﬁcally identiﬁed on American census reports as an ethnic category that may include several races (Grieco & Cassidy, 2001). The boundaries of the preexisting racial categories were thus undisturbed by the addition of this Mexican, racially mixed, mestizo population. Until the 2000 census, racial assignment for multiracial persons was a matter of choosing only one racial heritage with which to identify. Historically, the rule of hypodescent, that is, assignment of an individual to the racial group of lower status (Root, 1992a), deﬁned as black any one with “one drop” of black blood (i.e., a single black ancestor). The practice of assigning a black racial categorization to anyone with recognizable African features or known black ancestry developed in times of slavery and exerts inﬂuence on identity assignment and ascription even today. The last laws against interracial marriage (or sexual relations) were struck down, for the nation as a whole, only in 1967. However, even when racial boundaries became more permeable, people of mixed heritage were often pressured to identify as black to promote racial solidarity (Daniel, 2002). Throughout the early 20th century, immigration was mainly from Europe. In 1965, the same year in which Civil Rights legislation ended racial discrimination by law, immigration laws were radically changed. Since 1965, the major source of immigration to the United States has shifted from Europe to Latin America and Asia, reversing the trend since the founding of the nation. In many areas of the United States, the long-standing primary black/white

Phinney and Alipuria

215

racial dichotomy has become a more faceted racial matrix that now includes more Latin American, Asian, Middle Eastern, African, and Paciﬁc Island immigrants. The terms ethnicity, race, and culture refer to concepts that are used in diﬀerent ways over time and across geographical regions. We present distinctions important in the American context but caution that these concepts do not necessarily carry the same connotations in other settings or cultures.

Ethnicity An ethnic group, as deﬁned by sociologists Cornell and Hartmann (1998), implies a belief in a common ancestry and blood, that is, a sense of peoplehood. An ethnic group is formed by a population as an evolving consensus of the essential qualities, historical continuity, values, customs, and practices that “we” have and participate in but that “others” do not. “Ethnic groups are self-conscious populations; they see themselves as distinct” (p. 19). Ethnic groups change over time; they can increase or decrease in distinctiveness, or even be created where none existed (Roosens, 1989). In the United States, ethnicity has been used primarily to refer to non-European minority and immigrant groups and to European immigrant communities that have maintained themselves as separate from mainstream European American culture (Phinney, 1996). One’s ethnicity is a given, based on one’s ancestry, but the strength and quality of one’s relationship to that ethnic ancestry, that is, one’s ethnic identity, varies across contexts for a given individual and across individuals in the same group (Phinney, 1990).

Race Race, although not biologically deﬁnable (Zuckerman, 1990), is given signiﬁcance through social construction of meanings attributed to phenotypic diﬀerences. In the United States, race carries a stigma for the nonwhite person, giving the concept of race “a power in people’s lives that no mere logical argument can dispel” (Zack, 1993, p. 19). The stronger the stigma and the ﬁrmer the boundaries between races, the more relevant race is for society and the particular group and individuals in that group. Race, as constructed in the United States, is primarily an imposed category, not a dimension of selfcategorization. However, because of the variability in phenotypes, there is a wide variation in the extent to which a racial category is imposed. A racial identity deﬁnes how one engages the fact that by accident of birth one is either privileged or stigmatized within the racial matrix. According to Helms (1995), the racial identity task for nonwhites is to surmount “internalized racism in its various manifestations” (p. 24); for whites, the task is “abandonment of privilege” (p. 25). The distinction between racial categorization (identiﬁcation, whether self-chosen or imposed) and identity (used here as a

216

Multiple identities

response of the individual to the fact of racial stigma) is important to keep in mind in this discussion. Ethnicity and race, as deﬁned above, represent diﬀerent ways of dividing the human population into groups, and the resulting psychological challenges are somewhat diﬀerent. However, the terms are often used imprecisely and interchangeably. The term ethnicity is frequently used in the United States to refer to groups that vary racially as well as ethnically (Phinney, 1996). Latinos are typically of mixed indigenous Indian and European background, but also can be of African origin; Asian Americans diﬀer from European Americans both in terms of racial phenotype and cultural background and vary among themselves in terms of national origin or ethnicity. Because these distinctions are not clearly made in most cases, there is often confounding of the concepts of race and ethnicity, and the concepts of racial and ethnic identities are not clearly diﬀerentiated.

Culture Culture, according to Misra and Gergen (1993), is “a historically situated, collective product, constituted by the values, beliefs, symbols and other humanly created artifacts which are transmitted across generations” (p. 226). For the individual, culture is a far more malleable concept than ethnicity or race. Individuals can draw from their own ethnic culture and from others, to make sense of their own life and construct a cultural self. Thus the cultural categories used may or may not be tied to parental ethnic or racial groups. Cultural identity refers to self-constructed awareness of the cultural elements that inform one’s sense of self.

ETHNIC AND RACIAL IDENTITY FORMATION The study of multiple categorization within the social identity paradigm has focused on the ways in which individuals categorize, label, or identify themselves (Crisp & Hewstone, 2000). Multiracial and multicultural people can label themselves in a variety of ways, claiming membership in a single group, in two or more groups, or in a combined or alternative category. The choices they make in this regard, for example, the boxes that they check on forms, indicate an aspect of self-deﬁnition and also have social and political implications, as discussed later. In contrast to the focus on self-categorization, research on identity formation examines the process of constructing an identity that underlies these choices. A focus on identity formation provides a means of understanding multiple categorization from the perspective of multiracial or multicultural people themselves. Individuals confront their group-related identities as a part of the ego identity development process described by Erikson (1968). Erikson’s model of ego identity formation describes a process that takes

Phinney and Alipuria

217

place “in the core of the individual and yet also in the core of his communal culture” (p. 22). Identity, according to Erikson (1964), “depends on the support which the young individual receives from the collective sense of identity characterizing the social groups signiﬁcant to him: his class, his nation, his culture” (p. 93). This process involves the eﬀorts of individuals to integrate understandings of themselves and their place in their communities and societies and to achieve a sense of stability and continuity—an identity that can serve as a guide to the future. Because this process is similar to the formation of ethnic, racial, or multiracial identity, we discuss brieﬂy empirical research based on Erikson’s model. Central to the process of ego identity development as studied empirically by Marcia, Waterman, Matteson, Archer, and Orlofsky (1993) are the two processes of exploration and commitment. Exploration includes active learning and investigation into options in key areas of one’s life; and commitment involves a decision regarding these options. These areas, termed domains, include occupation, religion, politics, gender roles, and others. Individuals who have neither explored nor made commitments are termed identity diﬀuse. When the two dimensions of exploration and commitment are dichotomized and crossed in a two-by-two table, four identity types or statuses are deﬁned (see Figure 10.1). Those who have made commitments without having explored are described as foreclosed; they have accepted without question the attitudes and values of parents or parental ﬁgures. Those who are in the process of exploration and have not made commitments are in moratorium. Individuals who make commitments following a period of exploration are described as having an achieved identity. However, those who achieve a secure identity may later engage in further exploration and revise their commitments in response to internal or external pressures (Kroger, 2000). The validity of these identity statuses has been established in extensive research that documents diﬀerences in the psychological characteristics of people in diﬀerent statuses (Marcia et al., 1993). The process of ego identity formation has important parallels to the process of ethnic or racial identity formation. The study of group identities by

Figure 10.1 Identity statuses deﬁned by presence or absence of exploration and commitment.

218

Multiple identities

social psychologists has generally not considered developmental changes in group identity; the focus has rather been on the strength of the commitment that individuals have to their groups (e.g., Ellemers, Spears, & Doosje, 2002; see also Crisp; Hogg & Hornsey, this volume). The developmental model of ethnic identity (Phinney, 1989) based on the work of Marcia et al. (1993) gives an important role to commitment but notes that there are diﬀerent types of commitment, depending on the extent of exploration; foreclosure involves a commitment without exploration, whereas identity achievement is based on a commitment that follows exploration. To provide an understanding of this diﬀerence, we describe the developmental processes of ethnic and racial identity formation among members of ethnic minority groups. Ethnic and racial identities diﬀer in important ways from many other group identities. Most importantly, race and ethnicity are ascribed characteristics of a person that cannot be changed; one is considered to be in a particular racial or ethnic category on the basis of one’s parentage or appearance. In contrast, much social psychological research on group identity in real-world settings has examined identity in categories such as clubs, professional groups, or social groups that are chosen by the individual. Furthermore, for members of ethnic and racial minority groups, some of the ascribed identities carry with them a stigma that constitutes a limiting and often negative social identity. Ellemers et al. (2002) point out that when individuals are in situations where their group is low in status and they cannot—or do not wish to—leave the group, they may internalize a sense of inferiority or, alternatively, engage in creative strategies to aﬃrm the value of the group (Tajfel & Turner, 1986). These alternatives can be understood from a developmental perspective in terms of the ethnic or racial identity formation process. Developmental psychologists have described ethnic identity as a dynamic, evolving sense of self that changes with age (Phinney, 1989, 1990, 1993; Phinney & Kohatsu, 1997). Ethnic identity formation involves a process in which individuals explore the meaning of their given group membership and make a commitment to a way of relating to their group. Initially, young adolescents may show little interest in or knowledge about their ethnicity (identity diﬀusion) or may have internalized a conception of ethnicity based on messages from their parents or from society (foreclosure). Adolescents typically begin to explore and question the meaning and implications of their ethnicity as a result of an experience that highlights their ethnicity, such as discrimination or exposure to other groups, or as a part of the normal adolescent identity formation task (moratorium). This process can be highly emotional and disruptive, but it is assumed in most cases to lead to a secure, conﬁdent sense of one’s own ethnicity (identity achievement). A related sequence has been described for racial identity, although the emphasis in the racial identity literature for non-whites is on the process of overcoming internalized racism (foreclosure) through “immersion” in one’s group (Cross, 1991; Helms, 1990). Most of the literature on racial identity has dealt with African Americans,

Phinney and Alipuria

219

but a similar process has been described for Asian and Latino groups (Wijeyesinghe & Jackson, 2001). Adams (2001) identiﬁes core processes that are common across racial identity models. She points out that central to all models is a process (moratorium) that leads to a transformed consciousness and a redeﬁnition by which members of subordinated groups aﬃrm their own identity (identity achievement) against the stereotypes and discrimination they experience from the larger society. Atkinson, Morten, and Sue (1993) propose a generic model of minority identity development, consisting of ﬁve stages, from unquestioning conformity to cultural values of the mainstream (foreclosure), through stages of awareness, exploration, and introspection (moratorium), to a sense of self-fulﬁllment and resolution of conﬂicts regarding one’s ethnic or racial identity (achievement). Although research on the formation of a multiracial or multiethnic identity is limited, the process appears to be similar, with some important diﬀerences (Kich, 1992). Most notably, multiracial children are typically aware at an early age that they are diﬀerent from each of their parents, and they learn early that they do not fully belong to any of the obvious reference groups in their environment. They are therefore less likely to experience diﬀusion or foreclosure; rather, they engage in an extended moratorium. From their awareness of diﬀerence, they struggle for acceptance as biracial (Kich, 1992). However, lacking (in most cases) an existing biracial group with which to identify, they face a more diﬃcult task. Identiﬁcation with the group of one of their parents implies rejection of the other parent. An environment in which there are other biracial people can help the individual understand and accept his or her biraciality. According to Kich (1992), self-acceptance of a biracial identity, comparable to an achieved identity, is likely to occur during or after college or occupational decisions. The content of the identity that they achieve may take various forms, as discussed later in this chapter. Common to all the models of group identity formation is the idea that one’s identity changes over time and reﬂects diﬀering degrees of exploration and commitment. Particularly importantly for the present context is the idea that there are diﬀerent types of commitment. A commitment made without thoughtful consideration of the meaning and implications of the group membership, that is, foreclosure, is qualitatively diﬀerent from, and less secure or stable than, a commitment made after an exploration of one’s group membership, that is, identity achievement. The changes in selfdeﬁnition that take place in the identity formation process are relevant for intergroup relations because of their implications for attitudes toward other groups.

IDENTITY FORMATION AND INTERGROUP RELATIONS A key concept in the relationship between identity and intergroup relations is that of commitment. As Ellemers, Spears, and Doosje (1999) point out,

220

Multiple identities

“Group commitment not only predicts which kinds of inter- and intragroup relations people are likely to perceive as threatening, but it also determines the way they respond to such threats. . . . Group commitment . . . moderates group members’ social perceptions as well as their behavior on a range of indicators” (p. 3; see also Crisp; Hogg & Hornsey, this volume). The developmental perspective suggests that intergroup attitudes vary not only in relation to commitment but also in terms of exploration. A commitment based on exploration reﬂects a more secure group identity, which in turn is associated with more positive attitudes toward other groups (Phinney, Ferguson, & Tate, 1997). Thus developmental diﬀerences in ethnic or racial identity can provide one approach to understanding diﬀerences in attitudes toward other groups. Racial, ethnic, and minority identity models (Atkinson et al., 1993; Cross, 1991; Cross & Fhagen-Smith, 2001; Helms, 1990; Phinney, 1989) suggest diﬀerences in intergroup attitudes associated with changes in identity (see Table 10.1). When ethnic minority group members have an unformed or unexamined sense of their own group, they may have internalized negative societal images of their group and hold more positive attitudes toward the dominant group (foreclosure); they may in fact wish that they were “white” (Phinney, 1989). Subsequently, as individuals become immersed in their group and explore its history and culture (moratorium), they may feel anger and hostility toward the dominant group for past and present mistreatment and discrimination. It is at this point that members of minority groups may Table 10.1 Ethnic/racial identity formation and intergroup attitudes of minority group members Group identity status

Attitudes toward other groups

Diﬀusion

Individual has given little thought to the topic of diversity and is not very interested. Attitudes depend on socialization and intergroup context.

Foreclosure

Attitudes are based on internalization of parental or societal attitudes. Internalization of societal attitudes may lead to preference for the dominant group, and the desire to imitate majority group members. Alternatively, strong exclusive ties to one’s group may limit contact with other groups.

Moratorium

Increased awareness of a group’s history and discrimination may lead to distrust and anger toward the dominant group, along with feelings of shared concerns with other minority groups.

Achieved

Security in own group leads to acceptance of diversity. Selective liking for members of the dominant group combine with recognition of the reality of continuing prejudice.

Phinney and Alipuria

221

have the most negative feelings toward the dominant group. However, feelings toward other minorities may be sympathetic. According to developmental models, when individuals form a secure and conﬁdent sense of their own group membership (identity achievement), they are open to other groups, including the dominant group; their assessment of others now depends less on an individual’s race and more on individual characteristics. The statement of an adolescent in an interview study of intergroup attitudes reﬂects this level of understanding: “[Diversity] is good as long as you don’t forget yourself and at the same time you learn about other cultures. You have to know who you are, at the same time know what other people are. If you actually get to know them, they are a lot like everyone else. Every one has their traditions” (Oustinovskaya & Phinney, 2001). A study of intergroup attitudes among adolescents at ethnically diverse high schools (Phinney et al., 1997) provides support for the assumption of developmental models, that a secure sense of aﬃrmation of one’s own group is associated with more positive outgroup attitudes. Latino, Asian American, and African American high school students reported the strength of their ethnic identity, and their attitudes toward classmates from their own and other ethnic groups. Path analyses showed that a strong ethnic identity predicted adolescents’ positive attitudes toward peers from their own group, which in turn predicted positive attitudes toward other groups (see Figure 10.2). The results support the view that attitudes toward one’s own and other groups are positively related; a secure ethnic identity is accompanied by greater acceptance of other groups. Further evidence of this relationship comes from a recent study of 821 college freshmen from a predominantly ethnic minority university (Phinney, 2003b). Participants from Latino (58%), Asian American (22%), African American (7%), and European American (5%) backgrounds completed measures of ethnic identity and attitudes toward other groups. On the basis

Figure 10.2 Path model showing the prediction of outgroup attitudes from ethnic identity and ingroup attitudes among black and Latino adolescents. Beta weights are shown from Phinney et al. (1997). Reprinted with permission from Blackwell Publishing.

222

Multiple identities

of scoring high or low on exploration and commitment, they were assigned to one of the four ethnic identity statuses described earlier (see Figure 10.1). Analyses revealed a signiﬁcant association of identity status with attitudes toward other groups. Those with an achieved or secure identity status scored highest, signiﬁcantly higher than those with a foreclosed or diﬀuse status (see Figure 10.3). Even though the foreclosed participants scored high on commitment, they reported signiﬁcantly less positive attitudes toward other groups than the achieved participants. The identity-achieved participants did not diﬀer from those in the moratorium status, who also scored relatively high on other group attitudes. The positive attitudes among the moratorium participants suggest that the process of exploration, that is, thinking and learning about one’s own group, may make a contribution to other group attitudes. This ﬁnding may also reﬂect the context, as the participants lived and attended college in a predominantly minority setting where other group attitudes were primarily in reference to other minority groups, not to the dominant or white majority. Also of relevance to the topic of intergroup relations is evidence from research with children and adolescents for diﬀering levels of understanding ethnicity and culture and oneself in relation to these concepts. Quintana, Ybarra, Gonzalez-Doupe, and De Baessa (2000) have shown that Latino children and youth show systematic, age-related changes in the understanding of ethnicity. Young children view ethnicity in terms of literal, objective features such as food and language. With increasing age children become aware of social aspects of ethnicity, such as socioeconomic diﬀerences among groups, and still later develop an ethnic group consciousness. Finally, adolescents

Figure 10.3 Strength of positive orientation toward other groups in terms of ethnic identity status.

Phinney and Alipuria

223

develop a multicultural perspective that includes an enhanced ability to assume perspectives of other ethnic and racial groups. A diﬀerent but related approach is that of Bandlamudi (1994), who studied understandings of self and culture among ﬁrst-generation immigrant children from Asian, Indian, and European backgrounds. She described increasingly complex levels of understanding culture and oneself as a cultural being. Although she did not examine age changes, her categories show similarities to Quintana’s. Her categories include: a subjective, physical view of culture; a descriptive and comparative view of culture; a rule-governed view of culture as a code of conduct; a complex view of cultural systems that evolve; and a transformational view of self and culture as interactive, constantly creating each other over time. These studies and models have focused on individuals in culturally diverse settings, who have personal experience with two or more cultural or ethnic groups. They suggest that a secure understanding of one’s own group depends on suﬃcient experience or knowledge to see one’s own group within a larger context that includes other groups. For example, there is generally a stronger sense of ethnic identity among minority group members, who must accommodate to the dominant group, than among the majority white population (Phinney, 1990). Likewise, immigrants and minorities typically have a greater bicultural competence than do the majority (LaFromboise, Coleman, & Gerton, 1993). Individuals who are not familiar with other cultures or ethnic groups are likely to have less complex views of culture and be less aware of the perspectives of other groups. They may therefore more readily accept negative stereotypes and hold biases against them. Conversely, a secure achieved ethnic identity and the accompanying positive intergroup attitudes may depend on intercultural contact (Pettigrew, 1986). Because multiracial people, by virtue of their membership in two diﬀerent groups, have direct contact with two or more cultures and in fact represent the overlap of cultures, they should have a unique and broader perspective on group identity and intergroup relations than people who belong to only one group. The experiences of multiracial people, as they struggle to manage their dual backgrounds and form an identity, have important implications for the deﬁnitions of groups and for intergroup attitudes. These issues are explored in the next section.

MULTIRACIAL AND MULTIETHNIC PEOPLE Most of the research on multiracial individuals focuses on the eﬀorts of these people to establish a sense of identity and an understanding of their place in a society that imposes mutually exclusive racial categories (Root, 1992b, 1996b). The early literature was concerned with the question of whether multiracial people experienced psychological problems as a result of their

224

Multiple identities

mixed background. Often cited is the work of Stonequist (1961), who described the person of mixed ethnic or cultural heritage as the marginal man, the outsider who did not belong anywhere and as a result was expected to have psychological problems. Although early research with clinical samples (Gibbs, 1987) suggested that multiracial individuals had adjustment problems, more recent research (Grove, 1991; Phinney & Alipuria, 1996; Stephan & Stephan, 1991) generally shows they do not diﬀer in psychological well-being from those from a single racial or ethnic background. Current research also supports Park’s (1928) view that having a multiple heritage can be enriching, enabling a wider and more enlightened worldview. Of greater interest in the present context is the way in which these people question and stretch existing racial and ethnic categories. At the simplest level, namely, the label they use for themselves, most multiracial people resist being classiﬁed into a single group. Eﬀorts to modify the single categories used by the United States census came largely from parents in interracial relationships and their multiracial oﬀspring (Daniel, 2002). Following pressure for change during the late 1980s and the 1990s, the 2000 census, for the ﬁrst time, allowed individuals to check more than one box to indicate their race or ethnicity. This change has raised important questions of how to report the results when categories are not mutually exclusive. According to the Census Bureau, data on race are being reported in several diﬀerent ways. In some cases, mutually exclusive categories are used for six single “race” groups (white, black, American Indian/Alaska Native, Asian, Native Hawaiian/Paciﬁc Islander, Other race) and a mixed race group. In other cases, data will be shown for 63 racial categories—the same six groups alone and 57 possible combinations of these six categories. Finally, in still other cases, overlapping categories are used that include those who belong to a particular group both alone and in combination, e.g., either Asian or Asian/black. In the latter case, the actual numbers reported will be greater than the total population, since a biracial Asian/black individual could be included in both the Asian and mixed and the black and mixed categories (Daniel, 2002). These varied options highlight the complexity of the categorization process. The opportunity to select more than one category also creates pressures on multiracial individuals to make particular choices, because of the political and economic implications of racial tallies. Nationwide over 6.8 million people indicated in the 2000 American census that they were of mixed heritage, 2.4% of the total population (Jones & Smith, 2001). Since previous census data do not include information on multiracial people, we do not know the extent of change. However, intermarriage is increasing dramatically; in the past three decades the number of interracial marriages has increased from 300,000 to 1.4 million. Those who reported more than one race are younger than those selecting one race, suggesting that the number of interracial persons is rising (Jones & Smith, 2001). As a result, it will be increasingly diﬃcult to conﬁdently assign people to a single group.

Phinney and Alipuria

225

One issue of concern for researchers as well as demographers is the fact that multiracial individuals often fail to report their multiple heritages (Alipuria, 2002), thus leading to undercount. Furthermore, the use of a label varies with the way the question is asked. In a recent survey (Phinney & Alipuria, 1996), multiracial high school and college students were asked to indicate their ethnic group membership twice, once in an open-ended format at the beginning of the survey, and once in a checklist at the end that included the option “mixed”. Only 20% of biracial college students and 34% of biracial high school students spontaneously identiﬁed themselves as being of mixed heritage in the initial open-ended format. The remainder of the multiracial students could be identiﬁed as biracial only in a ﬁnal question asking speciﬁcally about the race or ethnicity of each parent. When given the option of indicating a mixed heritage later in the survey, 64% of the multiracial high school students reported being “mixed”. Nevertheless, 36% still continued to use a single group label. In addition, self-labels are not stable, and diﬀerent self-labels may be used in diﬀerent situations. The latter issue is illustrated in a survey of students of mixed white and Hispanic background carried out on two diﬀerent college campuses; one predominantly white and one ethnically diverse (Phinney & Alipuria, 1996). Results showed that the ethnic label used by these multiracial students varied across the two settings, with students more often calling themselves white on the predominantly white campus, and Latino on the ethnically diverse campus. Thus, in studies of diverse populations, there are likely to be many biracial individuals who are not identiﬁed, and the numbers will vary depending on the method of assessment. Researchers cannot assume that individuals who only check one box are in fact monoracial or monoethnic. Latinos in the United States present a particularly interesting case of multiple heritage people, cutting across racial, ethnic, and cultural categories (Ferdman & Gallegos, 2001). In Mexico, and Central and South America, the term mestizo is used by large numbers of people to reﬂect their mixed European and indigenous backgrounds. In the United States, people considered as Latino or Hispanic can be of any racial background or mix thereof (white, black, Native American, Asian) and from a range of cultural and ethnic backgrounds. The term “Latino” contains within it multiple categories of group membership. As noted above, the American census gives oﬃcial recognition that this ethnic category straddles all racial categories. Multiple racial identity in African Americans has a unique history, as noted earlier. The many children born with African American and Native American or white American parentage were historically treated as black (Root, 1992a), and this practice in most cases still requires multiracial blacks to be considered and treated simply as black (Rockquemore & Brunsma, 2002). Many black Americans who have a mixed racial heritage may not be aware of their non-black ancestry, and others may know of it but do not acknowledge it (Daniel, 2002). In the black community, a wide range of skin

226

Multiple identities

tones is accepted, and blacks are generally accepting of individuals of mixed background. African Americans of mixed black and white heritage may be more likely to consider themselves as belonging to a single group than mixed individuals from other mixed backgrounds, because of the history of exclusion by whites of people with any African phenotype or known African heritage. In a survey of multiracial high school students (Phinney & Alipuria, 1996), 77% of ethnically or racially mixed African American students labeled themselves as belonging to a single group (black), compared to only 30% of those who were of mixed Latino heritage. Across all combinations that included one white parent, students were far more likely to use a minority group label than a label as white, although this varied by the racial composition of the setting. In sum, the variability in the use of labels highlights the fuzziness of ethnic and racial boundaries and suggests that caution should be used in assuming that a particular label indicates single group membership (Daniel, 2002). Much of the literature on multiracial people has dealt with the question of how these people construct an identity for themselves in a society that seeks to assign them to a single group and does not support their attempts to express their complex heritage (Renn, 2000; Rockquemore & Brunsma, 2002; Root, 1992a, 1996b). The process of achieving a sense of how one relates to one’s mixed heritage is accomplished through a process comparable to that for achieving a monoracial or monoethnic identity, although more complex (Kich, 1992). However, the type of identity that is adopted varies across individuals. The considerable literature on this topic suggests at least four ways in which multiracial people experience and negotiate the borders between groups. First, identiﬁcation with only one of one’s heritage groups, that is, identifying as monoracial, is seen as a viable and common option among multiracial individuals (Renn, 2000; Rockquemore & Brunsma, 2002). Traditionally, at least for biracial black and white Americans, the only option was to consider oneself black, unless one were light-skinned enough to pass for white. Other multiracial people may also select a monoracial or monoethnic identity (Alipuria, 2002; Williams, 1997). Root (1996a) suggests, however, that this monoracial identity may change over the lifetime of the individual. A second option is the possibility of creating or aﬃrming a new category. In a study by Renn (2000), 20 out of 24 multiracial college students interviewed saw themselves as belonging to a new group, variously described as multiracial, mixed, or biracial (see Crisp & Hewstone; Smith, this volume, for some discussion of the psychological processes involved in forming new conjunctive categories). Root (1996a) describes these individuals as sitting on the border. A “border identity” is a “unique hybrid category of self-reference” (Rockquemore & Brunsma, 2002, p. 42). Individuals who have made eﬀorts to introduce a new multiracial option into the census and other demographic forms are likely to be those with this type of identity. When multiracial individuals assert that they represent a new “mixed” group, they may create

Phinney and Alipuria

227

links with other multiple heritage people from a range of backgrounds who see themselves as non-white. However, in enhancing solidarity with other minority groups, they may increase distance from whites (Daniel, 2002). Third, multiracial people may acknowledge membership in both groups and switch between them. Seven out of twenty-four students in Renn’s (2000) study adopted situational deﬁnitions, using two monoracial identities, as they moved between their heritage groups. Rockquemore and Brunsma (2002) observed similarly the ability of individuals to shift their identity according to the particular context or interaction (see McGarty, this volume, for a discussion of how context can inﬂuence selective self-categorization). Finally, studies of multiracial people have identiﬁed some people who choose not to identify along racial lines but rather think of themselves primarily, if not exclusively, in other terms (Renn, 2000; Rockquemore & Brunsma, 2002). This is likely to be possible only in a very diverse or open setting where race is not salient. Renn (2000) found that for some individuals exposure to postmodern theory allowed them to reject society’s construction of race and to deﬁne themselves in terms other than race. In sum, the evidence indicates that that there is a wide range in the kinds of identity that multiracial people may develop. The actual identities chosen depend on many factors, including the availability of various ethnic and racial communities and individuals from these communities, as well as individual preferences and phenotype. The setting is an important and often determining factor in the options available to multiracial individuals. An in-depth study of multiracial students on three diﬀerent university campuses (Renn, 2000) investigated how campus peer culture inﬂuenced the ways in which multiracial students made meaning of their racial identity. Although the decision as to how to identify is a private matter, Renn found that the three campuses had diﬀerent public spaces available for multiracial students to aﬃrm racial identity. Students reported that where they ﬁt in was determined largely by the messages they got from campus peer culture. For example, many multiracial participants found themselves unwelcome among monoracial ethnic minorities. If they chose to join a particular ethnic organization, they were often not accepted in the larger mainstream context. The size of the multiracial population on a campus was an important factor. On the one campus with a large enough group of multiracial students to form a critical mass, incoming multiracial students found an organized group to which to belong. At other campuses, the students explored other options in their eﬀorts to form an identity.

MULTIRACIAL PEOPLE AND INTERGROUP RELATIONS Multiracial individuals contain within themselves two (or more) diﬀering racial or ethnic backgrounds, representing the groups of each parent, but the extent of their interaction with these diﬀering groups varies depending on the

228

Multiple identities

context and their phenotype. If they are raised entirely within one ethnic or racial community, their intergroup attitudes will be shaped in part by their phenotype. An ambiguous appearance that raises questions about their legitimacy in the group is likely to spur eﬀorts to search out the other part of their heritage, whereas such eﬀorts will be minimized if they can “pass” as a group member. On the other hand, if they have contact with the families and communities of both parents, they will naturally be exposed to alternative cultural perspectives. On the basis of social-psychological research on perspective-taking (Galinsky & Moskowitz, 2000), they would be expected to be less biased toward either group. However, as noted earlier, neither of the groups of their parents is in fact an outgroup. A question of interest, then, is whether multiracial people would have more positive attitudes toward true outgroups, that is, the groups of neither parent. There has been little research on this question. However, several studies suggest that biracial individuals may be more open to other ethnic groups than those who are monoracial. Stephan & Stephan (1991) presented evidence suggesting that mixed heritage college students were more tolerant than students from a single heritage. Similarly, Phinney and Alipuria (1996) found that multiracial African Americans and multiethnic Latino males had more positive attitudes toward other ethnic groups in general than did their monoracial/monoethnic peers from the same demographic groups. Beyond their own attitudes, however, multiracial people may have a unique place in society because of the ways in which they challenge the predominant view that people can be classiﬁed into discrete categories. Because of their ambiguous phenotypic appearance, these individuals in most cases cannot easily be assigned to a single category. The human tendency to categorize is then disrupted, and people who meet phenotypically ambiguous individuals often feel uncomfortable. Mixed heritage people are pressured to choose a single designation (Alipuria, 2002; Omi & Winant, 1986). Root (1996a) notes that Americans are “overwhelmed by—or resistant to—the possibility that someone is both Black and White” (p. xxii). Root suggests that this discomfort may stem from the fact that individuals of mixed backgrounds represent both “us” and “them”. In meeting them, one wonders whether they can be seen as similar to one’s own group or considered diﬀerent. Omi and Winant (1986) refer to the crisis in racial meaning that occurs in the observer of phenotypically ambiguous people. Frequently this discomfort is expressed in the “What are you?” question. Multiracial adults who were interviewed in a recent dissertation reported being asked this question frequently, from childhood to the present (Alipuria, 2002). Williams (1997) conducted interviews exploring the experience of multiracial individuals and reported that multiracial people learn from an early age to expect this question, and to develop a variety of techniques for responding to it. The fact that it is asked, Williams suggested, reveals “the racial, social disorientation of the person asking the question” (p. 203). The interrogator seems to be saying, “You defy my limited understanding and

Phinney and Alipuria

229

social application of race. I have no label to ﬁt you, to pigeon-hole you, and therefore to make assumptions about you. I need to know what you are so I can ease this discomfort I feel for being unable to peg you” (p. 203). One multiracial respondent told Williams, “What I represent turns other people’s worlds upside down. Some [people] get really angry at me for not giving them the answer they want to hear, insisting that I belong to this group or that group” (p. 204). Although multiracial people report that the question can be tedious or annoying, they become aware that it gives them an opportunity to inﬂuence the interrogator. In such interactions, multiracial people are not simply passive recipients of the other person’s confusion or curiosity. Rather, they are in a position to create social reality, if only brieﬂy (Alipuria, 2002). A multiracial man reported that, “It took me a while to understand that . . . I have the power to determine—at least for that moment—who I am and how they are going to respond to me” (Williams, 1997, p. 204). Williams concluded, “Biracial individuals often create new meanings of race during social interaction, sometimes forcing those with whom they interact to rethink their assumptions and shift their understanding” (Williams, 1997, p. 208). The existence of multiracial people will increasingly require monoracial individuals to recognize the complexity and variability of ethnic and racial categories. Social psychologists have shown that the more subgroups that individuals perceive within a group, the greater the perceived variability of the group (Richards & Hewstone, 2001; see also Hogg & Hornsey, this volume). As a result of contact with multiracial individuals, people may become aware that ethnic minority group members cannot be neatly categorized into one of ﬁve or six groups. The American census identiﬁed 57 possible combinations of the six primary ethnic and racial groups used in the 2000 census. These combinations illustrate and highlight the variability in the American population and may contribute to a reduction of biases based on rigid categories.

MULTICULTURAL PEOPLE The identity issues of multicultural people are somewhat diﬀerent from those of multiracial people. Immigrants and their children, as well as indigenous peoples, represent the most obvious examples of multicultural people. These people are exposed to two or more diﬀerent cultural frameworks. Multicultural individuals have choices as to the extent to which they aﬃliate with their ethnic culture or the larger society (Berry, 1990). Like multiracial individuals, multicultural people cannot be assigned to ﬁxed categories; their position in society is dynamic and ﬂuid, varying over time and context depending on the categories available to them, the choices they make, and the ways in which others categorize them (Phinney, 2000). Multicultural people may in fact have a cultural identity diﬀerent from that of their parents. For

230

Multiple identities

example, several young adults from mixed American and East Indian backgrounds (Alipuria, 2002) identiﬁed primarily with neither parental culture. One embraced the Spanish culture and language; another, black and Hawaiian culture. Depending on the situation, immigrants to the United States may see themselves and be seen as foreigners, as ethnic minorities, or as members of particular racial groups, among other categories. Alternatively, they may think of themselves as being American, or as being “hyphenated” Americans, as in “Chinese-American.” The predominance of one or another of these possibilities depends on many factors, including generation of immigration, behavioral adaptation (language and customs), phenotype, personal preference, and bicultural competence (LaFromboise et al., 1993). Also important are the attitudes toward immigrants and toward their particular group that are prevalent in the larger society. Changes across generations of immigration in the labels used by and for members of American immigrant groups illustrate their use of dual categorization (Phinney, 2003a). Typically, members of the ﬁrst or immigrant generation think of themselves in terms of their country of origin, Chinese or Mexican. Second and later generations are American by virtue of their birth and generally consider themselves as American to some extent, but in most cases they retain their identity as members of a distinct group through dual labels such as Chinese American or Mexican American. The extent to which later generations retain an ethnic label depends in part on a phenotypic appearance that distinguishes them from Americans of European descent (Phinney, 2003a). Third- and fourth-generation Americans of Asian descent continue to be asked, “Where are you from?” and to be congratulated on how well they speak English (Uba, 1994). These experiences force them to continually reevaluate their position in American society. The various ways in which multicultural identities are played out can be conceptualized with reference to Berry’s (1990) acculturation model. Early perspectives on acculturation assumed a linear and negative relationship between ethnic and mainstream cultures, such that one had to give up one to become part of the other; the only choices were separation or assimilation. However, it is now evident, both conceptually and empirically, that individuals in multicultural societies can and do participate in, and identify with, both cultures (Berry, 1990). Berry suggests four types of acculturation that depend on the extent to which individuals value and are involved in their own culture and that of the larger society. Applied to ethnic or racial identity, the model suggests that individuals may be assimilated (identifying primarily as American), separated (identifying primarily as ethnic), integrated or bicultural (identifying as both American and ethnic), or marginal (identifying with no ethnic or cultural group). A conceptual model linking these acculturation types to ethnic or racial identity development has been proposed by Leong and Chou (1994). They suggest that foreclosure (pre-encounter, according to Cross, 1991) is equivalent

Phinney and Alipuria

231

to assimilation, in that individuals at this stage wish, and perhaps try, to be part of the larger society and may deny or downplay their own ethnicity. During the moratorium phase (immersion, according to Cross, 1991), individuals become deeply involved in exploring and understanding their own culture and may reject the dominant culture; thus they may appear separated. Finally, with ethnic identity achievement (or internalization), individuals accept and value both their own group and the larger society, and so appear bicultural or integrated. Leong and Chou (1994) provide no data to support the model, and there is no evidence that the acculturation strategies follow a sequence or change over time. Therefore, the parallels between the models are hypothetical. Nevertheless, acculturation and identity development models have in common the view that there are various ways of relating to one’s own ethnic group and to other groups and larger society. Both approaches assert the possibility of combining two cultures, although not everyone achieves this possibility. Whether members of American ethnic groups see themselves as ethnic, American, both, or neither is inﬂuenced by a variety of factors. In an interview study of African American and Mexican American adolescents, Phinney and Devich-Navarro (1997) found a wide variation in the extent to which the young people identiﬁed with their ethnic culture and with the wider society. The results showed two types of bicultural identity, blended or alternating, and a third type that was not bicultural but rather reﬂected a separated identity. No examples of assimilation or marginalization were found. Nearly 90% of the adolescents considered themselves to be bicultural, but the meaning of being bicultural had qualitatively diﬀerent interpretations. Adolescents with a blended bicultural identity had a strong sense of being American but also aﬃrmed their ethnicity. In interviews, they expressed strong feelings of inclusion in American culture, based on a view of America as a country characterized by diversity. They held positive attitudes as well about their own ethnic heritage, but typically did not see the two cultures as strongly diﬀerentiated. Belonging to two cultural categories was not problematic for them (Phinney & Devich-Navarro, 1997). In contrast, adolescents described as alternating biculturals saw the two cultures as distinct. They clearly acknowledged being American, but they based their biculturalism more strongly on their ethnic culture. They described their lives as moving between diﬀerent cultures; for example, being “ethnic” at home and “American” at school. They appeared to move comfortably between the two cultures, demonstrating bicultural competence (LaFromboise et al., 1993). Both the blended and alternating biculturals clearly identiﬁed with both their ethnic group and the mainstream society. A third pattern was seen in a group of adolescents who did not see themselves as bicultural (Phinney & Devich-Navarro, 1997). They reported a strong identiﬁcation with their ethnic culture, but expressed little or no identiﬁcation with America, which they considered to mean “white”. They

232

Multiple identities

reported experiences of discrimination and felt excluded from the wider society. This pattern was most common among the African American adolescents, who consistently reported experiencing more discrimination than other groups. For black Americans, it may be particularly diﬃcult to integrate their two identities. Du Bois spoke of the “double consciousness” of African Americans as both black and American (Du Bois, 1989 [1903], p. 5). Blacks, he stated, feel a “two-ness . . . two thoughts, two unreconciled strivings; two warring ideals in one Black body” (p. 6). These types of identity have clear implications for intergroup relations. An American identity has the potential of being a superordinate category that can serve to reduce ethnic and racial tensions among diverse ethnic and racial categories. Both bicultural and blended identities are examples of dual identities, described by Gonzalez and Brown (2003) as “simultaneous high superordinate and high subgroup categorization” (p. 195). In an experimental study, these authors showed that a dual identity condition not only reduced bias but also promoted generalized positive intergroup attitudes. An advantage of the dual identity model is that “the integrity of the original subgroup identities [was] not threatened within the superordinate category” (p. 210). Most minority group members wish to maintain their group identity, rather than losing it within the larger society. For them, the opportunity or encouragement for maintaining their own culture within the larger society can promote the achievement of a secure ethnic identity that in turn contributes to more positive intergroup attitudes. In contrast to bicultural or integrated identities, some adolescents in the Phinney and Devich-Navarro (1997) study felt separated from American society. To the extent that members of minority ethnic groups feel that “America” does not include them, the concept is of little use as an overarching category. A characteristic of the moratorium (or immersion) status for ethnic minorities is often anger and resentment due to unequal treatment of their group. Social-psychological work has shown that resentment regarding one’s situation in the larger context is positively associated with the strength of one’s subgroup categorization and with an inclination for public protest (Kessler & Mummendey, 2001). When members of subordinate groups are not equal participants in the society, attempts to include them in the superordinate category of the country as a whole are unlikely to contribute to better intergroup relations.

CONCLUSIONS AND IMPLICATIONS Multiple categorization as discussed in this chapter refers to people who encompass in themselves two groups within a single identity domain, such as race. This focus contrasts with multiple categorization as used by many social psychologists to refer to groupings based on two diﬀerent identity domains, such as race and gender. Various types of multiple categorization

Phinney and Alipuria

233

were examined. One type includes those who have parents from two diﬀerent ethnic or racial groups, such as black and white. For them, neither group is a true ingroup or outgroup; rather, they represent the overlap of two groups that have typically been seen as distinct and nonoverlapping. Another type includes those who have parents from one culture or ethnic group while growing up in another, usually the dominant culture of their country of residence. Although their ethnic group is an ingroup, the dominant culture may be seen as an ingroup, an outgroup, or a superordinate group. An additional type is represented by multiracial or monoethnic individuals who incorporate a culture other than that of their parents into their identity without necessarily moving to another cultural location. People who represent these types of multiple group categorization face numerous personal issues as they strive to develop a secure identity. They also present a challenge to a society that has historically tried to categorize people into a limited number of groups. Finally, because of their exposure to multiple perspectives from the diﬀering groups they belong to, they can provide models of the possible role of multiple categorization in reducing intergroup bias. Research on identity formation in multiracial and multicultural persons has demonstrated the complexity of the issues involved in multiple categorization. Racial, ethnic, and cultural dimensions continue to be widely treated by psychology and by society as though they exist in nature, and categories within these dimensions are typically seen as obvious and distinct. However, such divisions in society are increasingly understood to be constructed and maintained by individual and social processes and are challenged by a wide range of multiple identities. Increasing numbers and visibility of multiple heritage persons, of non-European immigrants, and of cultural subgroups and communities within society highlight the complex nature of group identities. Because these people do not ﬁt into the clearly distinct categories that society uses, they exemplify the ﬂuid boundaries between and among groups. The labels that people choose or are assigned change over time and setting and tell us little about the content of the group identities that are constructed in relation to race, ethnicity, or culture. Group labels rarely reﬂect accurately multiple heritages. Individuals who appear to be members of a particular group or who, as a result of social pressure, are required to declare a group aﬃliation may in fact have an identity quite diﬀerent from that which is ascribed. Furthermore, this identity is likely to change over time, as individuals consider and evaluate the meanings of relevant group memberships. Therefore researchers should use caution in making assumptions on the basis of group labels. Construction of group-related aspects of one’s identity is sensitive to factors in the immediate environment and in the wider social context (Alipuria, 2002; Renn, 2000; Roosens, 1989). Outcomes of identity processes are inﬂuenced by the “interaction between assignment (what others say individuals are) and assertion (who or what individuals claim to be)” (Daniel, 2002,

234

Multiple identities

p. 113). As particular geographical areas become more diverse and as the numbers of multiracial and multicultural individuals increases, the meanings of multiple category membership will also change. Consideration of these issues raises the question of whether the existence of a growing and increasingly vocal population of multiheritage persons in the United States can break down the distances and tensions among preexisting groups and improve intergroup relations. Multiracial and multicultural people can serve as bridges between groups that have been seen as distinct. The ability of some of these people to move comfortably among two or more cultures illustrates the possibility for this type of multiple categorization to improve intergroup relations. Oﬃcial recognition of multiracial and multicultural people is still being sought in many situations. The right to assert multiple heritage on oﬃcial forms is recent (Daniel, 2002). Availability of large enough groups of people of mixed heritage to aﬃrm varied identities is limited to large communities of such individuals, as in cities and on some university campuses. This is likely to change as multiethnic and multiracial cultures and communities increase and spread. It is not clear what form such changes will take and how they will inﬂuence the largely monoethnic structure of society. Yet the insistence of these individuals on their right to express their dual or multiple identities may eventually inﬂuence others to see that the boundaries and categories that have deﬁned American society are more ﬂuid than has been assumed. Although rigid categorization still exists in many situations, current trends in the United States may lead increasingly to recognition that virtually all cultural groups and all populations are in fact blends of many diverse strands.

ACKNOWLEDGEMENT Preparation of this chapter was supported in part by Grant S06 GM-08101 from the NIH MBRS SCORE Program.

REFERENCES Abrams, D. (1996). Social identity, self as structure and self as process. In W. P. Robinson (Ed.), Social groups and identities: Developing the legacy of Henri Tajfel (pp. 143–167). Woburn, MA: Butterworth-Heinemann. Adams, M. (2001). Core processes of racial identity development. In C. Wijeyesinghe & B. Jackson (Eds.), New perspectives on racial identity development (pp. 209–242). New York: New York University Press. Alipuria, L. (2002). Ethnic, racial, and cultural identity/self: An integrated theory of identity/self in relation to large scale social cleavages. Doctoral dissertation, Claremont Graduate University 2002. UMI Microform 3039092. Ann Arbor, MI: ProQuest.

Phinney and Alipuria

235

Atkinson, D., Morten, G., & Sue, D. (1993). Counseling American minorities (4th ed.). Dubuque, IA: Brown & Benchmark. Bandlamudi, L. (1994). Dialogics of understanding self/culture. Ethos, 22, 460–493. Berry, J. W. (1990). Psychology of acculturation. In J. J. Berman (Ed.), Crosscultural perspectives (pp. 201–234). Lincoln: University of Nebraska Press. Cornell, S., & Hartmann, D. (1998). Ethnicity and race: Making identities in a changing world. Thousand Oaks, CA: Pine Forge. Crisp, R., & Hewstone, M. (2000). Multiple categorization and social identity. In D. Capozza & R. Brown (Eds.), Social identity processes: Trends in theory and research (pp. 149–166). Thousand Oaks, CA: Sage International. Cross, W. (1991). Shades of black: Diversity in African-American identity. Philadelphia: Temple University Press. Cross, W., & Fhagen-Smith, P. (2001). Patterns of African American identity development: A life span perspective. In C. Wijeyesinghe & B. W. Jackson, III (Eds.), New perspectives on racial identity development: A theoretical and practical anthology (pp. 243–270). New York: New York University Press. Daniel, G. R. (2002). More than black? Multiracial identity and the new racial order. Philadelphia: Temple University Press. Du Bois, W. E. B. (1989 [1903]). The souls of black folk. New York: Penguin. Ellemers, N., Spears, R., & Doosje, B. (1999). Introduction. In N. Ellemers, B. Doosje, & R. Spears (Eds.), Social identity: Context, commitment, and content (pp. 1–5). Oxford: Blackwell. Ellemers, N., Spears, R., & Doosje, B. (2002). Self and social identity. Annual Review of Psychology, 53, 161–186. Erikson, E. H. (1968). Identity: Youth and crisis. New York: W. W. Norton & Company, Inc. Ferdman, B., & Gallegos, P. (2001). Racial identity development and Latinos in the United States. In C. Wijeyesinghe & B. W. III. Jackson (Eds.), New perspectives on racial identity development: A theoretical and practical anthology (pp. 32–66). New York: New York University Press. Galinsky, A., & Moskowitz, G. (2000). Perspective-taking: Decreasing stereotype expression, stereotype accessibility, and in-group favoritism. Journal of Personality and Social Psychology, 78, 708–724. Gibbs, J. (1987). Identity and marginality: Issues in the treatment of biracial adolescents. American Journal of Orthopsychiatry, 57, 265–278. Gonzalez, R., & Brown, R. (2003). Generalization of positive attitude as a function of subgroup and superordinate group identiﬁcations in intergroup contact. European Journal of Social Psychology, 33, 195–214. Grieco, E. M., & Cassidy, R. C. (2001). Overview of race and Hispanic origin: 2000. U.S. Census Bureau, Census 2000 Brief, CTKBR/01–1. Washington, DC: U.S. Government Printing Oﬃce. Grove, K. (1991). Identity development in interracial, Asian/white late adolescents: Must it be so problematic? Journal of Youth and Adolescence, 20(6), 617–628. Helms, J. (1990). Black and white racial identity: Theory, research, and practice. New York: Greenwood. Helms, J. E. (1995). An update of Helms’s white and people of color racial identity models. In J. G. Ponterotto, J. M. Casas, L. A. Suzuki, & C. M. Alexander (Eds.), Handbook of multicultural counseling (pp. 181–198). Thousand Oaks, CA: Sage Publications.

236

Multiple identities

Hermans, H., & Kempen, H. (1998). Moving cultures: The perilous problem of cultural dichotomies in a globalizing society. American Psychologist, 53, 1111–1120. Jones, N., & Smith, A. S. (2001). The two or more races population: 2000. U.S. Census Bureau, Census 2000 Brief. Washington, DC: U.S. Government Printing Oﬃce. Kessler, T., & Mummendey, A. (2001). Is there any scapegoat around? Determinants of intergroup conﬂicts at diﬀerent categorization levels. Journal of Personality and Social Psychology, 81, 1090–1102. Kich, G. (1992). The developmental process of asserting a biracial, bicultural identity. In M. Root (Ed.), Racially mixed people in America (pp. 304–320). Newbury Park, CA: Sage Publications. Kroger, J. (2000). Identity development: Adolescence through adulthood. Thousand Oaks, CA: Sage Publications. LaFromboise, T., Coleman, H., & Gerton, J. (1993). Psychological impact of biculturalism: Evidence and theory. Psychological Bulletin, 114, 395–412. Leong, F., & Chou, E. (1994). The role of ethnic identity and acculturation in the vocational behavior of Asian Americans: An integrative review. Journal of Vocational Behavior, 44, 155–172. Marcia, J., Waterman, A., Matteson, D., Archer, S., & Orlofsky, J. (1993). Ego identity: A handbook of psychosocial research. New York: Springer-Verlag. Misra, G., & Gergen, K. (1993). On the place of culture in psychological science. International Journal of Psychology, 28, 225–243. Omi, M., & Winant, H. (1986). Racial formation in the United States from the 1960s to the 1980s. New York: Routledge & Kegan Paul. Oustinovskaya, M., & Phinney, J. (2001). Commitment to multiculturalism among adolescents. Paper presented at the meeting of the Western Psychological Association, Hilo, Hawaii. Park, R. (1928). Human migration and the marginal man. American Journal of Sociology, 33, 881–893. Pettigrew, T. (1986). The intergroup contact hypothesis reconsidered. In M. Hewstone & R. Brown (Eds.), Contact and conﬂict in intergroup encounters (pp. 169–195). Oxford: Blackwell. Phinney, J. (1989). Stages of ethnic identity development in minority group adolescents. Journal of Early Adolescence, 9, 34–49. Phinney, J. (1990). Ethnic identity in adolescents and adults: A review of research. Psychological Bulletin, 108, 499–514. Phinney, J. (1993). A three-stage model of ethnic identity development. In M. Bernal & G. Knight (Eds.), Ethnic identity: Formation and transmission among Hispanics and other minorities (pp. 61–79). Albany, NY: State University of New York Press. Phinney, J. (1996). When we talk about American ethnic groups, what do we mean? American Psychologist, 51, 918–927. Phinney, J. (2000). Ethnic identity. In A. Kazdin (Ed.), Encyclopedia of psychology. Washington, DC: American Psychological Association. Phinney, J. (2003a). Ethnic identity and acculturation. In K. Chun, P. Organista, & G. Marin (Eds.), Acculturation: Advances in theory, measurement, and applied research (pp. 63–81). Washington, DC: American Psychological Association. Phinney, J. (2003b). Ethnic identity and attitudes toward other ethnic groups. Manuscript in preparation.

Phinney and Alipuria

237

Phinney, J., & Alipuria, L. (1996). At the interface of culture: Multiethnic/multiracial high school and college students. Journal of Social Psychology, 136, 139–158. Phinney, J., & Devich-Navarro, M. (1997). Variations in bicultural identiﬁcation among African American and Mexican American adolescents. Journal of Research on Adolescence, 7, 3–32. Phinney, J., Dupont, S., Espinosa, C., Revill, J., & Sanders, K. (1994). Ethnic identity and American identiﬁcation among ethnic minority youths. In A. Bouvy, F. van de Vijver, P. Boski, & P. Schmitz (Eds.), Journeys into cross-cultural psychology (pp. 167–183). Amsterdam: Swets & Zeitlinger. Phinney, J., Ferguson, D., & Tate, J. (1997). Intergroup attitudes among ethnic minority adolescents: A causal model. Child Development, 68, 955–969. Phinney, J., & Kohatsu, E. (1997). Ethnic and racial identity development and mental health. In J. Schulenberg, J. Maggs, & K. Hurrelman (Eds.), Health risks and developmental transitions in adolescence (pp. 420–443). New York: Cambridge University Press. Quintana, S., Ybarra, V., Gonzalez-Doupe, P., & De Baessa, Y. (2000). Cross-cultural evaluation of ethnic perspective-taking ability: An exploratory investigation with U.S. Latino and Guatemalan Ladino children. Cultural Diversity and Ethnic Minority Psychology, 6, 334–351. Renn, K. (2000). Patterns of situational identity among biracial and multiracial college students. Review of Higher Education, 23, 399–420. Richards, Z., & Hewstone, M. (2001). Subtyping and subgrouping: Processes for the prevention and promotion of stereotype change. Personality and Social Psychology Review, 5, 52–73. Rockquemore, K., & Brunsma, D. (2002). Beyond black: Biracial identity in America. Thousand Oaks, CA: Sage Publications. Roosens, E. (1989). Creating ethnicity: The process of ethnogenesis. Thousand Oaks, CA: Sage Publications. Root, M. (1992a). Within, between, and beyond race. In M. Root (Ed.), Racially mixed people in America (pp. 3–11). Newbury Park, CA: Sage Publications. Root, M. (Ed.) (1992b). Racially mixed people in America. Newbury Park, CA: Sage Publications. Root, M. (1996a). The multiracial experience: Racial borders as a signiﬁcant frontier in race relations. In M. Root (Ed.), The multiracial experience: Racial borders as the new frontier (pp. xiii–xxviii). Thousand Oaks, CA: Sage Publications. Root, M. (Ed.) (1996b). The multiracial experience: Racial borders as the new frontier. Thousand Oaks, CA: Sage Publications. Spickard, P. (1992). The illogic of American racial categories. In M. P. P. Root (Ed.), Racially mixed people in America (pp. 12–23). London: Sage Publications. Stephan, W., & Stephan, C. (1991). Intermarriage: Eﬀects on personality, adjustment, and intergroup relations in two samples of students. Journal of Marriage and the Family, 53, 241–250. Stonequist, E. (1961). The marginal man: A study in personality and culture conﬂict. New York: Russell & Russell. Tajfel, H., & Turner, J. (1986). The social identity theory of intergroup behavior. In S. Worchel & W. Austin (Eds.), Psychology of intergroup relations (pp. 7–24). Chicago: Nelson-Hall. Uba, L. (1994). Asian Americans: Personality patterns, identity, and mental health. New York: The Guilford Press.

238

Multiple identities

Wijeyesinghe, C., & Jackson, B. (Eds.) (2001). New perspectives on racial identity development: A theoretical and practical anthology. New York: New York University Press. Williams, P. (1997). Seeing a color-blind future: The paradox of race. New York: The Noonday Press. Williams-Leon, T., & Nakashima, C. (Eds.) (2001). The sum of our parts: Mixedheritage Asian Americans. Philadelphia: Temple University Press. Zack, N. (1993). Race and mixed race. Philadelphia: Temple University Press. Zuckerman, M. (1990). Some dubious premises in research and theory on racial diﬀerences. American Psychologist, 45, 1297–1303.

11 Political institutions and multiple social identities Neal A. Carter

Communal conﬂicts threaten political stability throughout the world. While conﬂicts such as those between Hutu and Tutsi or between Bosnian Serbs, Croats, and Muslims degenerated to genocide, others such as those between the Anglophone and Francophone Canadians or between Walloons and Flemings in Belgium have remained remarkably peaceful despite chronic tension. This diﬀerence in the level of hostility across cases of identity-based conﬂict presents a rich ﬁeld of investigation for political psychology. In addition, increased mobility and intergroup marriage has generated further political challenges to the notion of self-determination. It is increasingly difﬁcult to categorize people and to adapt politically to changing demands, as indicated by debate in several countries over identity categories in censuses. People are increasingly conscious of the fact that they belong to multiple politically salient groups that sometimes have conﬂicting goals. In light of these diﬃculties, I present a theoretic basis for studying this interaction. Political studies of communal conﬂict and psychological research on identity operate at diﬀerent levels of analysis, but can be combined to produce an explanation of the dynamic nature of identity politics. After a brief review of research in these areas, I present a model to adapt psychological theories of multiple identiﬁcation to the complex realm of political debate and legitimation. Although there is not space here for a full application of the combination, I use examples from Belgium and Canada to illustrate this perspective. This political psychological perspective can greatly contribute to our understanding of the dynamics of identity-based conﬂict and eﬀorts to intervene in such conﬂicts.

ETHNOTERRITORIAL POLITICS Despite many predictions that modernization would reduce ethnic tensions (Huntington, 1971), ethnic conﬂict has increased since the 1950s (Huntington, 1993; Lijphart, 1977; Zariski, 1989). Communal conﬂicts challenge the concept of the nation state (Gottlieb, 1993; Jacobson, 1997; Joireman, 2003). In France, ethnic groups such as the Bretons, the Basques, and the Catalans

240

Political institutions

demanded cultural linguistic accommodation, challenging the assimilationist policies linking French citizenship with membership in the French nation (Safran, 1989). Italy has addressed conﬂict between German and Italian speakers in the Alto Adige region (Eyck, 1991; Katzenstein, 1977). Spain granted regional autonomy based on regional and ethnonationalist criteria (Clark, 1989). Sri Lanka has been torn by a long-standing dispute between Sinhalese and Tamils (Horowitz, 1985; Pfaﬀenberger, 1991; Singer, 1991). Belgium drastically changed its constitution to recognize its three oﬃcial linguistic communities (Brassinne, 1994; Delmartino, 1986; Senelle, 1989). Canada struggles to address Quebec nationalism and alienation in its western provinces (Clark, 1994; Doran, 2001; McRoberts, 1997). The problems associated with ethnicity-based conﬂict are widespread indeed. The study of ethnoterritorialism focuses on political movements “that are derived from a group of people [with a distinct culture] . . . having some identiﬁable geographic base within the boundaries of an existing political system . . . [who] perceive themselves as being distinct from the broader population of the overall political system” (Rudolph & Thompson, 1989, p. 2). This subject is a gold mine for those interested in studying the eﬀects of multiple social identities. Many ethnoterritorial groups are geographically concentrated and share some combination of racial, religious, and linguistic traits as well as economic and political interests. Members of such groups tend to form communication and interaction networks within the group and to share clusters of values as a result of socialization. Enloe (1973) argued that modernization has increased the signiﬁcance of interaction networks at all levels, increasing the ability and desire of political leaders to make ethnoterritorial demands. As interaction and intermingling among various ethnic groups intensify, conﬂict over values and “ways of life” is likely to increase. Such conﬂict is inﬂuenced by the degree and importance of the “cultural distinctiveness” of the ethnic groups. This intensiﬁed interaction also increases the probability that individuals belong to multiple groups; as boundaries become more porous, they also tend to become more politically contested. Ethnoterritorial conﬂict is not necessarily destructive. Ethnoterritorialism can foster the development of a group, encouraging members’ self-esteem and personal growth by improving the image of the group to which the people belong (Leslie, 1989). Ethnonationalism should be approached with an awareness of the beneﬁts as well as the problems of viewing politics through an identity lens; it may increase the diﬃculty of uniﬁed action but also provide an important sense of distinctiveness to group members. As we learn more about the implications of multiple social identities, we may be able to apply those insights to the study of ethnoterritorial conﬂict.

Carter

241

SOURCES OF ETHNOTERRITORIAL CONFLICT Modern economic and social conditions have increased the permeability of borders, making it more diﬃcult to identify even small regions with ethnic or cultural homogeneity. Even regional groups that demand recognition due to ethnoterritorial conditions often experience internal conﬂicts. For example, several ethnic groups in Quebec, including Anglophones and Aboriginals, oppose the nationalist demands for Quebec sovereignty. The then Quebec Premier Parizeau’s comments on the night of the failed 1995 sovereignty referendum, blaming “money and the ethnic vote” (interpreted as “Jews and immigrants”) for the loss (Wilson-Smith, 1995), reﬂect the tensions involved in ethnoterritorial conﬂicts. This section examines political and sociological insights into the nature of ethnoterritorial conﬂict, in order to establish some of the parameters in which political psychologists can frame this issue. Especially important here is how changes in context aﬀect the political salience of identities. Lijphart (1977) posited that the resurgence of ethnic conﬂict in the 1960s and 1970s resulted from several phenomena. The rapidity of modernization in the post-World War II era produced a sense of loss of traditional values and a renewed interest in identity politics and maintaining culture. The “horizontalization of vertical ethnic groups” (Lijphart, 1977) created caste-like inequality among ethnic groups that had previously enjoyed internal class diversity, thereby increasing the material salience of ethnic divisions and reducing potential for multiple categorization. Increased state intervention in ethnic relations, especially in economic matters, was often perceived as reverse discrimination and increased majority group ethnic consciousness. The reduction of class and religious conﬂict permitted ethnic cleavages to become more salient. Expanded democratic participation increased the beneﬁts elites could gain by emphasizing ethnicity to mobilize constituents. Growth of “postbourgeois values”, including “the needs for belonging and for intellectual and aesthetic self-fulﬁllment” (Lijphart, 1977), limited the ability to appeal to all citizens simply as individuals and enhanced the value of ethnic identiﬁcation. The principle of self-determination gave political legitimacy to ethnic demands for autonomy. Further, the demonstration eﬀect of ethnic demands created a self-reinforcement loop: fulﬁllment of each demand encourages that and other groups to make more demands. With the end of the Cold War, the constraint of superpower contention has also been removed. As these causes reinforce the political utility of identity claims, politicians are increasingly likely to compete in their claims to speak for diﬀerent groups, thereby increasing the political salience of identity claims and the importance of multiple identities. With several potentially politically salient groupings, multiple identities enter into the political arena. Potential sources of ethnoterritorial conﬂict include social, economic, and political inequality; modes of institutional assimilation; cross-cultural contact; the perceived need to struggle for cultural survival of a group; and the

242

Political institutions

process of conﬂict resolution itself (Breton, Reitz, & Valentine, 1980). Some combination of these factors inﬂuences each ethnic group’s grievances. Each group has distinct characteristics, goals, and propensities for conﬂict. Any comparisons across groups can provide some means of ﬁnding “objective” diﬀerences that could be invoked in a conﬂictual setting (Connor, 1994). Since few regions have homogeneous populations, conﬂict is likely to arise over which group will control territory or political interactions. The intensity and depth of the conﬂict can change over time, due largely to the interaction between institutional arrangements of conﬂict management and the extent to which ethnoterritorial identities are salient in various political, social, cultural, and economic arenas. Various factors can be used to establish boundaries between groups, with the combination of factors providing both ﬂexibility of speciﬁc demands and stability of the divisions. Attempts to resolve conﬂicts must approach problems from a holistic perspective that considers how the interaction among material and psychological factors shapes conﬂict (Byrne & Carter, 1996; Carter & Byrne, 2000). Failure to understand this interaction could provoke simplistic suggestions for how to resolve the conﬂict, overreaction to developments whose eﬀects will be mitigated by the other factors, and an underestimation of the time required for the transformation of conﬂict (Byrne & Carter, 2002). An understanding of the eﬀects of multiple identities is increasingly important in attempts to intervene. Several researchers have investigated the importance of reinterpretations of politics via communal identiﬁcation and the use of identities in constituent mobilization. Horowitz (1985; 1991a; 1991b; 1994) used strategic interaction theory and social identity theory to examine how institutional factors such as the type of electoral system aﬀect instrumental incentives for using ethnic comparisons. Smith (1983; 1989) emphasized how leaders reinterpret the meaning and history of groups at various stages of their development. SearleWhite (2001) argued that a better understanding of the psychological implications of identity and human needs would help those trying to reduce ethnic conﬂict anywhere from the classroom to Sri Lanka. He reports on how the extended simulations he runs in the classroom mirror the psychological explanations of intergroup conﬂict. He emphasizes social categorization and such psychological processes as projection, group biases, and escalation in both the real world and in his own classroom experiment based on assigned groupings. Laitin (1998) argued that political context plays a signiﬁcant part in the choices that people make about which groups to identify with, focusing on ethnic identiﬁcation choices of Russian speakers in the former Soviet Union. Smith opposed both what he referred to as “gastronomic theory” and “geological theory” of ethnicity. The gastronomic perspective portrays the nation as “a piece of social engineering” (Smith, 1995) constructed by political leaders who simply combine a number of traits to create a nationality. This portrayal of the process, also known as “instrumentalism”, is too

Carter

243

simplistic because it fails to consider the constraints imposed by historical development and previously constituted populations. The “geological” theory, at the other end of the spectrum, considers the stages by which communities develop. Smith criticized this theory, also called “primordialism”, for assuming that the past dictates future constructions of group identity. Conversi (1995) provided a similar critique of the main theories, and argued that an essential factor of ethnoterritorial conﬂict is the creation and maintenance of psychological boundaries between groups, thus requiring a dynamic, psychologically sensitive treatment. Smith argued for a synthesis of the two perspectives, which he dubbed an “archeological” perspective that views nationalists as engaged in a “rediscovery and reinterpretation of the ethnic past and through it the regeneration of their national community” (Smith, 1995). The archeological approach recognizes the creativity of humans, who continually reconstruct their understanding of the world, within the historical and social constraints that bound the set of acceptable interpretations. Scholars should thus focus on the ways in which identity is used in conﬂicts. Connor (1994) noted that ethnonationalism is a psychological phenomenon. Like Smith (1995), Connor emphasized that nations are self-deﬁned, often on the basis of several factors that may change over time. As a mass phenomenon, nationalism depends on the notion of kinship or common heritage. According to Connor, this psychological bond is more important than speciﬁc material factors that often serve as a rallying point for conﬂict or shorthand for what diﬀerentiates one group from another. The diversity of groups’ self-deﬁnitions frustrates analytic attempts to deﬁne nationalism in a precise and consistent manner. Connor’s arguments emphasize both how political leaders discuss group identities and how conceptions of the nation interact with the political context. History helps deﬁne groups, but economic, demographic, and social challenges, as well as nationalistic discourse, also shape the interpretation of what it means to be a part of a particular nation or ethnic group. Demographic factors aﬀect group strategies. Regional minorities often establish their own interregional political and social organizations and may advocate universal recognition of minority rights. Regional ethnic majorities, on the other hand, may attempt to pursue both regional and ethnic goals, and favor decentralization. An ethnic majority of one region may favor a political agenda opposed by members of that ethnic community in another region (especially in matters of control of linguistic and educational policies). For example, the political goals of Quebec Francophones often oppose those of French-speaking Canadians in other provinces. Attempts to understand the multiplicity of goals and potential divisions are aided by a deeper understanding of the social psychology of the eﬀects of multiple identities. In like manner, social psychologists often note their studies should be able to address real groups as well as experimentally created “minimal” groups. Understanding the political context of intergroup interaction can thus beneﬁt

244

Political institutions

psychologists working to reduce bias. With this in mind, I turn now to review some psychological ﬁndings related to identity conﬂicts.

PSYCHOLOGICAL THEORIES OF IDENTITY There appears to be a wide array of possible causes and courses of development of conﬂict (Kriesberg, 2002). Much of the development depends on what leaders make of the situations and how they portray the context in their attempts to mobilize constituents. This section therefore examines the link between identity politics and elite attempts to aﬀect the identiﬁcation of their constituents. Social psychology provides useful insights for the analysis of group identity in politics. Group identiﬁcation aﬀects behavior and encourages the acceptance of group norms. Group identiﬁcation also provides a cognitive ﬁlter, creating selective perception of information and skewed evaluations of arguments. Several conditions encourage the formation of a psychologically signiﬁcant, and therefore potent, group (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). Similarity among individuals within a category must be salient since members of latent groups are unlikely to expend the energy to form a corporate entity unless something in the environment emphasizes their similarities. Members of the group must share the belief that they are subject to a common fate (that is, they will share rewards or penalties due to their commonality with members of the group). This belief is often supported by emphasis on group histories and traditions. The creation of a corporate entity is more likely when group members see themselves as interchangeable instances of the group category (Brewer & Schneider, 1990; Turner et al., 1987; see also McGarty, this volume). These characteristics of psychologically signiﬁcant groups help explain important facets of communal conﬂicts. The nature of groups changes in response to conﬂict. Political leaders focus on traits that distinguish their constituents from others, which reinforces the likelihood that constituents will see those traits as important. As politicians use traits to justify their decisions, group members come to share more closely a common fate. As groups become more polarized, group members identify more strongly with group norms and stereotypes, viewing members of groups as sharing essential traits (and therefore being interchangeable instances of the category). “Social identity refers to social categorizations of self and others, selfcategories that deﬁne the individual in terms of his or her shared similarities with members of certain social categories in contrast to other social categories” (Turner, Oakes, Haslam, & McGarty, 1994, p. 454). Social identity is a ﬂuid, context-dependent process rather than a ﬁxed set of criteria for categorizing people. Thus, social identity is both a process of ongoing construction and an outcome in the form of a group label. Much research has involved

Carter

245

laboratory experiments that indicate that even minimal (artiﬁcial) groups produce ingroup biases and can inﬂuence choices concerning the allocation of resources. Members often seek the greatest relative advantage for their group, even at the expense of greater absolute gains for more cooperative behavior (for a review, see Brown, 2000). Typically, individuals identify with a number of diﬀerent groups, and the salience of those groups depends on the context. Hofman (1988) clariﬁed several aspects of identiﬁcation. Each individual has a number of subidentities that relate to various groups with which she can identify. The salience, or visibility, of the subidentity inﬂuences the likelihood that the individual will respond to the situation in a manner consistent with membership in that group. According to “Tajfel’s law”, “as category memberships become salient, there will be a tendency to exaggerate the diﬀerences on criterial dimensions between individuals falling into distinct categories, and to minimize these diﬀerences within each of these categories” (Turner, 1982, p. 28). Prolonged salience may upgrade the psychological prominence of that identity and increase its centrality. Centrality involves a subidentity’s relative importance compared to other subidentities and the interconnectedness of subidentities. Further, the subidentity has a level of attractiveness, known as valence, for the individual. A highly positive valence increases the likelihood that an individual will identify with that group. Valence, salience, and centrality combine to determine potency, a subidentity’s ability to inﬂuence behavior (Hofman, 1988). Hofman’s use of subidentities is similar to current research on the eﬀects of multiple social identities (e.g., Brewer, 2001; Crisp & Hewstone, 2001; Gaertner & Dovidio, 2000; Gaertner, Dovidio, Anastasio, Bachman, & Rust, 1993; see Crisp & Hewstone; McGarty; Hogg & Hornsey; Dovidio et al., this volume). Rarely is only one identity salient at a given time. We therefore seek to understand what conditions will lead to our preferred identity being most potent. Leaders attempt to manipulate valence, salience, and centrality as they argue for particular actions. Much research has focused on how diﬀerent levels of abstraction or group size aﬀect the psychological potency of groups. Allport (1979) presented identities as concentric circles, with each circle representing a diﬀerent level of inclusiveness, as depicted in Figure 11.1. Allport considered smaller groups to be more potent. As groups become larger, the intensity of identiﬁcation, or “ingroup potency”, declines (as indicated by the thickness of the lines in Figure 11.1). The model rests on the assumption that potency relies on the intensity of interaction with group members and the ability of group members to see themselves as equally representative members of the group. As the group becomes larger, this ability decreases. Smaller groups could therefore be more easily called to collective action than larger ones. Brewer (1991) noted several problems with this notion of a simple correspondence between group size and ingroup potency. She argued that people judge themselves in terms of similarity and diﬀerence; they have a

246

Political institutions

Figure 11.1 Basic model of social identities. From Allport (1979, p. 43). Reproduced with permission from Basic Books, a member of Perseus Books, LLC.

psychological need to view the groups to which they belong as clearly distinct from other groups. These distinctions often play a role in the development of a positive self-image. While intergroup diﬀerences help create a feeling of being special, similarity within a group helps establish necessary bonds among its members, building a psychologically necessary feeling of belonging. Brewer’s optimal distinctiveness theory (see Figure 11.2) captures the importance of balancing the needs for similarity and diﬀerence. The relative strength of the needs for assimilation (belonging) and diﬀerentiation (distinction) depends on a number of factors, including individual predilections and the size of the salient groups. According to the theory, as group size increases, each group member’s need for assimilation is reduced, but her need for diﬀerentiation increases. Highly diﬀerentiated groups may be too small to provide a sustainable, meaningful group. Highly inclusive groups may address

Figure 11.2 Brewer’s model of optimal distinctiveness. From Brewer (1991, p. 477). Reproduced with permission from Sage Publications.

Carter

247

inadequately members’ need for diﬀerentiation, and therefore fail to encourage members to contribute to that group, since it commands limited commitment. Political elites can exploit the tension between uniformity and diﬀerence to mobilize followers to pursue political action. By playing on the needs for either distinctiveness or belongingness, leaders can support or oppose an institutional arrangement. Emphasizing the need for diﬀerentiation moves the diﬀerentiation line up the dissatisfaction axis, thereby reducing the size of the optimal group. Similarly, emphasizing the need for assimilation would raise that line, thereby increasing optimal group size. This strategy could be employed to support or oppose any particular institutional arrangement that will aﬀect the function or structure of groups. Combining optimal distinctiveness with structural changes occurring in conﬂict helps articulate more fully the importance of group identiﬁcation processes in conﬂict regulation. Brewer’s (1991) notion of optimal distinctiveness indicates that stable political agreements must attain a balance between the people’s needs for assimilation and diﬀerentiation. People seek arrangements that allow them to belong to, and identify with, groups that are meaningfully distinct from other groups. If the need for assimilation is not met, there will be too little common ground for collective action, and individual diﬀerences will obstruct cooperation. If the need for diﬀerentiation is not adequately addressed, the salience of the group is not high enough to warrant attention. People worry little about maintaining characteristics that all share. Politicians can most successfully mobilize support if they portray a desired outcome as supporting groups at the optimal level of distinctiveness.

CONTEXTUAL CONSIDERATIONS Ethnoterritorial movements generally function like other political groups (Banton, 1994) but rely more on social identities to mobilize their constituents. Elites often attempt to emphasize ethnoterritorial characteristics to provide constituents with a frame of reference and set of norms from which to judge the political issues at hand (Turner et al., 1987). This focus may serve as a cognitive shortcut for followers, providing cues about how to interpret various arguments. Emphasis on group identities and relations among groups provides political leaders with the ability to distribute solidary incentives for the mobilization of followers (Burton, 1985; Duﬀy & Frensley, 1991). Political research indicates that leaders use identity in their pursuit of political goals. In order to better understand leaders’ use of identity in their political interventions, it is useful to view their actions in the framework of strategic interaction. Models of strategic action used to explain the constraints that leaders face provide valuable insights into the dynamic nature of political action. They

248

Political institutions

present succinctly the interaction of context, goals, actions, and outcomes in complex situations. The actions of political leaders confronting constituents with multiple identities are best understood within a nested games and conﬂict processes framework that is sensitive to the multiple constraints they confront. Political elites involve themselves simultaneously in at least two sets of interactions, which may be characterized as games. In the ﬁrst set, elites seek support from their constituents by attempting to represent and enhance constituents’ interests while also helping to shape those interests. In the second set, elites vie among themselves for power. Because the relations are “nested”, a leader’s movement in one game aﬀects her standing in the other set of relations. Leaders negotiating intergroup settlements ﬁnd themselves constrained by elite constituent power structures (Panebianco, 1988; Putnam, 1988; Tsebelis, 1990). In other words, leaders need their decisions to “play at home”. As leaders vie for constituents, they may ﬁnd it useful to emphasize diﬀerent group identities. Constituents who belong to more than one category, as almost everyone does, may be inﬂuenced to give precedence to one category over others. This nested framework of decision-making may be a limiting factor for strategies to promote or reduce intergroup conﬂict, and thus should be of interest to researchers proposing strategies such as decategorization, recategorization, or mutual diﬀerentiation (discussed below). As conﬂict develops, elites alter their strategies to adapt to changes in the combination of these nested relationships (for Belgian examples, see Carter, 2003; Mughan, 1983). The notion of nested games provides a useful metaphor for the complex constraints within which leaders act. Leaders are aware of various contextual factors and usually have developed reasons for their choice of tactics and the way in which they use identity to appeal to constituents. Figure 11.3 depicts two basic games that are often combined in political negotiations and invoked to explain politics in identity-driven contention. In consociational politics, elites (represented in the ﬁgure by numerals) interact with their constituents (X with identifying subscript) and with other leaders. As discussed below, consociationalism involves elite-level accommodation among separated and isolated segments, or pillars, of society (Lijphart, 1969). The coalition game involves more complex patterns of elite interaction, since it entails the need for competing elites (represented by the numerals) to interact and resolve their diﬀerences within a coalition (represented by numerals within the same ellipse) in order to confront another coalition. This requires consideration of interaction both within and among coalitions. Their actions in this setting could be linked to mutual diﬀerentiation or the common ingroup strategy (discussed below). Combining the two models provides a more accurate portrayal of the relationships that aﬀect politicians’ decisions, as indicated in the “combined” diagram. This shows three nested games: elite constituents relations; intracoalition elite relations; and intercoalition relations. Because they are nested, each game is aﬀected by

Carter

249

Figure 11.3 Models of nested games. Adapted from Tsebelis (1990, p. 59), this ﬁgure represents the consociationalism and coalition models. We produced the combined model. Reprinted with permission from the University of California Press.

movement in every other relationship. There are thus several diﬀerent angles from which to appeal to constituents. Leaders also compete for constituents and face factional challenges within their parties, but this ﬁgure ignores such interactions for the sake of simplicity. As the interactions can be viewed as an iterative game, leaders will usually pursue strategies that strengthen their overall, long-term position (Elster, 1979). Strategies that produce success at an early stage in the evolution of a party may need to be changed to face new challenges to cohesion produced by that success (Rudolph, 1989). The size and cohesion of various movements are central to elite calculations, as they aﬀect the relative power of various leaders to impose their will on, or to reach acceptable accommodation with, others. Appearing to be challenged by internal factionalization may actually improve a leader’s horizontal bargaining position; leaders can state that internal challenges to their position reduce their room for maneuver and compromise while increasing their need to produce a settlement favorable to their constituents. However, they also need to convince other leaders that they can reasonably deliver the acceptance of the settlement by their constituents if the negotiation is to be successful (Putnam, 1988). Leaders need to balance their need to stress ingroup solidarity (often by emphasizing diﬀerences between groups), their need to deal with other leaders (often indicating some common ground), and their need to have agreements accepted once reached. Burton (1962, 1985) advocated an approach to conﬂict management in which teams from the parties (including oppositions

250

Political institutions

and hard-liners) meet in facilitated problem-solving workshops to break down intergroup hostilities. This approach makes use of the contact hypothesis, frequently debated among social psychologists (Brewer, 1996; Gaertner & Dovidio, 2000; Hewstone, 1996; Kelman, 1997). Political elites should be aware of the inﬂuence of identity on the conﬂict, and vice versa, and adjust their strategies accordingly. Models of sustained conﬂict should incorporate likely changes in the structure of such conﬂicts. The idea of structural change, based on the writings of Burton (1962), Coleman (1957), and Schumpeter (1955), includes the notion that conﬂictual interaction produces three types of structural change: (1) changes in the psychological states of participants due to “residues” of the interaction; (2) changes in the structure or function of groups due to mobilization; and (3) alterations of the complex social relations of the parties that further aﬀect the nature of the larger community (for a review of structural change, see Rubin, Pruitt, & Kim, 1994). During escalation, psychological states are marked by increasingly hostile and competitive goals, negative attitudes, and perceptions of the other group and intergroup interaction. This process can eventually develop into deindividuation, in which participants come to see themselves and others strictly as members of their group rather than as individuals. It can also become dehumanization, in which outgroup members are not seen as human but rather as vermin, diseases, or other undesirable objects (Kriesberg, 2002). Escalation can produce a greater need for intragroup cohesiveness, thereby increasing the eﬀect of group norms that can lead to polarization and enable more militant factions to gain inﬂuence, which in turn aﬀects the structure and nature of groups. In addition, conﬂicts often escalate to include more participants, as each side tries to bolster its position by mobilizing allies. In addition to the structural changes listed by Rubin et al. (1994), parties can also act to change the institutional framework in which their relations may be conducted. Those involved in the conﬂict are aware that constitutional changes will have multiple and lasting eﬀects on their power. Their bargaining sessions and negotiations are likely to produce residues that aﬀect future negotiations. Leaders consider the politics of constitutional reform as an iterated, learning process. Group leaders may pursue several objectives simultaneously. With any action, leaders may be addressing intergroup and intragroup conﬂict, attempting to gain new constituents, bolstering their positions within their groups and/or attacking their rivals. These models of strategic interaction can be used in a variety of settings. Nevertheless, ethnoterritorial conﬂicts should be viewed as qualitatively diﬀerent from other political issues. Ethnoterritorial identities provide an interpretive framework that helps explain a variety of policies. The emphasis on factors of identity as opposed to simply diﬀerences in preferences and interests sparks much more emotion within the political debate and increases the diﬃculty involved in de-escalating communal conﬂict.

Carter

251

POLITICAL REGULATION OF ETHNOTERRITORIAL CONFLICT States confronted with ethnoterritorial conﬂict often struggle to adopt political institutions that most of the involved groups consider legitimate. McGarry and O’Leary (1993) developed a taxonomy of approaches to the regulation of conﬂict, dividing policies into “methods for eliminating diﬀerences” (genocide, forced mass population transfers, partition and/or secession, and integration and/or assimilation) and “methods for managing diﬀerences” (hegemonic control, arbitration, cantonization and/or federalization, consociationalism or power-sharing). As they indicate, groups within a country may advocate diﬀerent policies, and the government may direct diﬀerent policies toward various groups. This typology is a general guide, not a strict ordering of policy options. The list of strategies is neither mutually exclusive nor collectively exhaustive. Diﬀerent groups may favor opposing goals as to the management or elimination of conﬂict, or may disagree on the precise strategy to be employed. Conﬂict strategies often pose a question of degree, such that leaders committed to institutional conﬂict can attempt to alter the institutions to change the balance between accommodation and elimination of conﬂict. A major diﬀerence between the nationalists and federalists in Quebec, for example, may be that nationalists desire diﬀerence elimination through secession, while federalists advocate management of diﬀerences through a combination of federalism and consociationalism. In Belgium, politicians have strengthened federalism and weakened consociationalism. While this could be viewed as a shift in strategy of conﬂict management, the shift from a unitary to a federal system could also be seen as an attempt to eliminate diﬀerence by shifting the locus of control and reducing the need for intercommunal cooperation (Carter, 2003). This typology, when viewed from a multiparty perspective, indicates which tactics groups are most likely to favor. If the involved parties favor diﬀerence accommodation, they are likely to create and use the institutional path of conﬂict. However, if either party favors the elimination of diﬀerence, conﬂict is likely to escalate to “hot conﬂict”. Diﬀerence accommodation is close to liberal (e.g., Rawls, 1971, 1993) or a more communitarian policies based on the politics of recognition (Taylor, 1991, 1993, 1994) and more successfully employs argumentation based on discourse ethics to regulate conﬂict (Habermas, 1990). Taking a slightly diﬀerent perspective, Horowitz (1985) listed four possible processes of ethnic fusion and ﬁssion. Amalgamation and incorporation are processes of assimilation. In amalgamation, groups combine to form a new group with a new identity, as attempted in Tito’s Yugoslavia. Incorporation occurs when members of one group assume the identity of another group, without the creation of a new group identity, as indicated by the French notion of nationality. Diﬀerentiation can occur by either division or

252

Political institutions

proliferation. Division indicates that one group divides into separate groups, abandoning the previous identity, as occurred in the breakup of Yugoslavia. Proliferation occurs when a faction forms a separate group, without the destruction of the original group identity, as, for example, the creation of pieds noirs (Algerian colonists) as opposed to the French. These four processes are to be seen as ideal types. As Horowitz cautioned, actual groups are more complex, and various levels of identiﬁcation can exist simultaneously. Polling evidence shows that members in a federal system can identify strongly with their province and with the overarching state (Vandekeere, Doutrelepont, & Jacquenain, 1994). These possibilities should have a familiar ring to psychologists interested in multiple identities. Ethnoterritorial conﬂicts can therefore end in a number of political outcomes, from an amicable form of mutual diﬀerentiation to the creation of separate countries, to genocide. Of particular interest to those interested in the peaceful regulation of conﬂict, federalism and consociationalism oﬀer frameworks of mutual diﬀerentiation within a single state, thus using a subgroup strategy (see Gaertner & Dovidio, 2000).

Federalism Federalism uses the subgroup strategy, in which various subgroups are territorially delineated within the overarching state. Federalism often has “been oﬀered as an institutional solution to the disruptive tendencies of intra-societal ethnic pluralism” (Long, 1991). Readily available examples include Switzerland, Belgium, and Canada. Ethnic groups frequently develop cohesive societies, and territorial division may roughly coincide with ethnic dispersion. In cases like Belgium and Canada, the existence of ethnic and regional diversity is a compelling factor in the creation of federal arrangements, and ﬁgures prominently in constitutional debates. According to Burgess, federalism is a belief that territorial devolution of some political responsibilities is a valuable method for creating a government responsive to diverse regional interests. A federation is “a speciﬁc organizational form which includes structures, institutions and techniques” (Burgess, 1986, p. 17). More simply, federation “combines self-rule with shared rule” (Canada, 1991, p. v). Federal systems generally have ﬁve principal elements, with some degree of variation. First, federations have written constitutions specifying and guaranteeing division of jurisdiction for the diﬀerent levels of government. Second, a bicameral legislature allows one chamber to represent the people in relation to population, while the second chamber is typically based on representation of constituent units, such as provinces or states. Third, overrepresentation of the smaller constituent units in the second chamber of the legislature is a common concession made to counteract the importance of population in the ﬁrst chamber. Fourth, the constituent units are involved in any changes in the constitution of the federation while they have the ability to

Carter

253

unilaterally change their own constitutions, which are subordinate to that of the federation. Fifth, constituent units have a large share of autonomy and responsibility (Elazar, 1987). In addition, a judicial system often settles disputes among the various governments. Elazar (1987) argued that federal principles develop from the idea that free people can freely create political associations to achieve common goals and protect certain rights while also protecting their own identities. It is also possible to form a covenant after conquest, which is the case for Canada. Furthermore, Belgian politicians federalized their unitary state to reduce the need for agreement and cooperation, contrary to Elazar’s characterization of federalism as a means of increasing association. Federations may result from the desire to accommodate or to eliminate diﬀerence, in McGarry and O’Leary’s (1993) terms, depending on the salient issue; federal conﬁgurations allow diﬀerent levels of government to address various issues. Further complicating this issue is that political divisions and boundaries tend to aﬀect identiﬁcation patterns among those who live within those boundaries, thereby shaping future identity-based politics (Carter, 2003). Federalism may be an eﬀective solution for addressing regional diﬀerences as it allows some issue areas to be under the jurisdiction of constituent units, but ethnic diﬀerences are rarely as easy to accommodate or to delineate territorially. It creates a system in which some combined activity is necessary, but subgroups can still conduct many of their own aﬀairs independently. Strict territorial division along ethnic lines is rarely the best policy, especially in cases where one ethnic group is signiﬁcantly larger than others. In such cases, it may be better to have several constituent units of roughly equal size, population, or wealth in order to break a tendency for collective ethnic political behavior and replace it with regional interests (Varty, 1991). Any proposed solution may not ﬁt the problem after a relatively short period, due to shifts in demographic factors. The changing linguistic composition of Brussels exempliﬁes the diﬃculties involved in attempts to implement ethnoterritorial policies (DeRidder & Fraga, 1986; Senelle, 1989; Zolberg, 1977). Given the multiplicity of political issues and the incentives to compete for constituents, political leaders will often use multiple sources of identiﬁcation in their mobilization attempts, and thus undermine previous arrangements. Furthermore, it “is not obvious that any particular allocation of responsibility among levels of government will ensure more tolerance for cultural diversity than any other” (Minow, 1990, p. 96). Indeed, federations may actually suppress ethnic diversity (Long, 1991). Some ethnic groups, especially those that are geographically dispersed, may ﬁnd their situation hindered by an accord of other groups, or by a constitution based on territoriality. Generalizations about federations are useful, but speciﬁc studies indicate that the basis for federalism is not universal. Because federalism is “a prescriptive guide to action” (Burgess, 1986, p. 14), diﬀerences in environment and cultural mindsets result in diﬀerent demands on the political system. In general: “Notions such as personalism, subsidiarity, corporatism, pluralism

254

Political institutions

. . . are as important in [the European] tradition as . . . territoriality which has typically dominated Anglo-American approaches to the study of federalism and federation” (Burgess, 1986, p. 26). It is therefore useful to consider the allocation of powers among levels of government along a continuum ranging from unitary to confederal arrangements. The precise allocation of powers and responsibilities may change over time within a country, reﬂecting moves for increased autonomy or increased harmonization. A shift from one general structure to another, such as occurred with Belgium’s switch from a unitary structure with some regional autonomy to a complex federal system with two distinct types of constituent units, may be less radical than expected, due to the ﬂexibility of power distribution within each category. Additionally, great structural change can occur within one institutional framework, as can be seen by the increased power of Canadian provinces vis-à-vis the federal government. Federalism can thus be viewed as a subgroup strategy to be used when identities are hierarchical, and the territorial delineation is most salient. Other identities may also require political accommodation, sometimes informally. In addition to the oﬃcial distribution of powers, the interaction of elites across groups signiﬁcantly aﬀects conﬂicts. Consociationalism, addressed in the next section, complements studies of power distribution by examining arrangements of power-sharing.

Consociationalism Consociationalism is another version of the subgroup strategy, in which the subgroups are culturally or ideologically distinct but still identify with the overarching state. It involves power-sharing by leaders of various groups to democratically manage their diﬀerences. The theory of consociational democracy provides a partial explanation of political practices in the countries for which it is often invoked, such as Switzerland (Church, 1989; Steiner, 1991), Belgium (Heisler, 1991; Murphy, 1988; Zolberg, 1977), and Canada (LaSelva, 1996; Smiley, 1977). Lijphart (1991) noted that the model is not expected to predict behavior or accurately describe all politics within those countries, but is rather intended as a normative model for conﬂict regulation. Consociational practices help ensure the continuation of government, especially in countries with parliamentary systems containing several parties. In consociational democracies, elites cooperate to counteract the “immobilizing and unstabilizing eﬀects of cultural fragmentation” (Lijphart, 1969, p. 212; emphasis omitted). The essence of consociationalism is not found in particular institutions, but rather in the eﬀort by the elites to pursue common goals. Consociational democracy requires that: (1) . . . elites have the ability to accommodate the divergent interests and demands of the subcultures. (2) This requires that they have the ability to

Carter

255

transcend cleavages and to join in a common eﬀort with the elites of rival subcultures. (3) This in turn depends on their commitment to the maintenance of the system and to the improvement of its cohesion and stability. (4) Finally, all of the above requirements are based on the assumption that the elites understand the perils of fragmentation. (Lijphart, 1969, p. 216) Consociational arrangement often results in a grand coalition of signiﬁcant groups, a mutual veto, a system of proportionality, and segmental autonomy (Halpern, 1986). Because elites across cultural groups need to share power, leaders tend to reject a simple majoritarian system in favor of coalitions. Coalitions provide some minority voice in the creation and implementation of policies, thereby contributing to the legitimacy of the government. A mutual veto provides additional protection for the segments, as leaders can prevent the implementation of policies considered unacceptable by their constituents, even if the majority of the population favors the measure. This provides protection from the tyranny of the majority. Segmental autonomy enables each group to control major aspects of political life, often including patronage systems. These eﬀects provide a great deal of protection for each group, thereby increasing their sense of security, which in turn increases their ability to cooperate. Without these guarantees, the quest for security could produce a defensive conﬂict spiral, restricting the potential for constructive cooperation. Three dominant aspects of consociationalism are the coincidence of politically salient cleavages, accommodation among leaders of subcultures as the major political process, and the avoidance of integration between the subgroups (Smiley, 1977). The coincidence of politically salient cleavages develops over time, as the same people ﬁnd themselves in the same groups on a number of diﬀerent issues. This division tends to segment the population as the identities have reinforced salience and therefore increase in centrality. These segments become the “pillars” of the consociational system. They are treated separately, with their leaders establishing policies in an intergroup environment while implementing these policies within their own segments (see Figure 11.3). Leaders thus engage in accommodation (diﬀerence management) in the horizontal game and partition (diﬀerence elimination) in the vertical game. The theory of consociational democracy has been criticized for “empirical inconsistencies [that] emanate . . . from the construction of the model” (Halpern, 1986, pp. 182–183) and for obscuring the development of more precise models that more accurately ﬁt a given situation (Church, 1989). Modernization and democratization have reduced the conditions favorable to consociationalism. Increased political participation of the general public combined with the inﬂuence of multiple identities greatly limits the authority previously attributed to leaders. Consociational democracy is compatible with federal and unitary governments. Federalism distributes power geographically in order to accommodate

256

Political institutions

regional needs. Consociationalism addresses problems of political, cultural, economic, ethnic, and ideological diﬀerences. Whether the form of government is federal or unitary, it is possible to establish a “consociational democracy [that] violates the principle of majority rule, but . . . does not deviate very much from the normative democratic theory” (Lijphart, 1969, p. 214). Consociationalism and federalism are thus diﬀerent methods for treating subgroups diﬀerently. Federalism, based on territory, tends to be more rigidly structured. Consociationalism, based on ideology, language, or ethnicity, tends to be less formal. An understanding of institutional and structural incentives is crucial to developing explanations of intergroup conﬂict. Keeping these incentives in mind helps researchers maintain sensitivity to the complexity of political interaction, and therefore a better understanding of the strategies being employed. Further, such sensitivity facilitates interpretation of political acts and political speech in light of the use of identity in political conﬂicts.

EMPHASIZING IDENTITY TO STUDY ETHNOTERRITORIAL CONFLICT In many respects, the McGarry and O’Leary (1993) and the Horowitz (1985) typologies of strategies for diﬀerence elimination and diﬀerence accommodation (discussed above) echo debate among psychologists over the best strategies to reduce bias, although they are operating at diﬀerent levels of analysis. Three basic strategies are often oﬀered as potential solutions to intergroup conﬂict (for a review see Brown, 2000). Decategorization discourages the emphasis on group identity, and emphasizes individual traits (see Miller et al., this volume). For example, several organizations try to ﬁnd youth activities such as summer camps or basketball leagues that will attract youth from all sides of a conﬂict. Some of the conditions of participating are that all group identity markers be removed and activities be integrated. Mutual diﬀerentiation advocates that interaction maintain group salience while encouraging cooperation (Hewstone & Brown, 1986; see Crisp; Hogg & Hornsey, this volume). In his work to improve Palestinian–Israeli relations, Kelman (1997, 1999) brings together small groups in problem-solving workshops based on mutual diﬀerentiation. Recategorization, in contrast, strives to create a new, overarching identity (see Crisp; Dovidio et al., this volume). For example, some place great hope in the European Union as a new identity that will help reduce tensions between Protestants and Catholics in Northern Ireland (Grove & Carter, 1999). The eﬀects of crossing categorizations oﬀer a great ﬁeld of investigation. If people are generally more biased against outgroup members, or at least more biased toward ingroup members (Gaertner & Dovidio, 2000), then there should be some merit in attempting to create groups of mixed membership. While some have attempted to repeatedly switch categorizations in the laboratory (Brewer, Ho, Lee, & Miller, 1987;

Carter

257

Deschamps & Doise, 1978), others have noted that the eﬀectiveness of this approach depends on the salience and dominance of the identities (Crisp & Hewstone, 1999, 2001; Migdal, Hewstone, & Mullen, 1998). The common ingroup identity model clariﬁes the likely causes and eﬀects of seeing others as all one group, two subgroups, two separate groups, or separate individuals (Gaertner & Dovidio, 2000; Gaertner et al., 1993). This model examines the interaction of cognitive and aﬀective implications of identity. There is no one right strategy for eliminating bias, but the diﬀerent strategies have some predictable characteristics. The single group strategy tends toward demands for homogenization and emphasis on group cohesion, and is thus a diﬀerence elimination strategy of amalgamation or incorporation much like the classic French approach to nationalism. Viewing people as individuals rather than members of groups is closer to the liberal approach of politics. In its way, it too focuses on diﬀerence elimination by ignoring group diﬀerences and insisting on the equal importance of all individuals. Ironically, members of the majority often favor this approach, since ignoring group diﬀerences can enhance the majority’s position without requiring blatant domination (Azzi, 1992). The subgroups’ strategy is a method of diﬀerence accommodation, as it marks an attempt to recognize both unity and diﬀerence simultaneously. The two main political mechanisms to attain this subgroup strategy are federalism and consociationalism. Horowitz’s (1985) and Brewer’s (1991) notions of assimilation and differentiation can be modiﬁed to examine identity politics. Brewer examined the psychological needs that may be targeted; Horowitz focused on outcomes that mirror these needs. In constitutional debates, political leaders enter claims based both on the outcomes they seek and the psychological needs of their constituents. Various aspects of political issues may enhance some bases of identiﬁcation at the expense of others. The salience of various bases of identiﬁcation can shift the cleavages within a system. Thus, debate will concern not only the degree of similarity and diﬀerence, but also the salience of issues that will provide the boundaries of comparison. Federal systems in particular pose an interesting case for this notion. Advocates of a federal system often use phrases like “unity in diversity” while advocates of a unitary system make claims such as “union makes power”. Those who oppose a federal arrangement can do so on the grounds either that it does not allow enough diversity (prevents suﬃcient diﬀerentiation) or that it allows too much diversity (prevents suﬃcient assimilation). Similarly, federal systems must balance various bases of identiﬁcation to shift some issues toward assimilation and others toward diﬀerentiation. In long-standing ethnoterritorial conﬂicts, stories, history, and tradition promote ingroup identity (Agnew, 1989). Fundamental features of the history of relationships among communities help us comprehend the meaning people attach to politics and group membership. Green (1993) used the image of “umbilical stories”, derived from the observation that paintings of Adam and Eve usually show their navels although they would not have needed

258

Political institutions

umbilical cords, to describe how people are socialized to feel connection with members of their social groups through myths of common descent and familial ties. This notion links large but important groups to the more psychologically potent category of family, thereby increasing their potency. Stern (1995) argued that leaders induce followers to sacriﬁce for a cause precisely by following this strategy. Tajfel (1982) and Turner et al. (1987) showed through psychological experiments that individuals identify with several groups simultaneously, but exhibit behavior typical of the group that is salient in a particular setting. Others have shown that identities often have a hierarchy of importance (Crisp, Hewstone, & Cairns, 2001; Hewstone, 1996; Migdal et al., 1998). Countervailing forces may weaken the inﬂuence of groups, but we expect members of cohesive, salient groups to exhibit similar patterns of behavior. Francophones and Anglophones in Quebec, for example, exhibit diﬀerent patterns of identiﬁcation with groups; Francophones identify strongly with groups having a tight cluster of values and common membership. Anglophones, a more heterogeneous group, exhibit a wider range of group membership, and greater personal distance from other group members (Taylor & Dubé, 1986). The diﬀerences between Francophones and Anglophones are not merely a product of language use. Rather, factors such as education, politics, religion, language, and friendship networks combine to shape perceptions of the relations between self and groups. These perceptions aﬀect individuals’ contributions to collective action. Diﬀerences in identiﬁcation have crucial implications in the establishment and legitimation of political agreements. If members of diﬀerent communities exhibit contrasting patterns of identiﬁcation, we may hypothesize that their understandings of society and ideals also diﬀer, and that they therefore are likely to advocate contrasting political institutions. Azzi (1992) provided experimental evidence that members of diﬀerent ethnic groups use diﬀerent criteria when advocating principles of justice by which to regulate communal conﬂicts. He noted the importance of group identiﬁcation in the preference of some normative principles (e.g., distribution based on merit) over others (e.g., distribution based on equality). Making speciﬁc identities salient during the evaluation of an institutional arrangement inﬂuences that evaluation. This process is especially crucial when analyzing debates over alterations of the constitutional framework that could have vast repercussions on the fate of those identifying with the various groups. Members of multiple groups may have diﬃculty balancing considerations for each group. For example, Aboriginal women’s commitment to women’s equality and Aboriginal self-government produced a conﬂict throughout the Canadian debates over the Charlottetown Agreement. Thus, instead of relying only on the concentric circle model of social identities, it may be better to emphasize the various bases of identiﬁcation with social groups. As depicted in Figure 11.4, several bases for social identiﬁcation overlap to varying

Carter

259

Figure 11.4 Model of competing bases of identiﬁcation. Note: Categories are only examples of several possible salient characteristics. Areas of overlap indicate compatibility of interests among those bases of identiﬁcation; areas without overlap indicate potential conﬂict.

degrees. In this portrayal, areas of overlap indicate compatibility of the bases of identiﬁcation, while areas that do not overlap indicate potential conﬂicts between the bases. This diﬀers from the concentric circle notion in that the groupings are not simply various levels of abstraction or group size. Instead, this model depicts qualitative diﬀerences in the characteristics made salient, rather than only the quantitative size of the optimal group. For example, Allport’s model represented in Figure 11.1 includes groups that are increasingly assimilated along a territorial dimension from a family (house) to mankind (Earth). It would also be possible to make a similar model based on language, with generational slang and regional dialects (e.g., Joual in Montreal) forming more narrow groupings of a particular language (French), and language families (such as Romance or Indo-European) forming the more highly assimilated levels. Other categorizations might be based on ideology, religion, or any number of possible delineations. Competing appeals made to diﬀerent bases of social identiﬁcation may alter the size of the optimal group and the relevant level of normative claim. For example, distribution of material goods tends to evoke diﬀerent bases of identiﬁcation than do questions of symbolic recognition of cultural practices, and therefore the former is likely to invoke appeals to individual-level principles of justice while the latter require group-level consideration (Azzi, 1992). Competition for constituents involves appeals aimed at representing political issues in a way that increases the salience of speciﬁc identities. Politicians seeking to mobilize constituents compete with others to deﬁne groups and shape collective identities (as can be presented in a nested games

260

Political institutions

framework discussed earlier). Here, the important question is how elites discuss groups: do they argue about the salient features of groups (i.e., the bases of identiﬁcation) and do they argue about the relative similarity or distinctiveness that should be sought along that dimension (i.e., level of assimilation)? Statements about these issues may be used to pinpoint disagreements based on a combination of the conceptions depicted in Figures 11.1, 11.2, and 11.4. Figure 11.5 combines level of assimilation (elite’s statement of level providing optimal distinctiveness) and bases of identiﬁcation (characteristics or issues emphasized by the elites). This conceptualization attempts to capture both the notion of levels of inclusiveness and the notion of bases of identiﬁcation. Each set of concentric circles indicates various levels of inclusiveness along one basis of identiﬁcation, such as religion. Smaller circles indicate lower levels of assimilation. The circles overlap to show possible combinations of bases to delineate optimal groups. Although the model appears to indicate otherwise, due to the limitations of two-dimensional representations, groups can be delimited with various bases of identiﬁcation at diﬀerent levels of assimilation. For example, a group could be delineated by combining highly diﬀerentiated physical attributes, moderately diﬀerentiated

Figure 11.5 Combining levels and bases of identiﬁcation. Note: Bases of identiﬁcation are only examples. The larger circles represent more inclusive groups. Characterizations of groups can move across bases (and combine them) and/or move to a diﬀerent level of inclusiveness.

Carter

261

linguistic patterns, and highly assimilated territorial criteria to produce a group—such as Albino Francophones of the World. When examining ethnoterritorial conﬂict, it is important to keep several factors in mind. A key aspect of political debate is disagreement over the potential eﬀects of changes. Changing the constitutional order may produce a shift of power from one group to another, or may alter the salience of various cleavages. In addition, political leaders may have incentives to inveigle, deceive, or obfuscate when discussing the legitimacy of various institutional arrangements and proposals. It is thus important to examine critically the speeches and writings of political leaders in light of their strategic interests, with the goal of better understanding the meaning and political importance of debates over the proper constitutional order. For example, Robert Bourassa (Premier of Quebec 1970–1976 and 1985–1993) shifted from portraying the Meech Lake Accord (the 1987 agreement to amend the constitution) as a major victory for Quebec (Bourassa, 1987) to representing it as a compromise in which Quebec received only its minimal demands (Bourassa, 1990). This apparent contradiction is more readily understood in the context of the arguments being deployed against the ratiﬁcation of the Accord. In the early stages of debate, he faced challenges that Quebec had not received enough from the rest of Canada to provide suﬃcient protection. Later, he was faced with challenges from outside the province to compromise in order to meet the new demands of other parties. In examining identity-based politics, several issues come to the foreground. Among those salient for political psychology, the portrayal of both ingroups and outgroups are prominent. To make use of the psychological insights, it would be important to examine how political elites discuss both ingroups and outgroups in their attempts to gain political support. First, what bases of delineation are emphasized? In other words, what are the fundamental cleavages portrayed by the leaders? Second, how inclusive or exclusive are the groups? This question will aﬀect the optimal distinctiveness of the desired grouping. Third, how are groups portrayed? What aﬀective tags, or epithets, are used to increase either negative or positive valence of the groups under discussion? This should aﬀect the potency of those identities, and thus aﬀect the willingness of constituents to do what the elites desire. Are the characteristics of the groups presented as innate or condition? This will aﬀect the ability of reaching institutional arrangements or other accommodations. What goals and motivations are imputed to the groups? This will aﬀect strategies and levels of trust. Are there potential areas where the groups could come together, perhaps seeing themselves as diﬀerentiated subgroups rather than distinct groups? Leaders will also argue about the likely outcomes of various options, and these provide great insight into their beliefs about the conﬂicts and potential solutions. It is important to remember that some leaders see it in their interest to continue a conﬂict, as it may give them other rewards. Answers to these questions can then be used to develop strategies for conﬂict regulation. If there are possibilities to shift to mutually held identities,

262

Political institutions

this may provide de-escalation. If members of the diﬀerent groups are being bombarded with dehumanized depictions of the enemy, perhaps some contact would reduce the ability of elites to make these claims. The possible combinations of salient identities that could be politically mobilized are legion. Political entrepreneurs may try new combinations to see what will gain constituents.

CONCLUSIONS This review shows that it is useful to consider the development of ethnoterritorial conﬂict in light of both strategic action and social identity theory. This perspective provides a framework of analysis. Combining an understanding of strategic interaction, political psychology, and the contextual setting, this framework is intended to help us map the viewpoints of diﬀerent parties in conﬂict. It is important to develop the constraints and strategic considerations the elites face in their attempts to pursue both horizontal and vertical interactions. Politics requires group identity, and thus there is much to be gained by incorporating such concepts as salience, centrality, and potency in the analysis. Political scientists and psychologists share an interest in understanding how individual and group interactions aﬀect behavior. The eﬀects need to be understood within particular settings. Psychologists can manipulate laboratory settings. Political scientists must try to explain the context suﬃciently to explain the behavior. This perspective could be used as an analytic framework to study previous events or potential interventions, or even as an early warning mechanism. In the ﬁrst case, it could be used to examine the dynamics of conﬂict, analyzing the structural changes occurring throughout various phases of a conﬂict. In the second case, one could examine contemporary discourse to investigate the various portrayals of groups and look for possible areas for mutual cooperation or even the development of a common ingroup identity. Research into the eﬀects of multiple identities could also play a large role in this area, as practitioners and academics might be able to intervene to help establish some cross-categorizations that could reduce intergroup hostility. In the third application, researchers could monitor political discourse, tracking use of group identities and the characteristics attributed to them. There may be more room to intervene to prevent escalation of conﬂict if we are aware of early attempts to mobilize constituents based for signiﬁcant confrontation. While this chapter has focused largely on ethnoterritorial politics and institutional mechanisms to deal with large-scale intergroup conﬂict, many of the same techniques operate at other levels and for other groups as well. Leaders choose their comparison groups. Many of the psychological mechanisms that inﬂuence political action are similar to those that inﬂuence other social actions. Limiting the scope of the analysis, however, facilitates an explanation of contextual factors essential to an understanding of the behaviors.

Carter

263

Many in the debate over how best to regulate conﬂict have concluded that the best method often depends on the current situation. Recategorization, decategorization, mutual diﬀerentiation, or other mechanisms may be appropriate in diﬀerent conditions. Building sophisticated models of the context may help us better match conﬂict regulation strategies to various conditions. Much work remains to be done to bring politics and psychology together. While laboratory-based group experiments present strong evidence for the importance of groups, political events present a vast ﬁeld of exploration. While methods and emphasis diﬀer, political scientists and psychologists could unite as political psychologists (or psychological political scientists). Some might prefer a common group, others diﬀerentiated subgroups, and still others separate groups working toward the same goal. In any event, much can be learned by combining insights.

ACKNOWLEDGEMENTS Many insights for this chapter were gained during the 2000 Summer Institute in Contentious Politics, held at Stanford University’s Center for the Advanced Study of the Behavioral Sciences and the 1994 Summer Institute in Political Psychology held at the Ohio State University. I would especially like to thank Marilynn Brewer, Richard J. Crisp, Gavan Duﬀy, Andrea Grove, Peg Hermann, Miles Hewstone, Doug McAdam, Deborah Martin, Byron Miller, Sydney Tarrow, and Charles Tilly for their inﬂuence on my thinking about these issues.

REFERENCES Agnew, J. (1989). Beyond reason: Spatial and temporal sources of ethnic conﬂicts. In L. Kriesberg, T. A. Northrup, & S. J. Thorson (Eds.), Intractable conﬂicts and their transformation. Syracuse, NY: Syracuse University Press. Allport, G. W. (1979). The nature of prejudice (25th anniversary ed.). Reading, MA: Addison-Wesley. Azzi, A. E. (1992). Procedural justice and the allocation of power in intergroup relations: Studies in the United States and South Africa. Personality and Social Psychology Bulletin, 18(6), 736–747. Banton, M. (1994). Modeling ethnic and national relations. Ethnic and Racial Studies, 17, 1–19. Bourassa, R. (1987). The case for the accord. Canadian Parliamentary Review (Autumn), 20–21. Bourassa, R. (1990, 12 March). Bourassa turns up the heat (Interview). Maclean’s, 22. Brassinne, J. (1994). La Belgique fédérale. Bruxelles: Éditions du Centre de Recherche et d’Information Socio-Politiques. Breton, R., Reitz, J. G., & Valentine, V. F. (1980). Cultural boundaries and the cohesion of Canada. Montreal: L’Institut de recherches politiques.

264

Political institutions

Brewer, M. B. (1991). The social self: On being the same and diﬀerent at the same time. Personality and Social Psychology Bulletin, 17, 475–482. Brewer, M. B. (1996). When contact is not enough: Social identity and intergroup cooperation. International Journal of Intercultural Relations, 20(3/4), 291–303. Brewer, M. B. (2001). The many faces of social identity: Implications for political psychology. Political Psychology, 22(1), 115–125. Brewer, M. B., Ho, H., Lee, J., & Miller, N. (1987). Social identity and social distance among Hong Kong schoolchildren. Personality and Social Psychology Bulletin, 13, 156–165. Brewer, M. B., & Schneider, S. K. (1990). Social identity and social dilemmas: A double-edged sword. In D. Abrams & M. A. Hogg (Eds.), Social identity theory. New York: Harvester Wheatsheaf. Brown, R. (2000). Group processes (2nd ed.). Oxford: Blackwell. Burgess, M. (Ed.) (1986). Federalism and federation in western Europe. London: Croom Helm. Burton, J. W. (1962). Peace theory. New York: Knopf. Burton, J. W. (1985). About winning. International Interactions, 12(1), 71–91. Byrne, S., & Carter, N. (1996). Social cubism: Six social forces of ethnoterritorial politics in Northern Ireland and Quebec. Peace and Conﬂict Studies, 3(2), 52–71. Canada (1991). Shaping Canada’s future together: Proposals. Ottawa: Minister of Supply and Services Canada. Carter, N. (2003). Political identity, territory, and institutional change in Belgium. Mobilization, 8(2), 205–220. Carter, N., & Byrne, S. (2000). The dynamics of social cubism: A view from Northern Ireland and Québec. In S. Byrne & C. Irvin (Eds.), Reconcilable diﬀerences: Turning points in ethnopolitical conﬂict (pp. 41–62). West Hartford, CT: Kumarian Press. Church, C. H. (1989). Behind the consociational screen: Politics in contemporary Switzerland. West European Politics, 12(April), 35–54. Clark, J., Rt. Hon. (1994). A nation too good to lose: Renewing the purpose of Canada. Toronto: Key Porter Books. Clark, R. P. (1989). Spanish democracy and regional autonomy: The autonomous community system and self-government for the ethnic homelands. In R. J. Thompson & J. R. Rudolph, Jr. (Eds.), Ethnoterritorial politics, policy and the Western World. Boulder, CO: L. Reinner Publishers. Coleman, J. S. (1957). Community conﬂict. New York: Free Press. Connor, W. (1994). Ethnonationalism: The quest for understanding. Princeton, NJ: Princeton University Press. Conversi, D. (1995). Reassessing current theories of nationalism: Nationalism as boundary maintenance and creation. Nationalism and Ethnic Politics, 1(1), 73–85. Crisp, R. J., & Hewstone, M. (1999). Diﬀerential evaluation of crossed category groups: Patterns, processes, and reducing intergroup bias. Group Processes and Intergroup Relations, 2(4), 307–333. Crisp, R. J., & Hewstone, M. (2001). Multiple categorization and implicit intergroup bias: Diﬀerential category dominance and the positive-negative asymmetry eﬀect. European Journal of Social Psychology, 31(1), 45–62. Crisp, R. J., Hewstone, M., & Cairns, E. (2001). Multiple identities in Northern Ireland: Hierarchical ordering in the representation of group membership. British Journal of Social Psychology, 40, 501–514.

Carter

265

Delmartino, F. (1986). Belgium: A regional state or a federal state in the making? In M. Burgess (Ed.), Federalism and federation in western Europe. London: Croom Helm. DeRidder, M., & Fraga, L. R. (1986). The Brussels issue in Belgian politics. West European Politics, 9, 376–392. Deschamps, J.-C., & Doise, W. (1978). Crossed category memberships in intergroup relations. In H. Tajfel (Ed.), Diﬀerentiation between social groups (pp. 141–158). Cambridge: Cambridge University Press. Doran, C. F. (2001). Why Canadian unity matters and why Americans care: Democratic pluralism at risk. Toronto: University of Toronto Press. Duﬀy, G., & Frensley, N. J. (1991). Community conﬂict processes: Mobilization and demobilization in Northern Ireland. In J. W. Lamare (Ed.), International crisis and domestic politics: Major political conﬂicts in the 1980s (pp. 99–135). New York: Praeger. Elazar, D. J. (1987). Exploring federalism. Tuscaloosa, AL: University of Alabama Press. Elster, J. (1979). Ulysses and the sirens: Studies in rationality and irrationality. Cambridge: Cambridge University Press. Enloe, C. H. (1973). Ethnic conﬂict and political development. Boston: Little, Brown and Company. Eyck, F. G. (1991). South Tyrol and multiethnic relationships. In J. V. Montville (Ed.), Conﬂict and peacemaking in multiethnic societies (pp. 219–238). New York: Lexington Books. Gaertner, S. L., & Dovidio, J. F. (2000). Reducing intergroup bias: The common ingroup identity model. Philadelphia: Psychology Press. Gaertner, S. L., Dovidio, J. F., Anastasio, P. A., Bachman, B. A., & Rust, M. C. (1993). The common ingroup identity model: Recategorization and the reduction of intergroup bias. European Review of Social Psychology, 4, 1–26. Gottlieb, G. (1993). Nation against state: A new approach to ethnic conﬂicts and the decline of sovereignty. New York: Council on Foreign Relations Press. Green, T. F. (1994). Public speech: The DeGarmo Lecture for 1993. Education Policy Analysis Archives, 2(5). Retrieved May 26, 2000, from http://epaa.asu.edu/epaa/ v2n5.html. Grove, A., & Carter, N. (1999). Not all blarney is cast in stone: International cultural conﬂict in Northern Ireland. Political Psychology, 20(4), 725–765. Habermas, J. (1990). Moral consciousness and communicative action (C. L. and S. W. Nicholsen, Trans.). Cambridge, MA: The MIT Press. Halpern, S. (1986). The disorderly universe of consociational democracy. Western European Politics, 9(April), 181–197. Heisler, M. O. (1991). Hyphenating Belgium: Changing state and regime to cope with cultural division. In J. V. Montville (Ed.), Conﬂict and peacemaking in multiethnic societies (pp. 179–195). New York: Lexington Books. Hewstone, M. (1996). Contact and categorization: Social psychological interventions to change intergroup relations. In C. N. Macrae, C. Stangor, & M. Hewstone (Eds.), Stereotypes and stereotyping (pp. 323–368). New York: The Guilford Press. Hewstone, M., & Brown, R. J. (1986). Contact is not enough: An intergroup perspective on the “contact hypothesis”. In M. Hewstone & R. J. Brown (Eds.), Contact and conﬂict in intergroup encounters (pp. 1–44). Oxford: Blackwell Publishing.

266

Political institutions

Hofman, J. E. (1988). Social identity and intergroup conﬂict: An Israeli view. In W. Stroebe, A. W. Kruglanski, D. Bar-Tal, & M. Hewstone (Eds.), The social psychology of intergroup conﬂict: Theory, research and applications (pp. 89–99). Berlin: Springer-Verlag. Horowitz, D. L. (1985). Ethnic groups in conﬂict. Berkeley and Los Angeles: University of California Press. Horowitz, D. L. (1991a). Ethnic conﬂict management for policymakers. In J. V. Montville (Ed.), Conﬂict and peacemaking in multiethnic societies (pp. 115–130). New York: Lexington Books. Horowitz, D. L. (1991b). Making moderation pay: The comparative politics of ethnic conﬂict management. In J. V. Montville (Ed.), Conﬂict and peacemaking in multiethnic societies (pp. 451–475). New York: Lexington Books. Horowitz, D. L. (1994). Ethnic and nationalist conﬂict. In M. Klare & D. Thomas (Eds.), World security (pp. 175–187). New York: St. Martin’s Press. Huntington, S. P. (1971). The change to change: Modernization, development and politics. Comparative Politics, 3, 283–322. Huntington, S. P. (1993). The clash of civilizations? Foreign Aﬀairs, (Summer), 22–49. Jacobson, D. (1997). New frontiers: Territory, social spaces, and the state. Sociological Forum, 12(1), 121–133. Joireman, S. F. (2003). Nationalism and political identity. New York: Continuum. Katzenstein, P. J. (1977). Ethnic political conﬂict in South Tyrol. In M. Esman (Ed.), Ethnic conﬂict in the Western World. Ithaca, NY: Cornell University Press. Kelman, H. C. (1997). Group processes in the resolution of international conﬂicts: Experiences for the Israeli-Palestinian case. American Psychologist, 52, 212–220. Kelman, H. C. (1999). The interdependence of Israeli and Palestinian national identities: The role of the Other in existential conﬂicts. Journal of Social Issues, 55(3), 581–600. Kriesberg, L. (2002). Constructive conﬂicts: From escalation to resolution (2nd ed.). Lanham, MD: Rowman & Littleﬁeld. Laitin, D. D. (1998). Identity in formation: The Russian-speaking populations in the near abroad. Ithaca, NY: Cornell University Press. LaSelva, S. V. (1996). The moral foundations of Canadian federalism: Paradoxes, achievements, and tragedies of nationhood. Montreal, Kingston, London, & Buﬀalo: McGill-Queen’s University Press. Leslie, P. M. (1989). Ethnonationalism in a federal state: The case of Canada. In R. J. Thompson & J. R. Rudolph (Eds.), Ethnoterritorial politics, policy and the western world. Boulder, CO: L. Reinner Publishers. Lijphart, A. (1969). Consociational democracy. World Politics, 21, 207–225. Lijphart, A. (1977). Political theories and the explanation of ethnic conﬂict in the western world: Falsiﬁed predictions and plausible postdictions. In M. Esman (Ed.), Ethnic conﬂict in the western World. Ithaca, NY: Cornell University Press. Lijphart, A. (1991). The power-sharing approach. In J. V. Montville (Ed.), Conﬂict and peacemaking in multiethnic societies (pp. 491–509). New York: Lexington Books. Long, J. A. (1991). Federalism and ethnic self-determination: Native Indians in Canada. Journal of Commonwealth & Comparative Politics, 29, 192–211. McGarry, J., & O’Leary, B. (Eds.) (1993). The politics of ethnic conﬂict regulation: Case studies of protracted ethnic conﬂicts. New York: Routledge. McRoberts, K. (1997). Misconceiving Canada: The struggle for nation unity. Toronto: Oxford University Press.

Carter

267

Migdal, M. J., Hewstone, M., & Mullen, B. (1998). The eﬀects of crossed categorization on intergroup evaluations: A meta-analysis. British Journal of Social Psychology, 37, 303–324. Minow, M. (1990). Putting up and putting down: Tolerance reconsidered. In M. Tushnet (Ed.), Comparative constitutional federalism: Europe and America. New York: Greenwood Press. Mughan, A. (1983). Accommodation or defusion in the management of linguistic conﬂict in Belgium? Political Studies, 31, 434–451. Murphy, A. B. (1988). The regional dynamics of language diﬀerentiation in Belgium: A study in cultural-political geography. Chicago: The University of Chicago Press. Panebianco, A. (1988). Political parties: Organization and power (M. Silver, Trans.). Cambridge: Cambridge University Press. Pfaﬀenberger, B. (1991). Ethnic conﬂict and youth insurgency in Sri Lanka: The social origins of Tamil separatism. In J. V. Montville (Ed.), Conﬂict and peacemaking in multiethnic societies (pp. 241–257). New York: Lexington Books. Putnam, R. D. (1988). Diplomacy and domestic politics: The logic of two-level games. International Organization, 42(3 Summer), 427–460. Rawls, J. (1971). A theory of justice. Cambridge: Belknap Press. Rawls, J. (1993). Political liberalism. New York: Columbia University Press. Rubin, J. Z., Pruitt, D. G., & Kim, S. H. (1994). Social conﬂict: Escalation, stalemate, and settlement (2nd ed.). New York: McGraw-Hill. Rudolph, J. R., Jr. (1989). Belgium: Variations on the theme of territorial accommodation. In R. J. Thompson & J. R. Rudolph (Eds.), Ethnoterritorial Politics, policy and the western world (pp. 91–113). Boulder, CO: L. Reinner Publishers. Rudolph, J. R., Jr., & Thompson, R. J. (Eds.). (1989). Ethnoterritorial politics, policy and the western world. Boulder, CO: Lynne Reinner Publishers. Safran, W. (1989). The French state and ethnic minority cultures: Policy dimensions and problems. In R. J. Thompson & J. R. Rudolph, Jr. (Eds.), Ethnoterritorial politics, policy and the western world (pp. 115–157). Boulder, CO: L. Reinner Publishers. Schumpeter, J. (1955). The sociology of imperialism. New York: Meridian Books. Searle-White, J. (2001). The psychology of nationalism. New York: Palgrave. Senelle, R. (1989). Constitutional reform in Belgium: From unitarism towards federalism. In M. Forsyth (Ed.), Federalism and nationalism (pp. 51–95). Leicester: Leicester University Press. Singer, M. R. (1991). Prospects for conﬂict management in the Sri Lankan ethnic crisis. In J. V. Montville (Ed.), Conﬂict and peacemaking in multiethnic societies (pp. 259–286). New York: Lexington Books. Smiley, D. V. (1977). French-English relations in Canada and consociational democracy. In M. Esman (Ed.), Ethnic conﬂict in the western world. Ithaca, NY: Cornell University Press. Smith, A. D. (1983). Theories of nationalism (2nd ed.). New York: Holmes & Meier Publishers. Smith, A. D. (1989). The origin of nations. Ethnic and Racial Studies, 12, 214–225. Smith, A. D. (1995). Gastronomy or geology? The role of nationalism in the reconstruction of nations. Nations and Nationalism, 1(1), 3–24. Steiner, J. (1991). Power sharing: Another Swiss “export product”? In J. V. Montville (Ed.), Conﬂict and peacemaking in multiethnic societies (pp. 107–114). New York: Lexington Books.

268

Political institutions

Stern, P. (1995). Why do people sacriﬁce for their nations? Political Psychology, 16, 217–235. Tajfel, H. (1982). Social identity and intergroup relations. New York: Cambridge University Press. Taylor, C. (1991). Shared and divergent values. In R. L. Watts & D. M. Brown (Eds.), Options for a new Canada (pp. 53–76). Toronto: University of Toronto Press. Taylor, C. (1993). Reconciling the solitudes: Essays on Canadian federalism and nationalism. Montreal and Kingston: McGill-Queen’s University Press. Taylor, C. (1994). The politics of recognition. In A. Gutman (Ed.), Multiculturalism: Examining the politics of recognition (pp. 25–73). Princeton, NJ: Princeton University Press. Taylor, D., & Dubé, L. (1986). Two faces of identity: The “I” and the “We”. Journal of Social Issues, 42, 89–98. Tsebelis, G. (1990). Nested games: Rational choice in comparative politics. Berkeley, CA: University of California Press. Turner, J. C. (1982). Towards a cognitive redeﬁnition of the social group. In H. Tajfel (Ed.), Social identity and intergroup relations. Cambridge: Cambridge University Press. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. Oxford: Blackwell Publishing. Turner, J. C., Oakes, P. J., Haslam, S. A., & McGarty, C. (1994). Self and collective: Cognition and social context. Personality and Social Psychology Bulletin, 20(5), 454–463. Vandekeere, M., Doutrelepont, R., & Jacquenain, M. (1994). Les déclinaisons de l’identité en Wallonie: Couplages et divorces entre électorats, appartenances et prises de position en matière communautaire et institutionnelle. In A.-P. Frognier & A.-M. Aish-van Vaerenbergh (Eds.), Élections le fêlure? Enquête sur le comportement électoral des Wallons et des francophones. Brussels: DeBoeck. Varty, D. L. (1991). Who gets Ungava? Vancouver: Varty & Company Printers. Wilson-Smith, A. (1995, 6 November). A house divided. Maclean’s, 108, 14–16. Zariski, R. (1989). Ethnic extremism among ethnoterritorial minorities in western Europe: Dimensions, causes, and institutional responses. Comparative Politics, 21(3), 253–272. Zolberg, A. R. (1977). Splitting the diﬀerence: Federalization without federalism in Belgium. In M. J. Esman (Ed.), Ethnic conﬂict in the western world. Ithaca, NY: Cornell University Press.

Part VI

Conclusion

12 Multiple social categorization Integrative themes and future research priorities Miles Hewstone, Rhiannon N. Turner, Jared B. Kenworthy, and Richard J. Crisp . . . there is no contradiction. I’m Muslim, I’m British, I’m Asian, I’m an imam, I’m a teacher. (The Guardian, 16 July 2005) There are no Hutus or Tutsis any more, only Rwandans. (The Economist, 27 March 2004)

In the beginning was categorization . . . The power of the categorization process in judgments of social stimuli has been well documented (e.g. Doise, 1978; Tajfel, 1978; Wilder, 1986), and work using the minimal group paradigm (Tajfel, Flament, Billig, & Bundy, 1971) has shown that as well as cultural norms or political/economic factors inﬂuencing discrimination, the mere act of categorization can lead to intergroup bias. Wilder (1986) provided a superb exposition of how mere categorization of persons into diﬀerent groups engages a series of assumptions that foster intergroup biases. He reviewed evidence of the consequences of social categorization including perceived intragroup similarity (but between-group diﬀerence), biased causal attributions and memory processes, and ingroup favoritism and outgroup derogation. Among the more telling of his own ﬁndings was Wilder’s demonstration of accentuation eﬀects with regard to expectations; when we believe that individuals share a salient group membership we expect them to be similar; but when believe that they are divided by categorization, then we expect them to be diﬀerent from each other (see Allen & Wilder, 1975). Social categorization matters especially because it is intimately linked to extreme forms of intergroup behavior and social conﬂict. Especially in highly segmented societies that are diﬀerentiated along a single primary categorization such as ethnicity or religion (e.g., Northern Ireland, Rwanda), Brewer (2001) predicts a strong relationship between intense ingroup favoritism and outgroup antagonism, especially if categorization is dichotomous. The problem with such segmentation is that it promotes social comparison and perceptions of conﬂict of interest in zero-sum terms, leading to negative

272

Themes and priorities

attitudes and high potential for conﬂict. By contrast, the potential for intergroup conﬂict may be reduced in societies that are more complex and differentiated along multiple dimensions that are not perfectly correlated, i.e., cross-cutting category dimensions (see Crisp & Hewstone, 1999; Miller et al., Vescio et al., & Singh, this volume). These societal arrangements reduce the intensity of the individual’s dependence on any particular ingroup for meeting psychological needs for inclusion (see Brewer, 1991, 1993); this in turn reduces the potential for polarizing loyalties along any single cleavage or group distinction and perhaps increases tolerance for outgroups in general. An especially pernicious aspect of categorical thinking is the human tendency to see at least some groups of people as kinds of people, and to believe that certain groups (e.g., groups deﬁned by race, gender and culture) have essences that diﬀerentiate them from other groups (e.g., Haslam, Rothschild, & Ernst, 2000). Rothbart and Taylor (1992) consider essentialist thinking about social categories, and focus on two elements: immutability and inductive potential. Hirschfeld (1996) lists three elements in his analysis of essentialist racial categorization: inherence, discreteness, and naturalness; and Yzerbyt, Rocher, and Schadron (1997) list four deﬁning features: immutability, inductive potential, homogeneity or coherence, and exclusivity. A major consequence of this kind of thinking is what Miller and Prentice (1999) termed the category-divide hypothesis; this refers to the tendency to attribute a disagreement between parties from diﬀerent social groups to group diﬀerences. Arguably, both tendencies would be stronger for multiple, convergent categorizations. Social categorization clearly contributes to the most extreme forms of intergroup bias. Zajonc’s (unpublished) analysis of massacres points to several social-psychological processes closely linked to intergroup bias and conﬂict. These include delegitimizing victims (assigning them to an extreme social category, which enjoys no protection; Bar-Tal, 1990), and morally excluding them (placing them outside the ingroup boundary of justice, fairness, and morality; Opotow, 1995; Staub, 2001). As Wenzel (2004) has argued, when people are considered to be outside the moral community or scope of justice, justice principles are not extended to them (Opotow, 1990; for a critical review see Hafer & Olson, 2003). Opotow (1990) used the term moral unconcern or moral exclusion to describe the fact that prosocial norms stop at ingroup boundaries, leading to diﬀerential treatment of outgroup members. If outgroups are excluded from the deﬁnition of human being (see Leyens et al., 2000), the moral and emotional constraints that usually inhibit extreme behaviors no longer operate. Inhumane behavior literally becomes more appropriate for outgroup members (Opotow, 1990). There is an interesting parallel to such essentialist thinking in some of the political science interpretations of intergroup conﬂict. The primordialist view asserts that social categories are natural, inevitable, and unchanging facts about the social world (i.e., social categories are ﬁxed by human nature, rather than by convention and practice). Speciﬁcally, ethnic boundaries are seen as

Hewstone et al.

273

strong, enduring, and pervasive aspects of human nature (see Geertz, 1973). This is a pessimistic view, whereby conﬂict, especially massacres, is seen as inevitable, resulting from ancient antipathies and antagonisms that are seen as enduring properties of ethnic groups. The chapters in this volume take a much less pessimistic view. Many of the authors are explicitly concerned with ﬁnding solutions to problems of intergroup conﬂict, based on a better understanding of social categorization. What they share is a concern with more complex forms of multiple as opposed to simple categorization, and a commitment to understanding them in all their complexity, ranging from cognitive underpinnings, to social interventions. Although our knowledge of the importance of multiple categorization phenomena is not new (anthropologists and sociologists provided early analyses; see Evans-Pritchard, 1939; Murphy, 1957; Simmel, 1950; see also Crisp & Hewstone, this volume), we have only recently paid these important phenomena the attention they deserve (but see Allport’s, 1954, acknowledgement that people have multiple social identities that are hierarchically organized in terms of increasing inclusiveness). There is now extensive empirical support for the view that people do have multiple group identities (e.g., Stryker & Statham, 1985; Tajfel, 1978; see Deaux, 1996, for a review). As soon as we think about it, in fact, we note that most realistic intergroup contexts involve several categorizations, some of which coincide and some of which cut across each other. Thus others may be outgroup on one dimension (e.g., blacks vs. whites in the United States), but ingroup on another (e.g., Southerners vs. Northerners; see Reed, 1982). Where others can be classiﬁed as outgroup members on multiple dimensions, Brewer and Campbell (1976) labeled the situation as one involving converging boundaries, where discrimination is likely to be increased. Many instances of intergroup conﬂict in the real world involve just such multiple converging social categorizations. For example, in Belfast, Northern Ireland, Catholics and Protestants tend to live in diﬀerent places (e.g., Ardoyne vs. Shankhill Road), espouse diﬀerent politics (Nationalist-Republican vs. Unionist-Loyalist), and even support diﬀerent football teams (e.g., Cliftonville vs. Linﬁeld); there are few social categories cross-cutting the religious dimension (Cairns & Mercer, 1984). Likewise in Nigeria, ethnic violence ﬂares periodically between the Yoruba and Hausa tribes. These two ethnic groups also diﬀer on the dimensions of religion and region; the Yoruba are mostly Christians and animists, indigenous to Lagos and southwestern Nigeria, whereas the Hausa are Muslims from the north of the country. We would expect all the associated phenomena of essentialist thinking and the category-divide hypothesis to be accentuated in the case of multiple, convergent categorizations. Roccas and Brewer (2002) proposed that individuals’ representations of their multiple group memberships can vary along a dimension they call complexity. Social identity complexity refers to an individual’s subjective representation of the interrelationships among his or her multiple group identities. It refers to the degree of overlap perceived to exist between groups

274

Themes and priorities

of which a person is simultaneously a member. When one’s representation is low in complexity, it is an indication that one perceives a high overlap between both the typical characteristics of one’s various social category memberships, as well as an overlap between the actual members of those same categories. In other words, low complexity is a high overlap between membership and characteristics. High complexity, by contrast, is the opposite. It implies that the representation of each ingroup category is distinct from the others, both in characteristics as well as membership. Roccas and Brewer also showed that low complexity is associated with lower outgroup tolerance, less openness, and greater intergroup bias. In the remainder of this chapter we strive to fulﬁll two goals. First, we identify some of what we see as major themes that crisscross the chapters in this volume. Second, we highlight some possible areas for future research, focusing on one more theoretical issue (identity ﬂuxion) and one more applied agenda (using research on multiple categorization to study biculturalism).

INTEGRATIVE THEMES: SEVEN PILLARS OF WISDOM? Multiple categorization is an idea whose time has come. Society has changed: there is increased mixing across the lines of social categorization, including intermarriage; it is now the case that many societies are increasingly multiethnic, multireligious, and multicultural; in addition, an increasing number of individuals claim more than one ethnic identity, racial identiﬁcation, and source of cultural competence (see Phinney & Alipuria, this volume). To deal with these new realities, respondents to national census questions are now permitted to assert their multiple identities on oﬃcial forms (see Daniel, 2002). The chapters in this volume clearly demonstrate that, in many contexts, multiple bases for social categorization can be simultaneously salient and both combined and used. Identities often, but not always, have a hierarchy of importance (e.g., Crisp, Hewstone, & Cairns, 2001), and multiple categorization, like single categorization, can be studied in terms of various dimensions. The following seven themes are by no means inclusive, but they represent some of what we see as the most interesting cross-currents that emerge from the diverse perspectives in these chapters.

Theme 1: Categorization and its (dis)contents Categorization might be thought of as a necessary evil. Certainly, it is necessary, and some have implied that it is evil. Categorization has been deﬁned as the process of understanding what something is by knowing what other things it is equivalent to and what other things it is diﬀerent from (McGarty, 1999, p. 1). This meaning-giving function of the categorization process was emphasized by Tajfel (1978; see also Oakes, 2001). As Dovidio, Gaertner, Hodson, Riek, Johnson, and Houlette point out in this volume, the ability

Hewstone et al.

275

to categorize, to sort people spontaneously, and with minimum eﬀort or awareness into a smaller number of meaningful categories, is a universal facet of human perception and cognition that is essential for eﬃcient functioning (see also McGarty, this volume). But Wilder (1986) also argued that categorization, per se, propels the individual down the road to bias (p. 292). Oakes (2001) oﬀered a rather diﬀerent view, based on self-categorization theory (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987), whereby categorization operates at varying levels of abstraction. From this perspective, categorization is a context-speciﬁc process rather than activation of a cognitive structure; attributes relevant to a particular categorization can only be deﬁned with reference to the perceiver’s purposes and primary understandings. Self-categorization theory sees categorization as operating to maximize perceived similarity within categories and perceived diﬀerentiation between categories. Categorization is, moreover, not suﬃcient for stereotyping, but merely necessary. This view sits comfortably with the way category deﬁnitions change over time (see below), and the fact that what to outsiders can seem very small diﬀerences can still be the focus of conﬂict between groups. In Northern Ireland, for example, McGarry and O’Leary (1995) argue that the social content of Protestant and Catholic labels has changed so much over the centuries that it would be diﬃcult to ﬁnd a set of long-standing cultural diﬀerences that do still separate the two populations (p. 250). Yet, while the content of such identities may be in ﬂux, the boundaries themselves (i.e., the criteria deﬁning membership) are long-standing and unquestioned. Oakes (2001) gives a robust defense of the view that categorization is not the cause of conﬂict; crucially, identiﬁcation with the ingroup is necessary to ensure the assumed sequence from categorization via social comparison to intergroup discrimination. Moreover, even if categorization has had devastating consequences for intergroup relations, and is a crucial part of any intergroup conﬂict, it also has a central role in reducing, as well as creating, intergroup bias (see Dovidio et al., this volume). A complete understanding of multiple categorization must include both categorization by others (or person perception; Macrae, Bodenhausen, & Milne, 1995; van Rijswijk & Ellemers, 2002) and aspects of self-categorization (Turner et al., 1987). Tajfel (1978) referred to this as the distinction between external categorizations (how people are categorized) and internal categorizations (how they see themselves). He pointed out that when people are repeatedly treated by others in terms of a particular group membership, they are likely to internalize this deﬁnition of themselves eventually (see Barreto & Ellemers, 2003). Phinney and Alipuria (this volume) draw the same distinction between what they term assignment (what others say individuals are) and assertion (who or what individuals claim to be; see Daniel, 2002).

276

Themes and priorities

Theme 2: Which categorization is used? How does the social mind deal with the problem of multiple category memberships? Self-categorization theory provides a precise prediction: the categorization will be chosen that subjectively gives meaning to the situation. Both meaning and category salience are determined by accessibility (i.e., readiness, e.g., manipulated via priming), structural ﬁt (cf. self-categorization theory’s meta-contrast ratio), and normative ﬁt (the match between the category and the content properties of its stimuli; see Oakes, 1987, 1996). The existence of multiple potential categorizations, whether they are self- or other categorizations, gives categorization what McGarty (this volume) calls its power. Because objects can be assigned to diﬀerent categories, there are, in fact, two options open to the categorizer. If we choose to categorize ourselves or others more speciﬁcally or exclusively (i.e., in terms of one category, but not others), we need to understand category selection or application processes. If, however, we choose to categorize ourselves or others more generally or inclusively (i.e., in terms of more than one category), then we need to understand how categories are combined or the category conjunction process. Sometimes perceivers evaluate others who are simultaneously members of two or more categories in terms of a compound category with emergent properties that are not predicted from the contributing categories considered separately (e.g., Hastie, Schroeder, & Weber, 1990; Kunda, Miller, & Claire, 1990; Smith & DeCoster, 1998; see Crisp & Hewstone; Smith, this volume). Although categories can sometimes be ordered hierarchically (as in the case of non-social objects; see Smith, this volume), there are problems associated with this hierarchical type of representation. McGarty (this volume) argues that it does not reﬂect the way that many (perhaps most) real social structures are organized; instead, he contends, social category systems are characterized by fuzzy boundaries and overlapping memberships that are subject to change as a function of the social context (see Theme 4, below). Smith (this volume) notes that the major limitation of hierarchical category representation is precisely that it does not allow for multiple cross-cutting categorizations. Whether categorization is by the self or by others, it can have both cognitive and more social consequences. As we argued in Chapter 1 (this volume), these aspects can be (and have been) studied with a focus on multiple category representation and multiple category evaluation. Cognitively, Doise (1978) argued that crossing categories would lead to cancellation. He proposed that simple categorization leads to an accentuation of diﬀerences between and similarities within categories. However, in a situation of crossed categorizations, these processes will work against one another (Deschamps & Doise, 1978). There will be convergence between the categories and divergence within each category. When categories are crossed, the simultaneous processes of between-category accentuation of diﬀerences and within-category assimilation of similarities that occur cancel each other out. Although there

Hewstone et al.

277

is evidence for this cancellation process (see Crisp & Hewstone, 1999), we have also learned that it is not always the case that both (or multiple) categorizations are equally salient, or indeed that both are used (e.g., Macrae et al., 1995; van Twuyver & van Knippenberg, 1999). But even if multiple categories are sometimes mutually exclusive, there is extensive evidence that people can and do process, represent, and use multiple categorizations simultaneously (e.g., Crisp & Hewstone, 2001; Stangor, Lynch, Duan, & Glass, 1992; see also Crisp & Hewstone, this volume). Kunda and Thagard (1996) argue that a number of applicable stereotypes can be activated simultaneously. But as Macrae and Bodenhausen (2001) point out, this task could be both daunting and counterproductive. Instead, targets may only be categorized in terms of multiple categories when perceivers have suﬃcient time, capacity, and motivation (e.g., no time pressure, single vs. dual task to perform, when pursuing complex interactional goals, such as accountability; see Pendry & Macrae, 1996). The default, Bodenhausen and Macrae (1998) suggest, is that basic inhibitory processes simplify processing of complex social information. When perceivers encounter a multiply categorizable target, all relevant categories are believed to be activated in parallel, but are subject to a competition for mental dominance. The category that wins out is likely to be the one that beneﬁts from category salience, chronic accessibility, and temporary goals states. Macrae and Bodenhausen (2001) argue that unselected categories are actively inhibited during the category selection process. This was shown in a clever study by Sinclair and Kunda (1999). After participants received favorable feedback from a black doctor, associations of the category black became signiﬁcantly less accessible in their minds, while associations of the category doctor became more accessible (they did the reverse when the black doctor provided negative feedback). The shifting standards model (e.g., Biernat & Thompson, 2002) focuses on how judgments of individual members of stereotyped groups are made, relative to expectations regarding the group as a whole. As they point out, there is a tendency for judges to evaluate members of certain category combinations most extremely in certain domains (e.g., from the combinations of black– white and athletic–caring, black males are judged most extremely athletic and white females are judged most extremely caring); these individuals may represent the prototypical carriers of the stereotypical attributes studied. Moreover, certain traits may call to mind certain category combinations, and traits may be interpreted with reference to particular types of people (e.g., athletic may call to mind black males, but the meaning of the same trait may change when applied to other targets (e.g., athletic white female). This last idea is taken up extensively in Kunda and Thagard’s (1996) parallel constraint satisfaction model that provides a theoretical account of impression formation. The model assumes that judgments of a target’s traits arise from the network of connections or associations that are activated or inhibited at the time of use. This process of activation or inhibition occurs

278

Themes and priorities

either simultaneously or in parallel. According to the model, traits change their meaning across contexts, as a speciﬁc subset of their network of associations is activated (see Kunda, Sinclair, & Griﬃn, 1997). Stereotypes, although given no special status in the model, are inﬂuential because their associations constrain the meanings of trait terms. Pendry and Macrae (1996) further demonstrated that the extent of stereotype activation is moderated by perceivers’ level of involvement with the target. If the level of involvement is rather low or uninvolved, stereotype activation typically takes places at the broadest level (i.e., superordinate categorization). However, when the perceiver holds complex interactional goals (e.g., accountability, outcome-dependence), then a target is categorized both in terms of a higher-order representation (e.g., woman) and a more diﬀerentiated category subgroup or subtype (e.g., businesswoman).

Theme 3: Categorization as a resource In terms of self-categorization, at least, identity is also a resource, and multiple identities can oﬀer multiple resources. Thus, in addition to any conﬂictreducing properties of crossed or multiple categorization (see below), membership in a large number of groups oﬀers an individual the possibility of alternative sources of positive social identity (Allen, Wilder, & Atkinson, 1983). These authors do, however, also acknowledge the following limitation, that the greater the number of group memberships possessed by an individual, the greater the likelihood of the occurrence of conﬂict, ambiguity, or other strains among them (p. 96). Golﬁng tyro Tiger Woods, in a famous example, has resolved any such ambiguity himself (as well as sidestepping pressures from constituencies who wished to see him assert himself as a person of color), by declaring himself a “Cablinasian”, a term that combines his Caucasian, African, Native American, and Asian heritage. Multiple categories can result in a more complex cognitive representation of individuals (Brewer, Dull, & Lui, 1981; Stangor et al., 1992); hence, they oﬀer the target of perception the possibility of being viewed, even stereotyped, in various ways, some perhaps with positive, others with negative, consequences. An example of identity as a resource at the level of the individual comes from work on stereotype threat. Shih, Pittinsky, and Ambady (1999) manipulated the salience of participants’ social identity (gender vs. ethnic) and assessed its eﬀects on performance. In line with the stereotype threat model, Shih et al. found that Asian-American women performed more poorly in a math test when their gender identity rather than their Asian identity was made salient (because math is related to negative stereotypes about women, but positive stereotypes about Asians). Thus, for these participants, on this task, their Asian identity was a resource that they were able to use to bolster their performance. An example at the level of the group comes from Hogg and Hornsey (this volume). They note that Tajfel’s emphasis on intergroup diﬀerentiation, or

Hewstone et al.

279

accentuation of diﬀerences, may take diﬀerent forms. Negatively, it may introduce harmful, ultimately destructive intragroup conﬂict; positively, it may celebrate diversity within the group, including its subgroups, and may permit groups to marshal diversity which can strengthen the groups when facing threat (e.g., on the advantages of within-group diversity, see Hewstone, Martin, Hammer-Hewstone, Crisp, & Voci, 2001). There are, of course, other advantages to membership in multiple groups. Most prominently, our group memberships contribute to establishing and maintaining a positive social identity (Tajfel, 1978). Identity derived from group memberships is valued; as Hogg and Hornsey (this volume) put it, groups and categories thrive to the extent that they provide their members with a clear and evaluatively positive sense of who they are and how they relate to others, and such groups typically have high entitativity and positive valence. Any so-called solution to problems of intergroup relations that threatens to deprive individuals of valued social identities is inherently undesirable (Hewstone, Rubin, & Willis, 2002). Phinney and Alipuria (this volume) also cite work showing that a secure sense of aﬃrmation of one’s own ethnic group is associated with more positive outgroup attitudes (Phinney, Ferguson, & Tate, 1997). Phinney and Alipuria (this volume) also point out that most socialpsychological research on multiple categorization has focused on simultaneous categorization in terms of two (or more) diﬀerent identity domains. Thus, for example, Hewstone, Islam, and Judd (1993) crossed religion (Hindu vs. Muslim) and nationality (Bangladesh vs. India), or religion, nationality, and language (Studies 1 and 2, respectively). Phinney and Alipuria, however, study a diﬀerent issue, that of multiple categorization within a single identity domain. This involves an analysis of individuals who are not black vs. white, or Hispanic vs. American, but rather black and white, or Hispanic and American. Phinney and Alipuria focus on three types of multiracial, multiethnic, or multicultural individuals: those who have parents from two diﬀerent ethnic or racial groups; those who have parents who both come from one and the same culture, but themselves grew up in another culture; and those who themselves may be either multiracial or monoethnic, but who incorporate into their identity a culture other than that of their parents. Phinney and Alipuria note that these individuals do not (have to) balance the importance or relevance of two distinct characteristics of the self, but rather they integrate and manage internal complexity, which may be potentially conﬂicting, but also potentially enriching. We believe that biculturalism will become one of the growth areas for research on multiple categorization, and we return to this topic, below, as one of our two highlighted areas for future research. Precisely because of these advantages of group memberships, we caution against solutions to problems of intergroup relations that involve category suppression, such as the colorblind approach (see Hewstone, 1996). Thus intergroup contact more eﬀectively reduces bias when two groups are provided with distinct roles that maintain their positive distinctiveness while cooperating (Dovidio, Gaertner, & Validzic, 1998; for a review, see Brown &

280

Themes and priorities

Hewstone, 2005). Wolsko, Park, Judd, and Wittenbrink (2000) also showed that suppressing categorization is ill-advised; exposure to multicultural vs. colorblind ideological messages not only was associated with reduced bias, but also permitted white perceivers to continue to diﬀerentiate black and white Americans in both more accurate and more positively valued ways (e.g., with respect to their attributes and their values). A topical case in which category suppression is currently being promulgated is Rwanda. President Paul Kagame’s government has abolished ethnic identity cards and forbids the breakdown of oﬃcial statistics by tribe. In “solidarity camps”, which all young Rwandans are obliged to attend, the message is taught that there are no more Hutus and Tutsis; all are Rwandans, who must learn to live together. After the devastating genocide of 1994, extreme solutions may well be called for. But from the literature on intergroup relations, it is questionable whether suppression of the categories Hutu and Tutsi, in favor of the superordinate Rwandan, will ultimately work.

Theme 4: The ﬂexibility of categorization and change over context and time As many of the chapters in this volume illustrate, there can often be multiple, cross-cutting, and conjunctive bases for social categorization, and as a result the processes and outcomes involved can be complex and multifaceted. Moreover, the context in which categories appear can cognitively change the way they are represented (Barsalou, 1987, 1989). Change of categorization (and identity) over context and time is also a theme highlighted in several chapters of this volume that represent the extremes of a more cognitive and a more societal approach to multiple categorization. From a cognitive perspective, McGarty (this volume) emphasizes that real systems of social categories shift in response to context, a point that is central to self-categorization theory, with its notion of categories being switched on and switched oﬀ (Turner et al., 1987). Smith (this volume) extends this idea by arguing that we should think of cognitive representations as states rather than things (see also Smith, 1998). For Smith, a representational state is deﬁned at a particular moment in time and will change as the perceiver shifts beliefs or focuses on new information. From a more social, and less cognitive, perspective on categorizations as states, Phinney and Alipuria (this volume) note that the meanings of multiple category membership will change as geographical areas become more diverse, and as the numbers of multiracial and multicultural individuals increase. We believe that further study of aspects of category ﬂux is so important that we have selected it as one of our two highlighted areas for future research (see below). A rather diﬀerent aspect of change is also identiﬁed by Phinney and Alipuria (this volume), and this one too needs more work in the future, namely research linking social and developmental psychology to study how identities develop and change with human development. Phinney and Alipuria’s

Hewstone et al.

281

perspective is unique in this volume, because it is based on Erikson’s (1968) model of identity development, rather than the cognitive-social theoretical underpinning of the majority of chapters in this volume.

Theme 5: Multiple categorization as a means of reducing intergroup bias As well as outlining the causes and consequences of categorization, Wilder (1986) wrote extensively on possibilities for reducing intergroup bias by altering the ingroup–outgroup representation. In particular, he considered two strategies: individuation of the outgroup and diminishing intergroup boundaries. The latter is of primary relevance here and includes: (1) weakening the boundary by weakening perceived belief dissimilarity between groups; (2) changing the social context by introducing a new outgroup; (3) having overlapping group memberships; and (4) eliminating the intergroup boundary. Many of the chapters in this volume deal in detail with aspects of the third possibility, overlapping group memberships, and yield new insights regarding two major approaches, crossed categorization and the common ingroup identity model, and a hybrid of the two, the dual identity model. Crossed categorization By contrast with highly segmented societies diﬀerentiated along a single primary categorization, the potential for conﬂict tends to be reduced in societies that are more complex and diﬀerentiated along multiple dimensions that are not perfectly correlated (Brewer, 2001; LeVine & Campbell, 1972). Cross-cutting category divisions reduce the intensity of the individual’s dependence on any particular ingroup for meeting the psychological need of inclusion (see Brewer, 1991, 1993). As Brewer points out, this, in turn, reduces the potential for polarizing loyalties along any single cleavage or group distinction, and may increase tolerance for outgroups in general. The crossed categorization approach is based on horizontal, cross-cutting categories, where others can be simultaneously classiﬁed as ingroup or outgroup members on multiple dimensions. Shared or overlapping category memberships should, in principle, reduce bias because they: (1) make social categorization more complex; (2) decrease the importance of any one ingroup– outgroup distinction; (3) make perceivers aware that the outgroup consists of diﬀerent subgroups; (4) increase classiﬁcation of others in terms of multiple dimensions; and (5) increase the degree of interpersonal interaction and trust across category boundaries (Brewer, 2000a; Brewer & Gaertner, 2001; Hewstone, 1996). Crossing two categories (i.e., the target is both ingroup and outgroup, or partial) does not typically result in less bias than is found in a simple categorization (i.e., target is only outgroup) condition (e.g., Crisp et al., 2001; Singh, Yeoh, Lim, & Lim, 1997; see meta-analysis by Mullen, Migdal, & Hewstone,

282

Themes and priorities

2001), but it does reduce the pronounced bias directed against double outgroup targets (i.e., outgroup on both available dimensions; e.g., Crisp et al., 2001; Vanman, Paul, Ito, & Miller, 1997; see review by Crisp & Hewstone, 1999, and meta-analyses by Migdal, Hewstone, & Mullen, 1998, and Urban & Miller, 1998). Cross-cutting social identities or role assignments manipulated as part of cooperative intergroup contact are especially eﬀective, increasing intracategory diﬀerentiation and decreasing perceived intercategory diﬀerences (Ensari & Miller, 2001; Marcus-Newhall, Miller, Holtz, & Brewer, 1993). Several diﬀerent patterns of bias can be found when categories are crossed (see Crisp & Hewstone, 1999, 2000a, 2000b; Hewstone et al., 1993), and bias is not always reduced. Singh (this volume) provides new insights into how we can best test between diﬀerent models of crossed categorization. Miller et al. (1993) have made a major contribution in detailing moderators of crossed categorization eﬀects (see Miller et al., this volume). Their work shows that diﬀerential category importance, mood, and personalized interaction moderate crossed categorization eﬀects. Mediators of crossed categorization have also been studied. Crisp, Ensari, Hewstone, and Miller (2002) reviewed evidence for underlying processes and concluded that both cognitive (i.e., diﬀerentiation) and aﬀective (e.g., emotion) factors can moderate how crossed categories aﬀect evaluations in diﬀerent contexts. Crisp et al. argue that diﬀerent mediators will likely operate in diﬀerent contexts (i.e., as a function of diﬀerent moderators). The need to study moderators is underscored by divergent ﬁndings in the crossed categorization literature. For instance, while Marcus-Newhall et al. (1993) found evidence that diﬀerentiation can mediate crossed categorization eﬀects on bias, recently Vescio, Judd, and Kwan (2004; see also this volume) found no evidence that diﬀerentiation mediated evaluations. It is therefore likely that categorization and bias are linked in complex ways, depending on moderating conditions (see below). Notwithstanding the positive eﬀects of crossed categorization, its eﬀectiveness in reducing bias is limited when: (1) one category dimension is functionally dominant (e.g., Crisp & Hewstone, 2001; Hewstone et al., 1993; Singh, this volume; Urada & Miller, 2000); (2) categories are correlated (Eurich-Fulcer & Schoﬁeld, 1995); and (3) groups are under threat, and social identities are deﬁned in more exclusive and less complex terms (Brewer, 2000a). Unfortunately, these are precisely the characteristics of many contexts characterized by severe intergroup conﬂict; as such they appear to limit the eﬀectiveness of crossed categorization as a conﬂict-reducing intervention (see also Themes 6 and 7, below). Recategorization and common ingroup identity The common ingroup identity (CII) model of recategorization seeks to alter which categorizations are used and to replace subordinate (us and them) with

Hewstone et al.

283

superordinate (we) categorizations (Gaertner & Dovidio, 2000). For example, a disagreement between two groups (such as police oﬃcers and nurses) could be resolved by uniting them in a more inclusive category (e.g., public services). There is extensive experimental support for the CII model (there is also support from survey research, but here the evidence for cognitive mediation is generally weaker; e.g., Gaertner, Rust, Dovidio, Bachman, & Anastasio, 1994). Key ﬁndings of these experiments are that: (1) inducing a one- versus two-group representation of intergroup relations (e.g., via cooperation) reduces bias via its eﬀect on cognitive representations of social categorization (Gaertner, Mann, Murrell, & Dovidio, 1989; Gaertner, Mann, Dovidio, Murrell, & Pomare, 1990); (2) distinctive two-group representations predicted more bias, and stronger superordinate representations predicted less bias (Dovidio, Gaertner, Isen, & Lowrance, 1995). Overall, this research ﬁnds, as predicted, that bias is reduced primarily by improving attitudes toward former outgroup members due to their recategorization from outgroup to ingroup. There are, however, two major limitations to the CII solution (Brewer & Gaertner, 2001; Hewstone, 1996). First, a common ingroup identity may only be short-lived, or unrealistic in the face of powerful ethnic and racial categorizations. For example, why did Tito’s regime fail, in the long term, to replace separate Serb, Croat, Macedonian, Albanian, and Muslim categories with the overarching Yugoslavian identity? Second, for groups with a history of antagonism, and for minorities who are likely to resist assimilation into a superordinate category that is dominated by a majority outgroup (van Oudenhoven, Prins, & Buunk, 1998), the prospect of a superordinate group identity may constitute a threat, which actually increases bias (Brewer 2000a, 2000b; Hornsey & Hogg, 1999; see also Theme 6, below). Perhaps the fundamental limitation of the recategorization model is that it threatens to deprive individuals of valued social identities in smaller, less inclusive groups (Brewer, 1999). By eradicating or replacing original categorizations, this model is unlikely to meet the needs of assimilation and diﬀerentiation, or of cognitive simplicity and uncertainty reduction (Brewer 2000b, 2001; Hogg & Abrams, 1993). Thus recategorization is a temporally unstable solution to the problem of intergroup discrimination (Brewer & Gaertner, 2001). The CII model is most likely to be a successful model for reducing conﬂict when symbols of superordinate category membership are highly salient, and individual needs for diﬀerentiation are not aroused. However, as Brewer (2000b, 2001) points out, the prospect of a superordinate common ingroup identity may constitute a threat rather than a solution when intergroup attitudes have evolved to the point of outgroup hate or overt conﬂict. In particular, under high levels of intergroup distrust, common ingroup identities are likely to be seen as threats; in contrast, when outgroup attitudes are characterized simply by lack of trust, rather than explicit distrust, then there is potential to reduce conﬂict by building on common interests and shared identities.

284

Themes and priorities

Dual identity Gaertner and Dovidio (2000) subsequently developed the dual-identity model, which recognizes that groups should not be made to forsake their identities (Dovidio et al., this volume). This hybrid model simultaneously recognizes both diﬀerent and common group memberships, and is a more complex form of common ingroup identity than a simple one-group representation. The dual-identity model aims to reduce bias between subgroups that share a common superordinate identity, rather than consider themselves as members of separate groups (Gaertner & Dovidio, 2000). Because the subgroups are both members of the same group at a superordinate level, bringing them together should not arouse motivations to achieve distinctiveness, increase perceived threat to identity, or exacerbate bias. Research has found that a dual identity led to more positive outgroup attitudes than did a superordinate identity alone (Crisp, this volume; Crisp, Stone, & Hall, 2006; Hornsey & Hogg, 2000; Hogg & Hornsey, this volume), especially if the superordinate category was too inclusive and did not aﬀord adequate distinctiveness (Hornsey & Hogg, 1999; see also Eggins, Haslam, & Reynolds, 2002, on the value of maintaining or restoring positive and distinct social identities for subgroups). A major problem remains, however, for the dual-identity approach; members of majority and minority groups may have diﬀerent preferences for what model of intergroup relations to adopt. Dominant majority ethnic groups tend to favor assimilation, whereas racial and ethnic minorities favor pluralistic integration (Berry, 1997; Dovidio, Gaertner, & Loux, 2000a; Dovidio, Major, & Crocker, 2000b; van Oudenhoven et al., 1998; Wolsko et al., 2000). Thus a dual identity may reduce bias for the minority, but not the majority (Zagefka & Brown, 2002). A further, fundamental problem for the dual-identity approach is that there are conditions under which multiple categorization approaches may be ineﬀective or, at worst, may even lead to heightened intergroup bias. To be successful a superordinate category and identity must be inclusive, and able to represent group diﬀerences in a complex way, rather than reﬂecting too strongly the superior characteristics of a dominant majority group (Mummendey & Wenzel, 1999). According to Mummendey and Wenzel’s theory of ingroup projection, the inclusion of subgroups within an inclusive superordinate identity can actually worsen intergroup relations. Comparisons between these groups become likely, group members tend to see their own subgroup as more typical of the superordinate group (they project their subgroup’s attributes onto the superordinate group), and judgments become more biased because group members perceive their subgroup as superior. According to studies reported by Dovidio et al. (this volume), this negative process may be more likely to occur for members of high- rather than low-status groups. A dual identity does not, however, always lead to increased bias (Gaertner, Rust, Dovidio, Bachman, & Anastasio, 1996), but ingroup

Hewstone et al.

285

projection and subsequent increased bias are more likely when the superordinate identity is relevant (see Mummendey, & Wenzel, 2002). Multiple categorization and the reduction of intergroup bias: Toward an integration The three main models of multiple categorization considered here (crossed categorization, common ingroup identity, and dual identity) should be seen as complementary and reciprocal, not competing and exclusive (Brewer, 2000b; Brewer & Gaertner, 2001; Hewstone, 1996; Hewstone et al., 2002). This integrative approach responds to the fact that each model can be eﬀective under particular conditions, but also has weaknesses and limitations, particularly with respect to ﬁnding an intervention that works for both majority and minority groups (Brewer & Gaertner, 2001; Gaertner & Dovidio, 2000; Hewstone, 1996; Pettigrew, 1998a, 1998b). Future research needs to specify how to combine multiple models into high-impact interventions. Future work will be aided by the progress already made in identifying moderators of both crossed categorization and recategorization eﬀect. Miller et al. (this volume) show that diﬀerential category importance, mood, and personalized interaction moderate crossed-categorization eﬀects, whereas Dovidio et al. (this volume) focus on category salience, importance, and relevance. As Dovidio et al. point out, this is theoretically and practically important. Theoretically, we need to improve our understanding of the relationship among multiple possible identities and their translation into thought, aﬀect, and action. Practically, this understanding will help practitioners to decide what kind of intervention is most likely to be eﬀective when (i.e., in a given situation), as well as which approaches are likely to be counterproductive. This acknowledgement that some kinds of interventions can sometimes exacerbate bias represents something of a loss of innocence, but theoreticians, researchers, and practitioners have to take this on board in selecting the optimal strategy for a given problem of intergroup relations. As Dovidio et al. (this volume) point out, the real challenge for future work will be to move beyond demonstrations about which technique is more eﬀective, or even which technique is more eﬀective under which circumstances. We will need to think more synthetically about how diﬀerent models, which exploit diﬀerent avenues to bias reduction and may operate via diﬀerent processes, can be integrated to reduce bias maximally. One very real problem that remains for all multiple categorization approaches is whether such overlapping groups (can) satisfy the needs for inclusion and diﬀerentiation (Brewer, 2001). As Brewer’s (1991, 1993) work attests, groups that are exclusive rather than inclusive, groups that have clear category boundaries, satisfy these needs. Brewer makes the excellent point that what we know of the psychology of assimilation and diﬀerentiation limits the extent to which strong social identiﬁcation can be indeﬁnitely extended to highly inclusive, superordinate social groups, or categories. Her

286

Themes and priorities

optimal distinctiveness theory implies that ingroup loyalty (and its concomitant depersonalized trust and cooperation) is most eﬀectively engaged by relatively small, distinctive groups, or social categories. If feelings of secure inclusion, ingroup loyalty, and optimal identity are dependent on the clarity of ingroup boundaries and intergroup distinctions, then cooperation with an outgroup threatens the basis for social identiﬁcation. Especially for individuals who are exclusively vested in a single group identity (as is the case in situations of category dominance, such as Northern Ireland, India, Cyprus, etc.), the threat of lost distinctiveness may override the pursuit of superordinate goals and lead to resistance to cooperation.

Theme 6: The central importance of threat Threat appears as a central explanatory concept in several key theories of intergroup bias (see Brewer, 1999; Hewstone et al., 2002). The threat can be perceived in terms of the ingroup’s social identity (Tajfel & Turner, 1979), its goals and values (Esses, Jackson, & Armstrong, 1998), its position in the hierarchy (Sidanius & Pratto, 1999), and even its existence (Greenberg, Simon, Pyszczynski, Solomon, & Chatel, 1992). This threat can be realistic or symbolic (Stephan & Stephan, 2000). Realistic threat includes threat associated with intergroup competition over scarce resources, as in realistic group conﬂict theory (see LeVine & Campbell, 1972). But Stephan and Stephan (2000) suggest a broader conception of realistic threat, extending beyond competition over scarce resources to include threat posed to the political and economic power of the ingroup, to its physical or material well-being, to the welfare of the group and its members, and to its very existence (Stephan & Stephan, 2000). Symbolic threat refers to perceived diﬀerences in morals, values, standards, belief, and attitudes (Stephan & Stephan, 2000); threat is posed to the worldview of the ingroup, and the outgroup is seen as blocking the ingroup’s values, customs, and traditions (Esses, Haddock, & Zanna, 1993; Sidanius, Devereux, & Pratto, 1992). These are all aspects of threat that come readily to mind when we consider the most troubling intergroup conﬂicts facing this world. Inﬂuenced by social identity theory, threat is frequently conceived, and experimentally manipulated, in terms of heightened intergroup similarity, which poses a potential threat to the ingroup’s distinctiveness. Several studies have yielded support for the prediction that a similar, relevant outgroup will be perceived as a threat to group identity, leading to attempts to diﬀerentiate from it (e.g., Jetten, Spears, & Manstead, 1996, 1997). Jetten, Spears, and Manstead (1998) found a curvilinear relation between distinctiveness and diﬀerentiation in both laboratory and natural groups, with strongest bias against an outgroup that was similar to but clearly separate from the ingroup. Diﬀerentiation was relatively low or nonsigniﬁcant when in- and outgroups were too similar or too dissimilar. It is this aspect of threat that is most likely provoked by interventions based

Hewstone et al.

287

on multiple categorization that involve cross-cutting categorization and superordinate identities. As we have noted several times, threat can be conceived in terms of both core needs in Brewer’s (1991, 1993) optimal distinctiveness theory. Threat to distinctiveness is, moreover, accompanied by emotions of fear and anger (Wilder & Simon, 2001) that trigger action against, rather than mere avoidance of, the outgroup (Mackie & Smith, 2002). Threat is central to the phenomena discussed by many of the chapters in this volume. Dovidio et al. (this volume) describe the common ingroup identity model as a model that recognizes the centrality of threat to collective identity for motivation, attitudes, and action. Hogg and Hornsey (this volume) ask, how do groups respond to threat? Their analysis is especially original in highlighting the consequences of threat for intragroup, rather than intergroup, relations and structures. They treat multiple categorization as a consequence of threat. It also evident from the contributions to this volume that how group members respond to threat is moderated by their level of social identiﬁcation. Thus interventions that seek to weaken intergroup boundaries could have divergent eﬀects for low and high identiﬁers (van Leeuwen, van Knippenberg, & Ellemers, 2003). For example, Crisp (this volume) argues that making a common ingroup identity salient (e.g., Europe for Britons) may only lead to reduced bias for low identiﬁers, whereas bias will remain or even be increased for high identiﬁers (see Crisp & Beck, 2005; Crisp et al., 2006). Crisp suggests that a better solution speciﬁcally for high identiﬁers may be maintaining the simultaneous salience of both superordinate (common ingroup) and subcategory group memberships (as outlined by Hornsey & Hogg, 2000); this form of multiple categorization protects against the threat to distinctiveness. This solution, however, may be less eﬀective than recategorization for low identiﬁers. Crisp proposes the intriguing possibility that recategorization may be the optimal strategy for low identiﬁers, whereas simultaneous categorization is the optimal strategy for high identiﬁers. But while this is theoretically appealing, it is practically onerous, if not impossible (how can we design interventions to ensure recategorization for one group, but simultaneous categorization for the other?). Thus Crisp argues that simultaneous categorization may represent the best all-round solution. Crisp’s chapter concludes that future work will need increased consideration of individual perceiver characteristics, to match multiple categorization strategy to target groups, so that we have a more complete understanding of the consequences of theoretically derived models and their likely impact in policy interventions. It will be especially necessary, albeit diﬃcult, to select the group-speciﬁc optimal strategy of bias reduction in such a way that it takes into account the major perceived threat to each group.

288

Themes and priorities

Theme 7: Multiple categorization and the reduction of intergroup conﬂict Most of the research on multiple categorization has focused on either a deeper understanding of its cognitive underpinnings, or on applying knowledge to interventions designed to reduce intergroup bias. Reducing bias, however, is not the same as reducing intergroup conﬂict. Social conﬂict is more complex than intergroup bias, and cannot be equated with the outcome of just one psychological process, nor should it be analyzed from just one disciplinary perspective (Hewstone et al., 2002). Real-world intergroup relations owe at least as much of their character to intergroup history, economics, politics, and ideology as they do to social-psychological variables such as self-esteem, ingroup identiﬁcation, group size, and group threat. Acknowledging these facts would suggest that we need a closer theoretical rapprochement between the disciplines of social psychology and political science. Carter (this volume) points out that social psychologists’ debates over the best strategies to reduce intergroup bias have a counterpart in political scientists’ typologies of strategies for elimination versus accommodation of ethnic group diﬀerences (e.g., Horowitz, 1985; McGarry & O’Leary, 1993). Of course, the two disciplines operate at diﬀerent levels of analysis, but each could learn from the other, and there are obvious parallels. For example, social psychology’s dual-identity or subgroup strategy has a counterpart in the political science notion of consociationalism (two subgroups are culturally or ideologically distinct, but still identify with the overarching state). The psychological literature could and should inform political scientists of the cognitive and social plausibility of, for example, imposing a particular constitution on a multiethnic state, whereas the political science literature could inform social psychologists of the likely applicability of various models in the real world of ethnic politics (e.g., Nigeria, Lebanon, or Northern Ireland; see Horowitz, 1985). Duckitt (1992) refers to the understated and undeveloped role of social scientists in designing constitutions and political systems to reduce or channel interethnic conﬂicts (e.g., Horowitz, 1985), which may seek to reinforce crossed, rather than converging, categorizations. The wider importance of crossed categorization as an intervention was recognized by political scientists when considering and designing constitutions and political–electoral systems to reduce or channel interethnic conﬂicts: Should, for example, territorial boundaries follow ethnic boundaries, or should they cross-cut them? (see Horowitz, 1985; Wallace, 1973). Horowitz’s (1985) classic text on analysis of ethnic conﬂict and violence in post-colonial Africa and Asia places due emphasis on both social-psychological and cognitive underpinnings of intergroup relations. His speciﬁcation of four possible processes of fusion and ﬁssion between ethnic groups (amalgamation, incorporation, division, and proliferation) is ripe for further cognitive-socialpsychological analysis. As knowledge grows about the cognitive bases and

Hewstone et al.

289

social consequences of models of multiple categorization, it is important that that knowledge is shared across group boundaries. Carter (this volume) also suggests using the analytic framework outlined in his chapter (which integrates a social-psychological/social identity approach with a political science/strategic action approach) to establish an earlywarning mechanism. This would involve monitoring of contemporary ethnic discourse, especially how various groups are portrayed, to intervene earlier and more eﬀectively (as a timely analysis of the mass media in both Rwanda and the former Yugoslavia would have suggested), but also to be alert to possible interventions to establish new cross-cutting categorizations to help reduce hostility.

TWO PRIORITIES FOR FUTURE RESEARCH ON MULTIPLE CATEGORIZATION Would you tell me, please, which way I ought to go from here? That depends a good deal on where you want to get to, said the Cat. Lewis Carroll (1865), Alice’s Adventures in Wonderland In this ﬁnal section we propose two topics worthy of intensiﬁed future research. The ﬁrst, biculturalism, is more practical, and the second, category ﬂuxion, is more theoretical; but both, in fact, have theoretical and practical implications and ramiﬁcations.

Contributions from theory and research on multiple categorization to the study of biculturalism Theory regarding how the categorization of groups can result in improved intergroup relations through crossed categorization, common ingroup identity, and dual identity models can be usefully applied to the study of immigration. This issue is of particular contemporary relevance, as many countries are now multicultural, made up of a diverse body of immigrants from all over the world, generated by increased geographic mobility and humanitarian crises. As well as coping with discrimination and racism, immigrants often ﬁnd themselves in a position where they have to deal with the presence of both their original culture and that of the host society (Phinney, Lochner, & Murphy, 1990), which may diﬀer in terms of values, attitudes, and customs (Berry & Annis, 1974). Minority or immigrant individuals are often faced with choosing between their own culture and identity, and those of the dominant culture (for an extended discussion, see Verkuyten, 2005a, 2005b). This dilemma is thought to have four potential outcomes, depending on the extent to which immigrants: (1) choose the host culture and reject their original culture (assimilation); (2) attempt to combine both cultures (biculturalism); (3) retain their original culture only and remain separate from the host

290

Themes and priorities

culture (separation); or (4) reject both cultures and live on the margins of society (marginalization) (see Berry, 1980; LaFromboise, Coleman, & Gerton, 1993; Phinney & Alipuria, this volume). Assimilation and biculturalism are typically thought to be the most adaptive strategies for immigrants. Assimilation and the common ingroup identity model Historically, assimilation was seen not only as inevitable (Gordon, 1964), but as optimal for mainstream socioeconomic success (Park, 1914) and social acceptance by the host culture (Ruiz, 1981). Gordon (1964) argued that maintaining diﬀerent cultures within one society would lead to segregation between groups; therefore groups should integrate into one for intergroup harmony to exist. According to the common ingroup identity model (Gaertner & Dovidio, 2000; Dovidio et al., this volume), viewing oneself as a member of one overarching group rather than as separate majority and minority groups should have a host of positive cognitive, aﬀective, and behavioral beneﬁts for intergroup relations (see also Phinney & Alipuria, this volume). Increasingly, however, both assimilation and the common ingroup identity model have faced criticism and are no longer considered optimal for intergroup relations in the immigration context. Even if minority group members were willing to give up their identity of origin, this might not be possible (e.g., non-white immigrants in a predominately white host society; Portes & Rumbaut, 1996). The emergence of a common ingroup identity is also dependent on the attitudes of the dominant group. Only if the dominant group accepts the minority into the host culture will assimilation be possible. A number of theorists (e.g., Vogt, 1957) have argued that assimilation will never be complete for minority groups. While individuals may participate in the majority community, they will always be identiﬁed as members of the minority culture (Ablon, 1964). As a result, minority individuals attempting to assimilate will often resign themselves to accepting second-class citizenship within the majority group (LaFromboise et al., 1993). Being in this position may lead minority group members to resent the majority, leading to poor intergroup relations between the two groups. Even for those minority individuals who do successfully assimilate into the majority culture, there may be socioeconomic and psychological costs (LaFromboise et al., 1993), including higher dropout rates from school compared to bicultural individuals (Feliciano, 2001), high risk of psychological maladjustment (Szapocznik & Kurtines, 1980), and alienation or isolation while transitioning between the two cultures (Johnson, 1976), bequeathing individuals poor relations with both the host society and the minority group. It is important to note here that assimilation does not reﬂect a true common ingroup identity where both groups change to accept the superordinate identity. Although in Britain members of diﬀerent ethnic groups may fall under a superordinate identity of being British, it is the minority groups who are expected to make the biggest change to their culture, values, attitudes, and

Hewstone et al.

291

behaviors. If, however, the superordinate group too strongly reﬂects the characteristics of the dominant majority group, this may lead minority groups to resent the majority group, with negative consequences for intergroup relations (see earlier discussion of Mummendey & Wenzel’s, 1999, ingroup projection model). The dual-identity model, crossed categorization, and the biculturalism approach As we have seen, the dual-identity model (see Dovidio et al., this volume) proposes that salience of initial groups should be maintained within a salient superordinate group for optimal intergroup relations between two groups. The model supports contemporary research on immigration which suggests that the optimal way for immigrants to deal with living in a new society is to understand and identify with both the culture of origin and the host culture. Immigrants who maintain their original sense of identity, while also sharing a superordinate nationality with the host society, and who are adept at dealing with both cultures (Ramirez, 1983), are known as bicultural. According to theories of dual identity and crossed categorization, bicultural individuals are less likely to suﬀer discrimination from the host society than those who retain only their culture and identity of origin (see Phinney & Alipuria, this volume). The relationship between bicultural individuals and the host community is also analogous to the partial ingroup in crossed categorization theory, whereby the outgroup is now a partial ingroup on a second dimension, the superordinate group. Thus, in the case of a minority group member, the host society may be outgroup because they are white, but ingroup because the minority group members and the host group members are all inhabitants of Britain. Consequently, they may experience less prejudice from the host group than full outgroup members (Deschamps & Doise, 1978). Theories of multiple categorization are largely tested in the laboratory using groups of little signiﬁcance, usually formed on an arbitrary basis (e.g., Deschamps & Doise, 1978). As a result, they only investigate simple examples of individuals who belong to more than one group simultaneously. Based on extensive research in the ﬁeld, however, immigration theorists propose several forms of biculturalism (Birman, 1994; LaFromboise et al., 1993), each with diﬀering implications for intergroup relations. We deal in more detail with two of these forms of biculturalism, alternation and integration. Alternation The alternation model suggests that it is possible for an individual to take part in two diﬀerent cultures by understanding the cultural assumptions that guide behavior and using this knowledge to think and behave appropriately in each (Bhawuk & Brislin, 1992; Yamada & Singelis, 1999). The model argues

292

Themes and priorities

that by alternating one’s cultural orientation depending on the situation, it is possible for an individual to have a sense of belonging in two cultures without compromising his or her sense of cultural identity (see Phinney & Alipuria, this volume). Alternation is likely to have a host of beneﬁts for intergroup relations. Bicultural students feel more at ease interacting with individuals from outside their ethnic minority, and have better problem-solving strategies and interpersonal skills (Buriel, Perez, De Ment, Chavez, & Moran, 1998). This should reduce anxiety about interacting with the host group (Stephan & Stephan, 1985), resulting in a more pleasant and comfortable interaction that will be perceived positively by both groups. Consequently, reciprocal intergroup attitudes should become more positive. Indeed, Schwarzer, Bowler, and Rauch (1985) found that minority students who were proﬁcient at communicating with the majority culture not only had higher levels of self-esteem than less bicultural individuals, but also reported less experience of racial tension and interethnic conﬂict. In addition, if minority individuals are able to switch eﬀectively between two groups, they will not be faced with the stress of having to choose between the two groups, which could lead to tensions with one or both groups. Lorenzo-Hernandez (1998) argued that the continuous contact bicultural individuals have with members of the dominant group leads them to become highly informed perceivers, who learn enough information about the outgroup to view them as a heterogeneous group, resulting in less stereotyping and conﬁrmatory biases, and more favorable outgroup attitudes. Extrapolating from dual-identity theory, bicultural individuals should be able to capitalize on the beneﬁts of holding a superordinate identity, by identifying with the host society, while avoiding any threat to the distinctiveness of their original group membership, which can be damaging to intergroup relations. LaFromboise et al. (1993) noted, however, that to adopt alternation, minority members must hold positive attitudes toward both groups, have the ability to communicate eﬀectively, and have their culture of origin strongly represented in the host society, to provide a support system and buﬀer the bicultural individual from stress (Birman, 1994). Integration Two distinct frameworks have been developed to describe how immigrants attempt to integrate with the dominant group (Birman, 1994). The cultural framework (Szapocznik, Kurtines, & Fernandez, 1980) is behavioral, focusing on gaining competence in the ways of the new culture and learning, while the identity framework (Helms, 1985; Phinney, 1989) considers whether minority individuals will identify with the culture of origin or the host culture. Berry (1986) proposed that in a multicultural society minority groups are encouraged to maintain, be conﬁdent of, and be proud of their original group identity, yet still maintain a relationship with the host group, by engaging in

Hewstone et al.

293

intergroup contact, learning the host language, and developing an acceptance of the host society. Both the alternation and integration approaches assume a relationship between the minority individual and both cultures, but while for alternation the relationship with both groups is identity-based, the integration model argues that it will be most adaptive for minority groups to identify with the minority group and merely learn the behavior or culture of the majority group (La Fromboise et al., 1993). This approach is thought to be especially relevant for immigrants and asylum seekers who need to quickly become competent in their new society in order to succeed. The integration model implies intergroup contact (Allport, 1954) between the minority and host groups, and therefore may have positive implications for intergroup relations. The lack of identiﬁcation with the host group, however, may be problematic. If the minority group do not identify with the host group, they are unlikely to see, or be seen, by the host group as a partial ingroup member, or as having a superordinate common identity in addition to their individual group memberships. As a result, they are likely to be considered less attractive than true ingroup members who hold a similar identity, placing limits on the impact of integration for intergroup relations. Combining multiple categorization and immigration perspectives Research on multiple categorization theories and immigration may be usefully combined. Assimilation and biculturalism research takes place in the real-life context of immigrant communities. Given that assimilation and biculturalism are forms of social categorization, ﬁeld research with immigrants provides an important opportunity to see if common ingroup identity, dualidentity, and crossed categorization models work in realistic contexts. It must, however, be noted that although immigration research provides indirect support for these models, there is little direct evidence that biculturalism leads to more positive outgroup attitudes and, if so, whether this is due to either a crossed categorization or dual-identity process. Further research is necessary to make the most of a combined approach considering immigration research and social categorization models. The social categorization literature also broadens our understanding of biculturalism. While immigration research focuses on outcomes for the individual and the self-contained group (e.g., Schiller, 1987), social categorization models place the experiences of immigrants in a broader theoretical context, dealing with the very important issue of how the strategies of assimilation and biculturalism might be beneﬁcial for intergroup relations. Findings from the crossed categorization literature provide a critique of some of the assumed beneﬁts of biculturalism, particularly regarding the role of threat. Brewer, Ho, Lee, and Miller (1987) predicted that in low-threat situations, crossing two group dimensions would result in both partial ingroups being seen as positively as the double ingroup. In high-threat situations,

294

Themes and priorities

however, only the double ingroup will be seen positively, with anyone who is in the outgroup on one dimension being seen as a threat. This is particularly relevant in the context of relations between host societies and asylum seekers in certain locations. Asylum seekers entering Britain are, for example, seen as a threat to jobs and security by the host community, leading to considerable animosity. In this case, biculturalism might be rejected by the host community, with asylum seekers and refugees viewed as outgroup members even if they come to consider themselves as British. Nevertheless, Brewer et al.’s (1987) predictions lend support to the alternation model of biculturalism. Those who alternate between two cultures have been shown to be able to switch, demonstrating the appropriate behavior, values, and attitudes of the host society, and thus are able to interact with the host group with comfort and ease (Yamada & Singelis, 1999). As a result, interactions with the host group are likely to be low in anxiety, and feelings of threat are likely to be low on both sides. According to Brewer et al.’s (1987) predictions, this would result in bicultural individuals (partial ingroup members) being evaluated as positively as full ingroup members. Summary In the context of immigration, the most positive consequences for intergroup relations are likely to arise when minority group members adopt biculturalism or integration, approaches analogous to the dual-identity model and crossed categorization. However, despite the similarities between the strategies employed by immigrants and social-psychological theories of categorization, little research has investigated the correspondence between the two. Our understanding of both the concerns facing immigrants and theories of intergroup relations will beneﬁt from an amalgamation of these approaches.

Identity ﬂuxion Much of the research and theory presented in this volume has dealt with the implications of multiple identities as they interact to form cognitive representations and evaluations of both self and others within a given context. In the broad literature on categorization, little attention has been paid thus far to the important and potentially fruitful topic of when, how, and why identities change across time and contexts, and what the implications are of such changes. We have used the word ﬂuxion, which is etymologically similar to the origin of the word ﬂuid. They both share a Latin root, ﬂuere, which means, in essence, to ﬂow. In this section, we want to elaborate on the ﬂowing, ﬂuid nature of identities, which, as the etymology suggests, have parts that can move and change their relative positions, and that can yield to pressure. Within the context of multiple identities, we propose that each singular identity forms a piece of the overall composite within an individual, and that the

Hewstone et al.

295

dimensions we discuss below are some of those pressures that lead to one identity yielding to another in a given context, or across time. Some recent work deals with identity change in sectarian Northern Ireland (Cassidy & Trew, 1998, 2002, 2004), a fascinating context in which to study social categorization and intergroup relations. In it, the authors discuss how identity with groups has important implications for one’s self-deﬁnition, and for one’s relationships with others. Thus, Cassidy and Trew argue, change and stability within the society can help us to understand the seemingly intractable intergroup conﬂict that has aﬀected the region for generations. They discuss identity change with respect to both short- and long-term contexts. Situational variables can temporarily increase or decrease feelings of identiﬁcation with a group (Waddell & Cairns, 1986), but these changes do not typically endure. There are, by contrast, persistent, ubiquitous factors that aﬀect identiﬁcation processes in Northern Ireland. For example, Cairns (1982) has argued that as long as a person remains within Northern Ireland, religious identity is structurally imposed and inescapable. Cassidy and Trew (2004) address the question of whether the meaning attached to identity in Northern Ireland changes over time. They followed a cohort of ﬁrst-year university students to examine this question. The nature of the sample is notable in that most young people in Northern Ireland attend segregated primary and secondary schools at which they typically interact with only members of their own religion; the transition to university life is thus a transition to cultural and religious heterogeneity. This vital ﬁrst year, therefore, provides an ideal crucible for observing the processes of change. At two diﬀerent points in time, once in the ﬁrst month of their ﬁrst term at university and again ﬁve months later, Cassidy and Trew (2004) measured identity salience, interactional commitment, and aﬀective commitment with respect to various social groups: family, friends, student, nationality, and religion. Identity salience was operationalized as the frequency with which respondents mentioned a particular identity when ﬁrst meeting someone (see also Carter, this volume, for use of similar measures in work by Hofman, 1988). Interactional commitment (Stryker & Serpe, 1982) measured the percentage of relationships that would be lost as a result of no longer belonging to a particular group, and aﬀective commitment (Stryker & Serpe, 1994) assessed how close and committed individuals were to the relationships that they had within a group. Whereas overall change in the measured identity processes remained at low to moderate levels (indicating relative stability), Cassidy and Trew (2004) found that for family, student, and religious categories, aﬀective commitment played a role in predicting such changes. Speciﬁcally, the more close and committed relationships that the individual had within a given category at Time 1, the more salience increased between Time 1 and Time 2. What eﬀects can increased salience (however measured) have on other intergroup processes? Some of our recent data (Kenworthy, Hewstone, & Cairns, 2004) indicate that high levels of salience and identiﬁcation are

296

Themes and priorities

associated with less liking for, and more distrust of, the outgroup. High identiﬁers were less likely to have high-quality contact with members of the outgroup and, as a consequence, threat remained at high levels, preventing the development of positive emotional reactions, attitudes, and trust toward the outgroup. We propose that longitudinal and cohort-type studies should now be conducted in Northern Ireland and similar settings, with the speciﬁc aim to track changes in identiﬁcation with one’s religious ingroup and the causal eﬀects that such changes produce on trust toward outgroup members. In the following subsections we explore issues of identity ﬂuxion under the following headings: (1) group status; (2) identity salience; (3) choice, essentialism, and group membership; and (4) commitment to the group. Group status Across time, the groups to which one belongs can, and often do, shift with respect to the relative levels of status and power within the superordinate society. Whites in South Africa have, in recent memory, gone from being a powerful minority to being a relatively powerless minority. Likewise, black South Africans have gone from being the disenfranchised majority to being the empowered majority. Catholics are the minority within Northern Ireland, but the majority on the island of Ireland. Protestants are the majority within Northern Ireland, but the minority on the island as a whole. The demographics of Northern Ireland are changing as time progresses. Catholics have historically been around 40% of the population of Northern Ireland, but this percentage is increasing. What can social scientists predict regarding the identiﬁcation of Catholics and Protestants with their respective religious groups, given the prospective shift in majority and minority numerical status? Work by Prislin, Limbert, and Bauer (2000) may help to answer this question. They have shown that majority groups who lose their status and become minorities are then less stable, and less attractive to the group members. The same eﬀects are not observed for minority groups who become majorities. In the latter condition, a new majority status produces only slight positive eﬀects on intragroup processes. Prislin et al.’s cumulative ﬁndings suggest to researchers that identity processes should be examined and tracked across time. When one loses majority status, there is a concomitant loss of identity with that group, to some degree. Does that loss of identiﬁcation create a sort of psychological vacuum, which is ﬁlled by another group identiﬁcation? Will the group membership that replaces the demoted one be of a similar nature, or will it be the next most important one in the hierarchically ranked array of group identities? Identity salience The psychological salience of our own and others’ identities is a function of both internal and external factors. Within the array of identities that each

Hewstone et al.

297

individual possesses, some are subjectively more important than others, and this importance may be stable across time. Nevertheless, external situational factors bear on the degree to which identities become salient and important in interactions. Waddell and Cairns (1986) showed that Catholics and Protestants in Northern Ireland generally feel primarily Irish and British, respectively, but that diﬀerent kinds of situational cue will produce variations from those baseline ethnic tendencies. The range of deviation from baseline levels of identiﬁcation was, however, greater for Protestant than for Catholic respondents. That is, certain kinds of situation made Northern Irish Protestants feel more Irish than British, but under no circumstances were Northern Irish Catholics made to feel more British than Irish. This is perhaps evidence that Protestants in Northern Ireland have a more complex identity structure due to competing ethnopolitical allegiances. Catholics, on the other hand, are the minority group in Northern Ireland, and as such exhibit more intragroup solidarity with respect to their identity (Simon, Aufderheide, & Kampmeier, 2001). This type of research has clear implications for further studies on crossed categorization and dual-identity models of intergroup relations. Research has already shown that importance and salience can aﬀect evaluations of crossed categorization targets (e.g., Urada & Miller, 2000). Further research might examine the moderating and mediating processes underlying the eﬀects of salience. Choice, essentialism, and group membership Research by Castano, Yzerbyt, and Bourguignon (2003) and Yzerbyt, Estrada, Corneille, Seron, and Demoulin (2004) has shown that perceptions of essentialism can aﬀect many other processes, including judgments of ingroup homogeneity, group polarization, self-anchoring, and entitativity. Although essentialistic beliefs can be created and fostered in many groups that people join voluntarily, it is likely that essentialistic beliefs spring naturally and spontaneously from those groups that people do not choose, those that we are often born into, such as gender, nationality, and race. Thus, although, as mentioned, essentialistic beliefs may be created ad hoc in voluntary groups, such beliefs can probably be undermined or disconﬁrmed more easily as well. By contrast, essentialistic beliefs concerning the involuntary groups are likely to be more enduring and resistant to disconﬁrming information or experience. These are, of course, speculations, albeit theoretically based ones. We think that an interesting line of future research would examine the interactions among the changes in individuals’ identiﬁcation processes with respect to their various ingroups that are voluntary versus involuntary, and which are accompanied by varying degrees of essentialistic beliefs.

298

Themes and priorities

Commitment to the group The survival and growth of a group depends not on its inherent value, morality, or ideology. Rather, its survival and growth depends on the extent to which it has internal mechanisms that are designed to ensure its long-term propagation. In other words, if a group can recruit new members and retain current ones, it is likely to survive regardless of whether it is socially acceptable (e.g., political parties, book clubs, sports teams) or unacceptable (e.g., religious cults, terrorist groups, pedophilia networks). Commitment is one way of increasing retention in groups. When people voluntarily commit to perform certain behaviors in the service of a group that they belong to, they will like the group more, and identify with it more strongly (Ellemers, Kortekaas, & Ouwerkerk, 1999). This voluntary commitment increases the likelihood of similar actions in the future as well. This is because of the rewards, both tangible and intangible, that accompany and reinforce such behaviors, but also because of the desire of people to be consistent (e.g., Aronson, 1999; Cialdini, Trost, & Newsom, 1995) in their behaviors. How do these processes bear on the discussion of multiple categorization? Greater commitment that may be demanded by membership in one group is likely to be at the expense of commitments to other group memberships. For example, a person who is a devoted son or daughter may be required to take increasing care of an elderly parent as that parent ages. This is also true of people whose children are disabled or in need of special care. In these circumstances, one’s commitment to family costs one the ability to commit to other group memberships, such as employment, or social, church, or political groups. This means that if a person has been a devoted community leader for decades, for example, the new demands of family commitments may require her or him to withdraw, either partially or completely, from the community group and its respective demands. What are the resulting implications for intraindividual and intragroup identity processes? Can one retain one’s identiﬁcation with a group if participation is curtailed; if eﬀectively interactional commitment is minimized, or even nulliﬁed? In many cases, the demands of competing group memberships may produce ambivalence. This is especially the case when membership in one group entails membership in others. For instance, belonging to a religion with which one identiﬁes strongly sometimes entails the taking of political positions with which one may or may not be comfortable. In the elections of the year 2000, members of the Mormon Church in California were explicitly asked by their leaders to support the proposed ballot measure to ban gay marriage by going door-to-door and asking people to vote in favor of it. This meant that at least a small minority of Mormons felt compelled to participate in a political campaign with which they disagreed. Likewise, many Catholics may feel pressure by their church to take a conservative stance on abortion rights (i.e., pro-life). How does the conﬂict of possessing simultaneous antagonistic group memberships (e.g., gay Mormons, pro-choice Catholics, pro-Palestinian

Hewstone et al.

299

Israeli Jews, etc.) aﬀect those respective identities? Can one retain high levels of commitment to a group whose ideology may diﬀer from that of another group to which one belongs? How intense will the cognitive dissonance be? Again, this is an important area of research that has received little attention. Perhaps the methodology of crossed categorization research, in particular, may inform us about the cognitive and aﬀective demands of such incongruent dual identities. Summary Clearly, then, many factors inﬂuence the ways in which the subjective value of group memberships changes ﬂuctuates across time. Little to no research has addressed these issues, especially within the context of multiple identities. But given that membership of groups is often a long-term commitment, and aspects of identity vary dynamically, we hope such research will form a greater part of future studies on multiple categorization.

CONCLUSION Many of the chapters in this volume seek, as is appropriate, to apply their ideas to intergroup relations, including extreme intergroup conﬂicts, that occur outside the laboratory. Caution should, however, be exercised. As Turner and Reynolds (2001) emphasize, social conﬂict is more complex than ingroup bias (see also Hewstone et al., 2002). Social conﬂict cannot be equated with the outcome of just one psychological process; rather, it must be understood in terms of the interplay of many processes, as shaped by historical, social, economic, and political structures of society (see also Billig, 1976). This caveat notwithstanding, we believe that issues concerning multiple categorization are of far more than purely theoretical interest and value. Categorization is, sometimes, quite literally a matter of life and death. We end this chapter, and this volume, with one new concept for the ﬁeld, and one compelling story. The new concept, new to psychology at least, is Ubuntu, an African concept used by Archbishop Desmond Tutu in his writings on reconciliation, forgiveness, and theology. He writes of this concept that it is very diﬃcult to express in a Western language, but that “it speaks to the very essence of being human; it means my humanity is caught up, inextricably bound up, in theirs” (i.e., other people’s; Tutu, 1999, pp. 34–35). Others have said that the concept recognizes that people and groups form their identities in relation to one another (Battle, 1997). It is a concept that seems particularly relevant to a book on multiple categorization, replete with its notions of common ingroup or superordinate identity and crossed categorization. It reminds us, simultaneously, that groups only exist and attain meaning in relation to each other, and that at a higher level we share a category membership, humanity.

300

Themes and priorities

That this matters is shown in the following story from the Rwandan genocide of 1994 (from The Guardian, 13 March 1999). During an attack in Gisenyi a group of schoolgirls at a convent boarding school were roused from their sleep and ordered to separate themselves into Hutus and Tutsis. The students refused, declaring that they were simply Rwandans. For this they were beaten and shot indiscriminately. The point of this story is not to argue the line that categorization is the root of all evil, a view that we have earlier disavowed. The moral, rather, is that had the hundreds of thousands who participated in the genocide had available a meaningful form of cross-cutting or superordinate identity, perhaps 800,000 people would never have been killed in 100 days. Understanding the role of social categorization in such cases is necessary; using our knowledge to help prevent future massacres is imperative.

ACKNOWLEDGEMENT Preparation of this chapter was supported by an Economic and Social Research Council grant (R000239382) to R. J. Crisp and M. Hewstone.

REFERENCES Ablon, G. (1964). Relocated American Indians in the San Francisco Bay area: Social interaction and Indian identity. Human Organization, 23, 296–304. Allen, V. L., & Wilder, D. A. (1975). Categorisation, belief similarity, and intergroup discrimination. Journal of Personality and Social Psychology, 32, 971–977. Allen, V. L., Wilder, D. A., & Atkinson, M. L. (1983). Multiple group membership and social identity. In T. R. Sarbin & K. E. Scheibe (Eds.), Studies in social identity (pp. 92–115). New York: Praeger. Allport, G. W. (1954). The nature of prejudice. Reading, MA: Addison-Wesley. Aronson, E. (1999). Dissonance, hypocrisy, and the self-concept. In E. Harmon-Jones & J. Mills (Eds.), Cognitive dissonance: Progress on a pivotal theory in social psychology (pp. 103–126). Science Conference Series. Washington, DC: American Psychological Association. Barsalou, L. W. (1987). The instability of graded structure: Implications for the nature of concepts. In U. Neisser (Ed.), Concepts and conceptual development (pp. 101–140). New York: Cambridge University Press. Barsalou, L. W. (1989). Intraconcept similarity and its implications for interconcept similarity. In S. Vosniadou & A. Ortony (Eds.), Similarity and analogical reasoning (pp. 76–121). Cambridge: Cambridge University Press. Bar-Tal, D. (1990). Causes and consequences of deligitimization: Models of conﬂict and ethnocentrism. Journal of Social Issues, 46, 65–81. Barreto, M., & Ellemers, N. (2003). The eﬀects of being categorised: The interplay between internal and external social identities. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 14, pp. 139–170). Hove, UK: Psychology Press.

Hewstone et al.

301

Battle, M. J. (1997). Reconciliation: The Ubuntu theology of Desmond Tutu. Cleveland, OH: Pilgrim Press. Berry, J. W. (1980). Acculturation as varieties of adaptation. In A. Padilla (Ed.), Acculturation: Theory, models and some new ﬁndings (pp. 9–25). Colorado, CO: Westview Press. Berry, J. W. (1986). Multiculturalism and psychology in plural societies. In L. H. Ekstrand (Ed.), Ethnic minorities and immigrants in a cross-cultural perspective (pp. 37–51). Berwyn, NY: Swets North America. Berry, J. W. (1997). Immigration, acculturation, and adaptation. Applied Psychology: An International Review, 46, 5–68. Berry, J. W., & Annis, R. C. (1974). Acculturation stress: The role of ecology, culture and diﬀerentiation. Journal of Cross-cultural Psychology, 5, 382–406. Bhawuk, D. P., & Brislin, R. (1992). The measurement of intercultural sensitivity using the concepts of individualism and collectivism. International Journal of Intercultural Relations, 16, 413–436. Biernat, M., & Thompson, E. R. (2002). Shifting standards and contextual variation in stereotyping. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 12, pp. 103–137). London: Wiley International. Billig, M. (1976). Social psychology and intergroup relations. London: Academic Press. Birman, D. (1994). Acculturation and human diversity in a multicultural society. In E. J. Trickett & R. J. Watts (Eds.), Human diversity: Perspectives on people in context (pp. 261–284). The Jossey-Bass Social and Behavioral Science Series. San Francisco, CA: Jossey-Bass/Pfeiﬀer. Bodenhausen, G. V., & Macrae, C. N. (1998). Stereotype activation and inhibition. In R. S. Wyer, Jr. (Ed.), Stereotype activation and inhibition: Advances in social cognition (Vol. 11, pp. 1–52). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Brewer, M. B. (1991). The social self: On being the same and diﬀerent at the same time. Personality and Social Psychology Bulletin, 17, 475–482. Brewer, M. B. (1993). The role of distinctiveness in social identity and group behavior. In M. Hogg & D. Abrams (Eds.), Group motivation: Social psychological perspectives (pp. 1–16). London: Harvester Wheatsheaf. Brewer, M. B. (1999). The psychology of prejudice: Ingroup love or outgroup hate? Journal of Social Issues, 55, 429–444. Brewer, M. B. (2000a). Reducing prejudice through cross-categorization: Eﬀects of multiple social identities. In S. Oskamp (Ed.), Reducing prejudice and discrimination (pp. 165–184). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Brewer, M. B. (2000b). Superordinate goals versus superordinate identity as bases of intergroup cooperation. In D. Capozza & R. Brown (Eds.), Social identity processes: Trends in theory and research (pp. 117–132). Thousand Oaks, CA: Sage Publications. Brewer, M. B. (2001). Ingroup identiﬁcation and intergroup conﬂict: When does ingroup love become outgroup hate? In R. D. Ashmore, L. Jussim, & D. Wilder (Eds.), Social identity, intergroup conﬂict, and conﬂict resolution (pp. 17–41). New York: Oxford University Press. Brewer, M. B., & Campbell, D. T. (1976). Ethnocentrism and intergroup attitudes: East African evidence. New York: Halstead Press. Brewer, M. B., Dull, V., & Lui, L. (1981). Perceptions of the elderly: Stereotypes as prototypes. Journal of Personality and Social Psychology, 41, 656–670.

302

Themes and priorities

Brewer, M. B., & Gaertner, S. L. (2001). Toward reduction of prejudice: intergroup contact and social categorization. In R. Brown & S. L. Gaertner (Eds.), Blackwell handbook of social psychology: Intergroup processes (pp. 451–474). Oxford: Blackwell Publishers. Brewer, M. B., Ho, H., Lee, J., & Miller, M. (1987). Social identity and social distance among Hong Kong schoolchildren. Personality and Social Psychology Bulletin, 13, 156–165. Brown, R., & Hewstone, M. (2005). An integrative theory of intergroup contact. In M. Zanna (Ed.), Advances in experimental social psychology (Vol. 37, pp. 255–343). San Diego, CA: Academic Press. Buriel, R., Perez, W., De Ment, T. L., Chavez, D. V., & Moran, V. R. (1998). The relationship of language brokering to academic performance, biculturalism, and self-eﬃcacy among Latino adolescents. Hispanic Journal of Behavioral Sciences, 20, 283–297. Cairns, E. (1982). Intergroup conﬂict in Northern Ireland. In H. Tajfel (Ed.), Social identity and intergroup relations (pp. 277–298). Cambridge: Cambridge University Press. Cairns, E., & Mercer, G. W. (1984). Social identity in Northern Ireland. Human Relations, 37, 1095–1102. Carroll, L. (1865). Alice’s Adventures in Wonderland. London: Macmillan. Cassidy, C., & Trew, K. (1998). Identities in Northern Ireland: A multidimensional approach. Journal of Social Issues, 54, 725–740. Cassidy, C., & Trew, K. (2002). Assessing identity change: A longitudinal study of the transition from school to college. Group Processes and Intergroup Relations, 4, 49–60. Cassidy, C., & Trew, K. (2004). Identity change in Northern Ireland: A longitudinal study of students’ transition to university. Journal of Social Issues, 60, 523–540. Castano, E., Yzerbyt, V. Y., & Bourguignon, D. (2003). We are one and I like it: The impact of group entitativity on group identiﬁcation. European Journal of Social Psychology, 33, 735–754. Cialdini, R. E., Trost, M. R., & Newsom, J. T. (1995). Preference for consistency: The development of a valid measure and the discovery of surprising behavioral implications. Journal of Personality and Social Psychology, 69, 318–328. Crisp, R. J., & Beck, S. R. (2005). Reducing intergroup bias: The moderating role of ingroup identiﬁcation. Group Processes and Intergroup Relations, 8, 173–185. Crisp, R. J., Ensari, N., Hewstone, M., & Miller, N. (2002). A dual-route model of crossed categorization eﬀects. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 13, pp. 35–74). Hove, UK & Philadelphia, PA: Psychology Press. Crisp, R. J., & Hewstone, M. (1999). Diﬀerential evaluation of crossed category groups: Patterns, processes, and reducing intergroup bias. Group Processes and Intergroup Relations, 2, 307–333. Crisp, R. J., & Hewstone, M. (2000a). Crossed categorization and intergroup bias: The moderating role of intergroup and aﬀective context. Journal of Experimental Social Psychology, 36, 357–383. Crisp, R. J., & Hewstone, M. (2000b). Multiple categorization and social identity. In D. Capozza & R. Brown (Eds.), Social identity theory: Trends in theory and research (pp. 149–166). Beverly Hills, CA: Sage Publications. Crisp, R. J., & Hewstone, M. (2001). Multiple categorization and implicit intergroup

Hewstone et al.

303

bias: Diﬀerential category dominance and the positive-negative asymmetry eﬀect. European Journal of Social Psychology, 31, 45–62. Crisp, R. J., Hewstone, M., & Cairns, E. (2001). Multiple identities in Northern Ireland: Hierarchical ordering in the representation of group membership. British Journal of Social Psychology, 40, 501–514. Crisp, R. J., Stone, C. H., & Hall, N. R. (2006). Recategorization and subgroup identiﬁcation: Predicting and preventing threats from common ingroups. Personality and Social Psychology Bulletin, 32, 230–243. Daniel, G. R. (2002). More than black? Multiracial identity and the new racial order. Philadelphia: Temple University Press. Deaux, K. (1996). Social identiﬁcation. In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology: Handbook of basic principles. New York: The Guilford Press. Deschamps, J.-C., & Doise, W. (1978). Crossed category membership in intergroup relations. In H. Tajfel (Ed.), Diﬀerentiation between social groups (pp. 141–158). Cambridge: Cambridge University Press. Doise, W. (1978). Groups and individuals: Explanations in social psychology. Cambridge: Cambridge University Press. Dovidio, J., Gaertner, S. L., & Loux, S. (2000). Subjective experiences and intergroup relations: The role of positive aﬀect. In H. Bless & J. P. Forgas (Eds.), The message within: The role of subjective experience in social cognition and behaviour (pp. 340–371). Philadelphia: Psychology Press. Dovidio, J. F., Gaertner, S. L., Isen, A. M., & Lowrance, R. (1995). Group representations and intergroup bias: Positive aﬀect, similarity, and group size. Personality and Social Psychology Bulletin, 21, 856–865. Dovidio, J. F., Gaertner, S. L., & Validzic, A. (1998). Intergroup bias, status, diﬀerentiation, and common ingroup identity. Journal of Personality and Social Psychology, 75, 109–120. Dovidio, J. F., Major, B., & Crocker, J. (2000). Stigma: Introduction and overview. In T. F. Heatherton, R. E. Kleck, M. R. Hebl, & J. G. Hull (Eds.), Stigma: Social psychological perspectives (pp. 1–28). New York: The Guilford Press. Duckitt, J. (1992). The social psychology of prejudice. New York: Praeger. Eggins, R. A., Haslam, S. A., & Reynolds, K. J. (2002). Social identity and negotiation: Subgroup representation and superordinate consensus. Personality and Social Psychology Bulletin, 28, 887–899. Ellemers, N., Kortekaas, P., & Ouwerkerk, J. (1999). Self-categorisation, commitment to the group and group self-esteem as related but distinct aspects of social identity. European Journal of Social Psychology, 29, 371–389. Ensari, N., & Miller, N. (2001). Decategorization and the reduction of bias in the crossed categorization paradigm. European Journal of Social Psychology, 31, 193–216. Erikson, E. (1968). Identity: Youth and crisis. London: Faber & Faber. Esses, V. M., Haddock, G., & Zanna, M. P. (1993). Values, stereotypes, and emotions as determinants of intergroup attitudes. In D. M. Mackie & D. L. Hamilton (Eds.), Aﬀect, cognition, and stereotyping: Interactive processes in group perception (pp. 137–166). San Diego, CA: Academic Press. Esses, V. M., Jackson, L. M., & Armstrong, T. L. (1998). Intergroup competition and attitudes toward immigrants and immigration: An instrumental model of group conﬂict. Journal of Social Issues, 54, 699–724.

304

Themes and priorities

Eurich-Fulcer, R., & Schoﬁeld, J. W. (1995). Correlated versus uncorrelated social categorizations: The eﬀect on intergroup bias. Personality and Social Psychology Bulletin, 21, 149–159. Evans-Pritchard, E. (1939). Nuer time reckoning. Africa, 12, 189–216. Feliciano, C. (2001). The beneﬁts of biculturalism: Exposure to immigrant culture and dropping out of school among Asian and Latino youths. Social Science Quarterly, 82, 865–878. Gaertner, S., & Dovidio, J. (2000). Reducing intergroup bias: The common ingroup identity model. Hove, UK: Psychology Press. Gaertner, S. L., Mann, J., Murrell, A., & Dovidio, J. F. (1989). Reducing intergroup bias: The beneﬁts of recategorization. Journal of Personality and Social Psychology, 57, 239–249. Gaertner, S. L., Mann, J. A., Dovidio, J. F., Murrell, A. J., & Pomare, M. (1990). How does cooperation reduce intergroup bias? Journal of Personality and Social Psychology, 59, 692–704. Gaertner, S. L., Rust, M. C., Dovidio, J. F., Bachman, B. A., & Anastasio, P. A. (1994). The contact hypothesis: The role of common ingroup identity on reducing intergroup bias. Small Group Research, 25, 224–249. Gaertner, S. L., Rust, M. C., Dovidio, J. F., Bachman, B. A., & Anastasio, P. A. (1996). The contact hypothesis: The role of a common ingroup identity on reducing intergroup bias among majority and minority group members. In J. L. Nye & A. M. Brower (Eds.), What’s social about social cognition? (pp. 230–360). Newbury Park, CA: Sage Publications. Geertz, C. (1973). The interpretation of cultures. New York: Basic Books. Gordon, M. M. (1964). Assimilation in American life. New York: Oxford University Press. Greenberg, J., Simon, L., Pyszczynski, T., Solomon, S., & Chatel, D. (1992). Terror management and tolerance: Does mortality salience always intensify negative reactions to others who threaten one’s worldview? Journal of Personality and Social Psychology, 63, 212–220. Hafer, C. L., & Olson, J. M. (2003). An analysis of empirical research on the scope of justice. Personality and Social Psychology Review, 7, 311–323. Haslam, N., Rothschild, L., & Ernst, D. (2000). Essentialist beliefs about social categories. British Journal of Social Psychology, 39, 113–127. Hastie, R., Schroeder, C., & Weber, R. (1990). Creating complex social conjunction categories from simple categories. Bulletin of the Psychonomic Society, 28, 242–247. Helms, J. (1985). Toward a theoretical explanation of the eﬀects of race on counseling: A black and white model. The Counseling Psychologist, 12, 153–165. Hewstone, M. (1996). Contact and categorization: Social psychological interventions to change intergroup relations. In N. Macrae, C. Stangor, & M. Hewstone (Eds.), Stereotypes and stereotyping (pp. 323–368). New York: The Guildford Press. Hewstone, M., Islam, M. R., & Judd, C. M. (1993). Models of crossed categorization and intergroup relations. Journal of Personality and Social Psychology, 64, 779–793. Hewstone, M., Martin, R., Hammer-Hewstone, C. M., Crisp, R. J., & Voci, A. (2001). Majority–minority relations in organizations: Challenges and opportunities. In M. A. Hogg & D. J. Terry (Eds.), Social identity processes in organizational contexts (pp. 67–86). Philadelphia, PA: Psychology Press. Hewstone, M., Rubin, M., & Willis, H. (2002). Intergroup bias. Annual Review of Psychology, 53, 575–604.

Hewstone et al.

305

Hirschfeld, L. (1996). Race in the making: Cognition, culture, and the child’s construction of human kinds. Cambridge, MA: The MIT Press. Hofman, J. E. (1988). Social identity and intergroup conﬂict: An Israeli view. In W. Stroebe, A. W. Kruglanski, D. Bar-Tal, & M. Hewstone (Eds.), The social psychology of intergroup conﬂict: Theory, research and applications (pp. 89–99). Berlin: Springer-Verlag. Hogg, M. A., & Abrams, D. (1993). Towards a single-process uncertainty-reduction model of social motivation in groups. In M. A. Hogg & D. Abrams (Eds.), Group motivation: Social psychological perspectives (pp. 173–90). London: Harvester Wheatsheaf. Hornsey, M. J., & Hogg, M. A. (1999). Subgroup diﬀerentiation as a response to an overly-inclusive group: A test of optimal distinctiveness theory. European Journal of Social Psychology, 29, 543–550. Hornsey, M. J., & Hogg, M. A. (2000). Subgroup relations: A comparison of mutual intergroup diﬀerentiation and common ingroup identity models of prejudice reduction. Personality and Social Psychology Bulletin, 26, 242–256. Horowitz, D. (1985). Ethnic groups in conﬂict. Berkeley, CA: University of California Press. Jetten, J., Spears, R., & Manstead, A. S. R. (1996). Intergroup norms and intergroup discrimination: Distinctive self-categorization and social identity eﬀects. Journal of Personality and Social Psychology, 71, 1222–1233. Jetten, J., Spears, R., & Manstead, A. S. R. (1997). Distinctiveness threat and prototypicality: Combined eﬀects on intergroup discrimination and collective self-esteem. European Journal of Social Psychology, 27, 635–657. Jetten, J., Spears, R., & Manstead, A. S. R. (1998). Dimensions of distinctiveness: Group variability makes a diﬀerence to diﬀerentiation. Journal of Personality and Social Psychology, 74, 1481–1492. Johnson, R. (1976). The concept of the “marginal man”: A reﬁnement of the term. Australian and New Zealand Journal of Science, 12, 145–147. Kenworthy, J., Hewstone, J., & Cairns, E. (2004). Unpublished data, University of Oxford, UK. Kunda, Z., Miller, D., & Claire, T. (1990). Combining social concepts: The role of causal reasoning. Cognitive Science, 14, 551–577. 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, 720–734. Kunda, Z., & Thagard, P. (1996). Forming impressions from stereotypes, traits, and behaviors: A parallel-constraint-satisfaction theory. Psychological Review, 103, 284–308. LaFromboise, T., Coleman, H. L., & Gerton, J. (1993). Psychological impact of biculturalism: Evidence and theory. Psychological Bulletin, 114, 395–412. LeVine, R. A., & Campbell, D. T. (1972). Ethnocentrism: Theories of conﬂict, ethnic attitudes and group behavior. New York: Wiley International. Leyens, J. Ph., Paladino, P. M., Rodriguez, R. T., Vaes, J., Demoulin, S., Rodriguez, A. P., & Gaunt, R. (2000). The emotional side of prejudice: The attribution of secondary emotions to ingroups and outgroups. Personality and Social Psychology Review, 4, 186–197. Lorenzo-Hernandez, J. (1998). How social categorization may inform the study of Hispanic immigration. Hispanic Journal of Behavioral Sciences, 20, 39–59.

306

Themes and priorities

Mackie, D. M., & Smith, E. R. (Eds.) (2002). From prejudice to intergroup emotions: Diﬀerentiated reactions to social groups. New York & Hove, UK: Psychology Press. Macrae, C. N., & Bodenhausen, G. V. (2001). Social cognition: Categorical person perception. British Journal of Psychology, 92, 239–255. Macrae, C. N., Bodenhausen, G. V., & Milne, A. B. (1995). The dissection of selection in social perception: Inhibitory processes in social stereotyping. Journal of Personality and Social Psychology, 69, 397–407. Marcus-Newhall, A., Miller, N., Holtz, R., & Brewer, M. B. (1993). Cross-cutting category membership with role assignment: A means of reducing intergroup bias. British Journal of Social Psychology, 32, 125–146. McGarry, J., & O’Leary, B. (1993). Introduction: The macro-political regulation of ethnic conﬂict. In J. McGarry & B. O’Leary (Eds.), The politics of ethnic conﬂict regulation: Case studies of protracted ethnic conﬂicts (pp. 1–40). London & New York: Routledge. McGarry, J., & O’Leary, B. (1995). Explaining Northern Ireland: Broken images. Oxford & Cambridge: Blackwell. McGarty, C. (1999). Categorization in social psychology. London: Sage Publications. Migdal, M., Hewstone, M., & Mullen, B. (1998). The eﬀects of crossed categorization on intergroup evaluations: A meta-analysis. British Journal of Social Psychology, 37, 303–324. Miller, D. T., & Prentice, D. A. (1999). Some consequences of a belief in group essence: The category divide hypothesis. In D. A. Prentice & D. T. Miller (Eds.), Cultural divides: Understanding and overcoming group conﬂict (pp. 213–238). New York: Russell Sage Foundation. Mullen, B., Migdal, M. J., & Hewstone, M. (2001). Crossed categorization vs simple categorization and intergroup evaluations: A meta-analysis. European Journal of Social Psychology, 31, 721–736. Mummendey, A., & Wenzel, M. (1999). Social discrimination and tolerance in intergroup relations: Reactions to intergroup diﬀerence. Personality and Social Psychology Review, 3, 158–174. Murphy, R. F. (1957). Intergroup hostility and social cohesion. American Anthropologist, 59, 1018–1035. Oakes, P. J. (1987). Salience and categorization. In J. C. Turner, M. A. Hogg, P. J. Oakes, S. D. Reicher, & M. S. Wetherell, Rediscovering the social group: A self-categorization theory. Oxford: Blackwell Publishing. Oakes, P. J. (1996). Stereotyping, individuation, and accuracy in intergroup perception. International Journal of Psychology, 31, 1637–1637. Oakes, P. (2001). The root of all evil in intergroup relations? Unearthing the categorization process. In R. Brown & S. L. Gaertner (Eds.), Blackwell handbook of social psychology: Intergroup processes (pp. 3–21). Malden, MA & Oxford, UK: Blackwell. Opotow, S. (1990). Moral exclusion and injustice: An introduction. Journal of Social Issues, 46, 1–20. Opotow, S. (1995). Drawing the line: Social categorization, moral exclusion, and the scope of justice. In B. B. Bunker & J. Z. Rubin (Eds.), Conﬂict, cooperation, and justice: Essays inspired by the work of Morton Deutsch (pp. 347–369). San Francisco, CA: Jossey-Bass. Park, R. E. (1914). Racial assimilation in secondary groups. American Journal of Sociology, 19, 6–23. Pendry, L. F., & Macrae, C. N. (1996). What the disinterested perceiver overlooks:

Hewstone et al.

307

Goal-directed social categorization. Personality and Social Psychology Bulletin, 22, 249–256. Pettigrew, T. F. (1998a). Intergroup contact theory. Annual Review of Psychology, 49, 65–85. Pettigrew, T. F. (1998b). Reactions toward the new minorities of western Europe. Annual Review of Sociology, 24, 77–103. Phinney, J. S. (1989). Stages of ethnic identity development in minority group adolescents. Journal of Early Adolescence, 9, 34–49. Phinney, J. S., Ferguson, D. L., & Tate, J. D. (1997). Intergroup attitudes among ethnic minority adolescents: A causal model. Child Development, 68, 955–969. Phinney, J. S., Lochner, B. T., & Murphy, R. (1990). Ethnic identity development and psychological adjustment in adolescence. In A. R. Stiﬀman & L. E. Davis (Eds.), Ethnic issues in adolescent mental health (pp. 53–72). Thousand Oaks, CA: Sage. Portes, A., & Rumbaut, R. G. (1996). Immigrant America: A portrait (2nd ed.). Berkeley, CA: University of California Press. Prislin, R., Limbert, W. M., & Bauer, E. (2000). From majority to minority and vice versa: The asymmetrical eﬀects of losing and gaining majority position within a group. Journal of Personality and Social Psychology, 79, 385–397. Ramirez, M. (1983). Psychology of the Americas: Mestizo perspectives on personality and mental health. New York: Pergamon. Reed, J. S. (1982). One south: An ethnic approach to regional culture. Baton Rouge, LA: Louisiana State University Press. Roccas, S., & Brewer, M. B. (2002). Social identity complexity. Personality and Social Psychology Review, 6, 88–106. Rothbart, M., & Taylor, M. (1992). Category labels and social reality: Do we view social categories as natural kinds? In G. Semin & K. Fiedler (Eds.), Language, interaction and social cognition (pp. 11–36). London: Sage Publications. Ruiz, R. (1981). Cultural and historical perspectives in counselling Hispanics. In D. Sue (Ed.), Counselling the culturally diﬀerent (pp. 186–215). New York: Wiley International. Schiller, P. M. (1987). Biculturalism and psychosocial adjustment among Native American university students. Dissertation Abstracts International, 48, 1542A (University Microﬁlms No. DA8720632). Schwarzer, R., Bowler, R., & Rauch, S. (1985). Psychological indicators of acculturation: Self-esteem, racial tension and inter-ethnic contact. In L. Ekstrand (Ed.), Ethnic minorities and immigrants in a cross-cultural perspective (pp. 211–229). Lisse, The Netherlands: Swets and Zeitlinger. Shih, M., Pittinsky, T. L., & Ambady, N. (1999). Stereotype susceptibility: Identity salience and shifts in quantitative performance. Psychological Science, 10, 80–83. Sidanius, J., Devereux, E., & Pratto, F. (1992). A comparison of symbolic racism theory and social dominance theory: Explanations for racial policy attitudes. Journal of Social Psychology, 132, 377–395. Sidanius, J., & Pratto, F. (1999). Social dominance: An intergroup theory of social hierarchy and oppression. New York: Cambridge University Press. Simmel, G. (1950). The sociology of Georg Simmel (Kurt Wolﬀ, Trans.). New York: Free Press (original work published 1908). Simon, B., Aufderheide, B., & Kampmeier, C. (2001). The social psychology of

308

Themes and priorities

minority–majority relations. In R. Brown & S. L. Gaertner (Eds.), Blackwell handbook of social psychology: Intergroup processes (pp. 303–323). Malden, MA: Blackwell Publishing. Sinclair, L., & Kunda, Z. (1999). Reactions to a black professional: Motivated inhibition and activation of conﬂicting stereotypes. Journal of Personality and Social Psychology, 77, 885–904. Singh, R., Yeoh, B. S. E., Lim, D. I., & Lim, K. K. (1997). Cross categorization eﬀects in intergroup discrimination: Adding versus averaging. British Journal of Social Psychology, 36, 121–159. Smith, E. R. (1998). Mental representation and memory. In D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), Handbook of social psychology (4th ed., Vol. 1, pp. 391–445). New York: McGraw-Hill. Smith, E. R., & DeCoster, J. (1998). Knowledge acquisition, accessibility, and use in person perception and stereotyping: Simulation with a recurrent connectionist network. Journal of Personality and Social Psychology, 74, 21–35. 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, 207–218. Staub, E. (2001). Ethnopolitical and other group violence: Origins and prevention. In D. Chirot & M. E. P. Seligman (Eds.), Ethnopolitical warfare: Causes, consequences, and possible solutions (pp. 289–304). Washington, DC: American Psychological Association. Stephan, W. G., & Stephan, C. (1985). Intergroup anxiety. Journal of Social Issues, 41, 157–175. Stephan, W. G., & Stephan, C. W. (2000). An integrated threat theory of prejudice. In S. Oskamp (Ed.), Reducing prejudice and discrimination (pp. 23–46). Mahwah, NJ: Lawrence Erlbaum Associates, Inc. Stryker, S., & Serpe, R. T. (1982). Commitment, identity salience and role behavior: Theory and research example. In W. Ickes & E. S. Knowles (Eds.), Personality, roles, and social behavior (pp. 199–218). New York: Springer-Verlag. Stryker, S., & Serpe, R. T. (1994). Identity salience and psychological centrality: Equivalent, overlapping, or complementary concepts. Social Psychology Quarterly, 51, 16–35. Stryker, A. & Statham, A. (1985). Symbolic interaction and role theory. In G. Lindsey & L. Aronson (Eds.), Handbook of social psychology (3rd ed., pp. 311–378). New York: McGraw-Hill. Szapocznik, J., & Kurtines, W. (1980). Acculturation, biculturalism and adjustment among Cuban-Americans. In A. M. Padilla (Ed.), Psychological dimensions on the acculturation process: Theory, models and some new ﬁndings (pp. 139–159). Boulder, CO: Westview Press. Szapocznik, J., Kurtines, W., & Fernandez, T. (1980). Bicultural involvement and adjustment in Hispanic-American youths. International Journal of Intercultural Relations, 4, 353–365. Tajfel, H. (1978). Interindividual and intergroup behaviour. In H. Tajfel (Ed.), Differentiation between social groups: Studies in the social psychology of intergroup relations (pp. 27–60). London: Academic Press. Tajfel, H., Flament, C., Billig, M., & Bundy, R. P. (1971). Social categorization and intergroup behavior. European Journal of Social Psychology, 1, 149–177. Tajfel, H., & Turner, J. (1979). An integrative theory of intergroup conﬂict. In

Hewstone et al.

309

W. G. Austin & S. Worchel (Eds.), The social psychology of intergroup relations (pp. 33–48). Monterey, CA: Brooks/Cole. Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Rediscovering the social group: A self-categorization theory. Oxford: Blackwell Publishing. Turner, J. C., & Reynolds, K. J. (2001). The social identity perspective in intergroup relations: Theories, themes, and controversies. In R. Brown & S. L. Gaertner (Eds.), Blackwell handbook of social psychology: Intergroup processes (pp. 133–152). Malden, MA & Oxford, UK: Blackwell. Tutu, D. (1999). No future without forgiveness. New York: Doubleday. Urada, D. I., & Miller, N. (2000). The impact of positive mood and category importance on crossed categorization eﬀects. Journal of Personality and Social Psychology, 78, 417–433. Urban, L. M., & Miller, N. M. (1998). A meta-analysis of crossed categorization eﬀects. Journal of Personality and Social Psychology, 74, 894–908. van Leeuwen, E., van Knippenberg, D., & Ellemers, N. (2003). Continuing and changing group identities: The eﬀects of merging on social identiﬁcation and ingroup bias. Personality and Social Psychology Bulletin, 29, 679–690. van Oudenhoven, J. P., Prins, K. S., & Buunk, B. P. (1998). Attitudes of minority and majority members towards adaptation of immigrants. European Journal of Social Psychology, 28, 995–1013. van Rijswijk, W., & Ellemers, N. (2002). Context eﬀects on the application of stereotype content to multiple categorizable targets. Personality and Social Psychology Bulletin, 28, 90–101. van Twuyver, M., & van Knippenberg, A. (1999). Social categorisation as a function of relative group size. British Journal of Social Psychology, 38, 135–156. Vanman, E. J., Paul, B. Y., Ito, T. A., & Miller, N. (1997). The modern face of prejudice and structural features that moderate the eﬀect of cooperation on aﬀect. Journal of Personality and Social Psychology, 73, 941–959. Verkuyten, M. (2005a). The social psychology of ethnic identity. Hove, UK: Psychology Press. Verkuyten, M. (2005b). Ethnic group identiﬁcation, and group evaluations among minority and majority groups: Testing the multiculturalism hypothesis. Journal of Personality and Social Psychology, 88, 121–138. Vescio, T. K., Judd, C. M., & Kwan, V. S. Y. (2004). The crossed-categorization hypothesis: Evidence of reductions in the strength of categorization, but not intergroup bias. Journal of Experimental Social Psychology, 40, 478–496. Vogt, E. Z. (1957). The acculturation of American Indians. Annals of the American Academy of Political and Social Science, 311, 137–146. Waddell, N., & Cairns, E. (1986). Situational perspective on social identity in Northern Ireland. British Journal of Social Psychology, 25, 25–31. Wallace, M. D. (1973). Alliance polarization, cross-cutting, and international war, 1815–1964. Journal of Conﬂict Resolution, 17, 575–604. Weber, U., Mummendey, A., & Waldzus, S. (2002). Perceived legitimacy of intergroup status diﬀerences: Its predictions by relative ingroup prototypicality. European Journal of Social Psychology, 32, 449–470. Wenzel, M. (2004). A social categorisation approach to justice. In W. Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 15, pp. 219–258). Hove, UK: Psychology Press.

310

Themes and priorities

Wilder, D. A. (1986). Social categorization: Implications for creation and reduction of intergroup bias. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 19, pp. 293–355). New York: Academic Press. Wilder, D., & Simon, A. F. (2001). Aﬀect as a cause of intergroup bias. In R. Brown & S. L. Gaertner (Eds.), Blackwell handbook of social psychology: Intergroup processes (pp. 153–72). Malden, MA & Oxford, UK: Blackwell. Wolsko, C., Park, B., Judd, C., & Wittenbrink, B. (2000). Framing interethnic ideology: Eﬀects of multicultural and color-blind perspectives on judgments of groups and individuals. Journal of Personality and Social Psychology, 78, 635–654. Yamada, A. M., & Singelis, T. M. (1999). Biculturalism and self-construal. International Journal of Intercultural Relations, 23, 697–709. Yzerbyt, V. Y., Estrada, C., Corneille, O., Seron, E., & Demoulin, S. (2004). Subjective essentialism in action: Self-anchoring and social control as consequences of fundamental social divides. In V. Y. Yzerbyt, C. M. Judd, & O. Corneille (Eds.), The psychology of group perception: Perceived variability, entitativity, and essentialism (pp. 101–126). Hove, UK: Psychology Press. Yzerbyt, V. Y., Rocher, S. J., & Schadron, G. (1997). Stereotypes as explanations: A subjective essentialistic view of group perception. In R. Spears, P. Oakes, N. Ellemers, & A. Haslam (Eds.), The psychology of stereotyping and group life (pp. 20–50). Oxford: Basil Blackwell. Zagefka, H., & Brown, R. (2002). The relationship between acculturation strategies, relative ﬁt and intergroup relations: immigrant–majority relations in Germany. European Journal of Social Psychology, 32, 171–188.

Author index

Abbey, A. 193, 194 Abele, A. E. 167 Abelson, R. P. 52 Ablon, G. 290 Abrams, D. 11, 66, 94, 95, 112, 113, 115, 116, 123, 211, 283 Adams, M. 219 Agnew, J. 257 Alicke, M. D. 179 Alipuria, L. 211, 212, 213, 224, 225, 226, 228, 229, 230, 233 Allen, V. L. 271, 278 Allport, G. W. xiv, 3, 67, 69, 245, 246, 273, 293 Ambady, N. 278 Anastasio, P. A. 6, 11, 14, 31, 32, 68, 73, 79, 82, 83, 92, 100, 104, 119, 120, 245, 257, 283, 284 Anderson, N. H. 9 Annis, R. C. 289 Archer, S. 217, 218 Arcuri, L. 10, 59, 93, 145, 165 Armstrong, K. 4 Armstrong, T. L. 286 Aron, A. 104, 121 Aronson, E. 298 Asch, S. E. 31 Atkinson, D. 219, 220 Atkinson, M. L. 278 Atran, S. 35 Aufderheide, B. 297 Axsom, D. 74, 92 Azzi, A. E. 257, 258, 259 Bachman, B. A. 6, 11, 14, 31, 68, 70, 73, 79, 82, 83, 92, 100, 104, 119, 120, 245, 257, 283, 284 Bandlamudi, L. 223 Banker, B. 70

Banker, B. S. 79, 104 Banton, M. 247 Bar-Tal, D. 272 Bargh, J. A. 12 Barreto, M. 275 Barsalou, L. 29, 35, 53, 54, 56 Barsalou, L. W. 280 Battle, M. J. 299 Bauer, E. 296 Baumeister, R. F. 115, 191, 194 Beck, S. R. 8, 98, 99, 287 Beike, D. R. 121 Bellour, F. 118 Berry, J. W. 15, 33, 229, 230, 284, 289, 290, 292 Best, D. L. 190 Bettencourt, B. A. 78 Bhawuk, D. P. 291 Biernat, M. 139, 145, 277 Billig, M. 13, 91, 139, 271, 299 Billig, M. G. 37 Birman, D. 291, 292 Blackwood, L. 120 Bless, H. 54, 166 Bliuc, A.-M. 40, 41 Bloom, Z. 97, 98, 101 Bodenhausen, G. V. 5, 8, 54, 169, 170, 275, 277 Borden, R. J. 124 Borgida, E. 179 Bourassa, R. 261 Bourguignon, D. 15, 297 Bourhis, R. Y. 126 Bower, G. H. 166, 170, 171, 172, 174, 177, 179 Bowler, R. 292 Bown, N. J. 123 Boyes-Braem, P. 28 Branscombe, N. R. 95, 113

312

Author index

Brassine, J. 240 Brazil, D. M. 66 Breinlinger, S. 42 Breitenbecher, D. L. 179 Brekke, N. 179 Breton, R. 242 Brewer, D. 170 Brewer, M. B. 4, 6, 7, 8, 11, 13, 14, 15, 31, 39, 65, 67, 68, 74, 77, 78, 79, 81, 91, 92, 93, 95, 102, 104, 113, 116, 117, 118, 119, 121, 139, 140, 152, 180, 181, 244, 245, 246, 247, 250, 256, 257, 271, 272, 273, 278, 281, 282, 283, 285, 286, 287, 293, 294 Brislin, R. 291 Broverman, D. M. 190, 204 Broverman, I. K. 190, 204 Brown, R. 13, 32, 33, 70, 71, 91, 95, 100, 163, 192, 195, 232, 245, 256, 279, 284 Brown, R. J. 11, 70, 71, 72, 76, 93, 94, 95, 100, 103, 105, 140, 143, 144, 156, 180, 256 Bruner, J. S. 27 Brunsma, D. 214, 225, 226, 227 Bundy, R. F. 13, 37, 91, 139, 271 Burgess, M. 252, 253, 254 Buriel, R. 292 Burton, J. W. 247, 249, 250 Butemeyer, J. 66 Buttram, R. T. 58, 60 Buunk, B. 81 Buunk, B. P. 283, 284 Byrne, D. 194 Byrne, S. 242 Cacioppo, J. T. 12 Cadinu, M. R. 12, 92 Cairns, E. 5, 8, 258, 273, 274, 281, 282, 295, 297 Callan, V. J. 119, 121 Campbell, D. T. 5, 10, 13, 67, 91, 114, 161, 273, 281, 286 Canada 252 Canales, C. J. 12, 13, 168, 172, 173, 174, 177, 178, 183 Cantor, N. 80, 167, 169, 170, 177, 178 Carey, C. J. 119, 121 Carlson, M. 174, 183 Carlston, D. E. 166 Carroll, L. 289 Carter, N. 242, 248, 251, 253, 256 Cassidy, C. 295 Cassidy, R. C. 214 Castano, E. 15, 118, 297

Castelli, L. 59, 114 Cecil, C. L. 66 Cejka, M. A. 59, 60 Chaiken, S. 3 Chan, E. S. 202 Chatel, D. 286 Chavez, D. V. 292 Chiker, V. 94 Choo, I. 191 Choo, W. M. 191, 196, 197 Chou, E. 230, 231 Church, C. H. 254, 255 Cialdini, R. B. 124 Cialdini, R. E. 298 Cihangir, S. 118 Claire, T. 5, 9, 56, 276 Clark, A. 53, 55, 56 Clark, J. Rt. Hon. 240 Clark, M. S. 13, 167 Clark, R. P. 240 Clarkson, F. E. 190, 204 Clement, R. W. 92 Clore, G. 166 Coleman, H. 223, 230, 231 Coleman, H. L. 15, 290, 291, 292, 293 Coleman, J. 95, 113 Coleman, J. S. 250 Coley, J. D. 35 Collins, A. M. 50, 51 Colquhoun, A. 6 Commins, B. 140 Connor, W. 242, 243 Conversi, D. 243 Cook, S. W. 67 Corneille, O. 15, 297 Cornell, S. 214, 215 Coser, L. A. 5, 161 Cotting, D. 191, 192, 193 Crisp, R. J. 5, 6, 7, 8, 9, 10, 12, 13, 27, 37, 74, 90, 92, 93, 95, 96, 97, 98, 99, 100, 101, 102, 104, 119, 140, 161, 163, 168, 169, 171, 172, 178, 180, 182, 194, 203, 211, 216, 245, 257, 258, 272, 274, 277, 279, 281, 282, 284, 287 Crocker, J. 117, 284 Crosby, F. 117 Cross, W. 218, 220, 230, 231 Cuddy, A. J. C. 191, 196, 204 Dalal, A. K. 192, 193, 197, 198, 199, 200, 202 Daniel, G. R. 211, 213, 214, 224, 225, 226, 227, 233, 234, 274, 275 Darley, J. 124, 125, 127

Author index Das, J. P. 12 Daubman, K. 167, 169 Day, E. A. 66 De Baessa, Y. 222 De Ment, T. L. 292 de Nicholas, M. E. 124 Deaux, K. 191, 192, 193, 203, 273 DeCoster, J. 5, 56, 276 Delmartino, F. 240 Demoulin, S. 15, 272, 297 DePaulo, B. 165 DeRidder, M. 253 Deschamps, J.-C. 6, 10, 11, 27, 70, 71, 74, 90, 92, 93, 139, 140, 141, 144, 149, 156, 160, 189, 190, 203, 204, 257, 276, 291 Deutsch, M. 5 Devereux, E. 286 Devich-Navarro, M. 231, 232 Devine, P. G. 189 Devos, T. 163, 172, 174 Diehl, M. 93, 95, 152 Diekman, A. B. 190 Dijksterhuis, A. 8 Doise, W. 6, 10, 11, 13, 27, 74, 90, 91, 92, 93, 97, 139, 140, 141, 144, 149, 157, 160, 189, 190, 203, 204, 257, 271, 276, 291 Doosje, B. 94, 95, 218, 219 Doran, C. F. 240 Doughtie, E. B. 170 Doutrelepont, R. 252 Dovidio, J. F. 6, 10, 11, 12, 13, 14, 31, 32, 33, 66, 67, 68, 69, 70, 71, 73, 74, 75, 76, 79, 80, 81, 82, 83, 84, 90, 91, 92, 93, 95, 99, 100, 103, 104, 119, 120, 169, 170, 175, 245, 250, 252, 256, 257, 279, 283, 284, 285, 290 Duan, C. 4, 5, 8, 145, 277, 278 Dubé, L. 258 DuBois, W. E. B. 232 Duckitt, J. 288 Duﬀy, G. 247 Dull, V. 7, 8, 278 Dunton, B. C. 12 Dupont, S. 212 Eagly, A. H. 190, 191, 196, 197, 204 Edwards, K. 102 Eggins, R. A. 33, 40, 284 Ehrenberg, K. 5 Elazar, D. J. 253 Ellemers, N. 5, 55, 57, 94, 95, 96, 113, 116, 162, 218, 219, 275, 287, 298 Elster, J. 249

313

Enloe, C. H. 240 Ensari, N. 5, 6, 12, 13, 74, 90, 92, 119, 161, 163, 164, 168, 175, 176, 177, 179, 180, 181, 183, 282 Erikson, E. 281 Erikson, E. H. 212, 216, 217 Ernst, D. 43, 272 Espinosa, C. 212 Esses, V. M. 95, 286 Estrada, C. 15, 297 Etcoﬀ, N. 145 Etcoﬀ, N. L. 8, 165 Eurich-Fulcer, R. 102, 140, 282 Evans, N. 12 Evans-Pritchard, E. 273 Evans-Pritchard, E. E. 5 Eyck, F. G. 240 Falcone, H.-T. 52 Fazio, R. H. 12, 59, 60 Feather, N. T. 124 Feliciano, C. 290 Ferdman, B. 225 Ferguson, D. 220, 221 Ferguson, D. L. 279 Fernandez, T. 292 Ferris, F. 74, 92 Fhagen-Smith, P. 220 Fielding, K. S. 124, 127 Finkel, E. J. 66 Fisher, J. 165 Fiske, S. T. 3, 8, 65, 67, 145, 152, 165, 191, 193, 194, 196, 204 Flament, C. 13, 37, 91, 139, 271 Forgas, J. P. 167, 168, 171, 172, 174 Foster-Fishman, P. G. 118 Fraga, L. R. 253 Frazier, S. 66, 70, 79 Freeman, S. 124 Frensley, N. J. 247 Frijda, N. H. 172 Gaertner, L. 66, 191, 194 Gaertner, S. L. 6, 10, 11, 14, 31, 32, 33, 66, 67, 68, 69, 70, 71, 73, 74, 75, 76, 79, 80, 81, 82, 83, 84, 90, 91, 92, 93, 99, 100, 103, 104, 119, 120, 169, 170, 175, 245, 250, 252, 256, 257, 279, 281, 283, 284, 285, 290 Galinsky, A. 213, 228 Gallegos, P. 225 Gardner, W. 113 Gaunt, R. 272

314

Author index

Geertz, C. 273 Gendolla, G. H. E. 167 Gentner, D. 31 Gergen, K. 216 Gerton, J. 15, 223, 230, 231, 290, 291, 292, 293 Gibbs, J. 224 Gilbert, D. T. 53, 126 Glass, B. 4, 5, 8, 145, 277, 278 Glick, P. 191, 193, 194, 196, 204 Gluckman, M. 5 Goh, H. Y. 202, 203 Goldstone, R. L. 31 Goldwin, L. 165 Golisano, V. 166 González, R. 93, 100, 103, 105, 174, 232 Gonzalez-Doupe, P. 222 Gooden, M. P. 117 Gordon, M. M. 290 Gottlieb, G. 239 Government of Canada 122 Gray, W. D. 28 Green, J. D. 174 Green, T. F. 257 Greenberg, J. 115, 286 Grieco, E. M. 214 Griﬃn, D. 278 Grove, A. 256 Grove, K. 224 Gschneidinger, E. 170 Guerra, P. 73, 80, 92, 99, 100, 103, 104 Gurtman, M. B. 12, 13 Habermas, J. 251 Haddock, G. 286 Hafer, C. L. 272 Hagendoorn, L. 66, 140, 161, 196 Hains, S. C. 66, 125, 189, 194 Hall, N. R. 96, 97, 100, 101, 102, 104, 284, 287 Halpern, S. 255 Hamilton, D. L. 43, 114, 116, 189 Hamlet, M. A. 124 Hammer-Hewstone, C. M. 279 Hampton, J. A. 9 Hart, D. 66 Hartmann, D. 214, 215 Haslam, N. 272 Haslam, N. O. 43 Haslam, S. 57 Haslam, S. A. 3, 8, 26, 30, 33, 36, 40, 78, 91, 139, 162, 244, 284 Hastie, R. 9, 276 Hayes, B. K. 26, 162

Heisler, M. O. 254 Helms, J. 218, 220, 292 Helms, J. E. 215 Henderson-King, D. 94 Henderson-King, E. 94 Hendres, D. 40, 41 Henke, R. 140, 161, 196 Henry, S. 118 Henson, M. 123 Hepburn, C. 179 Herbst, K. C. 194 Hermans, H. 213 Hewstone, J. 295 Hewstone, M. 4, 5, 6, 7, 8, 10, 12, 13, 27, 32, 33, 37, 39, 71, 72, 74, 79, 81, 90, 92, 93, 95, 99, 100, 102, 115, 119, 121, 140, 141, 144, 152, 156, 161, 163, 168, 169, 171, 172, 175, 178, 180, 181, 182, 190, 194, 196, 203, 204, 211, 216, 229, 245, 250, 256, 257, 258, 272, 274, 277, 279, 280, 281, 282, 283, 285, 286, 288, 289 Higgins, E. T. 13, 27 Hinkle, S. 95 Hinsz, V. B. 118 Hinton, G. E. 55 Hintzman, D. L. 54 Hirschfeld, L. 272 Ho, H. 39, 74, 140, 181, 256, 293, 294 Ho, H. K. 4, 6, 11 Ho, S. Y. 194 Hodges, J. A. 124 Hodson, G. 95 Hofman, J. E. 245, 295 Hogg, M. A. 10, 11, 13, 29, 32, 33, 35, 50, 65, 66, 70, 91, 93, 94, 95, 96, 99, 103, 104, 112, 113, 114, 115, 116, 117, 120, 121, 123, 124, 125, 127, 162, 180, 189, 194, 244, 247, 258, 275, 280, 283, 284, 287 Holtz, R. 14, 78, 81, 92, 93, 95, 140, 180, 282 Hopkins, N. 122 Hornsey, M. J. 11, 32, 33, 70, 93, 96, 99, 103, 104, 112, 119, 120, 121, 125, 180, 283, 284, 287 Horowitz, D. 288 Horowitz, D. L. 5, 240, 242, 251, 256, 257 Horton, R. S. 194 Houlette, M. 70, 104 Howard, J. M. 66 Hu, L. T. 79 Huntington, S. P. 239 Hutter, R. R. H. 9, 102

Author index Imani, A. 119, 125 Insko, C. A. 66, 191, 194 Isen, A. M. 13, 73, 80, 92, 99, 100, 103, 104, 167, 169, 170, 175, 177, 178, 283 Islam, M. R. 6, 8, 11, 39, 81, 141, 181, 196, 203, 279, 282 Ito, T. 168 Ito, T. A. 282 Jackman, M. R. 191 Jackson, B. 219 Jackson, J. R. 12 Jackson, L. M. 286 Jacobson, D. 239 Jacquenain, M. 252 Jetten, J. 91, 94, 95, 113, 125, 286 Johnson, B. 66, 70, 79 Johnson, D. M. 28 Johnson, J. D. 79 Johnson, K. M. 82 Johnson, R. 290 Joireman, S. F. 239 Jones, N. 224 Jost, J. T. 16 Judd, C. 280, 284 Judd, C. M. 6, 7, 8, 11, 39, 91, 141, 147, 148, 181, 196, 203, 279, 282 Kafati, G. 81 Kahn, A. 139 Kameda, T. 118 Kampmeier, C. 297 Kardes, F. R. 59 Karp, L. 13, 167 Katzenstein, P. J. 240 Kawakami, K. 67 Kelly, C. 42, 163 Kelman, H. C. 250, 256 Kempen, H. 213 Kenworthy, J. 295 Kenworthy, J. B. 12, 13, 168, 172, 173, 174, 178, 183 Kessler, T. 40, 78, 232 Keys, C. B. 118 Khalaf, A. 42, 43 Kich, G. 219, 226 Kim, K. K. 141 Kim, S. H. 250 King, G. 27 King, G. A. 13 Klandermans, B. 42 Klauer, K. C. 5, 147 Klotz, M. L. 179 Koay, P. S. 200, 202

315

Kohatsu, E. 218 Kortekaas, P. 298 Kozar, R. 66 Kramer, G. P. 169, 170 Kramer, R. M. 116 Kriesberg, L. 244, 250 Kroger, J. 217 Krueger, J. 92 Kuipers, P. 172 Kunda, Z. 3, 5, 9, 56, 276, 277, 278 Kurtines, W. 290, 292 Kwan, V. S. Y. 147, 148, 282 Lafrance, M. 191, 203 LaFromboise, T. 15, 223, 230, 231, 290, 291, 292, 293 Laitin, D. D. 242 Lampen, L. 165 Larson, J. R. 118 LaSelva, S. V. 254 Learly, M. R. 115, 191, 194 Lee, J. 39, 74, 140, 181, 256, 293, 294 Lee, J. Y. 4, 6, 11 Leong, F. 230, 231 Leslie, P. M. 240 LeVine, R. A. 5, 161, 281, 286 Lewis, A. 43 Leyens, J.-P. 118 Leyens, J. Ph. 272 Lickel, B. 43 Liebman, R. C. 122 Lijphart, A. 239, 241, 248, 254, 255, 256 Lilli, W. 140 Lim, D. I. 9, 93, 96, 141, 189, 196, 197, 203, 281 Lim, K. K. 9, 93, 96, 189, 196, 197, 203, 281 Limbert, W. M. 296 Lin, M. 65, 67 Lipset, S. M. 5 Lochner, B. T. 15, 289 Locksley, A. 179 Lockwood, J. 140 Long, J. A. 252, 253 Lord, C. G. 175 Lorenzo-Hernandez, J. 292 Loux, S. 284 Lowrance, R. 92, 169, 170, 175, 283 Lubin, B. 170 Lui, L. 7, 8, 278 Lynch, E. B. 35 Lynch, L. 4, 5, 8, 145, 277, 278 Maass, A. 12

316

Author index

Mackie, D. M. 139, 172, 174, 169, 287 Macrae, C. N. 5, 8, 275, 277, 278 Major, B. 117, 284 Malone, P. S. 126 Mann, J. 119, 120 Mann, J. A. 6, 11, 14, 69, 74, 75, 76, 90, 92, 99, 283 Manstead, A. S. R. 91, 94, 95, 113, 163, 286 Marcia, J. 217, 218 Marcus-Newhall, A. 14, 78, 81, 92, 93, 95, 140, 180, 183, 282 Marques, J. M. 113, 116, 123 Marshall-Goodell, B. S. 12 Martin, C. L. 189, 203 Martin, J. 194 Martin, R. 279 Mason, I. 125 Matoka, K. 66, 70, 79 Matteson, D. 217, 218 Maurer, K. L. 7 Mavor, K. I. 30 McAuliﬀe, B. J. 125 McClelland, J. L. 55 McGarry, J. 251, 253, 256, 275, 288 McGarty, C. 26, 27, 30, 32, 34, 36, 40, 41, 42, 43, 57, 78, 162, 244, 274 McLaughlin-Volpe, T. 104 McRoberts, K. 240 Medin, D. L. 31, 35, 54 Meenes, M. 43 Meiser, T. 83 Mercer, G. W. 5, 273 Mervis, C. B. 28 Messick, D. M. 139 Meyer, G. 27, 139 Migdal, M. 6, 12, 74, 92, 93, 140, 141, 144, 152, 156, 161, 181, 204, 257, 258, 281, 282 Miller, D. 276 Miller, D. T. 5, 9, 56, 119, 120, 128, 272, 282 Miller, N. 4, 5, 6, 11, 12, 13, 14, 39, 67, 68, 74, 78, 81, 90, 92, 93, 95, 102, 104, 119, 140, 141, 161, 162, 163, 164, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 183, 256, 282, 293, 294, 297 Miller, N. M. 6, 12, 92, 197, 282 Milne, A. B. 5, 8, 275, 277 Minow, M. 253 Misra, G. 216 Mizrahi, K. 191, 192, 193 Mladinic, A. 190, 191, 196, 204

Montoya, M. R. 66 Montoya, R. M. 194 Moran, V. R. 292 Moreno, K. N. 169 Morten, G. 219, 220 Moskowitz, G. 213, 228 Moskowitz, G. B. 9, 66 Mottola, G. 73 Mottola, G. R. 70, 104 Mughan, A. 248 Mugny, G. 123 Mulder, M. 126 Mullen, B. 6, 12, 13, 74, 79, 91, 92, 93, 140, 141, 144, 152, 156, 161, 163, 181, 204, 257, 258, 281, 282 Mullin, B. A. 94 Mummendey, A. 32, 40, 42, 78, 82, 83, 121, 232, 284, 285, 291 Murphy, A. B. 254 Murphy, G. L. 5, 9 Murphy, R. 15, 289 Murphy, R. F. 273 Murrell, A. 119, 120 Murrell, A. J. 6, 11, 14, 69, 74, 75, 76, 90, 92, 99, 283 Musgrove, L. 40, 41 Nakashima, C. 212 Nanda, P. C. 12 Nemeth, C. 119, 123 Neuberg, S. L. 3, 65, 67, 152 Newman, L. S. 9 Newman, R. 165 Newsom, J. T. 298 Ng, S. H. 126 Niedenthal, P. M. 80, 121, 167, 169, 170, 177, 178 Niemann, Y. F. 70 Nier, J. 70, 104 Noel, J. G. 95, 113 Oakes, P. J. 3, 8, 10, 13, 26, 27, 29, 30, 35, 36, 50, 57, 65, 66, 78, 91, 94, 113, 139, 162, 189, 244, 247, 258, 274, 275, 276, 280 O’Brien, A. 120 O’Leary, B. 251, 253, 256, 275, 288 Olson, J. M. 272 Omi, M. 228 Onorato, R. S. 66 Opotow, S. 272 Oppes, T. 119, 125 Orlofsky, J. 217, 218 Ostrom, T. M. 189

Author index Otten, S. 12, 13, 66 Oustinovskaya, M. 221 Ouwerkerk, J. 298 Owens, P. 119, 123 Páez, D. 113, 116, 123 Paladino, P. M. 272 Paladrino, M.-P. 118 Panebianco, A. 248 Paolini, S. 8, 12 Park, B. 7, 66, 91, 181, 280, 284 Park, R. 224 Park, R. E. 290 Paul, B. Y. 168, 282 Pederson, W. C. 174, 183 Pendry, L. F. 277, 278 Pepels, J. 145, 165 Perdue, C. W. 12, 13 Perez, W. 292 Pettigrew, T. 223 Pettigrew, T. F. 67, 79, 90, 119, 285 Petty, R. E. 12 Petzold, P. 167 Pfaﬀenberger, B. 240 Phinney, J. 212, 214, 215, 216, 218, 220, 221, 223, 224, 225, 226, 228, 229, 230, 231, 232 Phinney, J. S. 15, 279, 289, 292 Pickett, C. L. 118 Pinter, B. 66 Pittinsky, T. L. 278 Poh, L. L. 191, 196, 197 Pomare, M. 6, 11, 14, 69, 74, 90, 92, 99, 283 Portes, A. 290 Posokhova, S. 94 Postmes, T. 114, 118 Powell, M. C. 59 Pratto, F. 286 Prentice, D. A. 52, 119, 120, 128, 272, 282 Prins, K. S. 81, 283, 284 Prislin, R. 296 Pruitt, D. G. 250 Putnam, R. D. 248, 249 Pyszczynski, T. 115, 286 Quillian, M. 50, 51 Quintana, S. 222 Rabe, C. 166 Rae, D. W. 5 Ramirez, M. 291 Rauch, S. 292

317

Rawls, J. 251 Reed, J. S. 273 Rehm, J. 140 Reicher, S. D. 10, 13, 29, 35, 41, 50, 65, 66, 91, 94, 113, 114, 122, 162, 244, 247, 258, 275, 280 Reid, A. 191, 192, 193 Reid, S. A. 125 Reis, H. T. 191 Reitz, J. G. 242 Renn, K. 226, 227, 233 Revill, J. 212 Reynolds, K. J. 30, 33, 40, 41, 284, 299 Richards, Z. 7, 8, 12, 229 Riek, B. M. 83 Rips, L. J. 28 Roccas, S. 15, 94, 102, 121, 273 Rocher, S. J. 272 Rockquemore, K. 214, 225, 226, 227 Rodriguez, A. P. 272 Rodriguez, R. T. 272 Rogers-Croak, M. 78 Roosens, E. 215, 233 Root, M. 212, 214, 223, 225, 226, 228 Ropp, S. A. 104 Rosch, E. H. 28, 29, 35, 91 Roseman, I. J. 172, 174 Rosenkrant, P. S. 190, 204 Roskos-Ewoldsen, D. R. 59 Ross, L. 37, 126 Ross, M. H. 5 Rothbart, M. 7, 12, 66, 92, 181, 272 Rothgerber, H. 66, 119 Rothschild, L. 43, 272 Rubin, J. M. 119 Rubin, J. Z. 250 Rubin, M. 4, 7, 27, 37, 93, 95, 102, 115, 180, 190, 204, 279, 285, 286, 288, 299 Ruble, D. N. 189, 203 Ruderman, A. 145 Ruderman, A. J. 8, 165 Rudolph, J. R. 240, 249 Ruiz, R. 290 Rumbaut, R. G. 290 Rumelhart, D. E. 55 Russell, M. L. 13, 167 Rust, M. C. 6, 11, 14, 31, 32, 68, 70, 73, 79, 80, 82, 83, 84, 92, 99, 100, 103, 104, 119, 120, 245, 257, 283, 284 Ryan, C. S. 7 Ryen, A. H. 139 Sachdev, I. 126

318

Author index

Safran, W. 240 Samaha, C. 74, 92 Sanbonmatsu, D. M. 59 Sanders, K. 212 Sani, F. 41, 113, 122 Schachter, S. 125 Schadron, G. 272 Schaﬀer, M. M. 54 Schaller, M. 12 Scheepers, D. 95 Schiller, P. M. 293 Schneider, S. K. 244 Schoﬁeld, J. W. 102, 140, 282 Schopler, J. 66 Schroeder, C. 9, 276 Schumpeter, J. 250 Schwartz, S. H. 94 Schwarz, N. 54, 166, 169, 170, 177, 179 Schwarzer, R. 292 Schweitzer, S. 74, 92 Searle-White, J. 242 Sedikides, C. 115, 174 Senelle, R. 240, 253 Serodio, R. 123 Seron, E. 15, 297 Serpe, R. T. 295 Shalker, T. 13, 167 Sharmini 191 Sherif, M. 11, 67, 90, 115, 120 Sherman, S. J. 43, 114, 116 Shih, M. 278 Sidanius, J. 286 Sigelman, L. 124 Silver, M. 139 Simmel, G. 5, 161, 273 Simon, A. F. 169, 287 Simon, B. 42, 163, 297 Simon, L. 286 Sinclair, L. 3, 277, 278 Singelis, T. M. 291, 294 Singer, M. R. 240 Singh, R. 9, 93, 96, 141, 189, 191, 192, 193, 194, 196, 197, 198, 199, 200, 201, 202, 203, 281 Sloan, L. R. 124 Sloman, S. A. 34, 39 Smiley, D. V. 254, 255 Smith, A. D. 242, 243 Smith, A. S. 224 Smith, C. 13, 91, 163 Smith, E. R. 5, 52, 53, 54, 55, 56, 57, 58, 59, 60, 118, 172, 174, 276, 280, 287 Smithson, M. J. 35 Snider, K. 70

Solomon, S. 115, 286 Spears, R. 26, 91, 94, 95, 113, 114, 118, 163, 218, 219, 286 Spickard, P. 214 Srull, T. K. 52 Staats, A. W. 12 Staats, C. K. 12 Stangor, C. 4, 5, 8, 145, 277, 278 Stasser, G. 118 Statham, A. 273 Staub, E. 272 Staw, B. M. 119, 123 Steele, C. 117 Steiner, J. 254 Stemmler, G. 172 Stephan, C. 224, 228 Stephan, C. W. 286, 292 Stephan, W. 224, 228 Stephan, W. G. 286, 292 Stern, P. 258 Stewart, D. D. 118 Stewart, T. L. 58, 60 Stone, C. H. 96, 97, 99, 100, 101, 284, 287 Stonequist, E. 224 Strack, F. 170 Straw, J. 4 Strube, M. J. 115 Stryker, A. 273 Stryker, S. 295 Sue, D. 219, 220 Susser, K. 169, 170 Sutton, J. R. 122 Svensson, A. 119, 125 Swartz, T. S. 172, 174 Syllwasschy, J. 165 Szapocznik, J. 290, 292 Tajfel, H. 10, 13, 31, 37, 50, 65, 66, 67, 70, 83, 91, 92, 94, 113, 115, 116, 139, 163, 189, 192, 218, 258, 271, 273, 274, 275, 279, 286 Tan, G. L. 202 Tan, K. J. 200, 201, 202 Tassinary, L. G. 12 Tate, J. 220, 221 Tate, J. D. 279 Taylor, C. 251 Taylor, D. 258 Taylor, M. 5, 272 Taylor, S. E. 8, 52, 145, 165 Teasdale, J. D. 13, 167 Teoh, B. P. 194 ter Schure, E. 172

Author index Terry, D. J. 119, 121, 124 Thagard, P. 277 Thompson, E. R. 277 Thompson, R. J. 240 Thorne, A. 124 Tindale, R. S. 118 Trew, K. 295 Trope, Y. 3 Tropp, L. R. 67, 121 Trost, M. R. 298 Tsebelis, G. 248, 249 Turner, J. C. 3, 8, 10, 11, 13, 26, 29, 30, 31, 35, 36, 40, 50, 57, 65, 66, 67, 69, 70, 76, 78, 83, 91, 93, 94, 113, 115, 116, 139, 140, 143, 144, 156, 162, 163, 189, 192, 194, 218, 244, 245, 247, 258, 275, 280, 286, 299 Turner, J. W. 139 Tutu, D. 299 Tyler, R. B. 12, 13 Tyler, T. R. 126 Uba, L. 230 Uhles, A. N. 43 Uleman, J. S. 9 United Nations 189, 203 Urada, D. 12, 13 Urada, D. I. 78, 162, 169, 170, 171, 176, 178, 183, 282, 297 Urban, L. M. 6, 12, 74, 92, 141, 161, 168, 169, 172, 179, 181, 183, 197, 282 Vaes, J. 272 Valentine, V. F. 242 Validzic, A. 66, 70, 71, 79, 84, 93, 279 Van Eimeren, B. 140 van Knippenberg, A. 8, 10, 93, 145, 165, 277 van Knippenberg, D. 96, 112, 125, 287 van Leeuwen, E. 96, 287 van Oudenhoven, J. P. 81, 283, 284 van Rijswijk, W. 5, 55, 57, 162, 275 Van Twuyver, M. 8, 10, 93, 145, 165, 277 Vanbeselaere, N. 7, 74, 76, 78, 93, 95, 140, 163, 164 Vandekeere, M. 252 Vanman, E. J. 161, 168, 172, 282 Varty, D. L. 253 Verkuyten, M. 66, 289 Vescio, T. K. 119, 139, 145, 147, 148, 282 Voci, A. 279 Vogel, S. R. 190, 204 Vogt, E. Z. 290

319

von Hippel, W. 117 Vredenburg, D. S. 179 Wachtler, J. 119, 123 Waddell, N. 295, 297 Wade, G. 11, 70, 71, 180 Waenke, M. 54 Wagner, U. 165 Waldzus, S. 83 Walker, M. R. 124 Wallace, M. D. 5, 288 Wann, D. L. 95, 113, 124 Ward, C. 70 Ward, C. M. 70, 104 Ward, T. B. 9 Waterman, A. 217, 218 Wattenmaker, W. D. 9 Weaver, K. D. 12, 13, 168, 172, 173, 174, 178, 183 Weber, R. 9, 276 Wegener, I. 5, 147 Wentura, D. 12, 13 Wenzel, M. 32, 42, 82, 121, 272, 284, 285, 291 Wetherell, M. S. 10, 13, 29, 35, 50, 65, 66, 91, 94, 113, 162, 244, 247, 258, 275, 280 Wicker, A. W. 125 Wieczorkowska, G. 43 Wiest, C. 172, 174 Wijeyesinghe, C. 219 Wilder, D. 287 Wilder, D. A. 66, 68, 139, 163, 169, 271, 275, 278, 281 Wildschut, T. 66 Wilkenﬁeld, M. J. 9 Wilkes, A. L. 10, 13, 91, 92, 139 Williams, C. J. 12 Williams, J. E. 190 Williams, K. D. 113 Williams, P. 211, 226, 228, 229 Williams-Leon, T. 212 Willis, H. 4, 190, 204, 279, 285, 286, 288, 299 Wilson, T. M. 124 Wilson-Smith, A. 241 Winant, H. 228 Wittenbaum, G. M. 118 Wittenbrink, B. 280, 284 Wölk, M. 166 Wolsko, C. 280, 284 Wood, W. 190, 191, 196, 197 Worchel, S. 11, 66, 74, 92 Worth, L. T. 169

320

Author index

Wright, S. C. 104, 121 Wuthnow, R. 122 Wyer, R. S. 52, 166 Xu, J. 191, 196, 204 Yamada, A. M. 291, 294 Ybarra, V. 222 Yeoh, B. S. E. 9, 93, 96, 141, 189, 196, 197, 203, 281 Yurak, T. J. 179 Yzerbyt, V. 15, 118

Yzerbyt, V. Y. 15, 26, 272, 297 Zack, N. 215 Zadeh, L. A. 35 Zagefka, H. 284 Zanna, M. P. 286 Zárate, M. A. 54 Zariski, R. 239 Zhermer, N. 94 Zogmaister, C. 59 Zolberg, A. R. 253, 254 Zuckerman, M. 215

Subject index

Accentuation 140, 279 Accessibility 27 Acculturation 230–231 Achievement 217, 218, 219, 231 Additive model 196, 203 Adolescents 218, 221, 222–223, 231–232 Aﬀect infusion model 167–168, 171 Aﬀective arousal, 165–166, see also Mood Aﬀective commitment 295 African Americans 225–226 Age 10, 200–202, 222–223 Alternation 231, 291–292, 294 Amalgamation 251 Ambiguous ﬁgures 53 Anger 172–174, 183 Anthropology 5 Approach tendencies 172–174, 183 Archaeological perspective 243 Artistic movements 122 Assertion 275 Assignment 275 Assimilation 230, 246, 289, 290–291 Associative network theory 166, 170, 171, 177 Asylum seekers 293, 294 Attitudes 59, 118 Attributes 124, 126 Australian Aborigines 5 Avoidance tendencies 172–174, 183 Bases of social identity 258–260 Belgium 240, 248, 251, 252, 253, 254 Belonging 246 Biculturalism 15–16, 230, 231, 279, 289, 290, 291–294 Bigots 59 Biological taxonomy 27, 36 Biracial groups 219, 228, 229

“Black sheep” 123–124 Blank predicates 37 Blended bicultural identity 231 Blended multiple categorization 27 Border identity 226–227 Boys’ camp studies 120 Canada 240, 241, 243, 251, 252, 253, 254, 258, 261 Caste-like inequalities 241 Categorization beneﬁt 25–26 change over context and time 280–281 convergent 6–7 deﬁnition 274 determinants of on-line 57–59 divergent 7 dynamic and ﬂuid 3 external 275 ﬂexibility 280 horizontal 28 intermediate 28 internal 275 levels of abstraction 28, 275 meaning-giving function 274–275 power 25, 276 resource 278–280 simple 139 simultaneous 11, 99–101, 104–105, 287 social interaction 3 subordinate 28 superordinate 6, 14, 28, 82–83, 93, 99–100, 104, 119–122 suppression 279–280 Category conjunctions 8–10, 11–13, 14, 25, 56–57, 276 Category constituents 10–11, 13–14 Category diﬀerentiation model 10, 91, 155, 156

322

Subject index

Category-divide hypothesis 272 Category dominance model 196, 203 Category importance 162–166 Category membership, reﬁning 42–44 Category relevance 163–166 Category representation 50–53 Category selection 25, 276 Census 224 Centrality 245 Change 280–281 Charisma 126 Classical view of categories 28 Cleavage 127–128 Clinton, Bill 3 Coalition 248, 255 Coat color 77–78, 92 Cognitive diﬀerentiation 14 Cognitive overload 183 Coherence 272 Cold War 241 Collective action 41 Collective behavioural intentions 41–42 Collective guilt 41 Collective identity 66 Colorblind approach 279 Colored coats 77–78, 92 Commitment 217, 218, 219–220, 295, 298–299 Common fate 11, 244 Common ingroup identity model 10–11, 14, 31–32, 68–69, 71–73, 84–85, 91, 257, 282–283, 290 Comparison 203–204 Competition 192–194 Complexity 15, 273–274 Conﬂict processes framework 248, 249–250 Conjunctions 8–10, 11–13, 14, 25, 56–57, 276 Connectionist models 55 Consensus 26 Consociationalism 248, 254–256, 257, 288 Constituents 10–11, 13–14 Contact hypothesis 67, 100, 104, 250 Contact zones 213 Containment 27 Context 3, 227, 233, 245, 247–250, 280–281 Convergent categorization 6–7 Cooperation 32–33, 37, 92, 104 Cooperative interdependence 11 Corporate identity 244

Correspondence bias 126 Crossed categorization 73–79, 92–93, 99–100, 139–141, 160–161 additive model 196, 203 blended multiple categorization 27 cancellation 10, 140, 276–277 category accentuation process 140 category conjunctions 11–12 category diﬀerentiation model 10 category importance 162–166 category relevance 163–166 classic hypothesis 141–142, 145–146, 148–150, 153, 154 conﬂict reduction 5 convergent categorization 6–7 divergent categorization 7 dual-route model 12 full crossed categorization design 144–155 hierarchical rejection pattern 175–176 hypotheses 141–144 incidental aﬀect 169–172, 177–178, 182–183 integral aﬀect 169, 175–178, 183 intergroup bias reduction 140, 149–152, 154–155, 281–282 interpersonal attraction in men 194–197 mediators 282 meta-analysis 161 moderation hypothesis 7, 142–144, 145, 146–147, 149, 150–151, 153, 154 mood 13, 168–180, 182–183 patterns 180–182, 195–197 personalized contact 179–180, 183 processes 181–182 psychological signiﬁcance 152 social attraction in women 194–197 social stimuli 140 subgroups 119–122 Cults 122, 126 Cultural divides 128 Cultural identity 211 Culture 211, 213–215, 216 Decategorization 67–68, 69–70, 71, 76, 77, 102, 104, 256 Dehumanization 250 Deindividuation 250 Delegitimizing victims 272 Democracy 241 Depersonalized 114 Deviance 123–125

Subject index Dictatorships 126 Diﬀerence 246 accommodation 251–256 elimination 251–252 Diﬀerential inclusiveness 7–8 Diﬀerentiation 13–14, 101, 118–126, 246, 247, 278–279 Diﬀusion 217 Discreteness 272 Discrimination 32, 65 Disgust 172–174, 177, 183 Dissent 123–125 Dissolution 127 Distinctiveness 71, 73, 93, 94–95, 100, 101, 104, 120–121, 246–247, 287 Distributed models 50, 53, 55–56 Divergent categorization 7 Diversity 118–126 Division 252 Domains 217 Double consciousness 232 Dual identity 33, 74, 79–83, 232, 284–285, 291 Dual-route model 12 Ego identity formation 216–217 Emergent attributes 9, 56 Entitativity 15, 43, 114, 116, 117–118, 122 Environment 166–167, 233 Essentialism 15, 272, 297 Ethnic conﬂict 239–240 Ethnic identity 211, 217–218 Ethnicity 213–215, 222–223, 242–243 Ethnoterritorial conﬂict 241–244, 251–262 Ethnoterritorial politics 239–240 Evaluation 14, 114 Exclusivity 272, 276 Exemplar models 50, 53, 54–55 Expectations 271 Experience 57–59 Exploration 217, 222 External categorization 275 Federalism 252–254, 255–256, 257 Feelings as information 166, 177 Fluxion 294–299 Foreclosure 217, 218, 219, 230–231 France 239–240 Fundamental attribution error 126 Fuzzy boundaries 34–35, 91–92, 94 Fuzzy sets 35, 114

323

Gastronomic theory 242–243 Gays and lesbians 44 Gender 189–208, 211 acquisition 189 categorization 189–191 competition 192–194 dimensions of gender perception 192–194 immediately applied category 152 interpersonal attraction 189, 194–204 social attraction 189, 194–204 social cognition 189–190 social identity 189, 190 social interaction 192–194 Geological theory 242, 243 Goal-directed categories 35–36 Goal relations 11 Groups diﬀerentiation and diversity 118–126 group-type distinction 42–44 interactions 11 overlapping membership 281 polarization 15 size 125, 245–247 social category 43–44 social identity 113–114 status 296 Groupthink 118 Hierarchical categories beneﬁts 30–34 cognitive psychology 28–29 conceptual structures 39 containment 27 cooperation 32–33, 37 exception properties 51–52 identity 30–32 leadership 125–126 problems with 34–40, 52, 276 science 27 social psychology 29–30 tree-like structure 30, 50–51 Hierarchical rejection pattern 175–176 Highﬂyers 124 “Hispanic” population 214, 225 History 257 Homogeneity 15, 272 Homosexuals 44 Horizontal structure 28 Hot conﬂict 251 Humanity 299 Hybrid categories 226–227 “Hyphenated” Americans 230

324

Subject index

Hypodescent 214 Identiﬁcation 95–96, 275 Identity achievement 217, 218, 219, 231 categorization 25 crafting 40–42 diﬀuse 217 enhancement 191 ﬂuxion 294–299 formation 211, 216–223 hierarchies 30–32 psychological theories 244–247 status 217 threats 32–33 Immersion 218, 231, 232 Immigrants 212–213, 214, 223, 229, 289, 291, 292, 293–294 Immutability 272 Importance of categories 162–166 Incidental aﬀect 169–172, 177–178, 182–183 Inclusiveness 6, 7–8, 276 Incorporation 251 Indigenous people 229 Individuation 179 Inductive potential 272 Ingroups 4, 65, 66 bias 195 comparison 82–83 favoritism 14, 95 homogeneity 15 identiﬁcation 95, 275 potency 245 projection 82, 284 valence 12–13 Inherence 272 Inhumane behaviour 272 Instrumentalism 242 Integral aﬀect 169, 175–178, 183 Integration 230, 292–293 Interaction 227 Interactional commitment 295 Intergroup bias 4 evaluation 13–14 orientations 191–204 relations 37, 219–223, 227–229 similarity 13–14, 94 Intermarriage 224 Intermediate categorization 28 Internal categorization 275 Internalization 231 Interpersonal attraction 189, 194–204

Interpersonal orientations 191–204 Intervention 262, 289 Isen’s mood model 167, 171, 177, 178 Italy 240 Kinds of people 272 Large-scale social cleavages 211 Latinos 225 Leadership 125–126, 247–250, 261 Lesbians and gays 44 Levels of abstraction 28, 275 Liberalism 251, 257 Lists 35 Marginal 230 Marginal man 224 Marginalization 123–125, 290 Massacres 272 Meaning-giving 274–275 Memory 5, 52 Mental disorders 43, 44 Mestizo population 214, 225 Meta-contrast ratio 91 Minimal group paradigm 271 Minority groups 81, 213, 220–221, 223, 232, 243, 289, 290–291, 292, 293 Moderation hypothesis 7, 142–144, 145, 146–147, 149, 150–151, 153, 154 Modernization 241 Mood common ingroup representation 92 crossed categorization 13, 168–180, 182–183 environment 166–167 social judgment 166–168 Moral exclusion 272 Moral unconcern 272 Moratorium 217, 218, 219, 231, 232 Motivation 3, 94–95, 103, 115–118 Multicultural people 33, 213, 229–232, 233, 234, 279 Multiethnicity 212, 219, 223–227, 279 Multiple categorization 4 beneﬁts 26 blended 27 evaluating 11–14, 276 exemplar models 54–55 foundations 4–5 identiﬁcation 95–96 integrative approach 285–286 perception 5, 26

Subject index populations of interest 212–213 potential 276 psychological perspectives 5 reducing intergroup bias 281–286 reducing intergroup conﬂict 288–289 representing 8–11, 276 taxonomy 6–8 threat prediction 96–99 threat prevention 99–102 use 5 Multiracial people 212, 213, 219, 223–229, 233, 234, 279 Mutual diﬀerentiation 256 Mutual intergroup diﬀerentiation model 32, 71–73 Naturalness 43, 272 Nehru, Pandit 7 Nested games 248–249 Neural networks 55 New category formation 226–227 Nigeria 273 Nonsocial concepts 9 Northern Ireland 256, 273, 275, 295, 296, 297 On-line categorization 57–59 “One drop” of black blood 214 Opinion-based groups 40–42 Optimal distinctiveness theory 31, 246–247, 286 Organizational mergers and acquisitions 121 Outgroups 4, 12–13, 65, 66 Overachievers 124 Overlapping group memberships 281 Parallel constraint satisfaction model 277–278 Parallel distributed processing models 55 Patterns of evaluation 180–182, 195–197 Peer culture 227 Perception 5, 26, 114 Perceptual readiness 27 Person memory model 52 Person perception 5 Personal identity 66, 113 Personalization 179 Personalized contact 179–180, 183 Political elites 247, 248 Political leaders 247–250, 261 Political opinion-based groups 41

325

Politics 5, 120, 122, 123, 239–268, 272–273, 288 Positive social behaviour 33 “Postbourgeois values” 241 Potency 245 Potential categorizations 276 Power of categorization 25, 276 leadership 126 Practice 57–59 Pre-encounter 230 Prejudice 43, 65 Primordialism 242, 272–273 Processes 181–182 Proliferation 252 Prototypes 28–29, 114 Race 79–80, 92–93, 152, 211, 213–216 Racial identity 211, 217–219 Realistic threat 286 Recategorization 31, 67–68, 69–70, 71, 76, 77, 105, 256, 282–283 Reconstruction 53 Related categories 102 Relevance of categories 163–166 Religion 122, 123, 273, 298 Remodeling 117–118 Representation 14, 50–53 Representational states 52–57, 280 Resource 278–280 Russell, Bertrand 7 Rwanda 280, 300 Sadness 172–174, 183 Salience 41–42, 66, 79–81, 92–93, 105, 163, 245, 295, 296–297 Sameness 33 Schisms 122–123, 127–128 Schools of thought 122 Science 27 Scientiﬁc doctrines 122 Seating patterns 92 Selection 25, 276 Self-anchoring 15 Self-categories 37–39, 78, 114 Self-categorization theory 10, 27, 29–30, 31, 33, 36, 57, 66, 91, 275, 276 Self-determination 241 Self-enhancement 115 Separation 230, 290 September 11 attack 26 Sexual orientation 44 Shifting standards model 277 Similarity 13–14, 94, 246

326

Subject index

Simple categorization 139 Simultaneous categorization 11, 99–101, 104–105, 287 Situational deﬁnitions 227 Slavery 214 Social attraction 189, 194–204 Social categorization altering 67–70 bias 66–67 category diﬀerentiation 90–91 dynamic process 67 entitativity 43 groups 43–44 membership 4 multiple bases 4 multiple identities 73–83 naturalness 43 self-inclusion 4 social identity 114 social psychological account, xiv types 43–44 Social cognition, gender 189–190 Social conﬂict 299 Social context 233 Social identity bases 258–260 complexity 15, 273–274 ﬂuidity 244 gender 189, 190 groups 113–114 motivations 115–116 salience 66 Social identity theory 57, 66, 155, 156 Social interaction 3, 192–194 Social judgment 5, 166–168 South Africa 296 Spain 240 Splinter groups 122–123 Sports 6 Sri Lanka 240 Staﬃng theory 125 State intervention 241 States 52–57, 280 Status dual identity 81–82 groups 296 identity 217 Stereotypes 5, 9–10, 26, 43, 44, 54, 277–278 Storage bins 52

Stories 257 Subgrouping 7, 119–122 Subidentities 245 Subjective group dynamics 123 Subordinate categorization 28 Superordinate categorization 6, 14, 28, 82–83, 93, 99–100, 104, 119–122 Suppression of categories 279–280 Switzerland 252, 254 Symbolic threat 286 Systems of categories 25 “Tajfel’s law” 245 “Things” 52 Threat buﬀering 127–128 central importance 286–287 cooperation 32–33 distinctiveness 287 motivational responses 116–118 prediction 96–99 prevention 99–102 realistic 286 self-concept 113 symbolic 286 Tolerance 32 Totalism 125–126, 127 Tradition 257 Tree diagrams 30, 50–51 Trudeau, P. 122 Tutu, Desmond 299 Types 43–44 Ubuntu 299 Umbilical stories 257–258 Uncertainty reduction 115–116 Unrelated categories 102 Valence 12, 116–117, 122, 245 “What are you?” 212, 228–229 “Who said what?” paradigm 52, 145–147, 165 Witch-hunts 127, 128 “Women-are-wonderful eﬀect” 190–191, 195, 196, 197–198, 200, 202, 203 Woods, Tiger 278 Younge, Gary 3 Yugoslavia 251, 252