Distinctiveness threat and prototypicality: combined effects on intergroup discrimination and collective self-esteem

Similar documents
We re all individuals : group norms of individualism and collectivism, levels of identification and identity threat

A dynamic perspective on social identification: Predictors and consequences of identification during group formation

CHAPTER 4: PREJUDICE AND DISCRIMINATION

Gender Stereotypes Associated with Altruistic Acts

II. DISTRIBUTIONS distribution normal distribution. standard scores

Descriptive Statistics and Measurement Scales

Intergroup Helping as Status Relations: Effects of Status Stability, Identification, and Type of Help on Receptivity to High-Status Group s Help

Managing One s Group Image Dynamics of Group-Based Self-Esteem and Identity Management Strategies

interpretation and implication of Keogh, Barnes, Joiner, and Littleton s paper Gender,

Test Bias. As we have seen, psychological tests can be well-conceived and well-constructed, but

Introducing Social Psychology

Midterm Review Problems

Statistics 2014 Scoring Guidelines

Understanding Social Identity Maintenance: An application and extension of the Social Identity Approach

Research Methods & Experimental Design

Multivariate Analysis of Variance. The general purpose of multivariate analysis of variance (MANOVA) is to determine

Fairfield Public Schools

Descriptive statistics Statistical inference statistical inference, statistical induction and inferential statistics

Introduction to Analysis of Variance (ANOVA) Limitations of the t-test

Restaurant Tips 1. Restaurant Tips and Service Quality: A Weak Relationship or Just Weak Measurement? Michael Lynn. School of Hotel Administration

Sample Paper for Research Methods. Daren H. Kaiser. Indiana University Purdue University Fort Wayne

WHAT IS A JOURNAL CLUB?

Statistical tests for SPSS

UNDERSTANDING THE TWO-WAY ANOVA

Proposal of chapter for European Review of Social Psychology. A Social Identity Theory of Attitudes

IDENTITY THEORY AND SOCIAL IDENTITY THEORY

SELF-CATEGORIZATION THEORY AND THE DEVELOPMENT OF NATIONAL IDENTITY IN ENGLISH CHILDREN. Martyn Barrett, Hannah Wilson and Evanthia Lyons

MEASURES OF VARIATION

Specialisation Psychology

Horizontal Hostility: Relations Between Similar Minority Groups

Introduction to Quantitative Methods

QUANTITATIVE METHODS BIOLOGY FINAL HONOUR SCHOOL NON-PARAMETRIC TESTS

The Relationship between the Fundamental Attribution Bias, Relationship Quality, and Performance Appraisal

Chapter Seven. Multiple regression An introduction to multiple regression Performing a multiple regression on SPSS

COMPARISONS OF CUSTOMER LOYALTY: PUBLIC & PRIVATE INSURANCE COMPANIES.

The Personal Learning Insights Profile Research Report

HOW TO WRITE A LABORATORY REPORT

Lesson 4 Measures of Central Tendency

Sample Size and Power in Clinical Trials

How Professionally Skeptical are Certified Fraud Examiners? Joel Pike Gerald Smith

SCHOOL OF HEALTH AND HUMAN SCIENCES DON T FORGET TO RECODE YOUR MISSING VALUES

International Financial Reporting Standards (IFRS) Financial Instrument Accounting Survey. CFA Institute Member Survey

DRIVING TESTS: RELIABILITY, AND THE RELATIONSHIP BETWEEN TEST ERRORS AND ACCIDENTS

MULTIPLE REGRESSION WITH CATEGORICAL DATA

An Automated Test for Telepathy in Connection with s

Social Identity Theory and Self-categorization Theory: A Historical Review

Additional sources Compilation of sources:

Word count: 2,567 words (including front sheet, abstract, main text, references

Joseph Fordham. Kuo-Ting Huang. Corrie Strayer. Rabindra Ratan. Michigan State University

Chapter 13. Prejudice: Causes and Cures

Research of Female Consumer Behavior in Cosmetics Market Case Study of Female Consumers in Hsinchu Area Taiwan

Reflections on Probability vs Nonprobability Sampling

Session 7 Bivariate Data and Analysis

Research into Issues Surrounding Human Bones in Museums Prepared for

Sample Paper for Learning Research Proposal. Daren H. Kaiser. Indiana University Purdue University Fort Wayne

Statistics. Measurement. Scales of Measurement 7/18/2012

A successful market segmentation initiative answers the following critical business questions: * How can we a. Customer Status.

A. Introducing Social Psychology. Introduction

Measurement and Measurement Scales

Chapter 5 Analysis of variance SPSS Analysis of variance

Chapter 2. Sociological Investigation

Q FACTOR ANALYSIS (Q-METHODOLOGY) AS DATA ANALYSIS TECHNIQUE

I m good, she s good: The relationship between. Self-Esteem and Attitudes Towards Women in the Work Force. Alexandria Peach. McKendree University

research/scientific includes the following: statistical hypotheses: you have a null and alternative you accept one and reject the other

Why do people publish weblogs? An online survey of weblog authors in Japan

2 The Use of WAIS-III in HFA and Asperger Syndrome

The Relationship between Gender and Academic Success Online

The Relationship between Ethnicity and Academic Success in Online Education Courses

individualdifferences

Savvy software agents can encourage the use of second-order theory of mind by negotiators

Statistics Review PSY379

Creating a Culture of Inclusion

Running head: THE EFFECTS OF EXTRA-CURRICULAR ACTIVITIES

SELF THEORY HARRY ALBERT VAN BELLE

Chapter 6 Experiment Process

Measurement with Ratios

Chapter 2 Quantitative, Qualitative, and Mixed Research

Please note that this sample syllabus draws on a sample Social Psychology syllabus created by the faculty at Florida Tech University.

Chapter Eight: Quantitative Methods

Constructing a TpB Questionnaire: Conceptual and Methodological Considerations

Beyond Tops and Bottoms

A COMPARATIVE STUDY OF WORKFORCE DIVERSITY IN SERVICE AND MANUFACTURING SECTORS IN INDIA

Adult cognition of large-scale geometric facts

Inference for two Population Means

6.4 Normal Distribution

UNDERSTANDING ANALYSIS OF COVARIANCE (ANCOVA)

SAS Analyst for Windows Tutorial

DESCRIPTIVE STATISTICS. The purpose of statistics is to condense raw data to make it easier to answer specific questions; test hypotheses.

UNIVERSITY OF NAIROBI

Measures of Central Tendency and Variability: Summarizing your Data for Others

WMS III to WMS IV: Rationale for Change

Body adornment and body modification are universal

DESCRIPTIVE STATISTICS & DATA PRESENTATION*

This chapter discusses some of the basic concepts in inferential statistics.

TEACHERS AS ROLE MODELS FOR STUDENTS LEARNING STYLES

Statistiek II. John Nerbonne. October 1, Dept of Information Science

Relevant Concreteness and its Effects on Learning and Transfer

1/27/2013. PSY 512: Advanced Statistics for Psychological and Behavioral Research 2

THE ACT INTEREST INVENTORY AND THE WORLD-OF-WORK MAP

Transcription:

European Journal of Social Psychology, Vol. 27, 635±657 (1997) Distinctiveness threat and prototypicality: combined effects on intergroup discrimination and collective self-esteem JOLANDA JETTEN, RUSSELL SPEARS and ANTONY S. R. MANSTEAD Department of Social Psychology, University of Amsterdam, The Netherlands Abstract The combined influence on ingroup bias of threat to group distinctiveness and prototypicality as a group member was examined in two studies. It was predicted, in line with social identity theory, that threat to group distinctiveness would lead to more ingroup bias. In addition, on the basis of self-categorization theory it was predicted that protypical and peripheral group members would react differently to a threat to their group distinctiveness. Only group members who define themselves as prototypical group members should be motivated to defend their threatened distinctiveness by engaging in increased ingroup bias. This hypothesis was first supported in a modified minimal group setting in which threat was operationalized as overlapping group boundaries. These results were then replicated in a second study, using better-established groups, for whom distinctiveness threat was manipulated in terms of intergroup similarity. Moreover, some support was found in Study 2 for the prediction that the opportunity to engage in intergroup differentiation can, under restricted conditions, enhance group-related selfesteem. # 1997 John Wiley & Sons, Ltd No. of figures: 1 No. of Tables: 3 No. of References: 42 Addressee for correspondence: Jolanda Jetten, Department of Social Psychology, University of Amsterdam, Roetersstraat 15, 1018 WB Amsterdam, The Netherlands, tel: +31-20-5256714, fax: +31-20-6391896, e-mail: sp_jetten@macmail.psy.uva.nl. This research was supported by a University of Amsterdam research grant awarded to the second author. We would like to thank Nyla Branscombe, Daan van Knippenberg and an anonymous reviewer for their comments on an earlier draft of this paper and Patrice de Bruin, Iris van der Sluis, Britt Spaan, Petrouska Swinkels and Barbara Theunissen for their assistance during data collection. CCC 0046±2772/97/060635-23$17.50±00 Received 19 August 1996 #1997 John Wiley & Sons, Ltd. Accepted 7 November1996

636 J. Jetten, R. Spears and A. S. R. Manstead INTRODUCTION The present research investigates threats to distinctiveness produced by means of overlapping group boundaries in a modified minimal group setting (Study 1) and in terms of intergroup distance in natural groups (Study 2). Moreover, prototypicality is manipulated in both studies in order to investigate whether reaction to a distinctiveness threat is a function of the centrality of one's position within the group. We argue that perceived prototypicality of ingroup members is an important moderator of the effect of threats to distinctiveness on intergroup differentiation. According to social identity theory, threats to group distinctiveness and identity can lead to increased antagonism between groups. The argument underlying this phenomenon is that people are motivated to differentiate the ingroup from similar outgroups on relevant dimensions of comparison in order to maintain or enhance group distinctiveness and social identity (cf. Brown, 1984a; Tajfel, Billig, Bundy, & Flament, 1971; Tajfel, 1982; Tajfel & Turner, 1986; Turner, 1978). Given this emphasis of social identity theory on the importance of establishing intergroup differences, similarities between groups are likely to be important in instigating a search for distinctiveness, leading to enhanced ingroup bias (Allen & Wilder, 1975; Brewer, 1979; Brown, 1984a,b; Brown & Abrams, 1986; Diehl, 1988; Mummendey & Schreiber, 1984; Roccas & Schwartz, 1993; van Knippenberg & Ellemers, 1990; Tajfel, 1982; Turner, 1978). In this research tradition, identity or distinctiveness threats have typically been manipulated in terms of the similarity of intergroup status or attitudes (e.g. Dichl, 1988, Study 2). Moreover, intergroup differentiation as a function of group similarity has been observed in laboratory groups as well as natural groups. In the present research, the standard way in which similarity has previously been manipulated is extended and improved by providing information not only about the mean intergroup difference but also about the variability within each group. The issue addressed in the present research is not only how threat to distinctiveness affects the level of ingroup bias at the intergroup level, but also how intergroup bias is affected by more intragroup factors, namely prototypicality as a group member. In other words, we attempt not only to address the question of when a group displays increased ingroup bias but also the question of who within the group exhibits the most ingroup bias. Although we expect that the general group response to threat to distinctiveness will be heightened ingroup bias, we do not expect all group members to express ingroup bias to a similar extent. Past research has focused on the moderating effect of relatively stable individual difference variables, such as level of group identification and social value orientations, on a range of dependent variables. Moderating effects of group identification have been observed not only on measures of ingroup bias (Roccas & Schwartz, 1993), but also on levels of ingroup stereotyping (Spears, Doosje, & Ellemers, 1997), perceived group variability (Doosje, Ellemers, & Spears, 1995), group cohesiveness (Turner, Hogg, Oakes, & Smith, 1984), and evalauation of a disloyal ingroup member (Branscombe, Wann, Noel, & Coleman, 1993). For example, the study by Spears et al. (1997) showed that differences between low and high identifiers in level of ingroup stereotyping were #1997 John Wiley & Sons, Ltd.

Intergroup threat and intragroup position 637 especially apparent when social identity was threatened. Only high identifiers responded to a group threat with more ingroup stereotyping in an attempt to defend their distinctiveness, while low identifiers responded to this threat by distancing themselves from the group. It has also been shown that identity threats generated by low status led to the perception of more within-group similarities for high identifiers compared to low identifiers (Doosje et al., 1995), and to less cohesivenss after group failure for participants who had not chosen a particular group membership (Turner et al., 1984). In a similar vein, high identifiers (but not low identifiers) with a sports team were especially derogatory about disloyal group members when the ingroup was threatened by the loss of a game (Branscombe et al., 1993). In another line of research, social value theory (e.g. McClintock, 1972) has emphasized the influence of individual differences on intergroup bias. It has been found that ingroup bias varies as a function of the social value orientation of participants. Such value preferences are assumed to be relatively stable within individuals (Kuhlman, Camac, & Cunha, 1986). Originally applied to interpersonal relations, in an intergroup context it has been found that individuals with competitive social value orientations prefer relative ingroup gain allocation strategies over fairness, while those with individualistic orientations prefer absolute ingroup gain over fairness. Individuals with prosocial value orientations generally prefer fair over biased intergroup allocation rules (e.g. Platow, McClintock, & Liebrand, 1990). In sum, there is now considerable support for the effect of individual difference variables on level of ingroup bias. The present approach differs in some important theoretical respects from the research on moderating effects of identification and social values. Rather than focusing on individual differences, we argue that ingroup bias can be moderated by more contextually defined intragroup differences. That is, level of ingroup bias can be related to features of the comparative context, and specifically to the extent to which one sees oneself as prototypical of the group. Prototypicality is not a fixed or stable property of the individual but can vary with the comparative context, helping to explain situational variability in willingness to display ingroup bias. The present research can therefore be seen as an attempt to integrate specific motivational predictions derived from social identity theory concerning the effects of a threat to distinctiveness with the self-categorization theory concept of prototypicality as a group member (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). Level of identification with a group and prototypicality as a group member should be positively correlated and have similar effects on level of ingroup bias in most social contexts. However, these constructs differ in three theoretical respects. First, as mentioned before and as defined by self-categorization theory, the prototypicality concept explicitly emphasizes context dependency. Although group identification can vary with context to some extent (e.g. intergroup competition may temporarily heighten group identification), or change over time (e.g. Doosje, Spears, & Ellemers, 1997), it is likely to be more stable over time and less influenced by the comparative context than is perceived prototypicality. Second, identification with a group, in our view, reflects the extent to which the group as a category is integrated into the selfconcept, while prototypicality as a group member is related to the position of the self in relation to other group members, and reflects how central or peripheral the group member is to the group in any given intergroup context (Turner, 1991; Turner et al., 1987). Third, in contrast to identification, prototypicality is more concerned with #1997 John Wiley & Sons, Ltd.

638 J. Jetten, R. Spears and A. S. R. Manstead self-perception in relation to the group and is therefore less directly tied to affect (e.g. self-esteem) and motivational factors (see Spears et al., 1997). Drawing on self-categorization theory, the assumption is made that the prototypical position in a group (defined in terms of meta-contrast) represents the shared views of group members and the group as a whole. The more an individual differs from the outgroup and the less he or she differs from the ingroup, the more this individual is prototypical of the ingroup 1. The idea that there are differences in prototypicality of group members, and that prototypical group members will be more influential because they represent what the group has in common, has driven much recent research on group polarization (Hogg, Turner, & Davidson, 1990; Turner, Wetherell, & Hogg, 1989). We predict that in intragroup terms, the position of individual group members within the group should moderate the response to a distinctiveness threat. Combining the distinctiveness threat principle derived from social identity theory with the notion of prototypicality derived from selfcategorization theory, we predict greater ingroup bias when group distinctiveness is threatened by the similarity of a relevant outgroup, but that it will be above all group members who define themselves as prototypical who will react to the distinctiveness threat in this way. The argument underlying this prediction is that, compared to more peripheral group members, prototypical group members should define themselves more in group terms. When the distinctiveness of this group is threatened, these group members should be particularly motivated to defend their group distinctiveness by adopting a group strategy. Peripheral group members have less in common with other ingroup members and should therefore be less inclined to adopt a group strategy such as ingroup bias as a reaction to threatened distinctiveness. An early study by Allen and Wilder (1975) addresses some of the central concerns of this study. Participants were arbitrarily categorized into two groups and then led to believe their own attitudes were either similar or dissimilar to attitudes held by other ingroup members, and either similar or dissimilar to attitudes held by the outgroup. In accordance with the authors' prediction, it was found that only when participants received feedback that they held similar attitudes to other ingroup members was more discrimination towards the outgroup displayed. However, and unexpectedly, there was no effect of similarity or dissimilarity of own attitude with that of the outgroup. Allen and Wilder suggested that this non-significant effect of feedback concerning outgroup attitude may be due to features of their procedure (e.g. the minimal categorization, and the nature of the dependent measure). Another more fundamental reason for the absence of an outgroup attitude effect may be that crucial information about the similarity of the ingroup to the outgroup attitude at an intergroup level was lacking in this study. This probably led to a more interpersonal comparison process which may not have been perceived as threatening distinctiveness. 1 Dependent on the comparative context, the prototypical position can coincide with the group mean, can be less extreme than the group mean, or more extreme than the group mean. Intergroup threat will be defined in the present paper in terms of group distinctiveness and varying intergroup distances which implies different prototypical positions in terms of meta-contrast. However, in order to prevent confounds due to between-condition variation in extremity positions within the group, we will define the prototypical position as the mean position in the group. Realizing that, as a result, the prototypicality manipulation does not exactly match the meta-contrast position, the manipulation of prototypicality can best be understood as more and less prototypical positions. We will refer to the less prototypical as the peripheral position and to the more prototypical as the prototypical position. #1997 John Wiley & Sons, Ltd.

Intergroup threat and intragroup position 639 We introduce two methodological refinements in the present research. First, in order to make the intergroup context salient, we provide participants with visual information about the means and distributions of both ingroup and outgroup. Participants are then provided with feedback about their own position in relation to the ingroup. Second, in line with recent research on stereotyping (Park, Judd, & Ryan, 1991), we argue that assessment of intergroup similarity requires not only information about the central tendency of ingroup and outgroup, but also about the variability of ingroup and outgroup. For instance, the ingroup and outgroup means can be dissimilar but if group variability is large enough for the two distributions to overlap, the two groups might still be perceived as quite similar. We therefore extend previous research on similarity of intergroup status or attitudes by presenting participants with feedback concerning both similarity, defined as the mean difference between ingroup and outgroup, and distribution, defined as the variability within the two groups. Intergroup differentiation is one means by which group members maintain positive social identity (Tajfel, 1982; Tajfel & Turner, 1986). It is predicted on the basis of social identity theory that under distinctiveness-threatening conditions, only group members who are able to express ingroup bias in order to restore group distinctiveness will subsequently show increased self-esteem (Lemyre & Smith, 1985; Oakes & Turner, 1980, for reviews see Abrams & Hogg, 1988; Hogg & Abrams, 1990; Long & Spears, 1997; Messick & Mackie, 1989). In a study by Branscombe and Wann (1994) only high identifiers reported elevated collective self-esteem following derogation of an outgroup as a result of identity-threatening conditions. In the nothreat condition, amount of outgroup derogation did not significantly influence subsequent self-esteem. Following this line of reasoning, we predict that only those group members who express heightened ingroup bias in response to threat to distinctiveness will show an increase in self-esteem. Hence, under distinctivenessthreatening conditions, only prototypical group members are predicted to express higher levels of group-dependent self-esteem. STUDY 1 In this experiment we examine the influence of threat to distinctiveness and intragroup position on ingroup bias in a modified minimal group setting. Distinctiveness threat is operationalized as groups having non-overlapping boundaries (low distinctiveness threat) or overlapping boundaries (high distinctiveness threat), keeping the mean intragroup distance constant. Prototypical group members are positioned centrally in the group; peripheral group members are located in the tail of the distribution close to the outgroup, or in the opposite tail of the ingroup distribution, and thus far removed from the outgroup distribution. It should be noted that, as a result of the outgroup position, the meta-contrast should be slightly higher for peripheral group members further away from the outgroup, as compared to the peripheral group members close to the outgroup position (see footnote 1). Nevertheless, the two peripheral positions are comparable in the sense that they both deviate considerably from the central group position, which is higher #1997 John Wiley & Sons, Ltd.

640 J. Jetten, R. Spears and A. S. R. Manstead in meta-contrast and prototypicality. We therefore predict negligible differences on the dependent measures between `near' and `far' peripheral group members. Ingroup bias by means of a product evaluation task forms the primary dependent measure of this study. We also employ measures of ingroup stereotyping, outgroup stereotyping and collective self-esteem. The stereotyping measures are intended to supplement the ingroup bias measure and assess whether the predicted effects generalize to these measures. To summarize our predictions, we expect that differences between prototypical and peripheral group members will be especially apparent when group distinctiveness is threatened. First, we predict that prototypical group members will be more inclined than peripheral group members to display ingroup bias as a response to a distinctiveness threat. In a similar vein, we predict that, compared to peripheral group members, prototypical group members will define the situation more in intergroup terms and thus be more inclined to differentiate between the groups on stereotypic dimensions when their group distinctiveness is under threat. This should be reflected in more outgroup stereotyping and in more ingroup stereotyping. Prototypical and peripheral group members should be less likely to differ in levels of ingroup stereotyping and outgroup stereotyping when group distinctiveness is not threatened. Related to this is the prediction that heightened ingroup bias (especially expected for prototypical group members whose distinctiveness is under threat), will subsequently lead to increased levels of grouprelated self-esteem. In order to test predictions concerning self-esteem related to group membership, we measure identity-specific collective self-esteem, rather than personal self-esteem (cf. Long, Spears, & Manstead, 1994). Method Design and Participants The design was a 2 (distinctiveness threat: low versus high)63 (intragroup position: prototypical, far peripheral, near peripheral) factorial, with random allocation of participants to conditions. Participants were 101 students at the University of Amsterdam (49 females and 52 males, distributed evenly over conditions), with an average age of just under 23 years. They received 10 Dutch guilders (approximately $6) for their participation. Procedure The experiment was run on personal computers. Participants received a short instruction about how to work with the computer and then started the experiment. The experiment was introduced as an investigation into `modes of perceiving'. Participants were led to believe that previous research had shown that there are two kinds of perceivers: detailed and global. They were told that the purpose of the present study was to examine how these modes of perception are distributed and to investigate whether there is a clear-cut distinction between global and detailed perceivers (a categorical division), or whether there is considerable variation within #1997 John Wiley & Sons, Ltd.

Intergroup threat and intragroup position 641 modes of perceiving, making it necessary to examine the distribution of these two modes (a more continuous approach). Participants were told that the aim of the experiment was to address this question of whether the categorical or the continuous approach yields the better insight into the way people perceive. As clarification, a parallel was drawn with gender categorization. It was pointed out that while biological sex is typically a clear-cut categorization, with no overlap, masculinity± femininity is a continuum, with variation within men and women and overlap in their distributions. This extended cover story was needed in order to enhance the credibility of feedback concerning group distributions, to be given later in the experiment. Two tasks followed: the `dice recognition task' and the `dot estimation task'. In the dice recognition task participants had to compare an example dice with five dice. Participants could see three sides of each dice. The five dice shared attributes (dots, lines, etc.) with the example. On four separate trials, participants had to choose the dice they considered to be most similar to the example. Participants were informed that this task measured perceptual style in a categorical fashion: it made it possible to distinguish between detailed and global perceivers. After completion of this task the computer started `calculating' their score and after a couple of seconds participants received (false) feedback about their mode of perceiving (all were categorized as detailed perceivers) 2. Participants were informed that a second task was necessary to determine the extent to which they were a detailed perceiver. In this second task participants were presented with seven 3-second trials in which a large number of dots appeared on the computer screen (cf. Gerard & Hoyt, 1974). Participants had to estimate the number of dots (which ranged between 40 and 100), and were informed that the computer would integrate the results of the two tests and that they would then receive feedback concerning these integrated results. Next, in order to increase identification with the ingroup, a group task was performed (Doosje et al., 1995). Participants had to estimate the number of black squares that appeared for 4 seconds on the computer screen. After giving their initial estimation, they received (false) feedback about the estimates of three other ingroup members and had to give their final estimate. Participants were led to believe that the result of their group's performance would be compared to the results of a group of global perceivers. Independent Variables After the group task participants were informed that they would be provided with preliminary findings of the study. They would be shown the distribution of scores of participants (detailed and global perceivers) who had completed the two tasks previously. On a continuum ranging from detailed (0) to global perceiving (100), they saw a graphical presentation (i.e. a histogram) of the frequency distribution of their own group's scores and the other group's scores. In all conditions, the ingroup distribution was situated on the left and that of the outgroup on the right. Half of the participants were provided with histograms in which the distance between the ingroup and outgroup histograms was small but non-overlapping (low distinctiveness threat condition). The other half were provided with ingroup and 2 A pilot study (N=16) showed that the attractiveness of being categorized as a detailed perceiver (M=5.00) did not significantly differ from that of being categorized as a global perceiver (M=5.25); t(15)51. #1997 John Wiley & Sons, Ltd.

642 J. Jetten, R. Spears and A. S. R. Manstead outgroup histograms that overlapped by 10 points on the continuum (high distinctiveness threat condition). The intragroup range (i.e. the difference between the scores of the least and most detailed perceivers and the least and most global perceivers) was 31. In all conditions, the mean score of the detailed perceivers was 33 while that of the global perceivers was 67. In order to have identical group means in the low and high distinctiveness threat conditions, the histograms of ingroup and outgroup were slightly skewed. Next, participants were informed that they would see the histograms for a second time and that their own score, reflecting the integrated results of the two tests, would be projected onto the ingroup histogram. Intragroup position was manipulated by projecting the score of prototypical group members at position 33 (i.e. exactly at the mean of the ingroup), the score of participants in the `far' peripheral condition at position 21, and the score of participants in the `near' peripheral group member condition at position 48. Thus the score `near' peripheral group members was in the middle of the overlap region in the high distinctiveness threat condition. Dependent Variables Three items were used to check whether the two groups were seen as more distinctive in the low distinctiveness threat as compared to the high distinctiveness threat condition (`To what extent do you feel that detailed perceivers are distinguishable from global perceivers?' `To what extent do you feel that detailed perceivers are different from global perceivers?' and `To what extent do you feel that detailed perceivers form a well-defined group?'). We also checked whether participants in the high and low distinctiveness threat conditions perceived the difference between the means of detailed and global perceivers to be similar (`To what extent did you think the means of both groups (which you can find in the middle of the distribution) were different?'). The manipulation of intragroup position was checked by means of three items (`I am very similar to the average detailed perceiver', `I have a lot in common with detailed perceivers' and `I am a good example of a detailed perceiver'). Ingroup bias was measured by having participants evaluate the creativity of ingroup and outgroup products. After the manipulation of prototypicality and distinctiveness threat, participants were told that the first part of the experiment was complete, and that the level of creativity of detailed and global perceivers would be examined in the second part of the experiment. Their own creativity would be tested by means of a task they would perform, together with other members of their group, in another room, after finishing all the computer tasks. Their group would then receive seven figures of different shapes from the experimenter (triangles, squares, etc.). Their task as a group would be to construct a figure from these different components and after 10 minutes show the experimenter the most creative figure. In order to `prepare them for the upcoming group task' participants would first be provided with figures made by groups who had participated in the experiment previously. Participants were told that these figures had not yet been evaluated and that their task was to judge the creativity of these figures. To make sure that participants would know whether they were judging an ingroup or an outgroup figure, they were instructed to write on a sheet of paper the type of group (detailed or global), the number of the group, and the title of the figure, before giving their #1997 John Wiley & Sons, Ltd.

Intergroup threat and intragroup position 643 judgment. Participants saw 12 group products (six ingroup figures and six outgroup figures) projected on the computer screen and had to make their creativity judgment on a 9-point scale ranging from `not at all creative' (1) to `very creative' (9). On the basis of pilot research we constructed two sets of figures that were matched for level of creativity. A given set of figures was presented to half of the participants as being constructed by the ingroup, and to the remaining participants as having been made by the outgroup. Ingroup stereotyping and outgroup stereotyping were measured by means of semantic differentials. On the basis of a pilot study, traits were selected that were stereotypical for detailed and global perceivers. Participants had to indicate on a 100-point scale where they judged the average ingroup member and the average outgroup member to be. One endpoint always represented stereotypical traits for detailed perceivers (math-oriented, accurate, profound, rigid, cautious), while the other endpoint represented stereotypical traits for global perceivers (languageoriented, slovenly, superficial, flexible, adventurous). Additionally, collective selfesteem was measured using four items on 9-point scales ranging from (1) `not at all' to (9) `very much' (`I think the detailed perceivers group have little to be proud of,' `I feel good about the detailed perceivers group,' `I have little respect for the detailed perceivers group,' and `I would rather not let others know that I belong to the detailed perceivers group' (cf. Ellemers, Kortekaas, & Ouwekerk, 1995). Results Manipulation Checks The three items measuring group distinctiveness were averaged (a=0.84) and analysed in a 2 (distinctiveness threat: low versus high)63 (intragroup position: prototypical, far peripheral group member, near peripheral group member) betweensubjects ANOVA. The expected main effect for distinctiveness threat was significant, F(1,95)=4.87, p50.05, indicating that participants perceived more group distinctiveness in the low distinctiveness threat conditions (M=4.80) compared to the high distinctiveness threat conditions (M=4.00). Intergroup distance was not perceived differently over conditions (no significant main or interaction effects). An ANOVA on the combined manipulation check of prototypicality (a=0.82) revealed, as predicted, a main effect for intragroup position, F(2,95)=4.13, p50.05. Participants in the prototypical condition perceived themselves as more prototypical group members (M=5.53), compared to participants in the far peripheral group member condition (M=4.47) or the near peripheral group member condition (M=4.74). The interaction between prototypicality and distinctiveness threat was not significant, F(2,95)=1.05, n.s. Intergroup Differentiation Judgments of the creativity of the two sets differed reliably, F(1,99)=5.08, p50.05. Therefore, set was used as a covariate in all analyses of intergroup differentiation. The adjusted means are reported in Table 1. The ingroup creativity judgments and outgroup creativity judgments (i.e. ingroup bias) were analysed in a 2 (distinctiveness threat)63 (intragroup position)62 (target group) ANCOVA, with repeated #1997 John Wiley & Sons, Ltd.

644 J. Jetten, R. Spears and A. S. R. Manstead Table 1. Study 1: Adjusted mean evaluation of ingroup products, adjusted mean evaluation of outgroup products and ingroup bias Distinctiveness threat Low High Prototypical N 17 17 group member Ingroup evaluation 6.26 (0.68) 6.73 (0.63) Outgroup evaluation 6.21 (0.65) 6.33 (0.92) Ingroup bias 0.05 (0.58) 0.40* (0.88) Far peripheral N 16 16 group member Ingroup evaluation 6.33 (0.64) 6.31 (0.81) Outgroup evaluation 5.97 (0.96) 6.13 (0.84) Ingroup bias 0.37 (0.74) 0.18 (0.99) Near peripheral N 18 17 group member Ingroup evaluation 6.47 (0.56) 6.18 (0.76) Outgroup evaluation 6.29 (0.84) 6.36 (0.92) Ingroup bias 0.19 (0.60) 70.18 (0.90) Note. *Ingroup bias significantly greater than zero: p50.05. A higher mean indicates a higher creativity judgment (on a 9-point scale). measures on the last factor and set as a covariate. The predicted three-way interaction was only marginally significant, F(2,95)=1.97, p50.14. In order to maximize the power of the tests, the influence of distinctiveness threat and prototypicality on ingroup bias was analysed using planned comparisons (Rosenthal & Rosnow, 1984). First we predicted that ingroup bias would be higher in the high distinctiveness threat compared to the low distinctiveness threat condition, but only for prototypical group members. Second, we predicted no differences in ingroup bias between prototypical and peripheral group members (i.e. the average of far and near peripheral group members) in the low distinctiveness threat condition, but significant differences between prototypical and peripheral group members in the high distinctiveness threat condition. In contrast to the predictions tested in the first set of planned comparisons, ingroup bias was not significantly higher for prototypical group members in the high compared to the low distinctiveness threat condition. Ingroup bias did not differ between low and high distinctiveness threat for prototypical group members, far peripheral group members or near peripheral group members (F(1,95)=1.88, n.s.; F(1,95)51; and F(1,95)=1.71, n.s., respectively). However, ingroup bias was, as predicted, higher for prototypical as compared to peripheral group members within the high distinctiveness threat condition, although this effect was only marginally significant, F(1,95)=2.96, p50.09; bias in these groups did not differ significantly in the low distinctiveness threat condition, F(1,95)=1.00, n.s. Planned comparisons on ingroup and outgroup product evaluations revealed that differences were caused more by variations in ingroup evaluations between conditions than by variations in outgroup evaluations. Ingroup product evaluations were higher in the high distinctiveness threat compared to the low distinctiveness threat condition for prototypical group members, F(1,95)=4.16, p50.05, but not for far peripheral group members, F51, or for near peripheral group members, F(1,95)=1.51, n.s. Furthermore, the ingroup product evaluations of #1997 John Wiley & Sons, Ltd.

Intergroup threat and intragroup position 645 prototypical and peripheral group members did not differ under low distinctiveness threat (F51), but did so under high distinctiveness threat, F(1,95)=5.62, p50.05. Finally, ingroup bias did not differ between far and near peripheral group members in either the low or the high distinctiveness threat condition. Overall, ingroup bias was significant, F(1,95)=4.57, p50.05. Further analysis revealed that the mean ingroup evaluation significantly exceeded the mean outgroup evaluation only when distinctiveness threat was high and group membership was prototypical, F(1,95)=4.60, p50.05 (see Table 1). Group stereotyping Scores representing the difference between ingroup and outgroup evaluation on each of the five semantical differentials were submitted to a principal components analysis with varimax rotation. This analysis revealed two components, explaining 69.6 per cent of the variance. The first component (42.7 per cent of explained variance) can be interpreted as way of behaving and consisted of the scales `rigid±flexible' and `cautious±adventurous'. The first-named trait in each pair is stereotypical for ingroup (detailed perceivers), while the second-named trait is stereotypical for the outgroup (global perceivers). The loadings were 0.90 and 0.88, respectively. The second component (explaining 26.9 per cent of the variance) can be interpreted as way of thinking and consisted of the scales `math-oriented±language-oriented', `accurate±slovenly' and `profound±superficial'. The loadings were 0.83, 0.81 and 0.56, respectively. Judgments of the ingroup and outgroup on these dimensions were averaged for each component and analysed using the same set of planned comparisons as those reported above (the predicted three-way interactions did not reach conventional significance thresholds for the first, F(2,95)=2.23, p50.11, or the second component, F51). With respect to the first prediction, ingroup stereotyping was indeed higher for prototypical group members on the first component in the high distinctiveness threat as compared to the low distinctiveness threat condition, F(1,95)=3.92, p50.05. This was not true for either far peripheral group members, F51, or near peripheral group members, F51. Outgroup stereotyping on the first component did not vary as a function of distinctiveness threat for prototypical group members F(1,95)=1.90, n.s. or for peripheral group members (far peripheral group members and near peripheral group members, F51 and F(1,95)=1.90, n.s., respectively). Mean scores on the first component are presented in Table 2. Turning to the second set of planned comparisons, level of ingroup stereotyping and outgroup stereotyping on the first component did not differ between prototypical and peripheral group members within the low distinctiveness threat condition (F51). However, in the high distinctiveness threat condition, prototypical and peripheral group members differed marginally significantly for ingroup stereotyping, F(1,95)=2.99, p50.09 and significantly for outgroup stereotyping, F(1,95)=3.92, p50.05. As predicted, far peripheral group members did not differ from near peripheral group members with respect to either ingroup stereotyping or outgroup stereotyping, in either the low or the high distinctiveness threat condition. #1997 John Wiley & Sons, Ltd.

646 J. Jetten, R. Spears and A. S. R. Manstead Table 2. Study 1: Ingroup stereotyping and outgroup stereotyping Distinctiveness threat Low High Prototypical N 17 17 group member Ingroup stereotyping 54.03 (8.85) 62.76** (13.21) Outgroup stereotyping 43.00* (9.99) 37.47** (14.16) Far peripheral N 16 16 group member Ingroup stereotyping 54.53 (11.55) 58.56** (11.70) Outgroup stereotyping 42.72* (10.86) 42.59** (8.94) Near peripheral N 18 17 group member Ingroup stereotyping 55.83 (18.05) 53.62 (11.62) Outgroup stereotyping 40.72* (15.46) 46.18 (8.54) Note. A mean higher than 50 indicates more stereotyping on a stereotypical ingroup trait and a mean lower than 50 means more stereotyping on a stereotypical outgroup trait (on a 100-point scale). Significantly different from 50: *p50.05, **p50.01. Equivalent sets of planned comparisons on the second component did not reveal any significant effects. Collective Self-Esteem The items of the collective self-esteem scale were recoded and averaged (a=0.74). The influence of distinctiveness threat and prototypicality on collective self-esteem was analysed using planned comparisons. These analyses revealed no significant effects for the crucial comparisons. Furthermore, the correlation between level of ingroup bias and collective self-esteem was not significant (r=70.06, n.s.) and none of the within-condition correlations was significant. Discussion We predicted that differences between prototypical and peripheral group members would be especially apparent when group distinctiveness was threatened. The results supported this hypothesis. Only prototypical group members defended their threatened distinctiveness by engaging in increased levels of ingroup bias. Level of ingroup bias only reached significance in this condition, indicating that ingroup bias is not an inevitable consequence of intergroup comparisons. Furthermore, as often observed in previous research, ingroup bias was caused by a higher evaluation of the ingroup products and not by a devaluation of outgroup products (Brewer, 1979). Peripheral group members who did not identify themselves as typical group members showed less or no ingroup bias, presumably because they were less motivated to preserve their distinctive identity when this identity was threatened by overlapping group boundaries. Similar results were obtained on the group stereotyping measures. When distinctiveness was threatened, prototypical group members showed more ingroup #1997 John Wiley & Sons, Ltd.

Intergroup threat and intragroup position 647 and more outgroup stereotyping, compared to peripheral group members. As predicted, then, under distinctiveness threatening conditions, prototypical group members perceived group members in stereotypical terms. Noteworthy is that this heightened ingroup stereotyping occurred on dimensions that were not only stereotypical for the ingroup, but also negative (rigid and cautious), while heightened outgroup stereotyping took place on dimensions that were not only stereotypical for the outgroup but also positive (flexible and adventurous). Spears et al. (1997) obtained a similar finding in natural groups. No differences were found between peripheral group members located near to the outgroup and peripheral group members far from the outgroup. This rules out the possibility that ingroup bias was lower among the former participants because of any confusion about their group membership. Furthermore, this finding shows that it is not the absolute distance between oneself and the outgroup that determines reaction to a group distinctiveness threat, but rather the centrality of one's own position in the ingroup (but see footnote 1). No support was found for the hypothesis that collective self-esteem would increase as a direct function of elevated ingroup bias. We suspect that one reason for this non-significant result lies in the nature of the groups used in this modified minimal group setting. It has been suggested by Branscombe and Wann (1994) that when the threatened identity is not important to the self, as is typically the case in a minimal group context, the need to derogate should be lessened and self-esteem will not be dependent upon that identity. It is therefore probably useful to draw a distinction between minimal and natural group contexts concerning the effect of intergroup discrimination on collective self-esteem (Abrams & Hogg, 1988; Hogg & Abrams, 1990) 3. We expect that in better-established groups, in which group membership has more meaning for the individual and identity is more important, self-esteem will be more strongly related to distinctiveness threats, and as a consequence ingroup bias strategies will be used to restore group distinctiveness. This reasoning is tested in Study 2 where the effects of a distinctiveness threat and prototypicality as a group member are examined in a group to which individuals are more strongly attached and where the group identity is better established. STUDY 2 In this second study we attempted to replicate the findings of Study 1, this time using better-established, natural groups. There were two reasons for wanting to replicate these findings in real-life groups. First, as discussed above, we suspect that the nonsignificant relation between ingroup bias and level of collective self-esteem found in 3 Evidence for self-esteem as either the basis or result of intergroup discrimination in minimal group experiments is mixed. There are studies that report that opportunities to engage in intergroup discrimination can enhance self-esteem in minimal group contexts (e.g. Lemyre & Smith, 1985; Oakes & Turner, 1980). Other studies showed that in minimal groups, self-esteem was unrelated to differentiation (e.g. Wagner, Lampen, & Syllwasschy, 1986). It has been suggested that these inconsistent findings may be the result of problems with the appropriate measurement of self-esteem (Abrams & Hogg, 1988; Hogg & Abrams, 1990; Long et al., 1994). #1997 John Wiley & Sons, Ltd.

648 J. Jetten, R. Spears and A. S. R. Manstead Study 1 was due to the nature of the group. Our revised hypothesis is that collective self-esteem only varies as a function of threats to important identities. Secondly, it has been shown in previous research that patterns of ingroup bias found in artificial groups are not always replicated in real-life groups, where reverse patterns have even been found (Jetten, Spears, & Manstead, 1996; Mullen, Brown, & Smith, 1992). It is quite possible that such differences relate to variations between minimal and natural groups in level of identification and the resulting willingness to defend one's group (Spears et al., 1997). Replication of Study 1 using more natural groups would allow us to establish whether the effect of group distinctiveness threat and the intragroup position found in minimal groups extends to natural groups. In our second study participants are all members of the University of Amsterdam and this identity is made salient. The outgroup consists of students at a rival university in Amsterdam, namely the Free University. The nature of the groups was not the only difference between the two studies. We also manipulated distinctiveness threat slightly differently, by operationalizing it as intergroup similarity at one of three levels. The attitudes of ingroup and outgroup members were defined as being separated by a large, medium or small intergroup distance. In the small intergroup distance condition the groups partially overlapped. We expected this to be the most distinctiveness-threatening condition, because some outgroup members are described as having similar attitudes as some ingroup members. Furthermore, because we demonstrated in Study 1 that differences between peripheral and prototypical group members were due to centrality of the member's position in the group and not by absolute difference between the member's position and that of the outgroup, we only used the `near' peripheral position to implement peripheral group membership. Finally, feedback about the intergroup context and the position of the group member, did not, as in Study 1, relate to the dimension on which the group member was categorized, but rather to a separate attitude dimension. Method Participants and Design Participants were 106 students (39 female and 67 male) at the University of Amsterdam, with an average age of 22.5 years. They were recruited in the university canteen and participated on a voluntary basis. The experiment was a 3 (intergroup distance: large, medium, small)62 (intragroup position: prototypical versus peripheral group membership) between-subjects design. Procedure Participants were informed that the study involved a comparison between students of the University of Amsterdam and the Free University on the dimension `belief in supernatural phenomena'. In order to measure their own views on this topic, #1997 John Wiley & Sons, Ltd.

Intergroup threat and intragroup position 649 participants were asked to complete a 20-item questionnaire consisting of items such as `Reincarnation does not exist' and `I believe it is possible to have contact with the deceased' 4. They answered on 7-point scales, with 1 indicating `not at all' and 7 `very much'. After completing this questionnaire, participants were informed that the experimenter would need some time to calculate their scores and that they would in the meantime be provided with preliminary results showing the scores of University of Amsterdam (ingroup) and Free University (outgroup) students. Independent variables Participants were provided with a graphical presentation of the frequency distribution of ingroup and outgroup scores on the dimension `belief in supernatural phenomena' which they were able to study for approximately 3 minutes. The ingroup and outgroup scores were located on a continuum ranging from `no belief in supernatural phenomena' to `strong belief in supernatural phenomena'. The ingroup and outgroup were represented by two separate histograms that were equal in variability (50 mm). The ingroup distribution was situated to the left of that of the outgroup in all conditions, indicating that University of Amsterdam students believed in supernatural phenomena to a lesser extent than Free University students 5. Intergroup distance was manipulated by varying the means of the ingroup and outgroup frequency distributions. In the large intergroup distance condition the distance between the ingroup mean and the outgroup mean was 90 mm (the total continuum was 150 mm). The distance between the most extreme ingroup member and the most extreme outgroup member was 40 mm. The distance between the ingroup and outgroup mean in the medium intergroup distance condition was 50 mm, with only 5 mm separating the two distributions. In the small intergroup distance condition, there was a 35-mm difference between the ingroup and outgroup means, and a 10-mm overlap between the ingroup and outgroup distributions. After 3 minutes the experimenter announced that she had calculated the scores of the participants. Intragroup position was manipulated by providing participants with (false) feedback about their own score. The experimenter marked the position of each participant in the ingroup distribution. The mark was located exactly in the middle (the mean) of the ingroup distribution in the prototypical condition, and on the right-hand extreme of the distribution (i.e. closer to the outgroup) in the peripheral condition. As a consequence, the score of a participant labelled as a peripheral group member in the small intergroup condition was situated in the area where ingroup and outgroup overlapped. 4 A pilot study was performed to check whether the dimension `belief in supernatural phenomena' is neutral and value free. First, participants (N=15) had to complete the 20-item questionnaire about `belief in supernatural phenomena' and then they had to indicate in relation to five different positions on the continuum, varying from (1) `no belief in supernatural phenomena' to (5) `a lot of belief in supernatural phenomena', how pleasant they considered each of these five positions (7-point scale ranging from (1) `not at all pleasant' to (7) `very pleasant'). t-tests showed that there were no differences in rated attractiveness of these five conditions. All five positions were rated as moderately pleasant (M=4.03). 5 Our choice to situate the ingroup at the less extreme, and the outgroup at the more extreme end of the `belief in supernatural phenomena' continuum is consistent with existing stereotypes about these two universities. #1997 John Wiley & Sons, Ltd.