Reducing Intergroup Bias: the Moderating Role of Ingroup Identification Richard J

Reducing Intergroup Bias: The Moderating Role of Ingroup Identification Richard J. Crisp, Sarah R. Beck

To cite this version:

Richard J. Crisp, Sarah R. Beck. Reducing Intergroup Bias: The Moderating Role of Ingroup Iden- tification. Group Processes and Intergroup Relations, SAGE Publications, 2005, 8 (2), pp.173-185. 10.1177/1368430205051066. hal-00571601

HAL Id: hal-00571601 https://hal.archives-ouvertes.fr/hal-00571601 Submitted on 1 Mar 2011

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. 06 Crisp (bc/s) 29/3/05 1:11 pm Page 173

Group Processes & G Intergroup Relations P 2005 Vol 8(2) 173–185 I R Reducing Intergroup Bias: The Moderating Role of Ingroup Identiﬁcation

Richard J. Crisp and Sarah R. Beck University of Birmingham

Recent work developing interventions designed to reduce intergroup bias has sometimes yielded disparate findings. We tested whether the varying effectiveness of such interventions may have a motivational basis. In two experiments we examined whether differential ingroup identification moderated the effectiveness of a differentiation-reducing intervention strategy. In Experiment 1, thinking of characteristics shared between the ingroup and outgroup reduced ingroup favoritism to a greater extent for lower identifiers than for higher identifiers. In Experiment 2 we replicated this finding with different target groups and evaluative measures while controlling for information load. We discuss the implications of this work for developing social psychological models of bias-reduction.

keywords ingroup identiﬁcation, intergroup bias, reducing prejudice, social categorization

T HE EXTENT to which social categories— Hewstone, & Rubin, 2001; Deschamps & Doise, ingroups and outgroups—possess overlapping 1978) include the notion that, albeit via characteristics is a key determinant of how such different methods, emphasizing common groups are evaluated (Brewer, 1991; Tajfel & ground between groups and group members Turner, 1979; Turner, Hogg, Oakes, Reicher, & can be a good thing. Recently, however, some Wetherell, 1987). It is not surprising then that findings have raised the possibility that concepts of intergroup overlap and distinctive- reducing intergroup differentiation may not ness are central in deﬁning not only our under- always be the key to improved intergroup standing of intergroup relations but also social relations, and that in some cases doing so could psychologists’ attempts at reducing intergroup even make matters worse. In this article we bias. Models of contact (Brewer & Miller, 1984; examine this possibility and the moderating Hewstone & Brown, 1986; Miller, Brewer, & Edwards, 1985; Pettigrew, 1998), the formation of a common ingroup identity (Gaertner & Author’s note Dovidio, 2000; Gaertner, Mann, Dovidio, Address correspondence to R. J. Crisp, School Murrell, & Pomare, 1990; Gaertner, Mann, of Psychology, University of Birmingham, Murrell, & Dovidio, 1989), and crossed Edgbaston, Birmingham B15 2TT, UK categorization (Crisp & Hewstone, 1999; Crisp, [email: [email protected]]

Group Processes & Intergroup Relations 8(2)

conditions that may influence the effectiveness these models the principal aim is to reduce cog- of differentiation-based interventions for nitive differentiation between groups. Based on reducing intergroup bias. the theoretical accounts outlined above, this is a sensible strategy: if dichotomized categorical Social categorization and category differentiation is a key contributor to ingroup differentiation favoritism, then to tackle the latter we must address the former. Allport (1954) was among Ever since Sherif’s (1966) seminal summer the first to suggest that bringing groups closer camp studies, the link between category differ- together might provide the basis for improved entiation and ingroup favoritism has been an intergroup relations. In Sherif’s (1966) classic enduring feature of work on intergroup studies, it was cooperative contact, established relations. Merely distinguishing between people after the imposition of a categorical distinction, on the basis of their group affiliations appears that reduced ingroup favoritism. A sustained to be sufficient to produce ingroup favoritism. and extensive research program emerged from Tajfel’s research with the Minimal Group this notion that bringing groups together can Paradigm (Tajfel, Billig, Bundy, & Flament, be an important factor in creating more 1971) illustrated how even with no prior harmonious intergroup relations (e.g. Brewer contact, under conditions of anonymity, and & Miller, 1984; Hewstone & Brown, 1986; with meaningless social categories, the knowl- Pettigrew, 1997; Wright, Aron, McLaughlin- edge that ‘they’ are different to ‘us’ can trans- Volpe, & Ropp, 1997). late into evaluative differentiation (see Brewer, Although contact can have positive effects via 1979; Mullen, Brown, & Smith, 1992). affective mediators (e.g. Islam & Hewstone, Models have been proposed to explain this 1993), one key aspect of the approach is the impact of categorization on intergroup atti- establishment of perceived commonalities tudes. Campbell (1956) and Tajfel (1969; Tajfel between the ingroup and outgroup. This idea is & Wilkes, 1963) outlined accentuation and expressed in Gaertner and Dovidio’s (2000) differentiation principles that provide frame- Common Ingroup Identity Model. This model works for understanding the processes involved arose to explain why cooperation and contact in categorical perception. These ideas have can be successful at reducing intergroup bias. been expanded in various theoretical accounts In their 1989 and 1990 studies Gaertner and such as Doise’s (1978) Category Differentiation Dovidio (Gaertner et al., 1989, 1990) experi- Model and in the form of the meta-contrast mentally created intergroup contexts in which process (Oakes, Haslam, & Turner, 1994) groups were either segregated and distinct, or outlined by Self-Categorization Theory (Turner were cooperatively integrated. Their findings, et al., 1987). While the emphasis of these in line with the established effects of contact in accounts varies, what is common is the notion varied settings, were that cooperative contact that categorization affords a psychological basis was successful at reducing intergroup bias. for understanding ‘them’ to be different to ‘us’, More importantly, they established that the for- and it is this distinction between ingroups and mation of a common ingroup identity was the outgroups that provides the prerequisite for mediating process. Put another way, bias was intergroup discrimination. reduced following cooperative contact because psychological boundaries between the ingroup Methods and models for reducing and outgroup were broken down, and a new intergroup bias overarching group was formed that included former outgroup members. The theoretical emphasis on psychologically Subsequent work has expanded on the varied represented intergroup differences provides conditions that increase the psychological the focus for social psychological interventions overlap between the ingroup and outgroup, and targeted at reducing prejudice. For many of that have bias-reducing effects. Goal relations

174 06 Crisp (bc/s) 29/3/05 1:11 pm Page 175

Crisp & Beck reducing intergroup bias

and group interactions (Turner, 1982), percep- consequences, implications, and processes that tions of a common task or fate (Brown & might follow from social categorization and Abrams, 1986; Brown & Wade, 1987; Gaertner group affiliation. This perspective argues that et al., 1990), or even simply making a super- we are not simply passive members of social ordinate, or cross-cutting, categorization salient groups—some groups mean more to us than (Crisp, Ensari, Hewstone, & Miller, 2003; others, and, when they do, we use them as an Hornsey & Hogg, 2000) can all reduce the cog- important source of self-esteem (Abrams & nitive differentiation between ingroups and out- Hogg, 1998; Ellemers, Spears, & Doosje, 2002) groups, and thus, improve intergroup relations. or as a means to reduce subjective uncertainty (Hogg, 2000, 2001; see also Grieve & Hogg, Are boundaries always bad? 1999; Hogg & Mullin, 1999). Since such groups are valuable to self-conception, people want to Based on the work above, one might conclude maintain the perception of them as being (a) that we should always strive to emphasize the positive and (b) clearly distinguishable from overlapping characteristics of groups in order to other, relevant, comparison groups. create more harmonious intergroup relations. In the context of blurring intergroup bound- There are, however, some grounds for question- aries, the motivation to reduce subjective un- ing whether promoting overlap is always the best certainty is particularly applicable. The loss of way to reduce intergroup bias. Merging category prescriptive ingroup norms and defined boundaries and forming a superordinate outgroup stereotypes that accompany weakened representation has recently been found to some- categorical differentiation is an unpleasant times worsen intergroup relations (e.g. Hornsey state for perceivers invested in the self- & Hogg, 2000) or at least be a less effective definitional benefits of ingroup affiliation. strategy than alternative methods (Dovidio, Hogg (2000, 2001) has illustrated how meta- Gaertner, & Validzic, 1998; González & Brown, contrast and positive differentiation processes 2003). Convergent evidence can be found in are accentuated in such conditions that work on organizational mergers. Mergers can promote uncertainty in the social context. often cause the previously distinct subgroups to Given the opportunity, the best way to enhance engage in heightened ingroup favoritism (see the positive distinctiveness of the ingroup from Terry & Callan, 2001; Terry, Carey, & Callan, other groups is to differentially evaluate in 2001). These findings are clearly counter to the favor of the ingroup. Supporting this perspec- notion that recognizing overlapping character- tive, group members do indeed appear com- istics, reducing differentiation, and merging pelled to differentiate their ingroup from intergroup representations, can always improve similar outgroups on relevant dimensions of intergroup relations. It has recently been sug- comparison (see Brewer, 1979; Brown & gested that motivations to retain a distinctive Abrams, 1986; Roccas & Schwartz, 1993; Tajfel, social identity may provide an explanation for 1982; van Knippenberg & Ellemers, 1990). these divergent findings (see Hornsey & Hogg, Importantly, the application of such moti- 2000; see also Terry et al., 2001). It is these motivational processes appears to be contingent vations that we focus on here. upon the groups being perceived as important to self-definition. People who regard their Motivations to maintain ingroup as important—who are highly commit- distinctiveness ted to it—can be considered high in ingroup identification (Spears, Doosje, & Ellemers, 1997). Ingroup favoritism can not only be attributable Only when identification with the ingroup is to the psychological dichotomization of high does weakened differentiation appear to ingroups and outgroups. The Social Identity motivate increased ingroup favoritism (Hogg & perspective (Tajfel & Turner, 1979; Turner Abrams, 1990; Hogg & Turner, 1985; Jetten, et al., 1987) outlines a number of motivational Branscombe, Spears, & McKimmie, 2003; Jetten,

175 06 Crisp (bc/s) 29/3/05 1:11 pm Page 176

Group Processes & Intergroup Relations 8(2)

Spears, & Manstead, 2001; Spears, Jetten, & category overlap), or not (the baseline com- Scheepers, 2002). Indeed, in some recent parison group). All participants then completed research using minimal groups and a paradigm measures to assess ingroup favoritism. We designed to test the effects on group percep- expected this task to reduce ingroup favoritism, tion of a merger between the ingroup and contingent on the extent of participants’ outgroup, van Leeuwen, van Knippenberg, and ingroup identification. Ellemers (2003) found that pre-merger identification predicted post-merger intergroup atti- Method tudes. Specifically, prior to the merger Participants and design A total of 168 under- identification did not predict intergroup evalu- graduates at the University of Birmingham (66 ations. Following the merger, however, identifi- females, 72 males, with 30 participants failing to cation was positively correlated with ingroup state their gender, mean age 20) were allocated favoritism. These findings are clearly indicative across two between-subjects task conditions: that, specifically for higher identifiers, creating baseline versus overlapping groups. Identifi- category overlap could sometimes increase cation was a continuous variable, measured ingroup favoritism. prior to the manipulation. Birmingham Uni- versity students were the ingroup and students The current research from a local rival university, Aston University, were the outgroup. These two universities are While there is now evidence that, under some well-established in the UK Midlands and, due to circumstances, differentiation-based inter- geographical proximity, have an extensive ventions for reducing intergroup bias may be of history of intergroup comparison. As such, the variable effectiveness, the precise nature of these comparison and cover story (see below) were conditions remains unspecified, and potential relevant and credible to our participants. moderators empirically untested. From the Participants received a small monetary above review of social identity research on payment for their involvement. distinctiveness threat, we can predict, however, that the effectiveness of differentiation- Procedure Participants were approached reducing interventions will be related to pre- around campus by a female experimenter and existing levels of ingroup identification. We asked to complete a short questionnaire con- therefore hypothesized that a task designed to cerned with opinions of students at different increase the perceived overlap between the universities in the region. They were told that ingroup and outgroup would be more effective they would be asked some questions concerning at reducing intergroup bias for lower compared their attitudes and feelings toward the different to higher identifiers. In the two studies student groups. Participants first completed the reported here we tested this hypothesis. pre-manipulation measure of ingroup identification. The four items were adapted from Jetten Experiment 1 et al. (2003) and Luhtanen and Crocker (1992): ‘I identify strongly with other Birmingham The aim of Experiment 1 was to test the mod- students’, ‘Being a Birmingham student is an erating effect of ingroup identification on the important part of who I am’, ‘I feel strong ties effectiveness of a differentiation-reducing with other Birmingham students’, ‘I feel a strategy analogous to those developed in strong sense of solidarity with other Birming- the bias-reduction literature. We measured ham students’ (1 not at all, 9 very much so). These ingroup identification (with ‘University’ group items formed a reliable index (␣ = .879). membership) and then had participants simply Participants in the baseline condition then list characteristics shared between the ingroup simply completed the dependent measures. Prior and outgroup (as a means of reducing to this, those in the experimental condition were ingroup–outgroup differentiation and creating presented with the following instructions:

176 06 Crisp (bc/s) 29/3/05 1:11 pm Page 177

Crisp & Beck reducing intergroup bias

We would like you to think of up to five things that interaction variable by contrast coding task con- students at the University of Birmingham and dition as 0 and 1 (baseline vs. overlapping) and students at the University of Aston may have in multiplying it by the centered identification common (i.e. characteristics that are shared scores for each participant. We then entered between the two groups). this interaction variable into a multiple regres- Participants were instructed to write down sion on a second step, following the entry of the the characteristics they thought of, before com- task and identification factors independently at pleting the dependent measures and being step 1. This analysis revealed a significant inter- thanked and debriefed. action between identification and task at step 2 The effectiveness of this task for promoting 2(R Ch = .028, F(1, 164) = 4.84, p < .05). We the perception of increased intergroup overlap tested for differences between the baseline and has been established in our previous research. overlapping conditions at lower and higher Cocker and Crisp (2003, see also Cocker, 2004) levels of ingroup identification at –1 and +1 SDs gave 80 female undergraduate psychology respectively. This analysis revealed that at the students the task instructions described above, lower level of identification bias was signifi- but relating to Psychology and English subject cantly lower for participants in the overlapping (major) groups. They were then asked to compared to baseline condition (␤ = –.282, indicate the extent to which they perceived the t = –2.59, p < .05). At the higher level of identifi- ingroup and outgroup as ‘groups that have cation there was, however, no difference numerous overlapping characteristics’ (1 not at between task conditions (␤ = .062, t = .563, all, 9 very much so). Cocker and Crisp found that p = .574). We also tested whether identification following the overlapping characteristics task predicted bias within task conditions. Simple participants perceived the in- and outgroup as slopes analysis revealed that in the baseline having significantly more overlapping charac- condition identification was unrelated to bias teristics compared to participants in the (␤ = –.144, t = –1.30, p = .196), but following baseline condition. generation of overlapping characteristics identification was positively related to bias Dependent measures To measure ingroup (␤ = .213, t = 2.00, p < .05) (see Figure 1). favoritism participants were informed that To summarize, in line with our hypothesis we another aim of the research was to gain observed lower levels of bias for participants students’ opinions about how government- who completed the overlapping characteristics allocated money to universities in the local task—but only if they were lower identifiers. region should be divided. There followed a list Furthermore, within task condition analysis of 11 allocation options, ranging in 10% revealed that the effectiveness of the overlap- intervals from ‘100% to students at Birming- ping characteristics task at reducing bias was ham/0% to students at Aston’, through a related to participants’ level of identification: 50%/50% allocation, to ‘0% to students at the lower the level of pre-task identification Birmingham/100% to students at Aston’ reported by the participant, the greater the (adapted from Hornsey & Hogg, 2000). Partici- reduction in bias after completion of the task. pants ticked their preferred allocation. We coded the ingroup percentage allocation as a Experiment 2 measure of ingroup favoritism. Most work in the bias-reduction literature tests Results and discussion the effectiveness of interventions compared to We used moderated regression to assess the a no-intervention baseline condition (arguably interaction of our continuous identification the most appropriate in terms of testing the variable with the manipulation of group practical efficacy of an intervention). In Experi- characteristics overlap (Aiken & West, 1991; ment 1 this was also our strategy. It could be West, Aiken, & Krull, 1996). We computed an argued, however, that simply completing any

177 06 Crisp (bc/s) 29/3/05 1:11 pm Page 178

Group Processes & Intergroup Relations 8(2)

Baseline 80 Overlap

Ingroup favoritism 50

40 lower (–1 SD) higher (+1 SD) Identification

Figure 1. Ingroup favoritism as a function of overlapping characteristics and identiﬁcation, Experiment 1.

task in the experimental condition could have ments are more relevant, reacted negatively to had the same effects that we observed. This is the load imposed by our task, but that lower because there was, notwithstanding the specific identifiers were less sensitive to such load, the requirements of our task, an information load intergroup context being less important to differential between the baseline and experi- them. Using a baseline task that requires mental conditions. Urban and Miller (1998), exactly the same mental procedure as the with reference to the crossed categorization experimental condition, but with reference to literature, argued that complex task instructions irrelevant categories would provide a compara- can sometimes lead to mildly negative affect, tive test of equal load conditions. Observing an which can correspondingly lead to negative impact of differential identification only when judgments in intergroup contexts. With respect the task is related to the target groups would to the findings from Experiment 1, perhaps thus rule out the competing explanation that it higher identifiers, for whom intergroup judg- is load induced by the task that is responsible

178 06 Crisp (bc/s) 29/3/05 1:11 pm Page 179

Crisp & Beck reducing intergroup bias

for the effects. In Experiment 2 we therefore Dependent measures In this experiment we used a baseline condition in which participants used an alternative measure of ingroup generated overlapping characteristics between favoritism, in order to improve the generaliz- two categories that were unrelated to the target ability of our findings. Evaluations were groups they later evaluated. We also tested assessed with an attitude thermometer (see different target groups and used a different Haddock, Zanna, & Esses, 1993). Participants evaluative measure to increase the generaliz- were asked to indicate their general attitudes ability of our findings. toward Southerners and Northerners on separate thermometers (0 extremely unfavorable, Method 100 extremely favorable). They were asked to mark Participants and design Thirty-six high school a cross on the thermometers which reflected students (25 females, 11 males, mean age 17) their initial feeling that came to mind on were allocated across two between-subjects task reading the category labels ‘Southerners’ and conditions: baseline (overlapping unrelated) ‘Northerners’, and to write the exact number versus overlapping (related). Identification was beside the cross that they made. again a pre-manipulation measured variable. The target ingroup and outgroup were ‘South- Results and discussion erners’ and ‘Northerners’—relating to where Evaluations Again we used moderated regres- participants lived in the UK (all participants sion to assess the interactive effects on bias of were members of the former group). This is a identification and the characteristics task well-established and socially significant inter- manipulation. There was a significant inter- 2 group distinction in the UK, which is clearly action between identification and task (R Ch = delineated by accent. .099, F(1, 32) = 8.70, p < .01). As in Experiment 1 we tested for differences between the baseline Procedure Participants were tested in a class- and overlapping conditions at lower and higher room setting and informed by the male exper- levels of identification (–1 and +1 SDs respec- imenter that the aim of the survey was to gain tively). At the lower level of identification bias insight into the opinions that people from was again significantly lower for participants in different regions of the UK had of one the overlapping compared to baseline con- another. The procedure was as in Experiment dition (␤ = –.385, t = –2.46, p < .05). Interest- 1, except the identification items referred to ingly, at the higher level of identification there ‘Southerners’. The four identification items was also a marginally significant trend for bias formed a reliable scale (␣ = .929). Participants to be higher following the overlapping task generated 10 overlapping characteristics compared to baseline (␤ = .304, t = 1.90, between ‘Southerners’ and ‘Northerners’ in p = .067). Within task condition simple slopes the experimental condition and 10 overlap- analysis revealed that in the baseline condition ping characteristics between ‘Cars and identification was unrelated to bias (␤ = .366, Bicycles’ in the baseline condition. The latter t = 1.67, p = .112), but following generation of constituted a baseline task of equivalent cogni- overlapping characteristics identification was tive load, involving the same mental pro- positively related to bias (␤ = .854, t = 6.15, cedures as in the experimental condition, but p < .0005) (see Figure 2). focusing on categories unrelated to those In this experiment we therefore replicated evaluated in the later phase of the experiment. the effects observed in Experiment 1 with tasks We extended the number of characteristics of equivalent informational load. When partici- because the Northern/Southern distinction in pants thought about overlapping characteristics the UK was expected to be more pervasive than between the ingroup and outgroup bias was categorization based on University affiliation. reduced, but only for lower identifiers. For Hence we suspected a more extensive manipu- higher identifiers, thinking about overlapping lation would be required. characteristics did not reduce bias, and actually

179 06 Crisp (bc/s) 29/3/05 1:11 pm Page 180

Group Processes & Intergroup Relations 8(2)

Baseline 50 Overlap 40

-10

-20 Ingroup favoritism -30

-40

-50 lower (–1 SD) higher (+1 SD) Identification

Figure 2. Ingroup favoritism as a function of overlapping characteristics and identiﬁcation, Experiment 2.

led to a marginally significant increase. In sum, reduce intergroup bias may be differentially the effectiveness of the overlapping character- effective. In two experiments, using different istics task was again proportional to the level of social categories and contexts, and measures pre-task ingroup identification: the task was and manipulations, we found consistent more effective at reducing intergroup bias for evidence that the overlapping characteristics participants who reported lower levels of pre- task reduced bias for lower identifiers, but not task identification. for higher identifiers. Furthermore, thinking about characteristics that overlapped between General discussion the ingroup and outgroup reduced ingroup favoritism to an extent proportional to pre-task In this research we set out to examine the con- levels of ingroup identification. That is to say, ditions under which interventions designed to for participants who completed the overlapping

180 06 Crisp (bc/s) 29/3/05 1:11 pm Page 181

Crisp & Beck reducing intergroup bias

characteristics task, the lower their level of pre- that therefore serve to weaken this valued dis- task identification, the lower their bias subse- tinctiveness might, by virtue of the threat quent to the task. We discuss these findings in caused by the erosion of intergroup bound- the context of recent developments in the bias- aries, lead higher identifiers to maintain a high reduction literature. level of ingroup favoritism to protect threat- ened distinctiveness. Theoretical implications Higher identifiers in the two studies reported These findings provide strong evidence that the here appeared to do just that. In Experiment 1, extent of perceivers’ ingroup identification1 compared to baseline, they maintained a high can predict the effectiveness of interventions level of ingroup favoritism following the designed to reduce intergroup bias. Much work overlap task, and in Experiment 2 they even on contact, categorization, and conflict-resolu- became more ingroup favoring after the task.2 tion has incorporated the view that differentia- These studies thus take an important step tion is divisive, and that fostering intergroup forward in illustrating the importance of con- overlap is therefore one of the key elements for sidering social identity motivations when improving intergroup relations. There is a large developing intervention strategies for reducing corpus of empirical evidence in support of this, intergroup bias. however, there is also mounting evidence that We must, however, be careful not to take the effectiveness of such interventions can vary these findings as unequivocal support for the (Dovidio et al., 1998; González & Brown, 2003). distinctiveness-threat mechanism outlined by Indeed, sometimes such interventions may the social identity perspective. For instance, actually increase bias (e.g. Hornsey & Hogg, Diehl (1988) argues that increases in inter- 2000). This invites the question of what moder- group similarity (one potential consequence of ates these quite different outcomes. Ingroup our overlapping characteristics task) leads to identification, an affective commitment to increased bias against outgroups, whilst ingroups associated with motivations to increased interpersonal similarity reduces bias preserve positive distinctiveness, appears to be against outgroups. Here then, it is plausible one such moderator. In the two studies that following the overlapping characteristics reported here, we consistently observed the task lower identifiers switched to responding in bias-reducing effects of category overlap to be interpersonal terms (less bias), whereas higher moderated by ingroup identification. Ingroup identifiers responded in intergroup terms identification may therefore account for some (increased bias). Future research on distinc- previous divergence in the literature, and thus tiveness-threat should aim to include potential potentially play an important role in the future mediating measures in order to delineate more development of models of bias-reduction. fully the precise processes underlying such According to the Social Identity perspective, effects. some group affiliations are more important We should also acknowledge other competing than others, especially for some people, some of explanations for the effects we observed. In the time. Differences in identification should both experiments we found that while lower thus predict when social identity motivations identifiers were less ingroup favoring after the will come into play. For lower identifiers, group overlapping characteristics task, higher identi- membership contributes little to self-identity, fiers remained ingroup favoring. It may be that and so any weakening of categorical boundaries specifically for higher, but not lower identifiers, will likely lead to the formation of an inclusive additional conditions are required for inter- representation that includes former outgroup overlap to reduce intergroup bias. For groupers. For higher identifiers, however, group instance, perhaps it is not that higher identifiers membership provides a means of defining find the realization of overlapping character- oneself, in particular in comparison with a istics with the outgroup threatening, but rather relevant outgroup (Hogg, 2001). Interventions that they have a more solidified representation

181 06 Crisp (bc/s) 29/3/05 1:11 pm Page 182

Group Processes & Intergroup Relations 8(2)

of how different the ingroup is from the role. If higher identifiers react negatively to the outgroup. For higher identifiers it may be that erosion of intergroup boundaries, then one the augmenting conditions specified by the option is to strengthen those boundaries within contact hypothesis (e.g. cooperation, acquaint- the context of a more inclusive intergroup ance potential) are especially relevant. context. This ‘simultaneous categorization’ Finally, we should also acknowledge that approach (Hornsey & Hogg, 2000; or ‘Dual although our overlapping characteristics task identity approach’ Gaertner & Dovidio, 2000) was designed to capture the merged represen- can yield more positive effects for intergroup tation element of differentiation-reducing relations compared to the complete abandon- strategies, strategies such as contact and forming ment of subgroup identities (see González & a common ingroup identity incorporate other Brown, 2003). Interestingly, this parallels a elements as well. For instance, contact can similar appreciation of the importance of differ- reduce intergroup anxiety (see Islam & entiation in the contact literature. Hewstone & Hewstone, 1993), but this was not an element Brown’s (1986) Mutual Intergroup Differentia- in the overlapping characteristics task that we tion Model outlines the need to maintain employed in our experiments. Similarly, category distinctiveness in order to allow gener- research on the Common Ingroup Identity alization of positive affect from the individual Model recategorizes subgroup members at a outgroup member to the outgroup as a whole. higher level of inclusiveness, but this was not On the basis of the findings we report here, such necessarily a consequence of our overlapping ‘differentiation-preserving’ interventions may characteristics task. Future research will need to work particularly well for people who strongly establish whether, and how, ingroup identifi- identify with their ingroup. cation is relevant to specific intervention Research developing interventions to reduce strategies like contact and the formation of a intergroup bias has sometimes yielded dis- common ingroup identity. parate findings. In this research we tested whether motivations associated with maintain- Practical implications ing a distinct social identity could account for The finding that perceivers can react in such disparities. We hypothesized that differ- dramatically divergent ways to the same inter- ences in the extent to which participants vention has significant implications for the identified with their ingroup would predict application of social psychological models to whether, and when, reducing intergroup dif- real contexts of intergroup conflict. The ferentiation would reduce intergroup bias. findings we report suggest that a ‘one size fits Confirming these expectations, we found that all’ approach to reducing intergroup bias may our overlapping characteristics task reduced not be most effective. For some people (lower bias for lower identifiers, but not for higher identifiers) improved intergroup relations may identifiers. Furthermore, the extent to which be likely, but for others (higher identifiers) ingroup favoritism was reduced was pro- encouraging the dissolution of intergroup portional to pre-task levels of ingroup identifi- boundaries may be less effective, and possibly cation: the lower the level of pre-task even accentuate ingroup favoritism. identification, the lower was bias subsequent to The different ways in which lower and higher the task. The findings suggest that ingroup identifiers react to changes in intergroup dis- identification might play a central role in how tinctiveness calls for more tailored intervention future models of bias-reduction should develop strategies, suited to specific populations. Indeed, to tackle the complexity of intergroup affilia- recent approaches to bias-reduction have tions, motivations, and evaluations. Developing acknowledged that differentiation, in contrast to such multifaceted models for reducing inter- assimilation, can sometimes play an important group bias is the challenge for future work.

182 06 Crisp (bc/s) 29/3/05 1:11 pm Page 183

Crisp & Beck reducing intergroup bias

Notes Campbell, D. T. (1956). Enhancement of contrast as a composite habit. Journal of Abnormal and Social 1. We are keen to point out that identification is Psychology, 53, 350–355. not, as our conceptualization might imply, only Cocker, R. M. (2004). Intergroup overlap and intergroup an individual difference variable. People’s evaluation: The moderating role of ingroup identification with different social groups can identification. Unpublished MPhil dissertation. vary, and contextual factors exert a strong University of Birmingham, UK. influence on identification, and correspondingly, Cocker, R. M., & Crisp, R. J. (2003). Shared the tendency for people to behave in line with characteristics and perceived overlap. Unpublished social identity motivations. dataset, University of Birmingham, UK. 2. We note that the increase in bias observed Crisp, R. J., Ensari, N., Hewstone, M., & Miller, N. following the characteristics task at higher levels (2003). A dual-route model of crossed of identification could, in part, be attributable to categorization effects. In W. Stroebe & changes made to target groups and measures in M. Hewstone (Eds.), European review of social Experiment 2. psychology (Vol. 13, pp. 35–74). Philadelphia, PA: Psychology Press. Acknowledgments Crisp, R. J., & Hewstone, M. (1999). Differential evaluation of crossed category groups: Patterns, This research was funded by a Leverhulme Trust processes, and reducing intergroup bias. Group grant (F/00094/H) directed by R. J. Crisp. We thank Processes & Intergroup Relations, 2, 307–333. James Bacon for his assistance in data collection for Crisp, R. J., Hewstone, M., & Rubin, M. (2001). Experiment 2. Does multiple categorization reduce intergroup bias? Personality and Social Psychology Bulletin, 27, References 76–89. Deschamps, J. C., & Doise, W. (1978). Crossed Abrams, D., & Hogg, M. A. (1998). Prospects for category memberships in intergroup relations. In research in group processes and intergroup H. Tajfel (Eds.), Differentiation between social groups relations. Group Processes & Intergroup Relations, 1, (pp. 141–158). Cambridge, UK: Cambridge 7–20. University Press. Aiken, L. S., & West, S. G. (1991). Multiple regression: Diehl, M. (1988). Social identity and minimal Testing and interpreting interactions. London: Sage. groups—the effects of interpersonal and Allport, G. W. (1954). The nature of prejudice. Garden intergroup attitudinal similarity on intergroup City, NY: Doubleday. discrimination. British Journal of Social Psychology, Brewer, M. B. (1979). Ingroup bias in the minimal 27, 289–300. intergroup situation: A cognitive–motivational Doise, W. (1978). Groups and individuals: Explanations analysis. Psychological Bulletin, 86, 307–324. in social psychology. Cambridge, UK: Cambridge Brewer, M. B. (1991). The social self—on being the University Press. same and different at the same time. Personality Dovidio, J. F., Gaertner, S. L., & Validzic, A. (1998). and Social Psychology Bulletin, 17, 475–482. Intergroup bias: Status, differentiation and a Brewer, M. B., & Miller, N. (1984). Beyond the common ingroup identity. Journal of Personality and contact hypothesis: Theoretical perspectives on Social Psychology, 75, 109–120. desegregation. In N. Miller & M. B. Brewer (Eds.), Ellemers, N., Spears, R., & Doosje, B. (2002). Self Groups in contact: The psychology of desegregation and social identity. Annual Review of Psychology, 53, (pp. 281–302). Orlando, FL: Academic Press. 161–186. Brown, R. J., & Abrams, D. (1986). The effects of Gaertner, S. L., & Dovidio, J. F. (2000). Reducing intergroup similarity and goal interdependence intergroup bias: The common ingroup identity model. on intergroup attitudes and task performance. Philadelphia, PA: Psychology Press/Taylor & Journal of Experimental Social Psychology, 22, 78–92. Francis. Brown, R. J., & Wade, G. (1987). Superordinate Gaertner, S. L., Mann, J. A., Dovidio, J. F., Murrell, goals and intergroup behaviour: The effect of role A. J., & Pomare, M. (1990). How does cooperation ambiguity and status on intergroup attitudes and reduce intergroup bias? Journal of Personality and task performance. European Journal of Social Social Psychology, 59, 692–704. Psychology, 17, 131–142. Gaertner, S. L., Mann, J. A., Murrell, A. J., &

183 06 Crisp (bc/s) 29/3/05 1:11 pm Page 184

Group Processes & Intergroup Relations 8(2)

Dovidio, J. F. (1989). Reducing intergroup bias: contact as predictors of intergroup anxiety, The benefits of recategorization. Journal of perceived out-group variability, and out-group Personality and Social Psychology, 57, 239–249. attitude: An integrative model. Personality and González, R., & Brown, R. J. (2003). Generalization Social Psychology Bulletin, 19, 700–710. of positive attitude as a function of subgroup and Jetten, J., Branscombe, N. R., Spears, R., & superordinate group identifications in intergroup McKimmie, B. M. (2003). Predicting the paths of contact. European Journal of Social Psychology, 33, peripherals: The interaction of identification and 195–214. future possibilities. Personality and Social Psychology Grieve, P. G., & Hogg, M. A. (1999). Subjective Bulletin, 29, 130–140. uncertainty and intergroup discrimination in the Jetten, J., Spears, R., & Manstead, A. S. R. (2001). minimal group situation. Personality and Social Similarity as a source of discrimination: The role Psychology Bulletin, 25, 926–940. of group identification. European Journal of Social Haddock, G., Zanna, M. P., & Esses, E. M. (1993). Psychology, 31, 621–640. Asserting the structure of prejudicial attitudes: Luhtanen, R., & Crocker, J. (1992). A collective The case of attitudes towards homosexuals. Journal self-esteem scale: Self-evaluation of one’s social of Personality and Social Psychology, 65, 1105–1118. identity. Personality and Social Psychology Bulletin, 18, Hewstone, M., & Brown, R. (1986). Contact is not 302–318. enough: An intergroup perspective on the Miller, N., Brewer, M. B., & Edwards, K. (1985). ‘Contact Hypothesis’. In M. Hewstone & R. Brown Cooperative interaction in desegregated settings: (Eds.), Contact and conflict in intergroup encounters A laboratory analog. Journal of Social Psychology, 22, (pp. 1–44). Oxford, UK: Blackwell. 103–122. Hogg, M. A. (2000). Subjective uncertainty Mullen, B., Brown, R., & Smith, C. (1992). Ingroup reduction through self-categorization: A bias as a function of salience, relevance, and motivational theory of social identity processes. status: An integration. European Journal of Social European Review of Social Psychology, 11, 223–255. Psychology, 22, 103–122. Hogg, M. A. (2001). Self-categorization and Oakes, P. J., Haslam, S. A., & Turner, J. C. (1994). subjective uncertainty resolution: Cognitive and Stereotyping and social reality. Oxford, UK: Blackwell. motivational facets of social identity and group Pettigrew, T. F. (1997). Generalized intergroup membership. In J. P. Forgas, K. D. Williams, & contact effects on prejudice. Personality and Social L. Wheeler (Eds.), The social mind: Cognitive and Psychology Bulletin, 23, 173–185. motivational aspects of interpersonal behavior Pettigrew, T. F. (1998). Intergroup contact theory. (pp. 323–349). New York: Cambridge University Annual Review of Psychology, 49, 65–85. Press. Roccas, S., & Schwartz, S. H. (1993). Effects of Hogg, M. A., & Abrams, D. (1990). Social intergroup similarity on intergroup relations. motivation, self-esteem and social identity. In European Journal of Social Psychology, 23, 581–595. D. Abrams & M. A. Hogg (Eds.), Social identity Sherif, M. (1966). Group conflict and cooperation. theory: Constructive and critical advances (pp. 28–47). London: Routledge Kegan Paul. New York: Springer-Verlag. Spears, R., Doosje, B., & Ellemers, N. (1997). Hogg, M. A., & Mullin, B. A. (1999). Joining groups Self-stereotyping in the face of threats to group to reduce uncertainty: Subjective uncertainty status and distinctiveness: The role of group reduction and group identification. In identification. Personality and Social Psychology D. A. Abrams & M. A. Hogg (Eds.), Social identity Bulletin, 23, 538–553. and social cognition (pp. 249–279). Oxford, UK: Spears, R., Jetten, J., & Scheepers, D. (2002). Blackwell. Distinctiveness and the definition of collective self. Hogg, M. A., & Turner, J. C. (1985). Interpersonal In A. Tesser, D. A., Stapel, & J. Wood (Eds.), attraction, social identification and psychological Self and motivations: Emerging psychological group formation. European Journal of Social perspectives (pp. 147–171). Washington, DC: Psychology, 15, 51–66. American Psychological Association. Hornsey, M. J., & Hogg, M. A. (2000). Subgroup Tajfel, H. (1969). Cognitive aspects of prejudice. relations: A comparison of mutual intergroup Journal of Social Issues, 25, 79–97. differentiation and common ingroup identity Tajfel, H. (1982). Social psychology of intergroup models of prejudice reduction. Personality and relations. Annual Review of Psychology, 33, 1–39. Social Psychology Bulletin, 26, 242–256. Tajfel, H., Billig, M. G., Bundy, R. P., & Flament, C. Islam, M. R., & Hewstone, M. (1993). Dimensions of (1971). Social categorization and intergroup

184 06 Crisp (bc/s) 29/3/05 1:11 pm Page 185

Crisp & Beck reducing intergroup bias

behavior. European Journal of Social Psychology, 1, van Leeuwen, E., van Knippenberg, D., & Ellemers, 149–178. N. (2003). Continuing and changing group Tajfel, H., & Turner, J. C. (1979). An integrative identities: The effects of merging on social theory of intergroup conflict. In W. G. Austin & identification and ingroup bias. Personality and S. Worchel (Eds.), The social psychology of intergroup Social Psychology Bulletin, 29, 697–690. relations (pp. 33–47). Monterey, CA: Brooks/Cole. West, S. G., Aiken, L. S., & Krull, J. L. (1996). Tajfel, H., & Wilkes, A. L. (1963). Classification and Experimental personality designs: Analyzing quantitative judgment. British Journal of Psychology, categorical by continuous variable interactions. 54, 101–113. Journal of Personality, 64, 1–48. Terry, D. J., & Callan, V. J. (2001). In-group bias in Wright, S. C., Aron, A., McLaughlin-Volpe, T., & response to an organizational merger. Group Ropp, S. A. (1997). The extended contact effect: dynamics: Theory, research, and practice, 2, 67–81. Knowledge of cross-group friendships and Terry, D. J., Carey, C. J., & Callan, V. J. (2001). prejudice. Journal of Personality and Social Psychology, Employee adjustment to an organizational 73, 73–90. merger: An intergroup perspective. Personality and Social Psychology Bulletin, 27, 267–280. Paper received 1 May 2004; revised version accepted 4 Turner, J. C. (1982). Towards a cognitive October 2004. redefinition of the social group. In H. Tajfel (Ed.), Social identity and intergroup relations, Cambridge, UK: Cambridge University Press. Biographical notes Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S. (1987). Re-discovering the RICHARD J. CRISP is currently Reader in Social social group: A self-categorization theory. Oxford, UK: Psychology at the University of Birmingham. His Blackwell. research interests span social cognition and Urban, L. M., & Miller, N. M. (1998). A theoretical intergroup relations, with particular emphasis on analysis of crossed categorization effects: A meta- the antedecents, processes, and consequences of analysis. Journal of Personality and Social Psychology, multiple social categorization. 74, 894–908. van Knippenberg, A., & Ellemers, N. (1990). Social SARAH R. BECK is currently Lecturer in identity and intergroup differentiation process. In Developmental Psychology at the University of W. Stroebe & M. Hewstone (Eds.), European review Birmingham. Her research interests include of social psychology (Vol. 1, pp. 137–169). children’s and adults’ counterfactual thinking and Chichester, UK: Wiley. children’s cognitive development.

185