Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

507

British Journal of Social Psychology (2010), 49, 507523 q 2010 The British Psychological Society

The British Psychological Society

www.bpsjournals.co.uk

What will the others think? In-group norms as a mediator of the effects of intergroup contact
Pablo De Tezanos-Pinto1*, Christopher Bratt2 and Rupert Brown1
1 2

Department of Psychology, University of Sussex, Brighton, UK Centre for Child and Adolescent Mental Health, Oslo, Norway
The inuence of social norms in the context of intergroup relations has long been recognized by social psychologists, yet research on intergroup contact and social norms have usually remained disconnected. We explored the inuence of direct and indirect friendship on attitudes towards ethnic minorities in Norway, and in particular the role of in-group norms about the social approval of intergroup contact as a mechanism that distinguishes direct from indirect contact. Using a sample of school students from 89 classrooms N 823, we tested this hypothesis with both one level and multi-level structural equation modelling (ML-SEM), where the amount of contact of other classroom members was considered as a form of indirect contact. The results suggest that the intergroup contact of other in-group members (in-group friends or classmates) affects attitudes towards the out-group by changing the perception of in-group norms and by reducing intergroup anxiety. In contrast, direct contact (or contact at the individual level in the case of ML-SEM), improved attitudes only by reducing intergroup anxiety, and did not affect the perception of in-group norms.

The idea that promoting positive contact between members of different groups could contribute to the reduction of prejudice has proved a fertile source of inspiration for researchers and policy makers alike. First properly formalized by Allport (1954), this contact hypothesis has been tested in over 500 independent studies according to a recent meta-analysis (Pettigrew & Tropp, 2006) and played a not insignicant role in the famous US Supreme Court decision to outlaw ethnically segregated schools (Allport et al., 1953). Today, few doubt that the main tenets of the contact hypothesis have been empirically supported (Brown & Hewstone, 2005; Pettigrew & Tropp, 2006), and its central ideas are implicated in many intervention programmes to reduce prejudice in schools and elsewhere (e.g., Aboud & Levy, 1999; Paluck & Green, 2009). Nevertheless, despite this impressive contribution, signicant gaps remain in our understanding of intergroup contact effects. One of these gaps concerns the extent to which direct contact between members of different groups is necessary for prejudice reduction to

* Correspondence should be addressed to Pablo De Tezanos-Pinto, Department of Psychology, University of Sussex, Brighton BN9 1QH, UK (e-mail: p.de-tezanos-pinto@sussex.ac.uk).
DOI:10.1348/014466609X471020

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

508 Pablo De Tezanos-Pinto et al.

occur or, instead, whether indirect or extended contact could sufce (Wright, Aron, McLaughlin-Volpe, & Ropp, 1997); another concerns the mediating mechanisms underlying both direct and indirect contact effects, an issue that has begun to preoccupy researchers only over the last decade or so (Brown & Hewstone, 2005). In this paper, we seek to make a contribution to lling both these lacunae by studying the roles that in-group norms and intergroup anxiety play in the mediation of contact effects. We do so with the aid of structural equation modelling (SEM) and multi-level analysis of a relatively large sample of school students, an analysis that also sheds some light on the relationship between direct and indirect contact. In-group norms and intergroup relations The inuence of social norms in the context of intergroup relations has long been recognized by social psychologists (e.g., Allport, 1954; Sherif, 1966). The amount of prejudice that people express towards different groups is highly correlated with the social approval of that expression (Crandall, Eshleman, & OBrien, 2002) and has been found to be affected by manipulations of this social approval using confederates (Blanchard, Crandall, Brigham, & Vaughn, 1994; Blanchard, Lilly, & Vaughn, 1991; Monteith, Deneen, & Tooman, 1996). Stangor, Sechrist, and Jost (2001), observed that manipulating the apparent consensus of in-group attitudes towards Blacks changed the beliefs of White participants about the stereotypes of that group, a change that persisted in an unrelated session a week later. Furthermore, when in-group norms were consistent with the participants original stereotypes, their attitudes were held more strongly and were more resistant to change. Manipulating the out-group consensus, on the other hand, was signicantly less effective. This is consistent with the literature about the inuence in social norms in general (i.e., Smith & Louis, 2008; Terry & Hogg, 1996), and underlines the importance of social identication as a fundamental mechanism by which group norms produce changes on individuals attitudes and behaviours. Apparently, this sensitivity to social norms develops quite early in life. Rutland, Cameron, Milne, and McGeorge (2005) observed that the expression of prejudiced attitudes in young children was affected by the salience of self-presentation concerns (a camera allegedly recording their responses). In another study, Nesdale, Maass, Durkin, and Grifths (2005) found that children who were told that their team members liked to work with children on other teams had more favourable intergroup attitudes than children who were told that their team members did not like people in other teams. A eld experiment in Rwanda, a country with a recent history of genocide, showed that a long-term media intervention can reinforce social norms and behaviours regarding positive intergroup relations (Paluck, 2009). Participants were randomly assigned to listen to different radio programmes (either a soap opera about intergroup relations or one about health issues). The programme about intergroup relations was designed to promote positive beliefs about intergroup interactions and trauma healing, and the ctional characters displayed positive social norms regarding cross-group relationships. The intervention lasted for over a year, and was particularly effective in promoting trust, empathy, cooperation, trauma healing, and positive norms about interethnic marriage. In sum, if people perceive that the in-group has a positive view of the out-group they will be more willing and able to see the out-group in a positive light, perhaps because the out-group, the in-group and the self become psychologically linked. Even though it has been argued that intergroup contact should be considered from a normative perspective (Pettigrew, 1991, 1998, 2008), research on the contact

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

In-group norms and intergroup contact

509

hypothesis and research on social norms have usually remained disconnected. As is well known, Allport (1954) included institutional support as one of the four facilitating conditions for contact to be effective in reducing prejudice, but it remains the least studied of the four. As Pettigrew (1998) noted, intergroup contact can both affect and be affected by social norms about the relationship between two groups. Whilst the effects of norms on contact (or the desire for contact) can plausibly be assumed to operate at an individual level, and can be usefully analysed as such, it is our contention that the effects of contact on norms (about future contact with out-group members) are better conceived as operating at a group or aggregate level. That is, if members of the in-group on average have more (or less) contact with out-group members, this is likely to generate group norms about the acceptability (or unacceptability) of further such contact. Individuals own level of contact with the out-group is unlikely to provide this information since it says little about what is expected or accepted by others in the group. Such a contention implies incorporating an aggregate or group level of analysis into the research design. We return to this point shortly. Studying racially integrated housing projects in the USA, Wilner, Walkley, and Cook (1952) found that the expectation that other Whites would approve of intimate contact with Blacks was related both to the attitudes towards Blacks and to the amount and quality of intergroup contact that people experienced. Women who perceived the social climate to favour interracial association and engaged in friendly interaction with Blacks were the most likely to have positive attitudes towards Blacks. Fendrich (1967) found that the perceived peer support for interracial association was a relevant predictor of both White students attitudes towards Blacks and behaviours related to interracial interaction. Perceived peer support was measured using a range of reference groups that were close to the participants: their friends, family, and people whom they admired. The role of expectations about social approval of intergroup contact with members of other groups was also underlined by Stephan and Stephan (1985) in their concept of intergroup anxiety. They argued that one of the reasons that people might experience anxiety when interacting with out-group members is the fear that other in-group members would not approve of that contact: people would fear that they themselves could be subject of discrimination because of their association with members of the outgroup. Indeed, if strong in-group norms proscribe too close a relation with out-group members, these fears could be well founded (Abrams, Rutland, & Cameron, 2003; Marques & Paez, 1994). As we shall note shortly, intergroup anxiety itself can play an important mediating role in the effects of contact. The contextual inuence of in-group norms was also integral to the extended contact hypothesis (Wright et al., 1997). This hypothesis states that the mere knowledge that fellow in-group members have a close relation with members of the outgroup can reduce prejudice. Since its rst promulgation, considerable correlational and experimental evidence have been adduced in its favour (Cameron, Rutland, Brown, & Douch, 2006; Liebkind & McAlister, 1999; Paolini, Hewstone, & Cairns, 2007; Paolini, Hewstone, Cairns, & Voci, 2004; Pettigrew, Christ, Wagner, & Stellmacher, 2007; Turner, Hewstone, & Voci, 2007; Turner, Hewstone, Voci, & Vonofakou, 2008; Wright et al., 1997). In setting out the rationale for their hypothesis, Wright et al. (1997) argued, amongst other things, that such extended or indirect contact can generate changes in perceived in-group norms about the acceptability of positive intergroup relationships and that these changed in-group norms facilitate an improvement of the attitudes towards the out-group. There is some evidence that supports this hypothesis (Pettigrew et al., 2007; Turner et al., 2008). With the extended contact

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

510 Pablo De Tezanos-Pinto et al.

hypothesis, Wright et al. (1997) introduced for the rst time the inuence that the direct contact of other in-group members (i.e., indirect contact) have on the attitudes towards the out-group, thus paving the way for a normative perspective in the study of intergroup contact.

Intergroup anxiety Islam and Hewstone (1993) were one of the rst to document the mediating role of intergroup anxiety. They observed that high-quality contact between Muslims and Hindus in Bangladesh led to reduced intergroup anxiety, which in turn led to more favourable intergroup attitudes. Since then, as Brown and Hewstone (2005) have noted, several other studies have established that reduced intergroup anxiety is one important mediating process by which contact has its effects (Binder et al., 2009; Greenland & Brown, 1999; Harwood, Hewstone, Paolini, & Voci, 2005; Hewstone, Cairns, Voci, Hamberger, & Niens, 2006; Hewstone et al., 2004, 2005; Voci & Hewstone, 2003). According to Wright et al. (1997), intergroup anxiety is also implicated in extended contact effects. Observing that others in the in-group maintain positive relationships with out-group members can reduce the likelihood in the perceivers that their own intergroup encounters will be fraught with anxiety. Indeed, there is evidence for precisely this process (Paolini et al., 2004; Turner et al., 2007, 2008). The research by Turner et al. (2008) is particularly noteworthy since it tested simultaneously all the mediators of extended contact that Wright et al. (1997) had proposed: in-group norms, out-group norms, intergroup anxiety, and inclusion of the out-group in the self. In two cross-sectional studies, Turner et al. (2008) found evidence for all four of Wright et al.s (1997) hypothesized mediators and, in both studies, the links between indirect crossgroup friendship and, respectively, intergroup anxiety and in-group norms were especially strong. As would be expected, direct contact did not affect the perception of in-group norms, probably because it provides very limited information about the in-group as a whole. Yet direct contact also failed to have a signicant effect on intergroup anxiety in these studies, and its effect was mediated only by the inclusion of the out-group in the self. The present study seeks to build on these results by providing further evidence that in-group norms about contact with the out-group are affected by extended, but not by direct, intergroup contact. We also introduced a new measure of in-group norms about intergroup contact that differs for the one used by Turner et al. (2008) in that it focuses on injunctive norms (what other in-group members approve of) rather than descriptive norms (what other in-group members do; Cialdini, Kallgren, & Reno, 1991). This emphasis on the perception of the social approval of having contact with the out-group is consistent with much previous research and theorizing. Stephan and Stephan (1985) argued that one of the possible antecedents of intergroup anxiety was indeed the fear of social sanctions by other in-group members, an aspect that was also present in Wright et al.s (1997) conceptualization of in-group norms. Wilner et al. (1952) also found that the perceived approval of interracial association by other neighbours was related to the opportunity to observe actual contact between in-group and out-group members. We used a large sample to give greater statistical power, and we based the study in a little researched and quite complex intergroup setting (Norway) where several different ethnic groups reside (e.g., Turkish, Indian, Pakistani, in addition to indigenous Norwegians). Importantly, we tested not only a model in which direct and extended

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

In-group norms and intergroup contact

511

contact were considered as predictors at an individual level, but also a multi-level model in which the amount of group-level direct contact (in the classroom) was considered as a predictor. Multi-level analyses are particularly well suited to study normative inuences because they separate the individual effects of the variables studied from the group-level effect of these variables (Pettigrew, 2006). In the case of intergroup contact, the effect of having one or more out-group friends (individual level) may not be the whole story: belonging to a classroom in which many individuals have out-group friends (group level) can have an effect on attitudes that is independent of the actual intergroup contact of each individual. The choice of the classroom as the group level unit of analysis is also noteworthy. We suspect that for adolescent school students, this micro-level social aggregate is a particularly powerful source of social norms. Thus, the main hypothesis of this study is that both direct and extended contacts (or the aggregate amount of contact in the classroom) have an effect on attitudes towards the out-group, but they rely on different mechanisms to achieve this effect. In particular, direct contact will improve attitudes by reducing the levels of intergroup anxiety, while extended contact will improve intergroup attitudes by affecting both the perception of in-group norms about contact with the out-group and by reducing intergroup anxiety. This hypothesized model is shown in Figure 1.

Method
Participants Eight hundred and twenty-three ethnic Norwegian school students (51% F; mean age, 13.8 years) from 89 classrooms voluntarily participated in the study. The study took place in the town of Drammen, Norway. This town has a considerable number of people belonging to ethnic minorities (18%, mainly Turkish and Pakistani, but also Indian and other minority groups), providing a suitable setting to explore the effects of cross-group friendship. Procedure Data collection was organized in close cooperation with the local authorities in the municipality of Drammen. All the students from the towns six schools for grades 810 were invited to participate. A letter stating the objectives of the study was sent to the parents, and they were asked to give written consent regarding their
Indirect contact (or class level contact)

()

Norms against contact

()

() (+) (+) Positive attitudes () () Intergroup anxiety

Direct contact

Figure 1. Hypothesized relations between the variables considered in the study.

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

512 Pablo De Tezanos-Pinto et al.

son/daughters participation.1 Students also received a letter stating the objectives of the study, the voluntary nature of their participation, and guarantying the anonymity of their responses. To increase the response rate, we included a draw for one mobile phone and two mp3 players. In total 1,129, students completed the questionnaire, but our analysis include only ethnic Norwegian students. The questionnaire was answered on-line, via a web-based interface. Students used their own laptop to answer the questionnaire (laptops are routinely provided by their schools) and the application was overseen by the teachers of each class.

Measures The questionnaire focused on three ethnic minorities that are salient in Drammen: Turkish, Pakistani, and Indian.2

Cross-group friendship We used measures of friendship for intergroup contact, since friendship meets the necessary conditions proposed by Allport (1954) and therefore has a stronger effect on attitudes (Pettigrew & Tropp, 2006). Direct cross-group friendship was measured for each of the ethnic minorities considered in the study and for male and female targets separately [How many boy (Turkish/Pakistani/Indian) friends do you have?; How many girl (Turkish/Pakistani/Indian) friends do you have?]. These six items were combined in a single indicator, by adding the number of friends in the different out-groups a :90. Indirect cross-group friendship was measured with one item (How many Norwegian friends do you have who have friends from another ethnicity?).

Norms against contact The perception of in-group norms about contact was measured using three items: (1) I believe that friends in my ethnic group prefer that I am not together with youths from other ethnic groups, (2) I believe that friends in my ethnic group think it is a bit uncool if I hang around with boys or girls from another ethnic group, and (3) (reversed) I believe that friends in my ethnic group think it is cool if I have a close friend from another ethnic group (1 completely disagree to 7 completely agree; a :68).

Intergroup anxiety Intergroup anxiety was measured with the following question: if you were the only ethnic Norwegian working with a group of (Turkish/Pakistani/Indian), how would you feel? Participants answered on two items in a scale of 17: (1) nervous and (2) condent (reversed). For the analysis, we used two indicators by combining responses to the different targets for each of these two items (nervous, a :95; condent, a :96).
Where applicable, this letter was translated to minority languages. All items reported here have been translated from the Norwegian, and the questionnaire also included other measurements that are not relevant for this study. Original versions of the items, and details of these other measurements, can be obtained from the second author.
2 1

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

In-group norms and intergroup contact

513

Attitudes towards the out-group As with direct friendship, attitudes were measured for each of the ethnic groups and for male and female targets separately (Turkish boys, Turkish girls, Pakistani boys, Pakistani girls, Indian boys, Indian girls). Participants responded by marking in a feeling thermometer how they felt towards each target, using a scale from 0 to 10. For the analysis, responses for male and female targets were combined for each of the outgroups (Turkish, r :82; Pakistani, r :87; Indian, r :87).

Results
The means, standard deviations, and the correlations between the variables of this study are presented in Table 1. Sixty-one per cent of the participants reported having at least one out-group friend, and 83% reported having at least one indirect out-group friend. Both direct and indirect cross-group friendship were severely positively skewed (Skew 3:5, Kurt 16:2 for direct friendship; Skew 3:9, Kurt 17:2 for indirect friendship), so for the analysis we performed a log transformation of the scores. In addition to normalizing the distribution of direct and indirect cross-group friendship (Skew 0:8, Kurt 20:1 for direct friendship; Skew 0:3, Kurt 20:04 for indirect friendship), this transformation increased the correlation between these variables and the other variables considered in this study. This is consistent with the idea that adding a new out-group friend has a stronger effect when people have fewer cross-group friendships than when they already have several out-group friends.
Table 1. Means, standard deviation, correlations, and sample size for the variables used in the study N 1. Direct friendship (log) 2. Indirect friendship (log) 3. Norms against contact 4. Intergroup anxiety 5. Attitudes 695 665 806 783 747 Mean 4.2 (1.0) 8.4 (1.6) 2.9 4.1 4.6 SD 8.2 (1.0) 15.1 (1.1) 1.3 1.5 2.8 2 .33 (.46) 3 2 .11 (2 .16) 2 .21 (2 .31) 4 2 .20 (2 .23) 2 .20 (2 .30) .38 5 .21 (.24) .16 (.30) 2 .38 2 .49

Note. All correlations are signicant p , :001. The scales of norms against contact and intergroup anxiety range from 1 to 7, while the scale of attitudes ranges from 0 to 10. Values in parenthesis correspond to the log transformation of the scores.

As can be seen in Table 1, in-group norms about contact with the out-group is a relevant variable in this context. It correlates both with attitudes towards the out-group and with the anxiety expected by participants in an imagined intergroup encounter. Moreover, it showed a higher correlation with indirect cross-group friendship than with direct cross-group friendship, which is consistent with the idea that indirect friendship can change the perceptions of contact norms. SEM analysis To test the hypothesized model we performed SEM analysis using Mplus (version 5.2; n & Muthe n, 19982007). All the SEM models were run using robust maximum Muthe likelihood estimators and full information maximum-likelihood (FIML) for missing data.3
3

As can be observed in Table 1, the proportion of missing data in this study was not irrelevant (between 2% for in-group norms and 19% for indirect friendship). FIML has been shown to be superior to listwise deletion, pairwise deletion, and similar response pattern imputation (Enders & Bandalos, 2001).

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

514 Pablo De Tezanos-Pinto et al.

Because the questionnaire did not include some of the variables that have been found to mediate the effect of cross-group friendship, such as inclusion of the out-group in the self (Aron et al., 2004) or empathy and perspective taking (Batson et al., 1997; Pettigrew & Tropp, 2008), we allowed a direct relation between cross-group friendship and attitudes, hypothesizing that intergroup anxiety will partially mediate this effect. The analyses were performed using a partial disaggregation approach for intergroup anxiety and attitudes (Bagozzi & Heatherton, 1994; Little, Cunningham, Shahar, & Widaman, 2002). Intergroup anxiety was measured with two indicators, nervous and condent (reversed), and attitudes were measured with three, by averaging attitudes towards boys and girls that belonged to each of the three out-groups. Norms about contact with the out-group were measured using three items, and direct cross-group friendship and indirect cross-group friendship were single-item constructs. We used the criteria proposed by Hu and Bentler (1999; CFI . :95, RMSEA , :06, and SRMR , :08) to assess model t. Because of the large sample size, we considered a signicant chi-squared to be acceptable if it was less than double the degrees of freedom (Jo reskog, 1969). According to these criteria, the measurement model provided an insufcient t to the data (x2 27 97:78, p , :01; RMSEA :056; SRMR :049; CFI :971) because the x2/df was considerably larger than two. But the t was signicantly improved when an error covariance between the two negative items of ingroup norms was introduced (x2 26 61:49, p , :01; RMSEA :041; SRMR :018; CFI :986; Dx2 1 36:29, p , :001). These two items not only differ from the third in that they are negatively worded, but also in that they refer to casual contact whereas the third item refers to the development of cross-group friendships. Therefore, we included this error covariance in all subsequent models. The results for our hypothesized model (Model 1) can be observed in Figure 2. The model presented a good t with the data (x2 27 61:68, p , :01; RMSEA :040; SRMR :018; CFI :985), and showed the expected relations between the variables. In the model, indirect cross-group friendship predicted the attitudes towards ethnic minorities by affecting both contact norms (b 20:442, p , :001) and intergroup anxiety (b 20:244, p , :001), while direct cross-group friendship predicted

y1

y2

y3

R 2 = .22
.38

y6

y7

y8

Indirect cross-group friendship

.44

R 2 = .40
Norms against contact .54 Positive attitudes

.29

.47 .24 Direct cross-group friendship .13 Intergroup anxiety

.10

R = .11
y4 y5

Figure 2. Results for one-level SEM model.

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

In-group norms and intergroup contact

515

attitudes both directly (b 0:098, p , :001) and indirectly, by reducing intergroup anxiety (b 20:128, p , :001). As expected, the relation between direct friendship and contact norms was negligible and not statistically signicant (b 20:043, p :5). Both intergroup anxiety and contact norms predicted attitudes towards ethnic minorities (b 20:289, p , :001 and b 20:380, p , :001, respectively), and the model accounted for 40.2% of the variance of attitudes. The inclusion of a direct path from indirect friendship to attitudes did not improve the model t (Dx2 1 0:187, p :665) and showed this path to be non-signicant (b 0:033, p :575), suggesting that the effect of indirect friendship was fully mediated by in-group norms about contact and intergroup anxiety. An effect decomposition analysis (see Loehlin, 2003) showed that the total effect (TE) of indirect cross-group friendship on attitudes towards ethnic minorities (TE :239, p , :001) was greater than the TE of direct cross-group friendship (TE :151, p , :001). We analysed alternative models in order to provide some support for the causal relationships proposed in the model. We tested two alternative models by varying the relative position of cross-group friendship, the mediators, and attitudes towards ethnic minorities. For the comparison, we used a model that excluded the direct path from direct cross-group friendship to attitudes (Model 2). Even though the exclusion of this path produced a signicant increase in the chi-squared (Dx2 1 6:83, p , :001), the model still presented an acceptable t with the data (x2 28 68:51, p , :01, RMSEA :042; SRMR :023; CFI :983). The rst alternative model inverted the relation between attitudes and the mediators, so that attitudes mediate the effect of cross-group friendship on intergroup anxiety and contact norms. In this view, cross-group friendship would affect attitudes towards the out-group directly, and participants with more negative attitudes would expect higher levels of intergroup anxiety in a contact scenario and perceive their in-group to be less tolerant about having contact with the out-group. This model did not t the data as well as Model 2 (x2 30 94:83, p , :01, RMSEA :051; SRMR :042; CFI :973). The second alternative model inverted the relation between attitudes and crossgroup friendship, so that direct and indirect cross-group friendship mediate the effect of attitudes on intergroup anxiety and in-group norms. This model reects the alternative hypothesis that there is a self-selection process in the development of cross-group friendships, where people with more negative attitudes towards ethnic minorities are less likely to have out-group friends, or in-group friends who have out-group friends. This model did not t the data well (x2 28 214:44, p , :01, RMSEA :090; SRMR :125; CFI :924).

Multi-level SEM We had hypothesized that the number of direct cross-group friendships at an aggregate level can form the basis for generating in-group norms about the acceptability of such intergroup contact. Our data included participants from 89 classrooms in different schools, so we had the opportunity to test this group level effect of cross-group friendship. The use of multi-level SEM (ML-SEM) takes into account the hierarchical structure of these data, specifying a measurement model, and relations between latent variables both at the individual or within level and at the classroom or between level. Most importantly, the use of ML-SEM will allow us to separate the effect of direct cross-group friendship at the individual level (having actual out-group

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

516 Pablo De Tezanos-Pinto et al.

friends) from its effect at the classroom level (belonging to a classroom with others that have out-group friends). We propose that the number of out-group friends in each classroom is similar to the traditional measure of indirect contact in that it reects the inuence that the amount of intergroup contact of other in-group members has on each individual. In this case, rather than the behaviour of friends, we are measuring the behaviour of other classroom members. Although students are unlikely to be friends with all their classmates, the frequency of their interactions given by their physical proximity makes the classroom a relevant portion of the in-group. This propinquity effect has been well documented in sociology and social psychology (e.g., Byrne, 1961; Priest & Sawyer, 1967). Therefore, much in the same way as with indirect cross-group friendship, individuals that belong to a class with several out-group friends should perceive in-group norms to be more tolerant and should expect less intergroup anxiety when anticipating intergroup contact. Individual cross-group friendship, on the other hand, is not particularly informative about in-group norms, and should affect attitudes only by reducing intergroup anxiety. To test these hypotheses, some simplications had to be made to the model. The error variance for the rst item of intergroup anxiety had to be xed to zero for the model to converge (u1 2:007; Hox, 2002). The measurement model provided an insufcient t with the data, particularly at the between level (x2 35 121:83, p , :01, RMSEA :055; SRMRwithin :057; SRMRbetween :121; CFI :97). We simplied the measurement model by removing the positively worded item of in-group norms, and this model provided a good t with the data (x2 23 42:78, p , :01, RMSEA :032; SRMRwithin :011; SRMRbetween :042; CFI :99). The intra-class correlation (ICC) for attitudes towards ethnic minorities was .155, indicating that 15.5% of the variation in attitudes was at the classroom level, and the remaining 84.5% of the variation corresponded to the individual level. In-group norms and intergroup anxiety had relatively smaller ICC coefcients, but there was still considerable variance at the classroom level (ICC in-group norms :070; ICC intergroup anxiety :104).4 The model with the hypothesized relations showed a good t with the data ( x2 31 73:39, p , :01, RMSEA :044; SRMRwithin :018; SRMRbetween :057; CFI :982), and is presented in Figure 3. At the individual (within) level, the results are consistent with the results obtained in Model 1, but now the effect of direct friendship is fully mediated by intergroup anxiety. Direct friendship predicted intergroup anxiety (b 20:179, p , :001), but did not show a signicant relation with contact norms (b 20:054, p :227) or a direct path to attitudes towards ethnic minorities (b 0:052, p :173). Both contact norms and intergroup anxiety predicted attitudes, but now intergroup anxiety showed a greater effect (b 20:447, p , :001, for intergroup anxiety; and b 20:154, p , :001, for contact norms). The model explained 28% of the variance of attitudes at the individual level. At the classroom (between) level, cross-group friendship predicted both contact norms (b 20:718, p , :001) and intergroup anxiety (b 20:567, p , :001), and the direct relation between cross-group friendship and attitudes was not signicant

4 n (1991), restricting the factor loadings in the The ICC coefcients for latent variables were calculated as proposed by Muthe measurement model to be equal at the between and within level. This restricted model presented a good t with the data (x2 28 62:99, p , :01, RMSEA :039; SRMRwithin :014; SRMRbetween :029; CFI :986).

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

In-group norms and intergroup contact

y1 y2

517

R 2 = .00
Within level - individual Norms against contact .32 Direct cross-group friendship .18 Intergroup anxiety .45 Positive attitudes .15 y6 y7 y8

R 2 = .28

R 2 = .03

y4

y5

y1

y2

R 2 = .52
.72 Norms against contact .93 Direct cross-group friendship .57 Intergroup anxiety Positive attitudes y6 y7 y8

Between level - classroom

R 2 = .85

R 2 = .32

y4

y5

Figure 3. Results for multi-level SEM model.

(b 0:181, p :659). The relationships between the mediators and attitudes towards ethnic minorities were also non-signicant at this level, though the model explained 85% of the between level variance of attitudes (the lack of signicant relationships in this case may be due to the high correlation between in-group norms and intergroup anxiety at the between level). It is worth remarking that the effect of cross-group friendship on in-group norms and intergroup anxiety at this level is in actuality a crosslevel effect; that is, the amount of cross-group friendship in the classroom predicts the intercept of the regressions at the within level. Thus, individuals that belong to a classroom with more out-group friends have a more tolerant perception of in-group norms and lower levels of intergroup anxiety. These results give additional support to the idea that intergroup contact by other in-group members has an effect on the attitudes towards the out-group. In this case, instead of measuring the intergroup contact of fellow in-group friends (indirect crossgroup friendship), we used the total amount of out-group friendships reported in the classroom. Furthermore, the mechanisms by which cross-group friendship works at the individual level seem to differ from the mechanisms at the classroom level. At an individual level, the effect of having out-group friends was mediated by intergroup anxiety, but not by in-group norms. Belonging to a classroom that reported more outgroup friends, on the other hand, affected both intergroup anxiety and the perception of in-group norms.

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

518 Pablo De Tezanos-Pinto et al.

Discussion
Giving further evidence for the extended contact hypothesis, the results presented here support the idea that the effect of intergroup contact is not limited to the personal experience with a member of the out-group. Intergroup contact experienced by other in-group members (in-group friends or classmates) also affects intergroup attitudes. In addition, this study suggests that this effect can be explained by a change in the perception of in-group norms about contact with the out-group, using both a traditional measure of indirect cross-group friendship and multi-level analysis of direct cross-group friendship. Using one-level SEM analysis including both direct and indirect cross-group friendships as predictors, we replicated the results obtained by Turner et al. (2008), showing that the effect of indirect cross-group friendship is mediated by in-group norms about contact with the out-group. The relation between indirect cross-group friendship and in-group norms about contact was particularly strong, and the path from in-group norms to attitudes towards the out-group was comparable to the effect of intergroup anxiety, a well-established mediator of intergroup contact (Brown & Hewstone, 2005; Pettigrew & Tropp, 2008). The results also supported the same theoretical idea with the use of ML-SEM, in which direct cross-group friendship affected the mediators differently at the individual level compared to the relationships observed at the classroom level. As hypothesized, direct friendship at the individual level did not affect in-group norms, but in-group norms were perceived to be more tolerant when individuals belonged to a classroom where out-group friendships were more prevalent. These results underline the importance of including contextual variables in a theory of intergroup contact. The fact that classrooms with more contact with out-group members also report more positive attitudes cannot be accounted for by the individual effect that direct contact has on attitudes; this effect is controlled in the within part of the model. The relationship is explained by the contact of other classroom members, which affects both intergroup anxiety and the perception of in-group norms. As stated earlier, most studies related to the contact hypothesis do not consider this effect and, until the extended contact hypothesis was introduced, the theory of intergroup contact was based solely on the effect that contact has directly on each individual. This exclusion of the inuence that rest of the in-group has in the attitudes and behaviours of individuals probably results in an overestimation of the individual effect of intergroup contact, and more importantly, it results in an underestimation of the total benets that intergroup contact can provide for the improvement of intergroup relations. To our knowledge, only one other study has reported a multi-level effect of intergroup contact, but using districts as the between level unit of analysis. Wagner, Christ, Pettigrew, Stellmacher, and Wolf (2006) set out to evaluate the effect of the proportion of foreigners on prejudice, using a large probability sample in Germany. Contrary to what would be expected by the threat hypothesis (Blalock, 1967), which states that a large percentage of foreigners will increase prejudice, Wagner et al. (2006) found that a larger percentage of foreigners in the districts was associated with more tolerant attitudes and that this effect was mediated by intergroup contact (cf. Stein, Post, & Rinden, 2000). More important for the purpose of this discussion is the fact that the effect of contact with foreigners was signicant both at the individual and at the district level. People who lived in districts that reported more contact with foreigners were less prejudiced, regardless of their own level of contact. Even though this relationship was not explored further in their paper, it seems reasonable to hypothesize that this effect of

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

In-group norms and intergroup contact

519

contact at the district level is related to the perception of more or less tolerant social norms about contact with foreigners. People who live in districts where there is more contact between members of the majority and foreigners are more likely to perceive that such contact is normal and that relations between the majority and foreign people are more harmonious. As we mentioned before, this study focused on injunctive rather than descriptive ingroup norms. It is interesting to note that extended contact is indeed very close to a measure of descriptive norms (the amount of in-group members that are known to have out-group friends), and therefore is somewhat more surprising that individuals would infer from this a perception of the social approval of such behaviour. It is also likely that descriptive and injunctive norms about contact have independent effects on attitudes towards the out-group, and future research should explore these issues by measuring or manipulating these norms separately. Another interesting result in this study is the comparison between the effects of direct and indirect cross-group friendship. Indirect friendship had a stronger inuence on attitudes than direct friendship when they were both included in the analysis. Even though the typical result when comparing direct and indirect friendship is for the former to have a larger effect, this is not the rst study in which indirect friendship proves to be a superior predictor (Paolini et al., 2007, Study 3; Turner et al., 2008, 2007, Study 2). The study of the circumstances that can affect the relative importance of these two variables is still in its infancy, yet the literature suggests two interesting explanations. When Turner et al. (2008) found a larger effect of indirect friendship, they considered the lack of direct contact with out-group members in their samples as the main reason for this result. Indeed, there is some evidence that in contexts where direct contact is scarce, indirect contact can have a more powerful effect (Christ et al., 2009). A second factor that has been proposed relates to the idea that the effect of direct contact is of a rather more emotional nature, while the effect of indirect contact is more cognitive. Paolini et al. (2007) showed that direct friendship had a stronger effect on attitudes towards out-groups that generated strong affective responding while indirect friendship only had a signicant effect on the attitudes towards engineering students, the out-group that generated the strongest cognitive responding. But neither of these explanations can satisfactorily account for the results of the present study. Attitudes towards ethnic minorities can hardly be considered particularly cognitive, and the measure of attitudes towards these groups used in this study a 10-point adaptation of the feeling thermometer is actually an emotional measure. Also, this study took place in a setting with a large proportion of ethnic minorities, and this was reected on the number of cross-group friendship reported by participants (61% reported having friends from other ethnic groups). A third factor that could be included in future research is the relevance of the attitudes towards the out-group for obtaining in-group approval and the importance attributed by participants to such approval. Teenagers have been found to be particularly susceptible to peer inuence, especially during early adolescence (Costanzo & Shaw, 1966; Steinberg & Monahan, 2007), and extended friendship does seem to be more relevant in adolescent samples. Of the three other studies that have measured both direct and indirect friendship in adolescents, two have reported a larger effect of indirect friendship (Turner et al., 2007, Study 2; Turner et al., 2008, Study 2) and the other found similar effects for both direct and indirect friendship (Turner et al., 2007, Study 3). Because our study used cross-sectional data, the causal relations proposed should be treated with caution. Although we tested alternative models that focused on different

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

520 Pablo De Tezanos-Pinto et al.

theoretical views, future research should focus on the clarication of these relations using both longitudinal studies and experiments. There has been some debate over the causal direction of the relation between intergroup contact and attitudes. In the case of direct intergroup contact, while there is no doubt that contact brings about attitude change, there is also probably some selection process by which more tolerant people are more likely to have out-group friends. There is evidence for both causal directions (Binder et al., 2009; Brown, Eller, Leeds, & Stace, 2007; Levin, van Laar, & Sidanius, 2003; Pettigrew & Tropp, 2006). In the case of extended contact, Wright et al. (1997) were quick to point out that people had little power in choosing the friends of other ingroup members. But it still could be argued that people are likely to have in-group friends with similar views, and therefore less tolerant people would be less likely to have indirect cross-group friends. This argument does not apply to multi-level analyses of intergroup contact in the classroom though, because students cannot choose their classmates. The clustering in attitudes and levels of intergroup contact observed in the classrooms is not a product of self-selection, but the product of a normative process. Another important issue regarding causality is the inuence that in-group norms about contact can have on intergroup behaviour, and particularly on the amount and quality of contact with out-group members. Fendrich (1967) already observed this relationship using a behavioural measure and we intend to pursue this idea further by collecting longitudinal data. This would also effectively include indirect cross-group friendship as a precursor of changes in direct cross-group friendship later on, further increasing its relevance. Finally, it is worth briey considering the policy implications of our ndings. The present ndings, together with those from a recent study of British elementary schoolchildren in institutions of varying ethnic diversity (Brown, 2008), suggest that heterogeneous classrooms and schools, with their greater potential for increasing contact opportunities and developing norms in favour of cross-group relationships, will promote more favourable intergroup attitudes than educational environments that are ethnically more homogeneous. In a political climate where some are doubting the wisdom of such diversity policies, these ndings are a forceful reminder of the social benets of increased intergroup contact, as Allport (1954) so wisely advocated many years ago.

Acknowledgements
Work done by Christopher Bratt was funded by the Norwegian Research Council (project number 185731). We wish to thank local authorities in Drammen for funding the data collection and Conexus for developing the web-based version of the questionnaire.

References
Aboud, F. E., & Levy, S. R. (1999). Are we ready to translate research into programs? Journal of Social Issues, 55(4), 621625. Abrams, D., Rutland, A., & Cameron, L. (2003). The development of subjective group dynamics: Childrens judgments of normative and deviant in-group and out-group individuals. Child Development, 74(6), 18401856. Allport, F. H., Allport, G. W., Babcock, C., Bernard, V. W., Bruner, J. S., Cantril, H., et al. (1953). The effects of segregation and the consequences of desegregation: A social science statement. Minnesota Law Review, 37, 429440.

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

In-group norms and intergroup contact

521

Allport, G. W. (1954). The nature of prejudice. Reading, MA: Addison-wesley. Aron, A., McLaughlin-Volpe, T., Mashek, D., Lewandowski, G., Wright, S. C., & Aron, E. N. (2004). Including others in the self. European Review of Social Psychology, 15(4), 101132. Bagozzi, R. P., & Heatherton, T. F. (1994). A general approach to representing multifaceted personality constructs: Application to state self-esteem. Structural Equation Modeling, 1, 3637. Batson, C. D., Polycarpou, M. P., Harmond-Jones, E., Imhoff, H. J., Mitchener, E. C., Bednar, L. L., et al. (1997). Empathy and attitudes: Can feeling for a member of a group improve feelings towards the group? Journal of Personality and Social Psychology, 72(1), 105118. Binder, J., Zagefka, H., Brown, R., Funke, F., Kessler, T., Mummendey, A., et al. (2009). Does contact reduce prejudice or does prejudice reduce contact? A longitudinal test of the contact hypothesis amongst majority and minority groups in three European countries. Journal of Personality and Social Psychology, 96(4), 843856. Blalock, H. M. (1967). Percent non-White and discrimination in the south. American Sociological Review, 22(6), 677682. Blanchard, F., Crandall, C., Brigham, J., & Vaughn, L. (1994). Condemning and condoning racism: A social context approach to interracial settings. Journal of Applied Psychology, 79(6), 993997. Blanchard, F., Lilly, T., & Vaughn, L. A. (1991). Reducing the expression of racial prejudice. Psychological Science, 2(2), 101105. Brown, R. (2008). Acculturation processes and social relations in young children: Why integration is (mostly) good for you. Paper presented at EAESP General Meeting, Opatje, Croatia. Brown, R., Eller, A., Leeds, S., & Stace, K. (2007). Intergroup contact and intergroup attitudes: A longitudinal study. European Journal of Social Psychology, 37(4), 692703. Brown, R., & Hewstone, M. (2005). An integrative theory of intergroup contact. Advances in Experimental Social Psychology, 37, 255343. Byrne, D. (1961). The inuence of propinquity and opportunities for interaction on classroom relationships. Human Relations, 14(1), 6369. Cameron, L., Rutland, A., Brown, R., & Douch, R. (2006). Changing childrens intergroup attitudes toward refugees: Testing different models of extended contact. Child Development, 77(5), 12081219. Christ, O., Tausch, N., Hewstone, M., Wagner, U., Hughes, J., & Cairns, E. (2009). The benets and limitations of direct and indirect intergroup contact. Manuscript submitted for publication. Cialdini, R. B., Kallgren, C., & Reno, R. (1991). A focus theory of normative conduct: A theoretical renement and reevaluation of the role of norms in human behavior. Advances in Experimental Social Psychology, 24, 201234. Costanzo, P. R., & Shaw, M. E. (1966). Conformity as a function of age level. Child Development, 37(4), 967975. Crandall, C. S., Eshleman, A., & OBrien, L. (2002). Social norms and the expression and suppression of prejudice: The struggle for internalization. Journal of Personality and Social Psychology, 82(3), 359378. Enders, C. K., & Bandalos, D. L. (2001). The relative performance of full information maximum likelihood estimation for missing data in structural equation models. Structural Equation Modeling: A Multidisciplinary Journal, 8(3), 430457. Fendrich, J. M. (1967). Perceived reference group support: Racial attitudes and overt behavior. American Sociological Review, 32(6), 960970. Greenland, K., & Brown, R. (1999). Categorization and intergroup anxiety in contact between British and Japanese nationals. European Journal of Social Psychology, 29(4), 503522. Harwood, J., Hewstone, M., Paolini, S., & Voci, A. (2005). Grandparentgrandchild contact and attitudes toward older adults: Moderator and mediator effects. Personality and Social Psychology Bulletin, 31(3), 393406.

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

522 Pablo De Tezanos-Pinto et al. Hewstone, M., Cairns, E., Voci, A., Hamberger, J., & Niens, U. (2006). Intergroup contact, forgiveness, and experience of The Troubles in Northern Ireland. Journal of Social Issues, 62(1), 99120. Hewstone, M., Cairns, E., Voci, A., McLernon, F., Niens, U., & Noor, M. (2004). Intergroup forgiveness and guilt in Northern Ireland: Social psychological dimensions of The Troubles. In N. Branscombe & B. Doosje (Eds.), Collective guilt: International perspectives (pp. 193215). New York: Cambridge University Press. Hewstone, M., Cairns, E., Voci, A., Paolini, S., McLernon, F., & Crisp, R. J. (2005). Intergroup contact in a divided society: Challenging segregation in Northern Ireland. In D. Abrams, J. M. Marques, & M. A. Hogg (Eds.), The social psychology of inclusion and exclusion (pp. 265292). Philadelphia, PA: Psychology Press. Hox, J. J. (2002). Multilevel analysis: Techniques and applications. Mahwah, NJ: Erlbaum. Hu, L., & Bentler, P. M. (1999). Cutoff criteria for t indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6(1), 155. Islam, M. R., & Hewstone, M. (1993). Dimensions of contact as predictors of intergroup anxiety, perceived outgroup variability, and outgroup attitude: An integrative model. Personality and Social Psychology Bulletin, 19(6), 700710. Jo reskog, K. (1969). A general approach to conrmatory maximum likelihood factor analysis. Psychometrika, 34(2), 183202. Levin, S., van Laar, C., & Sidanius, J. (2003). The effects of ingroup and outgroup friendships on ethnic attitudes in college: A longitudinal study. Group Processes and Intergroup Relations, 6, 7692. Liebkind, K., & McAlister, A. (1999). Extended contact through peer modelling to promote tolerance in Finland. European Journal of Social Psychology, 29(56), 765780. Little, T. D., Cunningham, W. A., Shahar, G., & Widaman, K. F. (2002). To parcel or not to parcel: Exploring the question, weighing the merits. Structural Equation Modeling: A Multidisciplinary Journal, 9(2), 151173. Loehlin, J. C. (2003). Latent variable models: An introduction to factor, path, and structural analysis (4th ed.). Mahwah, NJ: Erlbaum. Marques, J. M., & Paez, D. (1994). The black sheep effect: Social categorization, rejection of ingroup deviates, and perception of group variability. European Review of Social Psychology, 5(1), 3768. Monteith, M. J., Deneen, N. E., & Tooman, G. D. (1996). The effect of social norm activation on the expression of opinions concerning gay men and Blacks. Basic and Applied Social Psychology, 18(3), 267288. n, B. O. (1991). Multilevel factor analysis of class and student achievement components. Muthe Journal of Educational Measurement, 28(4), 338354. n, L. K., & Muthe n, B. O. (19982007). Mplus users guide (5th ed.). Los Angeles, CA: Muthe n & Muthe n. Muthe Nesdale, D., Maass, A., Durkin, K., & Grifths, J. (2005). Group norms, threat, and childrens racial prejudice. Child Development, 76(3), 652663. Paluck, E. L. (2009). Reducing intergroup prejudice and conict using the media: A eld experiment in Rwanda. Journal of Personality and Social Psychology, 96(3), 574587. Paluck, E. L., & Green, D. P. (2009). Prejudice reduction: What works? A review and assessment of research and practice. Annual Review of Psychology, 60(1), 339367. Paolini, S., Hewstone, M., & Cairns, E. (2007). Direct and indirect intergroup friendship effects: Testing the moderating role of the affective-cognitive bases of prejudice. Personality and Social Psychology Bulletin, 33(10), 14061420. Paolini, S., Hewstone, M., Cairns, E., & Voci, A. (2004). Effects of direct and indirect cross-group friendships on judgments of Catholics and Protestants in Northern Ireland: The mediating role of an anxiety-reduction mechanism. Personality and Social Psychology Bulletin, 30(6), 770786.

Copyright The British Psychological Society

Reproduction in any form (including the internet) is prohibited without prior permission from the Society

In-group norms and intergroup contact

523

Pettigrew, T. F. (1991). Normative theory in intergroup relations: Explaining both harmony and conict. Psychology and Developing Societies, 3(1), 316. Pettigrew, T. F. (1998). Intergroup contact theory. Annual Review of Psychology, 49, 6585. Pettigrew, T. F. (2006). The advantages of multilevel approaches. Journal of Social Issues, 62(3), 615620. Pettigrew, T. F. (2008). Future directions for intergroup contact theory and research. International Journal of Intercultural Relations, 32(3), 187199. Pettigrew, T. F., Christ, O., Wagner, U., & Stellmacher, J. (2007). Direct and indirect intergroup contact effects on prejudice: A normative interpretation. International Journal of Intercultural Relations, 31(4), 411425. Pettigrew, T. F., & Tropp, L. R. (2006). A meta-analytic test of intergroup contact theory. Journal of Personality and Social Psychology, 90(5), 751783. Pettigrew, T. F., & Tropp, L. R. (2008). How does intergroup contact reduce prejudice? Metaanalytic tests of three mediators. European Journal of Social Psychology, 38(6), 922934. Priest, R. F., & Sawyer, J. (1967). Proximity and peership: Bases of balance in interpersonal attraction. American Journal of Sociology, 72(6), 633649. Rutland, A., Cameron, L., Milne, A., & McGeorge, P. (2005). Social norms and self-presentation: Childrens implicit and explicit intergroup attitudes. Child Development, 76(2), 451466. Sherif, M. (1966). Group conict and cooperation. London: Routledge. Smith, J. R., & Louis, W. R. (2008). Do as we say and as we do: The interplay of descriptive and injunctive group norms in the attitudebehaviour relationship. British Journal of Social Psychology, 47(4), 647666. doi:10.1348/014466607X269748 Stangor, C., Sechrist, G. B., & Jost, J. T. (2001). Changing racial beliefs by providing consensus information. Personality and Social Psychology Bulletin, 27(4), 486496. Stein, R., Post, S., & Rinden, A. (2000). Reconciling context and contact effects on racial attitudes. Political Research Quarterly, 53(2), 285303. Steinberg, L., & Monahan, K. C. (2007). Age differences in resistance to peer inuence. Developmental Psychology, 43(6), 15311543. Stephan, W. G., & Stephan, C. (1985). Intergroup anxiety. Journal of Social Issues, 41(3), 157175. Terry, D. J., & Hogg, M. A. (1996). Group norms and the attitudebehavior relationship: A role for group identication. Personality and Social Psychology Bulletin, 22(8), 776793. Turner, R. N., Hewstone, M., & Voci, A. (2007). Reducing explicit and implicit outgroup prejudice via direct and extended contact: The mediating role of self-disclosure and intergroup anxiety. Journal of Personality and Social Psychology, 93(3), 369388. Turner, R. N., Hewstone, M., Voci, A., & Vonofakou, C. (2008). A test of the extended intergroup contact hypothesis: The mediating role of intergroup anxiety, perceived ingroup and outgroup norms, and inclusion of the outgroup in the self. Journal of Personality and Social Psychology, 95(4), 843860. Voci, A., & Hewstone, M. (2003). Intergroup contact and prejudice towards immigrants in Italy: The mediational role of anxiety and the moderational role of group salience. Group Processes and Intergroup Relations, 6(1), 3754. Wagner, U., Christ, O., Pettigrew, T. F., Stellmacher, J., & Wolf, C. (2006). Prejudice and minority proportion: Contact instead of threat effects. Social Psychology Quarterly, 69(4), 380390. Wilner, D. M., Walkley, R. P., & Cook, S. W. (1952). Residential proximity and intergroup relations in public housing projects. Journal of Social Issues, 8(1), 4569. Wright, S. C., Aron, A., McLaughlin-Volpe, T., & Ropp, S. A. (1997). The extended contact effect: Knowledge of cross-group friendships and prejudice. Journal of Personality and Social Psychology, 73(1), 7390. Received 24 February 2009; revised version received 29 July 2009