15
JOURNAL OF EXPERIMENTAL SOCIAL PSYCHOLOGY 22, 78-92 (1986) The Effects of Intergroup Similarity and Goal interdependence on intergroup Attitudes and Task Performance RUPERT BROWN AND DOMINIC ABRAMS University of Kent. United Kingdom Received February 15. 1985 The concept of similarity occupies an important place in several theories of social relations. An experiment was designed to examine the effects of both status and attitudinal similarity under intergroup competition or cooperation. Previous research had indicated that the usual attractive effects of similarity would be most evident in cooperative contexts, while similarity might lead to divergence under competition due to threats to group identity. Two hundred eight school children participated in a 2 x 3 x 2 factorial experiment (nature of Task x outgroup Status x outgroup Attitudes). Results indicated that, contrary to hypothesis, simple attitude similarity led to a general increase in liking for and cooperativeness toward the outgroup, unaffected by goal relations. However, on ratings of group performance, status and attitudinal similarity combined did lead to increased intergroup differentiation, again unaffected by goal relations. Finally, subjects’ performances on the experimental task (a verbal and arithmetical reasoning test) were reliably affected by goal relations: highest performance being observed under intergroup cooperation, particularly with same or lower status outgroups, or with outgroups which had different attitudes. It is concluded that the need for positive distinctiveness in intergroup relations may compromise or even reverse the usual relationship between similarity and attraction. 0 1986 Academic Press, Inc. This paper is concerned with the effects of intergroup similarity on intergroup attitudes and task performance under conditions of cooperation or competition. Consistent with earlier research (Brown, 1984a) we have The authors gratefully acknowledge the invaluable assistance of Peter Allen, Robert Crowther and Russell Newcombe who acted as fellow experimenters, the children and teachers of the two schools which participated in the research, and the Faculty of Social Sciences at the University of Kent for supporting the research financially. Requests for reprints should be addressed to Dr. Rupert Brown, Social Psychology Research Unit, University of Kent, Beverley Farm, Canterbury, Kent, United Kingdom. Dr. Abrams is now at the Department of Psychology, University of Dundee. 78 0022-1031/86 $3.00 Copyright 0 1986 by Academic Press. Inc. All rights of reproduction in any form reserved.

The effects of intergroup similarity and goal interdependence on intergroup attitudes and task performance

  • Upload
    kent

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

JOURNAL OF EXPERIMENTAL SOCIAL PSYCHOLOGY 22, 78-92 (1986)

The Effects of Intergroup Similarity and Goal interdependence on intergroup Attitudes and Task

Performance

RUPERT BROWN AND DOMINIC ABRAMS

University of Kent. United Kingdom

Received February 15. 1985

The concept of similarity occupies an important place in several theories of social relations. An experiment was designed to examine the effects of both status and attitudinal similarity under intergroup competition or cooperation. Previous research had indicated that the usual attractive effects of similarity would be most evident in cooperative contexts, while similarity might lead to divergence under competition due to threats to group identity. Two hundred eight school children participated in a 2 x 3 x 2 factorial experiment (nature of Task x outgroup Status x outgroup Attitudes). Results indicated that, contrary to hypothesis, simple attitude similarity led to a general increase in liking for and cooperativeness toward the outgroup, unaffected by goal relations. However, on ratings of group performance, status and attitudinal similarity combined did lead to increased intergroup differentiation, again unaffected by goal relations. Finally, subjects’ performances on the experimental task (a verbal and arithmetical reasoning test) were reliably affected by goal relations: highest performance being observed under intergroup cooperation, particularly with same or lower status outgroups, or with outgroups which had different attitudes. It is concluded that the need for positive distinctiveness in intergroup relations may compromise or even reverse the usual relationship between similarity and attraction. 0 1986

Academic Press, Inc.

This paper is concerned with the effects of intergroup similarity on intergroup attitudes and task performance under conditions of cooperation or competition. Consistent with earlier research (Brown, 1984a) we have

The authors gratefully acknowledge the invaluable assistance of Peter Allen, Robert Crowther and Russell Newcombe who acted as fellow experimenters, the children and teachers of the two schools which participated in the research, and the Faculty of Social Sciences at the University of Kent for supporting the research financially. Requests for reprints should be addressed to Dr. Rupert Brown, Social Psychology Research Unit, University of Kent, Beverley Farm, Canterbury, Kent, United Kingdom. Dr. Abrams is now at the Department of Psychology, University of Dundee.

78 0022-1031/86 $3.00 Copyright 0 1986 by Academic Press. Inc. All rights of reproduction in any form reserved.

THE EFFECTS OF INTERGROUP SIMILARITY 79

concentrated particularly on two forms of Gmilasity~ttitudinaI similarity and status similarity-since both of these are relevant to several important social psychological’ theories.

In a recent review it was argued that there are two main theoretical positions regarding the effects of similarity (Brown, 1984b). The first, and by far the most predominant view, holds that similarity has convergent effects on social behavior: individuals are hypothesized to seek out those similar to them, and the resulting relationship is thought to be one of mutual attraction (e.g., Festinger, 1954; Newcomb, 1961). There is con- siderable empirical support for this view, at least in the area of interpersonal attraction (e.g., Byrne, 1971; Duck, 1977). Against this prevailing or- thodoxy, Snyder and Fromkin (1980) have argued that most research investigating the similarity-attraction relationship has explored only moderate levels of similarity. In their uniqueness theory Snyder and Fromkin propose that extreme similarity may actually be aversive because of people’s need to retain some unique components of their identity. Although Snyder and Fromkin are primarily concerned with interpersonal relationships, they also extend their argument to the group domain (see, e.g., Snyder & Fromkin, 1980, pp. 65-72). This point of view is entirely congruent with Tajfel’s (1978) theory of social identity which proposes that an important source of people’s identity actually derives from their membership in social groups and, in particular, from comparisons which they make between their own group and others. The main aim of these comparisons is thought to be to establish some kind of positive distinc- tiveness for the ingroupi.e., to make it as different as possible in postively valued ways. Thus, groups which become too similar may make greater efforts to distinguish themselves from one another, with a resulting increase in discrimination and dislike. Theoretically speaking, then, sim- ilarity can be both attractive or repellent depending on the manner in which identity is implicated.

What has research into intergroup behavior contributed to this theoretical controversy over the effects of similarity? Brown (1984b) concluded that the evidence is rather equivocal. On the one hand a number of field studies have been reported in which intergroup similarity-usually similarity between different cultural or subcultural groups-has been associated with more favorable evaluations and higher ratings of attraction (e.g., Berry, Kahn, & Taylor, 1977; Brewer & Campbell, 1976). On the other hand, the evidence from experimental research has been more ambivalent. For example, Turner (1978) and Mummendey and Schreiber (1984) found that competing groups of students showed more ingroup favoritism when confronted with a similar outgroup than when they met a different outgroup. In contrast, Brown (1984a, Experiment l), in a setting in which groups anticipated cooperation with each other, obtained evidence supporting the similarity-attraction view. In an attempt to reconcile these findings Brown (1984a, Experiment 2) manipulated anticipated cooperation and

80 BROWN AND ABRAMS

competition, reasoning that the similarity-attraction relationship might be more applicable in cooperative situations since it is there that the consensual-validation function of attitudinal similarity-the basis of most similarity-attraction theories-is most relevant. For it is difficult to see how cooperation can occur without some agreement over aims and ob- jectives. Where, however, the situation is competitive it was thought that similarity might increase rivalry, first, because of the implied lack of uniqueness and, second, because competition is facilitated if basic values are shared. Two schools are likely to show much more rivalry toward one another if each places a premium on success at football than if one values football and the other musical achievement. A parallel argument was made for status similarity. In cooperation, the need may be for status equality so that participants can regard themselves as con- tributing equally to the joint achievement (Deutsch, 1975). Under com- petition, however, similarity implies less distinctiveness and hence dif- ferentiation, and dislike may be stimulated. Two teams adjacent in a league are likely to show stronger rivalry than two teams at opposite ends of the table. There was some support for this hypothesis since more competitive subjects did indeed react to attitudinal similarity by expressing less liking toward the outgroup, while less competitive subjects showed the more usual pattern. However, the force of this result was weakened by the fact that it only emerged in a post hoc secondary analysis of the data; the actual experimental manipulations had not successfully altered subjects’ feelings of competitiveness and also had failed to elicit the predicted response. In the light of this evidence it may be wise to reconsider the view that the need for positive distinctiveness prevails in all intergroup contexts. Instead, we are led to ask what kind of intergroup relations increase or decrease the attractiveness of similar outgroups, and which characteristics of the outgroup are most important?

The experiment reported below was designed to investigate the impact of status and attitude similarity within cooperative and competitive in- tergroup goal relationships. The design was a modification of that adopted by Brown (1984a) in order to create a clearer manipulation of intergroup interdependence. Accordingly, it was decided to use actual competition and cooperation, rather than the anticipation thereof, as we had previously. Following Rabbie and de Brey (1971), who have extended the pioneering work of Sherif (1966), it seemed likely that actual competition and co- operation should be more effective in altering people’s goal orientations and intergroup attitudes than merely expecting to compete or cooperate. In line with this tradition of work, it was also predicted that competition would evoke more differentiation and less liking than cooperation. In- tergroup differentiation is simply a measure of the magnitude and direction of difference in rating of aspects of the ingroup versus the outgroup. From Sherif (1966) onward, such indices of differentiation (or bias) have

THE EFFECTS OF INTERGROUP SIMILARITY 81

been among the most commonly used dependent measures in intergroup relations research. They are particularly relevant to social identity theory since, as already noted, social identity is thought to be maintained by the establishment of ingroup superiority over outgroups on valued di- mensions of comparison (Tajfel, 1978).

While the predictions for the effects of different goals are quite straight- forward, the effects of status and attitudinal similarity are more contentious. First, we should note that, in general, intergroup evaluations will tend to reflect socially defined status relations between groups. For example, Brown (1984a, Experiment 2) found that, where there was clear consensus about status differences, higher status groups were evaluated more fa- vorably than lower status groups (see also van Knippenberg, 1984). Only when that consensus is absent, due to some perception of instability or illegitimacy, does this unconditional preference for higher status outgroups become modified. (Caddick, 1982; Turner & Brown, 1978). In fact, in the present experiment some instability was deliberately introduced into the status manipulation to allow the subjects some psychological room for maneuver in their intergroup ratings. Therefore, while we predicted that the experimental variation of the groups’ relative status would have some general linear effect on subjective judgments, (i.e., superior status groups would be likely to receive higher ratings than lower status groups), we hypothesized that these judgments would not perfectly reflect the objectively defined hierarchy-both because of the instability in the status rankings and because of a general tendency for ingroup status to be upgraded (cf. Festinger, 1954). The effects of status similarity may be tested by comparing conditions of outgroup status equivalence against the combined conditions of outgroup inferiority and superiority, thus controlling for any effects of status per se (Brown, 1984a). Social identity and uniqueness theory predict that the greatest intergroup differentiation and least attraction will arise when the outgroup is similar (either in status, attitudes, or both), particularly in competitive contexts. In contrast, similarity may be more likely to lead to attraction in cooperative contexts.

Another objective of the experiment was to investigate the effects of these same variables on task performance. This has been a little studied issue. Most intergroup research has concentrated on attitudinal and cog- nitive aspects of intergroup situations (see Tajfel, 1982. and Brewer & Kramer, 1985, for reviews of recent work) and has rather neglected the implications of different intergroup relations for group productivity. John- son, Maruyama, Johnson, and Skon’s (1981) comprehensive review con- cluded that cooperation is generally more effective than competition in promoting productivity, but their meta-analysis did not directly compare intergroup cooperation with intergroup competition. In two of the few studies to do so, Workie (1974) and Goldman et al. .( 1977) found higher performance under intergroup cooperation than under intergroup com-

82 BROWN AND ABRAMS

petition, and on the basis of these we hypothesized a similar result in our study. No predictions were made concerning the effects of similarity, since this aspect of the experiment was largely exploratory.

METHOD

Subjects Two hundred eight If-year-old boys and girls (95 male. 113 female) from two secondary

schools in Kent, England, acted as subjects. Almost equal numbers came from each school (103 and 105).

Overview of Design Three independent variables were factorially combined in a 2 x 3 x, 2 independent

groups design. These were the nature of the Task (Competitive or Cooperative), the presumed Status of the outgroup (Higher than, Equal to, or Lower than the ingroup), and the perceived Attitudes of the outgroup (Similar to or Different from those in the ingroup). Subjects were randomly assigned to each condition yielding cell sizes ranging between 15 and 19 with a mean of 17.3.

Procedure Within each school the children were gathered together and reintroduced to the exper-

imenters, one of whom they had met a few weeks earlier during a pretesting session to establish the relevant dimensions of status and over which issues there was a fair degree of attitudinal consensus in the two schools. It was explained that the purpose of the experiment was to “research into a number of aspects of school life.” The children were then explicitly randomly divided into four groups of 25-30 and taken off to separate classrooms where they were seated well apart from each other and each given an envelope containing the experimental materials. It was administratively more convenient to keep all subjects in any one Task condition together. Subjects were told that we were trying out a new kind of General Abilities (GA) Test “especially designed to test how good people are at Maths and English” (these two disciplines had been rated as the two most important in pretesting). It was emphasized that scores would be kept confidential and that no teacher would see their answers.

At this point the Task manipulation was introduced. In the Cooperative condition subjects were told that we were “interested in how working ~lith another school affects people’s performance.” It was explained that the test had been designed so that half of it had to be answered by one school and half by another so that they would be cooperating with another school in order to answer it. The experimenter questioned the class to ensure that they understood what the word cooperate meant. In the Competitive condition, by contrast, subjects were told of our interest in the effects of competition on performance and that the schools’ relative performances would be compared. Again, the class was questioned to ensure that they understood the meaning of compete. In both conditions it was revealed that we would be giving out some prize money depending either on the schools’ joint (in Cooperation) or relative (in Competition) performances. Although the exact amount was not specified it was implied that the reward was fairly substantial by indicating that the head teacher would be using it to purchase some expensive new sports equipment for the school.’ In Sherif’s (1%6) terms, therefore, the school groups were clearly and materially interdependent-positively in Cooperative conditions, negatively in Competitive conditions.

’ In fact, a sum of f25 was given to each school at the conclusion of the experiment.

THE EFFECTS OF INTERGROUP SIMILARITY 83

Indeed, the nature of the manipulations meant that subjects in the Cooperation condition had no means of competing while those in the Competition condition could not cooperate. In both conditions less effort would result in a poorer outcome for the ingroup.

The perceived nature of the other school was then varied in a similar way to that used by Brown (1984a). The experimenter indicated that it was from another part of the country and thus would not be known to the children. However, they were told that some information about the academic standards and general attitudes of the two schools had been collected and that this was summarized on the information sheets contained in the envelopes in front of them. (This story was plausible in view of the pretest session some weeks earlier). The children were guided through these information sheets without, however, revealing to them that the sheets were not all identical. The information about their own school corresponded more or less to their own perceptions which we had elicited in the pretest. Thus, their school was said to be “a bit above average at Maths and English,” coming somewhere between 35th and 40th out of 100 other schools; also it was revealed that “nearly 9 out of 10 of the children at your school think that Maths and English are the most important subjects to learn,” and this was supported by three plausible reasons attributed to some people at their school. These mentioned the vocational, academic, and general importance of these disciplines. This information was then contrasted with some corresponding information about the “other school,” which varied according to experimental condition.

The Status of the other school in relation to its overall competence at Maths and English was varied by describing it as “quite a bit better than,” “about the same as,” or “quite a bit worse than” their own school with corresponding rank orders of 20-25th, 35-40th. and 50-55th out of 100. In addition to the slight numerical imprecision in these rankings, it was noted that we could not be very sure that “we are exactly right about how good the schools are because schools change from year to year.” The purpose of introducing this ambiguity was to introduce an element of instability into the hierarchy so that the status differences would not seem too immutable.

The Attitudes of the other school toward the task relevant skills of Maths and English were varied in a similar way. In the Similar attitudes condition it was observed that “nearly 9 out of 10” of the children in the outgroup agreed with the children’s own high evaluation of the importance of Maths and English, and some plausible “quotes” were added for realism. In the Different attitudes condition it was reported that a similar majority actually believed that science and practical subjects were the most important ones to learn, again with some accompanying justifications.

Since the envelopes containing this information had been distributed randomly at the start of the session, it was possible to run all six combinations of the Status and Attitudes variables simultaneously.

Subjects were then reminded about the Cooperative or Competitive nature of the task and the precise nature of the test was revealed. It consisted of 32 IQ test type items (both verbal and arithmetical) and was so designed that each question had two answers. The test form was divided in the middle with “Form A” and “Form B” printed on either side. Their job was to answer Form B since the other school were said to have already answered Form A. This deception was reinforced by the experimenter drawing a completed questionnaire from a box in front of him, showing one half already filled in. He explained how each of the children would be matched up at random with someone from this other school and their joint answers examined. This randomness was emphasized since it meant that all members of the group would have to try equally hard, and could not rely on being paired with a “poor performer” from the outgroup. The encounter was thereby maintained at an intergroup level with little room for subjects to individualize their outgroup partners (see Turner, 1981, for elaboration of this methodological device in the minimal group paradigm). In the Cooperative conditions both halves of each answer had to be correct to score a point, and the total points gained by both schools would be compared to the national

84 BROWN AND ABRAMS

average in order to assess the amount of prize money they would be awarded. In the Competitive conditions each answer was to be scored independently but the total points gained by each school were to be compared with each other, and the winning school would gain the larger share of the prize money. In both conditions it was stressed that it was the school’s performance as a whole which we were interested in, and now how well they did individually.

Subjects were given exactly 5 min for the test. After this, they completed the dependent measure booklet which contained 14 randomly ordered questions relating to their perceptions of the experiment and the schools taking part. In this, as throughout, it was emphasized that their answers would be confidential. Finally, once all sessions had been completed the groups were reconvened, debriefed thoroughly, and thanked for their participation.

Dependent Measures

All the measures, with one exception,’ consisted of bipolar scales represented by a IO- cm unmarked line. This format was used because in previous research it has been found to be more easily introduced and hence more suitable for use by subjects of this age group than are numbered scales (e.g., Brown, 1984a; Brown & Ross. 1982). Subjects simply had to mark the line to give their response, and these were later converted to a IO-point scale. The principal items consisted of various measures of social or affective orientation (e.g., friendliness, cooperativeness, and competitiveness toward the outgroup) and more directly evaluative measures (e.g., estimated test performance of both schools). There were also some checks on the experimental manipulations.

RESULTS

The data were analyzed using the SPSS ANOVA subprogram with the regression option for unequal n’s. All significance levels reported are for two-tailed tests unless otherwise noted. Where questions had been omitted or improperly answered, these observations were omitted, resulting in reduced degrees of freedom in some analyses. Preliminary analyses revealed few significant effects due to sex of subject or school affiliation, and those which did occur did not affect the principal findings reported below. Accordingly, both sex and school factors have been collapsed in analyses which follow.

Manipulation Checks

Checks on all three independent variables revealed that they had been successfully manipulated. Subjects in Competitive conditions were more likely to see the two schools as being on “opposite sides” rather than as “two teams working together” than were those in Cooperative conditions (6.4 vs 2.5; F(1, 189) = 104.2, p < .OOl). This is the critical manipulation check for the Task variable, since it taps how the objective interdependence between the groups was perceived. However, it is worth noting that subjective feelings of competitiveness and cooperativeness also showed

’ The exception was a simple payoff matrix in which subjects were invited to distribute a hypothetical f60 between the two schools. The results from this and other less directly relevant measures are available from the authors on request.

THE EFFECTS OF INTERGROUP SIMILARITY 85

some correspondence to the externally defined goal relations. Thus subjects felt more competitive and less cooperative in the Competitive conditions (6.1 vs 5.4, F(1, 193) = 4.0, p < .05; 4.9 vs 5.7, F(1, 195) = 6.4, p < .025). They also judged that the two groups would work less well together (4.9 vs 5.9, F(1, 194) = 11.2, p < JOI). The fact that these subjective orientations, while clearly different in the two task conditions, did not perfectly match the objective task interdependence is a matter of some theoretical interest (see Turner, 1981) but does not jeopardize the effec- tiveness of the manipulation itself. The perceived relative competence of the other school3 in the two task relevant skills (maths and English)- our check on the Status manipulation-varied strongly and approximately linearly with the Status variable (xH = 0.3,, Es = l.l,, XL = 2.jb; F(2, 179) = 13.3, p < .OOl). Means with a different subscript were significantly different from one another using the Newman-Keuls procedure). No other independent variables had a significant effect on this measure. Note again that these perceived status differences, although in line with the experimentally defined relations, are somewhat compressed upward. This is understandable in view of the instability of the status positions and the resultant desire to avoid a negative evaluation of the ingroup (Tajfel, 1978).4 Finally, those in the Similar attitudes condition saw the two schools as more alike in the “things they think important to learn at school” than those in Different (5.9 vs 2.7; F(1, 194) = 78.0, p < .OOl). Again, this was the only significant effect on this measure.

Intergroup Differentiation

The principal measure of differentiation was derived from subjects’ estimations of the two groups’ performances on the test. A “within subjects” ANOVA (with a fourth Ingroup-Outgroup factor) revealed several significant effects (see Table 1). Taking first the combined ratings of both groups (“between subjects” stratum): ratings increased as a function of both the status of the outgroup (x, = 6.2, Es = 5.9, x, = 5.4) and its attitudinal similarity (Xsim = 6.1, Sr;,, = 5.5). More relevant to our hypotheses, however, were the effects in the “within-subjects” stratum. As is clear, there was consistent positive intergroup differentiation (x1, = 6.3, TOUT = 5.4), but this was qualified by interactions with Status, and with Status and Attitudes combined. The former interaction is represented in the last two columns of Table 1, where it can be seen that the differentiation (or bias) displayed varied inversely with the status

3 This was measured by taking the difference between the rating given to the ingroup and that given to the outgroup (on the same scale) in answer to the question “How good do you think the two schools are at Maths and English.” The higher the score the greater the perceived difference in status.

4 This result is also consistent with Festinger’s (1954) hypothetical “unidirectional drive upwards” in interpersonal comparisons.

86 BROWN AND ABRAMS

TABLE 1 DIFIXRENTIATION IN ESTIMATIONS OF TASK PERFORMANCE’

Competitive Cooperative

Similar Different Similar Different Means

In Out In Out In Out In Out In out

Higher 6.4 6.4 6.6 4.6 6.1 6.7 6.5 6.4 6.4 6.0 Same 6.8 5.8 5.5 4.9 7.1 5.9 5.7 5.3 6.3 5.5 Lower 6.0 4.7 6.1 4.1 6.2 5.4 6.3 4.3 6.1 4.6

Means 6.4 5.7 6.0 4.6 6.5 6.0 6.2 5.4 6.3 5.4

Note. ANOVA: Status (S), F(2, 190) = 4.9, p < .Ol; Attitudes (A), F(1, 190) = 7.9, p < .Ol; MSE between = 4.4. Ingroup-Outgroup Bias (B), F(1. 190) = 29.6, p < ,001; B x S, F(2, 190) = 4.1, p < .025; B x S x A, F(2. 190) = 3.3, p < .05; MSE within = 2.6. All other interactions were nonsignificant.

’ Based on answers to the question, “In the test you have just done, how well do you think your school/the other school did?”

of the group with whom the task was undertaken: least bias was shown against the Higher status outgroup, most against the Lower outgroup, with the Same status condition falling between these two. There was no sign of any depression or enhancement in this latter condition due to a status similarity effect, the comparison of Same vs (Higher + Lower)/2 failing to achieve significance (t(190) = 0.4, n.s.). This Status effect was qualified by a Status x Attitudes interaction shown in Fig. 1. As can be seen, for Same status outgroups similarity of attitudes increased dif-

+2-

+1-

. Similar Attitudes

FIG. 1. The effects of Status and Attitude similarity on intergroup differentiation in performance estimations. To simplify the presentation of this interaction, outgroup scores have been subtracted from ingroup scores to produce an index of differentiation (or bias).

THE EFFECTS OF INTERGROUP SIMILARITY 87

ferentiation, while in the other conditions it had the reverse effect. The predicted interactions with the Task variables were absent. These subjective estimates of performance bore little relation to their own actual perfor- mance. An analysis of covariance using the subject’s own performance score as covariate revealed that there was no significant covariate effect and that the interactions in Table I were still visible at the same levels of significance.

Intergroup Liking and Cooperativeness

Liking for and cooperativeness toward the outgroup were both greater when the outgroup was thought to have Similar attitudes than when it appeared to be Different (5.4 vs 4.8 and 5.8 vs 4.9; F(1, 194) = 4.3, p < .05 and F( 1, 195) = 8.5, p < .005, respectively). These findings support the similarity-attraction position. Contrary to prediction, there were no interactions between Task and Attitudes on these measures, and the main effect for Task on liking, while in the predicted direction, was rather weak (Xcomp = 4.8, ZcooP = 5.3; F(1, 194) = 2.9, p < .05, one tail). However, as already noted, subjects in the Competitive condition did feel less cooperative and less ready to see the two groups working well together than did those in the Cooperative condition. It should also be noted that all the above results still held even when using task per- formance as a covariate. In order to make a direct comparison with analyses conducted in an earlier study (Brown, 1984a, Experiment 2), subjects were partitioned (using a median split) into High (x = 8.2) and Low (x = 3.2) competitive groups on the basis of their scores on the measure “How competitive toward the other school did you feel?“. Those 34 subjects whose scores fell at the median (=6) were excluded from the sample.

Analysis revealed a significant Competitiveness x Attitudes interaction on the liking measure (F(1, 159) = 8.0, p < .005): similarity led to increased liking only for subjects scoring low in competitiveness; for highly competitive subjects the reverse was true. Although this finding must be viewed with caution in view of its post hoc and correlational nature, it nevertheless replicates a similar result obtained in the previous experiment (Brown, 1984a).

Test Performance

Subjects’ performance on the test varied systematically with all three independent variables (see Table 2). Those taking the test under Cooperative instructions scored higher than those under Competitive instructions (18.6 vs 16.7), but this main effect was qualified by two significant interactions. The interaction between Task and Status indicated that performance under Cooperative conditions was superior only when participating with a Same or Lower status outgroup; with a Higher status outgroup per-

88 BROWN AND ABRAMS

TABLE 2 MEAN TEST PERFORMANCE SCORES

Similar

Competitive

Different Mean Similar

Cooperative

Different Mean

Higher 19.7 18.2 18.9 16.3 18.4 17.4 Same 17.8 14.9 16.4 18.6 21.8 20.2 Lower 14.9 14.2 14.5 17.5 18.9 18.3

Means 17.5 15.8 17.5 19.6

Note. Possible range, O-32. ANOVA: Task, F(1, 196) = 4.0, p < .05; Task x Status, F(2, 196) = 3.2, p < .05 Task x Attitudes, F(1, 196) = 3.9, p < .05; MSE = 53.3. All other effects were nonsignificant.

formance was slightly better under Competitive conditions. Similarly, the Attitudes X Task interaction revealed that it was only in the Different attitudes condition that the superiority of intergroup cooperation was evident.

DISCUSSION

The purpose of this experiment was to examine the competing claims of two different theoretical approaches concerning the likely effects of intergroup similarity on intergroup differentiation and liking, following the suggestive evidence from some earlier research in which goal rela- tionships had been implicated as a possible mediating variable (Brown, 1984a). In discussing our findings we focus on three main issues.

The first and major conclusion is that the effects of similarity at the intergroup level are not as straightforward as they are often assumed to be by similarity-attraction theorists. On the one hand, there are clearly some parallels: on our liking and cooperativeness measures we obtained unambiguous evidence that attitude similarity leads to attraction, regardless of the type of goal relation. This was contrary to our hypothesis and clearly supports the similarity-attraction position, suggesting that some of our earlier caution about the dangers of extrapolating from interpersonal to intergroup relations may not have been completely warranted (Brown & Turner, 1981). On the other hand, there was little sign that the similarity- attraction approach is so readily applicable to situations involving dif- ferential status relations or measures tapping the more evaluative aspects of intergroup behavior. Thus, on the ratings of group performance ingroup bias was enhanced when an outgroup was both similar in outlook and status. That such a “doubling up” of similarity should have led to intergroup divergence is clearly more consistent with social identity theory and uniqueness theory than similarity-attraction. The occurrence of such radically different results on these two different kinds of measures raises

THE EFFECTS OF INTERGROUP SIMILARITY 89

the important question of the relationship between them. There is often a tendency among workers in intergroup relations to assume that the two types of measure simply covary, but the evidence from this experiment and others suggest otherwise (Brewer, 1979; Brown, 1984b). In other words, we should not assume that just because two groups enjoy friendly or cooperative relations that they will necessarily not seek ways to derogate each other by making ingroup-favoring judgments on other subjectively important dimensions of intergroup comparison (van Knippenberg, 1984).

If these data provide a partial fit with the distinctiveness postulates of social identity and uniqueness theory they may also be explicable in terms of Tesser and Campbell’s (1980) model of self-evaluation main- tenance. This model holds that, when an individual compares him or herself with others on dimensions which are relevant to self-definition, the other’s performance can be regarded as posing a potential threat. This threat is thought to be particularly great when the other is “close,” i.e., similar in origin, age, or abilities; or physically close (Tesser & Paulhus, 1983). Thus, Tesser and Campbell (1980) found that subjects who were outperformed on a task later devalued that task more if the competitor was close on the relevant attributes than if the competitor was close on irrelevant attributes. In the present study both attitude and status dimensions were experimentally defined as being relevant to task performance. When similarity between groups occurred on only one of these dimensions, subjects could maintain psychological distance from the outgroup in terms of the other. When, however, both attitudes and status were similar, distance could only be maintained by differentiating from the outgroup on another dimension, i.e., evaluation of its performance. The interpretation of these data in terms of self-evaluation maintenance is also supported by the subjective Competitiveness x Attitudes interaction. If one regards competitiveness as being an indicator of a desire to distance oneself from others, then it makes sense that more competitive subjects liked similar outgroups less than different ones while less competitive subjects showed the opposite tendency. It therefore seems that the same social psychological processes which maintain personal aspects of identity may impinge on intergroup relations through their impact on identifications with a social group.

The implications of this experiment for the theoretical ideas which inspired it seem clear enough. In accordance with previous research (Turner (1978), Mummendey and Schreiber (1984), and Brown (1984a)), there was no evidence that intergroup similarity per se led to reduced liking; most of the research indicates the reverse. Where similarity does have divergent effects these seem to center on evaluative rather than affective dimensions. The interaction between Status and Attitudes on the task evaluations which was observed here fits closely with the data from Turner (1978) and Mummendey and Schreiber (1984), using similar

90 BROWN AND ABRAMS

measures and employing designs in which status and value (or attitudinal) similarity were confounded (see Brown, 1984b). In those experiments, as here, increased differentiation was observed against a similar status outgroup who happened to share similar values.

A second interesting aspect of our data concerns the distinction between objective goal relations and subjective orientation. By this we mean the difference between the actual interdependence linking groups and the group members’ subsequent competitive or cooperative response. This is highlighted by responses to the task manipulation checks and by the subsidiary finding of an interaction between Attitudes and Competitiveness on the liking measure. This interaction, in addition to replicating a similar result in Brown (1984a), leads us to conclude that the objective/subjective distinction is more useful than that between anticipated and actual in- tergroup goal relations (Rabbie & de Brey, 1971). For those subjects who felt motivated to compete on behalf of their group (and, by implication, enhance social identity), similarity of the outgroup failed to exert its usual attractive effects. For those who were less competitive (for whom identity may have been less important), similarity led to attraction. In sum, while the mere existence of two groups may lead to some intergroup differentiation (Turner, 1981), it seems to be subjective, and not objective, goal orientations which influence the relationship between similarity and attraction. Finally, the impact of status and attitudinal similarity does not seem to be contingent on actual task performance. This was shown by the complete absence of any alteration to the observed relationships when controlling for task performance (in covariance analyses). Moreover, our experiment also suggests that task performance itself can be influenced by knowledge about the attributes of the outgroup. Goldman, Stockbauer, and McAuliffe (1977) found that intergroup cooperation can enhance performance. Our data support that view but reveal that this is most likely when the outgroup is of equal or lower status, or holds different attitudes from the ingroup.

In conclusion, this experiment illustrates that the effects of similarity in intergroup contexts are more complex than is often assumed. If the positive distinctiveness of social identity is threatened by too great a similarity on important dimensions of comparison, the reaction may be to increase the level of ingroup bias on another dimension.

REFERENCES

Berry, J. W., Kahn, R., & Taylor, D. M. (1977). Mulriculturalism and ethnic attitudes in Canada. Ottawa: Supply and Services Canada.

Brewer, M. B. (1979). In-group bias in the minimal intergroup situation: A cognitive- motivational analysis. Psychological Bulletin, 86, 307-24.

Brewer, M. B.. & Campbell, D. T. (1976). Ethnocentrism and infergroup attitudes: East African evidence. New York: Sage.

THE EFFECTS OF INTERGROUP SIMILARITY 91

Brewer, M. B., & Kramer, R. M. (1985). The psychology of intergroup attitudes and behaviour. Annual Review of Psychology, 36, 219-243.

Brown, R. J. (1984a). The effects of intergroup similarity and cooperative vs. competitive orientation on intergroup discrimination. British Journal of Social Psychology, 23,21- 33.

Brown, R. J. (1984b). The role of similarity in intergroup relations. In H. Tajfel (Ed.), The

social dimension: European developments in social psychology. London: Cambridge Univ. Press.

Brown, R. J., & Ross, G. F. (1982). The battle for acceptance: An exploration into the dynamics of intergroup behaviour. In H. Tajfel (Ed.), Social identity and intergroup

relations. London: Cambridge Univ. Press. Brown, R. J., & Turner, J. C. (1981). Interpersonal and intergroup behaviour. In J. C.

Turner & H. Giles (Eds.), Intergroup behaviour. Oxford: Blackwell. Byrne, D. (1971). The attraction paradigm. New York: Academic Press. Caddick, B. (1982). Perceived illegitimacy and intergroup relations. In H. Tajfel (Ed.).

Social identity and intergroup relations. London: Cambridge Univ. Press. Deschamps, J.-C., & Brown, R. J. (1983). Superordinate goals and intergroup conflict.

British Journal of Social Psychology, 22, 189-195. Deutsch, M. (1975). Equity, equality and need: What determines which value will be used

as the basis for distributive justice? Journal of Social Issues, 31, 137-149. Duck, S. W. (Ed.). (1977). Theory andpractice in interpersonal attraction. New York/London:

Academic Press. Festinger, L. (1954). A theory of social comparison processes. Human Relations. 7, 117-

140. Goldman, M., Stockbauer, J. W., & McAuliffe, T. G. (1977). Intergroup and intragroup

competition and cooperation. Journal of Experimental Social Psychology, 13, 81-88. Johnson, D. W., Maruyama, G., Johnson, R., Nelson, D. & Skon, L. (1981). Effects of

cooperative, competitive and individualistic goal structures on achievement: a meta- analysis. Psychological Bulletin, 89, 47-62.

Mummendey, A., & Schreiber, H. D. (1984). Social comparison, similarity and ingroup favouritism. European Journal of Social Psychology, 14, 231-233.

Newcomb, T. M. (1961). The acquaintance process. New York: Holt, Rinehart & Winston. Rabbie, J. M., & de Brey, J. H. C. (1971). The anticipation of intergroup cooperation and

competition under private and public conditions. International Journal of Group Tensions, 1, 230-251.

Sherif, M. (1966). Group conflict and cooperation. London: Routledge & Kegan Paul. Snyder, C. R., & Fromkin, H. L. (1980). Uniqueness: The human pursuit of difference.

New York: Plenum. Tajfel, H. (1978). Differentiation between social groups: Studies in the social psychology

of intergroup relations. New York/London: Academic Press. Tajfel, H. (1982). Social psychology of intergroup relations. Annual Review OfPsychology,

33, I-30. Tesser. A., & Campbell, J. (1980). Self-definition: the impact of relative performance and

similarity of others. Social Psychology Quarterly, 43, 341-347.

Tesser, A., & Paulhus, D. (1983). The definition of self: Private and public self-evaluation management strategies. Journal of Personality and Social Psychotogy, 44, 672-682.

Turner, J. C. (1978). Social comparison, similarity and ingroup favouritism. In H. Tajfel (Ed.), Differentiation between social groups: Studies in the social psychology of intergroup relafions. New York/London: Academic Press.

Turner, J. C. (1981). The experimental social psychology of intergroup behaviour. In J. C. Turner & H. Giles (Eds.), Intergroup behaviour. Oxford: Blackwell.

92 BROWN AND ABRAMS

Turner, J. C., & Brown, R. J. (1978). Social status, cognitive alternatives, and intergroup relations. In H. Tajfel (Ed.), Differentiation between social groups: Studies in the

social psychology of intergroup relations. New York/London: Academic Press. van Knippenberg, A. F. M. (1984). Intergroup differences in group perceptions. In H.

Tajfel (Ed.), The social dimension. London: Cambridge Univ. Press. Workie, A. (1974). The relative productivity of cooperation and competition. Journal of

Social Psychology, 92, 225-230.