Journal of Personality and Social Psychology
1996, Vol. 7 I, No. 6, 1222 - 1233
Copyright 1996 by the American Psychological Association, Inc.
0022-3514/96/$3.00
Intergroup Norms and Intergroup Discrimination:
Distinctive Self-Categorization and Social Identity Effects
Jolanda Jetten, Russell Spears, and Antony S. R. Manstead
University of Amsterdam
Manipulation of in-group and out-group norms of discrimination and fairness allowed for the operation of competing social identity principles concerning in-group bias, conformity, and group distinctiveness. The combined effects of these principles on in-group bias were first examined in a
modified minimal-group setting (Study 1). Results demonstrated that participants' allocation strategies were in accord with the in-group norm. Furthermore, dissimilar norms resulted in greater use
of positive differentiation allocation strategies. However,in natural groups (Study 2 ), more in-group
bias was found when both group norms were similar and discriminatory. The results confirm the
importance of in-group norms and demonstrate differences between experimental and natural
groups in the applicabilityof competing social identity and self-categorization principles.
The power o f group norms to influence the behavior of group
members has been a recurring theme in social psychology since
its earliest days (e.g., Asch, 1956; Newcomb, 1943; Sherif,
1936). Classical studies of conformity focused on intragroup
processes such as normative influence, attraction, and interdependence (e.g., Deutsch & Gerard, 1955; Lott & Lott, 1965).
However, when we consider the role of norms relating to prejudice and discrimination, this clearly moves us from the intragroup to the intergroup realm. Pioneering research by Pettigrew
(1958) revealed the importance of social norms and values in
influencing prejudiced attitudes, throwing into question the
earlier psychodynamic and thus individualistic explanations of
intergroup discrimination. The social explanation of discrimination has continued more recently with social identity theory
(Tajfel, 1978; Tajfel & Turner, 1986), although, for reasons discussed below, the role o f group norms is curiously absent from
this framework. Self-categorization theory (e.g., Turner, Hogg,
Oakes, Reicher, & Wetherell, 1987) has made the normative basis of group behavior much more central, but it has tended to
neglect issues of intergroup discrimination covered by social
identity theory. In investigating the role played by group norms
in intergroup discrimination in the present research, we aimed
to both integrate and set in opposition some of the potentially
conflicting principles raised by social identity and self-categorization theories. As well as extending our understanding of the
phenomenon of intergroup discrimination, our objective was
to gain greater theoretical specification of when and how these
theoretical principles may combine.
The lack of attention to the role of group norms in social
identity theory can at least partly be explained by the fact that
the theory was originally formulated to explain discrimination
in ostensibly normless "minimal" groups and arose in contrast
to the "generic norm" explanation of in-group bias (Billig,
1973; Tajfel, 1970). The generic-norm explanation fell from
favor, among other reasons because of the difficulty of predicting which of a number of other social norms (e.g., fairness) will
be activated in a given situation (Brown, 1988; Tajfel, 1978).
However, the idea that specific group norms might help account
for discrimination more generally was also neglected. Although
self-categorization theory has been more concerned with norms
in an intergroup context, this framework has largely moved
away from the original focus on intergroup discrimination to
consider broader issues of social influence and stereotyping.
The present research represents an effort to reconcile these diverging theoretical agendas.
Despite the increasing laboratory and field research on discrimination, the social identity literature reflects the theoretical
focus described above, with few researchers measuring or manipulating group norms or using the concept in more than a
descriptive sense. Although some researchers have contrasted
social identity and equity principles by manipulating outgroup
norms (Diehl, 1989; Vivian & Berkowitz, 1992, 1993), in line
with self-categorization principles (Turner et al., 1987) in the
present research we anticipated that the in-group norm would
be a stronger and more direct determinant of reward allocation
strategies than would the out-group norm. This is because conformity to the in-group norm expresses one's salient social identity, assuming some degree of identification with the in-group.
Specifically, we predicted that an in-group norm of fairness
would lead to more fairness and an in-group norm of discrimination would lead to more discrimination toward the outgroup.
As implied above, however, this prediction had to be tempered by the robust finding that people tend to favor the in-
Jolanda Jetten, Russell Spears, and Antony S. R. Manstead, Department of Social Psychology, University of Amsterdam, Amsterdam, the
Netherlands.
This research was supported by a University of Amsterdam research
grant. We would like to thank Nyla Branscombe, Bertjan Doosje, Ernestine Gordijn, Craig McCarty, Tom Postmes, and Krystyna Rojahn,
for their comments on a draft of this article.
Correspondence concerning this article should be addressed to Jolanda Jetten, Department of Social Psychology, University of Amsterdam, Roetersstraat 15, 1018 WB Amsterdam, the Netherlands. Electronic mail may be sent via Internet to sp [email protected].
1222
SPECIAL ISSUE: GROUP NORMS AND INTERGROUP RELATIONS
group over the out-group in their reward allocations (e.g.,
Brewer, 1979; Mullen, Brown, & Smith, 1992; Turner, 1975,
1981 ). According to social identity theory, group members exhibit such in-group bias because they are motivated to seek or
maintain a positive identity and group distinctiveness, and one
important way to achieve this is by favoring the in-group over
the out-group (Taj fel, 1978; Tajfel & Turner, 1986). How would
these two processes, the tendency to show in-group bias and
conformity to in-group norms, relate to each other? An ingroup norm of discrimination would be in accordance with the
tendency to show in-group bias and would therefore complement it. However, an in-group norm of fairness would conflict
with the tendency to show in-group bias. One of the objectives
of the present research, therefore, was to examine the relative
strengths of the processes of conformity and in-group bias
when they were in conflict. Thus we manipulated orthogonally
both in-group and out-group norms of fairness versus
discrimination.
Manipulating in-group and out-group norms would vary not
only the sort of behavior they sanction or prescribe for the respective groups, but also the relation of similarity or dissimilarity of in-group and out-group norms. Further predictions
about the similarity of group norms on in-group bias can also
be derived from social identity and self-categorization theories.
Classical social identity theory predicts that intergroup similarity will enhance comparability and competition, leading to
greater in-group bias (Allen & Wilder, 1975; Brown, 1984a,
1984b; Brown & Abrams, 1986; Diehl, 1988; van Knippenberg
& EUemers, 1990; Mummendey & Schreiber, 1984; Roccas &
Schwartz, 1993; Tajfel, 1982; Turner, 1978 ). The argument underlying this phenomenon is that people are motivated to
(positively) differentiate the in-group from similar out-groups
on relevant dimensions of comparison in order to maintain or
enhance group distinctiveness and social identity. Research in
this tradition has typically manipulated the similarity of intergroup attitudes or status. For instance, Diehl ( 1988, Study 2)
investigated the effects of intergroup attitudinal similarity on
intergroup discrimination in a modified minimal-group paradigm. The results showed, in accordance with social identity
theory, that more discrimination was exhibited when the outgroup members had similar attitudes to in-group members. It
therefore seems possible that the similarity of group norms with
respect to discriminatory behavior (including fairness) may
also lead to an analogous process of positive differentiation for
groups whose norms are similar.~ This effect would be expected
to combine with the tendencies to display in-group bias and
conform to in-group norms described earlier. A consideration
is that when the in-group and out-group norms reflect fairness,
discrimination against the out-group directed at achieving positive differentiation would work against the content of the
group norm, creating a theoretical tension between conformity
and enhancing group distinctiveness. If the similaritydifferentiation hypothesis holds, it seems reasonable to expect
greater positive differentiation when the similar norms are discriminatory rather than fair.
Self-categorization theory also introduces another theoretical
principle that might be predicted to crosscut the threat to distinctiveness caused by the similarity of intergroup norms. This
concerns the analysis of group differentiation based on compar-
1223
ative fit and the metacontrast principle (Turner et al., 1987).
According to this view, the fit between the categorization and
group norms will be higher when these norms are different for
the in-group and out-group, allowing maximal differentiation
between them. Under these conditions, group norms may provide one way of defining and differentiating the group identity
more clearly. Norms do not have this property when they are
identical for the in-group and out-group and may therefore be
less central t o t h e identity of the group (Turner et al., 1987).
One interpretation of this analysis is that distinctive groups may
thereby encourage or elicit greater positive differentiation. However, once again, positive differentiation in terms of in-group
bias may be compromised if the in-group norm implies fairness.
This tension between positive differentiation and an in-group
norm of fairness suggests that in-group bias should be stronger
when the in-group norm is discriminatory and the out-group
norm fair than when the in-group norm is fair and the outgroup
norm discriminatory.
This analysis of social identity and self-categorization principles with regard to the similarity and dissimilarity of group
norms seems to lead to opposite predictions. How can these two
tendencies be reconciled? It is possible that a certain degree of
group distinctiveness is necessary for groups to define themselves (the self-categorization prediction), after which point
group similarity may result in attempts to restore distinctive
identity (the social identity prediction). In view of these conflicting possibilities, it seems prudent to treat the effect of norm
similarity as an empirical issue that may shed light on the alternative accounts generated by social identity and self-categorization theories.
To summarize, our goal in Study 1 was to examine the influence of in-group and out-group norms on levels of in-group bias
in a modified minimal-group setting. Processes relating to conformity to in-group norms and the tendency to differentiate the
in-group positively in terms of rewards in response to the similarity or dissimilarity of intergroup norms have been identified,
and social identity theory holds that social identity is enhanced
by in-group bias. Varying the content and relation of in-group
to out-group norms inevitably means that these different principles will come into conflict. Our aim was to investigate how
these combine or counteract each other.
Study 1
In-group bias and positive differentiation as measured by the
Tajfel reward matrices (Tajfel, Billig, Bundy, & Flament, 1971 )
The Tajfel matrices are designed to some extent to distinguish strategies of in-group favoritism (maximizing the in-group's profit) from
strategies of positive differentiation (maximum difference) (Tajfel et al.,
1971). We frame this prediction in terms of positive differentiation
rather than in-group bias more generally because positive differentiation
is closest to the motive of maintaining positive distinctiveness. It is also
appropriate at this point to clarify our terminology. We use in-group
bias to refer collectively to in-group favoritism and positive differentiation as defined above and operationalized it as allocation of rewards to
the in-group minus allocation of rewards to the out-group. In terms of
the strategies or "pulls" identified by the Tajfel matrices, we defined ingroup favoritism as strategies that maximized in-group profit on its own
or in combination with maximum differentiation strategies. Positive
differentiation was defined as maximizing the positive difference in re-
1224
JETTEN, SPEARS, AND MANSTEAD
were the p r i m a r y d e p e n d e n t m e a s u r e s in this study. Specifically,
we p r e d i c t e d t h a t the i n - g r o u p n o r m would have a disproport i o n a t e influence o n allocation strategies, a l t h o u g h a n i n - g r o u p
n o r m o f d i s c r i m i n a t i o n would b e m o r e influential t h a n o n e o f
fairness, because the latter conflicts with the e n h a n c e m e n t m o tive o f social identity theory. We were u n c e r t a i n w h e t h e r similar
or different n o r m s would underlie greater positive differentiation o f the in-group, given t h e alternative p r e d i c t i o n s o f the social identity a n d self-categorization theories. However, we predicted t h a t once again positive differentiation would b e higher
w h e n the i n - g r o u p n o r m was d i s c r i m i n a t o r y t h a n w h e n it was
fair, given the e n h a n c e m e n t motive o f social identity theory.
We also used m e a s u r e s o f i n t e r g r o u p evaluation, i n - g r o u p
identification, o u t - g r o u p identification, a n d prototypicality. 2
T h e evaluation m e a s u r e was i n t e n d e d to s u p p l e m e n t the m a t r i ces a n d assess w h e t h e r the p r e d i c t e d effects generalized to this
m e a s u r e . Moreover, because t h e n o r m m a n i p u l a t i o n referred
specifically to allocation strategies, there would be less conflict
between positive differentiation o n this m e a s u r e a n d a n o r m
o f i n - g r o u p fairness in particular. A n y m o d e r a t i o n o f positive
differentiation p r e d i c t e d for reward allocations would therefore
b e less likely o n this evaluative measure. We m e a s u r e d i n - g r o u p
identification, o u t - g r o u p identification, a n d prototypicality to
d e t e r m i n e w h e t h e r they reflected the degree o f i n - g r o u p bias
a n d w h e t h e r they were affected b y possible effects o f similarity
or dissimilarity entailed b y the g r o u p n o r m s . O n c e again the
conflicting p r i n c i p l e s a n d p r e d i c t i o n s o f the social identity a n d
self-categorization theories would apply.
Method
Design and participants. The design was a 2 (in-group norm: fairness vs. discrimination) × 2 (out-group norm: fairness vs.
discrimination) factorial with random allocation of participants to conditions. Participants were 75 students from the University of Amsterdam, ages 17-34. There were 50 females and 25 males, distributed
evenly over the conditions. They received course credit for their
participation.
Procedure. We conducted the experiment on personal computers.
Participants received brief instructions on how to use the computer and
then began the experiment. We introduced the experiment as an investigation into "modes of perceiving" and led participants to believe that
previous research had shown that there are two kinds of perceivers: detailed and global. We told them that the purpose of the present study
was to examine which kind of perceiver they themselves were.
A "dot estimation task" followed. In seven trials participants were
exposed to constellations of small and large dots presented for 3 s on a
computer (cf. Gerard & Hoyt, 1974) and were asked to estimate the
number of dots in each constellation (between 40 and I00). After they
had completed this task the computer started "calculating" their score,
and after a couple of seconds participants received false feedback about
their mode of perceiving--all of them were categorized as detailed perceivers. 3 We told participants that they were classified as detailed perceivers because they had been influenced more by the number of small
dots than by the number of large dots in making their estimations. To
increase their identification with the in-group, we then introduced a
group task. Participants had to estimate the number of black squares
wards between the in-group and out-group, in its clearest form in opposition to maximizing in-group profit or maximizing joint profit.
that appeared for 4 s on the computer screen. After giving their initial
estimation, they received false feedback about the estimates of 3 other
in-group members and had to give their final estimate. We led participants to believe that the result of their group work would be compared
with the result of a group of global perceivers.
After the group task, we informed participants that they had completed the first part of the study and the second part would follow. This
took the form of a distribution task. The purpose of this task was to
investigate whether detailed and global perceivers used different strategies in dividing resources (e.g., money) between two people (a detailed
perceiver and a global perceiver). Participants saw an example of an
allocation matrix and were instructed to use it to allocate money between two anonymous participants: a detailed perceiver and a global
perceiver. We stressed that they could not allocate money to themselves.
To reinforce the salience of group identity, we first asked participants to
indicate whether they had been categorized as a detailed or a global
perceiver in the first part of the experiment.
Independent variables. Following the procedure used by Diehl
(1989), we informed participants that they could use four allocation
strategies to distribute the money: (a) maximizing the joint profit of the
detailed and global perceivers, (b) allocating the same amount of money
to the detailed and global perceivers (i.e., fairness), (c) maximizing the
profit for the detailed perceiver, and (d) maximizing the profit for the
global perceiver. 4 We explained these strategies by showing examples of
each. Next, participants had to indicate which strategy they expected
detailed and global perceivers would use. The questions were "What
strategy do you think the majority of the detailed perceivers will use
when distributing money?" and "What strategy do you think the majority of the global perceivers will use when distributing money?". We
informed participants that they would receive feedback on their responses to these two questions so that they could examine whether their
assumptions about the in-group and out-group had been correct.
The false feedback consisted of information about the strategies used
by 15 detailed perceivers and 15 global perceivers who had already participated in this experiment. Participants in the fairness condition (ingroup or out-group) were told that l0 of 15 detailed or global perceivers
had allocated the money equally. In the discrimination condition (ingroup or out-group), participants received feedback indicating that 10
of 15 detailed or global perceivers had generally favored their own group
members. In all conditions, the number of in-group or out-group members who used a maximizing joint profit strategy was a clear minority,
with their number varying from 2 to 4 persons. Furthermore, the feedback indicated that no detailed or global perceiver had used an outgroup favoritism strategy. After this extended instruction, participants
had to complete l0 allocation matrices.
Dependent variables. The dependent variables consisted of manipulation checks, allocation matrices, evaluative trait ratings of the two
groups, social identification with the in-group and out-group, and prototypicality measures. All ratings except traits were made on a 9-point
scale, ranging from not at all ( 1 ) to very much (9). We measured evalu-
2 The out-group identification measure can be seen as complementary to the in-group identification measure. Given their limited involvement in minimal groups, participants may have been more prepared to
claim disidentification with the out-group than to profess identification
with the in-group.
3A pilot study (N = 16) showed that the attractiveness of being
categorized as a detailed perceiver ( M = 5.00) did not significantly
differ from that of being categorized as a global perceiver ( M = 5.25),
t(15) < 1.
4 Note that the explained strategies are not exactly identical to the
method of Diehl (1989). Diehl told participants in his study that they
could use four strategies: maximizing in-group profit, fairness, maximizing joint profit, and maximizing differences in favor of the in-group.
SPECIAL ISSUE: GROUP NORMS AND INTERGROUP RELATIONS
ations of the in-group and out-group on several traits. On a 100-point
scale, participants indicated detailed and global perceivers"tendency to
be nice, easily irritated, stiff, intelligent, creative, accurate, and knowledgeable about human character. We measured social identification
with four items (e.g., "I identify with other detailed perceivers"). We
asked the same items with respect to the out-group, that is, the global
perceivers (e.g., "I identify with global perceivers" ). The prototypicality
scale consisted of four items (e.g., "I like to think of myself as a typical
detailed perceiver").
Matrices. To assess behavioral strategies of distribution of money
between members of the two groups, we used three types of matrices:
(a) maximizing difference and maximizing in-group profit versus fairness ("MD & MIP versus F") (Billig & Tajfel, 1973), which distinguishes between strategies of in-group favoritism and fairness; (b) maximizing difference versus maximizing joint profit and maximizing ingroup profit ("MD versus MJP & MIP") (Tajfel et al., 1971; Turner,
Brown, & Taj fel, 1979), which distinguishes between strategies of maximizing the difference between the in-group and the out-group and strategies of joint profit and maximum in-group profit; and (c) maximizing
joint profit versus maximizing difference and maximizing in-group
profit ("MJP versus MD & MIP") (Tajfel et al., 1971 ), which distinguishes between strategies of joint profit and strategies of in-group favoritism. We presented the MD versus MJP & MIP and MJP versus
MD & MIP matrices twice: once in the version used by Tajfel et al.
( 1971 ) and once in the version used by Turner et al. ( i 979). This made
for a total of five matrices. Following standard practice, we presented
each of the five matrices twice: the labels "in-group" and "out-group"
were reversed in the second presentation of each matrix.
Results
Manipulation checks. Immediately after the manipulation
of the in-group and out-group norms, we asked participants two
manipulation check questions: "What strategy was predominantly used by the 15 detailed perceivers when allocating the
money?" and " W h a t strategy was predominantly used by the
15 global perceivers when allocating the money?". The manipulation proved to be successful; only 5 participants failed to respond to these questions in accordance with the manipulation.
We decided to keep these participants' data in the analyses, because exclusion of their data did not affect the results.
The expectations about the allocation behavior of the ingroup and out-group (measured before the manipulation of
norms) did not differ between conditions. Overall, the majority
of participants expected that in-group and out-group members
would" allocate more money to their own group members
(respectively, 68% and 60% of the participants); 13.3% of the
participants expected that in-group members would use a maximizing joint profit strategy, and 18.7% expected that they
would use a fairness strategy. Similarly, 18.7% of the participants expected that out-group members would use a maximizing joint profit strategy and 21.3% predicted they would use a
fairness strategy.
In-group bias, in-group favoritism, and positive differentiation. We computed the allocation strategies in two ways. The
first method consisted o f summing across all matrices all the
money allocated to the in-group and all the money allocated to
the out-group. The difference between the two sums provided
an indication of behavioral differentiation: The greater the
difference, the less fairness and the more in-group bias (Diehl,
1988). The second method was identical to the one used by
1225
Table 1
Study h Mean In-Group and Out-Group Profit
and In-Group Bias
In-group/out-group norm
Reward
allocations
n
In-group profit
M
SD
Out-group profit
M
SD
In-group bias
M
SD
Fair/
fair
Fair/
disc
Disc/
fair
Disc/
disc
19
20
18
18
17.17
2.56
17.55
2.89
18.20
2.74
19.31
2.32
15.06
2.49
13.23
2.75
12.53
2.66
13.24
3.46
2. l 1*
3.94
4.32**
5.23
5.67**
5.19
6.07**
4.77
Note. Disc = discrimination.
* Significantly greater than zero, p < .05. ** Significantly greater than
zero, p < .01.
Tajfel et al. ( 1971 ). In order to measure the intensity of individual distribution strategies, pull scores can be calculated
(Bourhis, Sachdev, & Gagnon, 1994). Both methods measure
in-group bias (Mlicki, 1993).
We performed the summation method first, finding in-group
bias in all four conditions. As shown in Table 1, the difference
between the total allocated in-group points and out-group
points was significantly different from zero in all conditions. To
investigate the influence of group norms on the level of in-group
bias, we submitted the allocated in-group points and out-group
points to a 2 (in-group norm: fairness vs. discrimination) X 2
(out-group norm: fairness vs. discrimination) X 2 (target
group: in-group vs. out-group) analysis of variance (ANOVA),
with repeated measures on the last factor. This analysis revealed
an interaction effect between in-group norm and target group,
F ( 1, 74) = 5.69, p < .05. The means are presented in Table I.
The difference scores in Table 1 show that an in-group norm
of fairness led to less in-group bias in money allocation than
did an in-group norm of discrimination; in other words, the InGroup Norm X Target Group interaction was equivalent to a
main effect of in-group norm on in-group bias, as given by the
difference score. We found no significant effects for out-group
norm, indicating that it was less important in determining the
strategy used to allocate money between an in-group member
and an out-group member.
We then calculated pull scores (Taj fel et al., 1971 ), performing a principal-components analysis on these scores to simplify
analysis and interpretation and reduce error variance (cf.
Mlicki, 1993). We chose an orthogonal solution to try to tease
apart the in-group favoritism and maximizing differentiation
strategies measured most purely by the MIP and M D pulls.
This analysis revealed three components that together explained 78.6% of the variance. We interpreted the first component, which explained 43.1% of the variance, as in-group
favoritism; it consisted of two M D & MIP versus MJP and two
M D versus MJP & MIP strategies, all loading highly ( .91, .89,
.79, and .65). This component thus also contained elements of
1226
JETTEN, SPEARS, AND MANSTEAD
Table 2
Study 1: Mean 1n-Group Favoritism and Maximizing Joint
Profit Versus Positive Differentiation Component Scores
In-group/out-group norm
Allocation strategies
n
In-group favoritism
M
SD
Maximizing vs. positive
differentiation
M
SD
Fair/
fair
Fair/
disc
Disc/
fair
Disc/
disc
19
20
18
18
-.50
.81
-.06
1.0
.20
1.05
.39
.96
.20
.91
-.12
.87
-.42
.82
.35
1.26
Note. The more negative the score on the maximizing versus positive
differentiation component, the greater the positive differentiation. Disc
= discrimination.
positive differentiation, as indicated by the latter two strategies.
The M D & MIP versus F pull also loaded positively (.88),
whereas, as expected, the F versus M D & MIP pull loaded negatively ( - . 8 4 ) . The second component, which explained 24%
of the variance, was labeled maximizing versus positive differentiation and consisted of two MIP & MJP versus M D strategies, each loading positively (.87 and .80). Because this component most clearly separated M D and MIP (unlike the first
component), it represented the purest form of positive differentiation derived from the factor solution (we refer to this component as positive differentiation for short). Negative scores on
this component represented greater positive differentiation. The
third component consisted of two MJP versus M D & MIP pulls,
each loading positively. Because the third component explained
only 11.5% of the variance and preliminary analyses revealed
no effects for this component, we excluded it from further
analyses.
Analyses of variance on the component scores revealed a
main effect for in-group norm on in-group favoritism, F ( 1, 74)
= 6.60, p < .05. The mean component scores are presented in
Table 2. An in-group norm of fairness led, in comparison with
an in-group norm of discrimination, to less use of differentiating strategies. This effect was similar to the main effect found
with the summation measure.
Furthermore, we found a significant interaction on the positive differentiation component, F ( 1, 74) = 5.87, p < .05. As
shown in Table 2, the positive differentiation strategy (negative
loadings) was stronger in the two conditions in which the ingroup and out-group norms differed from each other, and the
maximizing strategy was stronger in the two conditions in which
the in-group and out-group norms were identical. More specifically, simple main effects showed that positive differentiation
strategies were used most when the in-group norm represented
discrimination and the out-group norm fairness; they were used
significantly more in this condition than in the discrimination/
discrimination condition, F ( 1, 71 ) = 5.64, p < .05, and marginally significantly more in this condition than in the fairness/
fairness condition, F ( 1, 71 ) = 3.65, p = .06. Simple main effects
also showed that the combination o f an in-group norm of fair-
ness and an out-group norm of discrimination had a smaller
impact on the use of positive differentiation strategies than did
the discrimination/discrimination condition, F ( 1, 71 ) = 2.13,
p < . 15, and was not significantly different from the fairness/
fairness condition, F ( 1, 71 ) = 1.03, ns. In sum, positive differentiation strategies were stronger than maximizing strategies
when in-group and out-group norms differed, but this was only
significant when the in-group norm represented discrimination
and the out-group norm fairness.
Participants" evaluations of the in-group and out-group. The
scale we used to measure participants' evaluations of their own
group members and out-group members as nice, intelligent,
creative, and knowledgeable about human character had acceptable reliability for judgments of the in-group, a = .67, and
for judgments of the out-group, a = .77. We computed the average of the traits separately for the in-group and out-group. To
investigate the influence of group norms, we submitted the trait
judgments of the in-group and out-group to a 2 (in-group norm:
fairness vs. discrimination) X 2 (out-group norm: fairness vs.
discrimination ) × 2 (target group: in-group vs. out-group ) ANOVA, with repeated measures on target group. This analysis revealed a main effect of target group, F ( 1, 65) = 8.98, p < .01.
Overall, the in-group was evaluated more positively than the
out-group. This main effect was qualified by a marginally significant interaction between in-group norm and target group,
F ( l, 65) = 3.29, p = .07. The average difference scores (on a
100-point scale in-group minus out-group evaluation) showed
that the in-group was evaluated more positively when its norm
represented fairness ( M = 5.52) and less positively when its
norm implied discrimination ( M = 1.36).
In-group and out-group identification and prototypicality.
The reliabilities of the in-group identification, out-group identification, and prototypicaiity scales were satisfactory (as = .82,
.86, and .87, respectively). We averaged the scores on each of
these three scales. Because these three scales were conceptually
related but not identical, we performed a 2 (in-group norm) X
2 (out-group norm) multivariate ANOVA with in-group identification, out-group identification, and prototypicality as dependent measures. This analysis revealed a marginally significant
multivariate interaction between in-group norm and out-group
norm, F(3, 69) = 2.62, p < .06, and no other effects. Univariately, this interaction was significant only for the prototypicality
scale, F ( l, 71 ) = 5.64, p < .05. The mean in-group identification, out-group identification, and prototypicality scores are
presented in Table 3.
Simple main effects revealed that the level of prototypicality
when both the in-group and out-group norms were discrimination was significantly lower than it was in the discrimination/
fairness condition, F ( l, 71 ) = 4.81, p < .05, and the fairness/
discrimination condition, F ( l , 71) = 4.32, p < .05. The
fairness/fairness condition did not significantly differ from the
fairness/discrimination condition, F ( 1, 71 ) = 1.33, ns, or the
discrimination/fairness condition, F ( l, 71 ) = 1.55, ns, in level
of prototypicality.
Discussion
In line with predictions, in-group norm proved to be a more
important factor for the allocation of rewards than did out-
SPECIAL ISSUE: GROUP NORMS AND INTERGROUP RELATIONS
Table 3
Study 1: Mean In-Group and Out-Group
Identification and Prototypicality
In-group/out-group norm
Group identification
and prototypicallity
n
In-group identification
M
SD
Out-group identification
M
SD
Prototypicality
M
SD
Note.
Fair/
fair
Fair/
disc
Disc/
fair
Disc/
disc
19
20
18
18
4.13
1.63
4.68
.91
4.62
1.37
3.86
1.60
3.40
1.45
3.49
1.23
3.08
1.38
3.61
1.46
4.33
1.53
4.90
1.55
4.84
.79
3.89
1.57
Disc = discrimination.
group norm. Participants' allocation of money to the in-group
and out-group was influenced by the in-group norm, although
there was still some residual in-group bias in absolute terms (as
reflected by the difference score) when the in-group norm reflected fairness. This bias presumably reflected the general tendency to favor the in-group in reward allocations, as predicted
by social identity theory (e.g., Brewer, 1979; Tajfel & Turner,
1986), and this was also reflected by the overall tendency to
evaluate the in-group more positively than the out-group on the
evaluative scales.
Perhaps more interesting, however, is the fact that this ingroup bias effect could be dramatically reduced when the norm
pointed against this strategy. The out-group norm had no significant effect, suggesting that equity principles (e.g., treat the
out-group as they treat you) may have been overshadowed by
conformity and in-group bias effects in this study. Moreover,
conformity to the norm was found despite the dominant expectation that group members would favor their own group in reward allocations. This baseline expectation finding is interesting in its own right and seems to imply that the minimal-groups
situation evokes a competitive norm, akin to the generic norm
of discrimination rejected by social identity theory. However,
this expectation may also itself reflect the tendency to enhance
the in-group predicted by social identity theory and would once
again help to explain the significant level of in-group bias across
conditions. The fact that positive evaluative differentiation was
just as strong, if not stronger, when the norms reflected fairness
shows that conformity to the fairness norm only moderated ingroup bias on the reward allocation dimension. Thus norms
against discrimination do not necessarily undermine group enhancement as such, but they may well influence the way in
which it is expressed or channeled.
A further important aspect of these results concerns the consequences of the relationship between in-group and out-group
norms. We found no evidence for the prediction derived from
social identity theory that similarity of group norms should lead
to greater positive differentiation in the similar-norm conditions, particularly the discrimination/discrimination condition. On the contrary, we found clear support for the hypothesis,
1227
based on self-categorization principles, that high comparative
fit reflected by dissimilar norms produces a sense of group distinctiveness that results in positive differentiation. This was also
confirmed by a measure of perceived group distinctiveness.5 Interaction effects on the use of maximizing versus positive
differentiation strategies showed that, compared with when the
in-group and out-group had similar norms, different norms resulted in greater positive differentiation, although this was especially true for the condition in which the in-group norm represented discrimination and the out-group norm fairness. This is
also logical, because, as noted earlier, in this condition a strategy
of positive differentiation did not conflict with the in-group
norm. We found this interaction effect only in the analyses of
the pull scores, which allow for the separation of positive differentiation from more general in-group bias or favoritism. Although the validity of pull scores has been debated (Bornstein
et al., 1983a, 1983b; Turner 1983a, 1983b), in the present study
pull scores proved to be a useful way of disentangling specific
strategies of positive differentiation from more general maximizing strategies and in-group bias.
The pattern of greater positive differentiation in the conditions with dissimilar norms was also reinforced by the findings
on the group prototypicality measure: Participants described
themselves as more typical of their group when the norms were
different than when they both reflected discrimination
(although not when they both reflected fairness). This finding
fits with the general interpretation of the positive differentiation
effect in terms of self-categorization principles: People not only
positively differentiate groups more when they see them as normatively distinct, but also see themselves more in group terms
("self-stereotyping"). Although the pattern of means was similar for the closely related concept of group identification, this
effect did not attain significance.
Clearly, the pattern of results found for positive differentiation and prototypicality is inconsistent with research that has
shown intergroup similarity to enhance intergroup differentia-
s We also measured group distinctivenesswith three items (e.g., "To
what extent do you feel that detailed perceiversare different from global
perceivers?"). The measure of perceived distinctiveness provided a
more direct test of the degree to which the groups were seen as more
distinct because of either (a) the threat to distinctivenessposited by social identity theory or (b) the metacontrast principle of self-categorization theory. We expected that this measure, like evaluation judgments,
would be less affected by the fairness norm undermining (positive)
differentiation, as in the case of the reward allocation matrices. An
ANOVAon the group distinctivenessscale (a = .82) revealedan interaction between in-group norm and out-group norm, F(I, 74) = 7.88, p <
.01. Simple main effects revealed that participants were more likely to
see themselves as a distinguishable group when the in-group norm was
fairness and out-group norm differentiation (M = 4.95), than when
both group norms implied fairness (M = 4.00), F(I, 71) = 3.94, p <
.05, or when both group norms implied differentiation (M = 3.48 ), F(1,
71 ) = 8.87, p < .01. The in-group was seen as more distinctive when its
norm was differentiation and the out-group norm was fairness (M =
4.50), than when both group norms implied differentiation, F( 1, 71 ) =
4.06, p < .05, but not when both group norms implied fairness, F( 1,
71 ) = 1.01, ns. In sum, higher levels of group distinctiveness were expressed when both group norms v~re different, especiallywhen the ingroup norm was fairness and the out-group norm differentiation.
1228
JETTEN, SPEARS, AND MANSTEAD
tion and even discrimination. It is possible that distinctiveness
implied by different norms may have raised the sense of group
definition to a level necessary to elicit differentiation by starting
from a low baseline (i.e., minimal groups). One explanation for
the finding that groups positively differentiate more and show
greater prototypicality when group norms are different may
therefore lie in the nature of the groups. Group identity as a
detailed perceiver is clearly not a well-established identity. In
longer lasting natural groups, in which group identity and group
boundaries are better established and identification as a group
member is higher, the resulting pattern could be quite different. In this case, group members' allocation behavior is perhaps
more likely to be influenced by intergroup similarity than
by distinctiveness, in line with the original social identity
predictions.
There are some findings in the literature that are in keeping
with this distinction between experimental and natural groups.
Important differences (e.g., asymmetries in the effects of group
status) between the patterns of in-group bias in artificial groups
and those in natural groups were found in a meta-analysis by
Mullen et al. (1992). It is quite possible that such differences
relate to variations between minimal and natural groups in level
of identification and the resulting willingness to defend one's
group ( Spears, Doosje, & Ellemers, in press). Research by Worchel, Coutant-Sassic, and Grossman (1992) on the effects of the
developmental stage of a group is also generally consistent with
the present analysis. According to these authors, when a group
has just developed, its members are mainly concerned with
drawing clear boundaries between their group and out-groups,
and at this stage group norms may serve an important distinguishing function. The distinctiveness check (see Footnote 5 )
and prototypicality results o f Study 1 are consistent with this
notion. The idea is that at later stages intergroup similarity becomes more of a threat to the separate identity of the group,
with the result that members attempt to differentiate their
group from the out-group, in line with social identity theory. In
Study 2, we investigated this idea by replicating Study 1 using a
well-established natural group categorization.
Study 2
Our goal in Study 2 was to test our earlier hypotheses concerning conformity and in-group bias resulting from similar
versus dissimilar group norms in the context of natural social
groups. Thus the only difference between Study 1 and Study
2 was the nature o f the groups. The participants, or in-group
members, were all students at the University of Amsterdam, and
the out-group was defined as students at the rival university in
Amsterdam, the Free University. (These two universities are the
only universities in Amsterdam, and they are approximately
equal in status.) We assumed that the participants' identity as
University o f Amsterdam students would be stronger and better
developed than the group identities based on the quasi-minimal
groups of Study 1. We hypothesized that if group identity was
well established, feedback that in-group and out-group norms
were similar would be experienced as more of a threat to group
identity and would instigate a search for distinctiveness, resulting in greater in-group bias for similar than for dissimilar group
norm conditions (especially when the in-group norm was
discrimination). In addition, we predicted that in-group norms
would have the most influence, especially when they corresponded to group enhancement (i.e., the discrimination norm),
as we had predicted in the first study. The measures used in this
study were similar to those used in Study l, the main difference
being that we were open to the possibility that the similarity of
group norms might enhance positive differentiation, in line
with social identity principles.
Method
Design and participants. The design was identical to that of Study
1. Participants were 71 students ( 35 female and 36 male) from the University of Amsterdam. Their ages varied from 18 to 31. Half of the sample participated for course credits and the other half received 10 Dutch
guilders (approximately $6 ) for their participation.
Procedure. We followed a procedure similar to that of Study 1. In
this second study, we told participants that the research involved a comparison between students oftbe University of Amsterdam and students
of Free University. All participants were led to believe that this study
was also being conducted at the Free University. We manipulated the ingroup and out-group norms in the same way as in Study 1 and used the
same dependent measures.
Results
Manipulation checks. As in Study 1, we asked two manipulation check questions immediately after manipulation of the
in-group and out-group norms: "What strategy was predominantly used by the 15 University of Amsterdam students while
allocating the money?" and "What strategy was predominantly
used by the 15 Free University students while allocating the
money?". Four participants responded in a way that did not accord with the manipulation. As in the earlier study, we decided
to keep these participants' data in the analyses, because the resuits were minimally influenced by their exclusion. Two participants agreed not at all with the statement "I know more students at the University of Amsterdam than at Free University."
The identification measures revealed that these 2 participants
identified more as students of Free University than as students
of the University of Amsterdam. Given the centrality of the
identification construct in our reasoning, we decided to exclude
their data from the analyses.
We assumed that participants would identify more with better established, more meaningful groups. Analyses of the ingroup identification data from the first and second studies revealed that participants did indeed identify more with meaningful groups ( M = 5.02) than with minimal groups ( M =
4.33), t(142) = 2.72,p < .01.
Expectations about the allocation behavior of the in-group
and out-group did not differ between conditions. Overall, participants expected from in-group and out-group members approximately just as much in-group favoritism (50.7% and
47.8% of participants, respectively) as fairness (33.3% and
37.7% of participants, respectively) and maximizing of joint
profit ( 19.9% and 14.5% of participants, respectively).
In-group bias, in-group favoritism, and positive differentiation. As in Study l, we computed the allocation strategies in
two ways. First, we separately summed across all the matrices
all money allocated to the in-group and all money allocated to
1229
SPECIAL ISSUE: GROUP NORMS AND INTERGROUP RELATIONS
Table 4
Study 2: Mean In-Group and Out-Group Profit
and In-Group Bias
In-group/out-group norm
Reward
allocations
n
In-group profit
M
SD
Out-group profit
M
SD
In-group bias
M
SD
Fair/
fair
Fair/
disc
Disc/
fair
18
17
17
17
18.14
2.29
17.13
2.38
16.96
2.32
18.14
2.78
15.59
1.98
14.99
2.46
15.35
2.27
12.81
2.52
2.55**
3.48
2.14"
3.76
1.61"
3.14
Disc/
disc
5.33**
4.90
Note. Disc = discrimination.
* Significantlygreater than zero, p < .05. ** Significantlygreater than
zero, p < .01.
the out-group. Again, we found in-group bias in all conditions.
The difference between the total allocated in-group points and
out-group points differed significantly from zero in all conditions (Table 4). We submitted the allocated in-group and outgroup sums of money to a 2 (in-group norm: fairness
vs. discrimination) × 2 (out-group norm: fairness vs.
discrimination) × 2 (target group: in-group vs. out-group)
ANOVA with repeated measures on the last factor. The interaction effect found in Study 1 between in-group norm and target
group was not replicated, F( 1, 65) = 1.46, ns. However, a marginally significant interaction effect was found between outgroup norm and target group, F ( 1 , 6 5 ) = 3.15, p < . 10. As may
be seen in Table 4, participants showed slightly more bias when
the out-group norm was discrimination.
This interaction was qualified by a three-way interaction
among in-group norm, out-group norm, and target group, F( 1,
65) = 4.90, p < .05. As predicted for these groups, whose membership was well established, in-group bias tended to be higher
when the in-group and out-group norms were similar. This was
particularly true when both group norms implied discrimination. Simple main effects revealed that participants showed
more in-group bias (in-group minus out-group profit) when
both group norms implied discrimination than when the ingroup norm was discrimination and the out-group norm was
fairness, F( 1, 65) = 7.84,p < .01, and when the in-group norm
was fairness and the out-group norm was discrimination, F( 1,
65) = 5.77, p < .05. However, in-group bias when both group
norms implied fairness was not significantly higher than it
was when the in-group norm was discrimination and the outgroup norm was fairness, F( 1, 65) < 1, or when the in-group
norm was fairness and the out-group norm discrimination,
F(1, 65) < 1.
As in Study 1, we also calculated pull scores. Three components that explained 85.7% of the variance were extracted in a
principal-componentsanalysis. These components were almost
identical to those extracted in Study 1, and we interpreted and
labeled them in the same ways. We interpreted the first compo-
nent (explaining 45.2% of the variance) as in-group favoritism;
it consisted of two MD & MIP versus MJP and one MD versus
MJP & MIP pull, all loading positively (.94, .93, and .66,
respectively); an MD & MIP versus F pull that loaded positively
(.96); and the pull F versus MD & MIP negatively ( - . 9 6 ) . The
second component (explaining 28.5% of the variance) was labeled maximizing versus positive differentiation. Two MJP &
MIP versus MD pulls loaded positively (.82 and .81 ), and an
MD versus MJP & MIP pull loaded negatively ( - . 6 6 ) . The
third component consisted of two MJP versus MD & MIP strategies. Because this component explained only 12% of the variance and preliminary analyses of this component revealed no
effects, we excluded it from further analyses.
An ANOVA revealed a significant interaction between ingroup and out-group norms on the in-group favoritism component, F( 1, 68) = 5.72, p < .05. This effect was identical to the
interaction effect we found with the summation measure. Participants displayed more in-group favoritism when their own
group norm was identical to the outgroup norm. More in-group
favoritism was shown when both group norms implied discrimination than when the in-group norm was discrimination and
the out-group norm was fairness, F( 1, 65) = 7.34, p < .01, and
when the in-group norm was fairness and the out-group norm
was discrimination, F( 1, 65) = 4.87, p < .05. In-group favoritism when both group norms implied fairness was not significantly higher than it was when the in-group norm was discrimination and the out-group norm was fairness, F( 1, 65) = 1.30,
ns, or when the in-group norm was fairness and the out-group
norm was discrimination, F( 1, 65) < 1. This pattern was identical to the overall in-group bias measure.
An ANOVA also revealed a main effect for in-group norm on
the positive differentiation component, F( 1, 68) = 5.30, p <
.05. The mean component scores are presented in Table 5.
Compared with an in-group norm of discrimination, an ingroup norm of fairness led to greater use of the maximizing
joint profit strategy than the positive differentiation strategy.
Participants' evaluations of the in-group and out-group. After receding of the negative traits, the reliability of the scale participants used to evaluate their own group members and out-
Table 5
Study 2: Mean In-Group Favoritism and Maximizing Joint
Profit Versus Positive Differentiation Component Scores
In-group/out-group norm
Allocation strategies
n
In-group favoritism
M
SD
Maximizing vs. positive
differentiation
M
SD
Fair/
fair
Fair/
disc
Disc/
fair
Disc/
disc
18
17
17
17
.02
.92
-.20
.93
-.36
.79
.53
!.18
.51
.87
.01
.93
-.13
.79
-.42
1.22
Note. The more negativethe score on the maximizing versus positive
differentiation component, the greater the positive differentiation. Disc
= discrimination.
1230
JETTEN, SPEARS, AND MANSTEAD
group members as nice, easily irritated, stiff, intelligent, creative, accurate, and knowledgeable about human character was
acceptable (for judgments of the in-group, a = .75, and for judgments of the out-group, a = .72). We computed the average of
the traits separately for the in-group and out-group. We submitted the trait ratings of the in-group and out-group to a 2 (ingroup norm: fairness vs. discrimination) × 2 (out-group norm:
fairness vs. discrimination) × 2 (target group: in-group vs. outgroup) ANOVA with repeated measures on target group.
This analysis revealed a main effect of target group, F( 1, 65)
= 19.54, p < .001. In all conditions, the in-group was evaluated
more positively than the out-group (Table 6). There was also a
significant interaction among in-group norm, out-group norm,
and target group, F( 1, 65) = 4.68, p < .05. The means show
that the difference between in-group and out-group evaluations
was higher when the in-group's norm was identical to the outgroup's. The difference between in-group and out-group evaluations was higher when the two groups shared a norm of fairness
than when the in-group norm represented discrimination and
the out-group norm implied fairness, F ( 1, 65) = 4.24, p < .05,
but not when fairness was the in-group norm and discrimination the out-group norm, F( 1, 65) = 1.61, ns. The difference
between in-group and out-group evaluations was marginally
higher when the two groups shared a norm of discrimination
than when discrimination was the in-group norm and fairness
the out-group norm, F( 1,65 ) = 3.24, p < . 10, but not when the
in-group norm was fairness and the out-group norm discrimination, F( 1, 65) = 1.00, ns.
In-group and out-group identification and prototypicality.
The reliabilities of the in-group identification, out-group identification, and prototypicality scales were high (as = .95, .87,
and .90, respectively), allowing us to average the items into
three scales. As in Study 1, we performed a multivariate
ANOVA with in-group identification, out-group identification,
and prototypicality as dependent measures. This analysis revealed a multivariate interaction effect between in-group norm
and out-group norm, F(3, 63) = 4.31, p < .01. In univariate
Table 6
Study 2: Mean In-Group and Out-Group Evaluations
and the Mean Difference Between Them
In-group/out-group norm
Group evaluations
n
In-group evaluation
M
SD
Out-group evaluation
M
SD
Difference between in-group
and out-group
evaluations
M
SD
Fair/
fair
Fair/
disc
Disc/
fair
Disc/
disc
18
17
17
17
61.15
8.23
54.26
4.44
56.19
8.23
57.63
7.92
55.94
7.73
51.71
5.19
55.35
7.72
52.93
4.94
5.21
2.02
2.56
1.94
.84
2.08
4.70
2.44
Note. The higher the mean, the more positive the evaluation. Disc =
discrimination.
Table 7
Study 2: Mean In-Group and Out-Group
Identification and Prototypicality
In-group/out-group norm
Group identification
and prototypicallity
n
In-group identification
M
SD
Out-group identification
M
SD
Prototypicality
M
SD
Fair/
fair
Fair/
disc
Disc/
fair
Disc/
disc
18
17
17
17
5.51
1.57
4.90
1.12
4.57
1.68
4.99
2.16
1.83
1.24
2.41
1.11
2.25
1.89
1.02
.65
5.46
t.50
4.41
1.36
4.18
1.51
4.96
2.07
Note. Disc= discrimination.
terms, this interaction was significant for out-group identification, F( 1, 65) = 8.40, p < .01, and prototypicality, F( 1, 65)
= 5.40, p < .05. The mean in-group identification, out-group
identification, and prototypicality scores are presented in Table 7.
Identification with students of Free University, the out-group,
was lowest when the in-group and out-group norms were similar. Outgroup identification was significantly lower when both
group norms were discrimination than when the in-group
norm was fairness and the out-group norm was discrimination,
F( 1, 65) = 9.73, p < .01, or when the in-group norm was discrimination and the out-group norm fairness, F ( 1 , 6 5 ) = 7.67,
p < .01. The fairness/fairness condition did not significantly
differ in level of out-group identification from the fairness/
discrimination condition, F(1, 71) = 1.66, ns, or the
discrimination/fairnesscondition ( F < 1).
Furthermore, simple main effects revealed that group members' level ofprototypicality was higher when both group norms
implied fairness than when the in-group norm was fairness and
the out-group norm was discrimination, F ( 1, 65) = 5.44, p <
.05, or when the in-group norm was discrimination and the outgroup norm was fairness, F(1, 65) = 3.67, p < .06.
The discrimination/discrimination condition did not significantly differ in the level of prototypicality from the fairness/
discrimination condition, F(1, 71) = 1.94, ns, or the discrimination/ fairness condition ( F < 1).
Discussion
The results of Study 2 showed that for better established
groups the use of the maximizing joint profit versus positive
differentiation strategy was in accordance with the in-group
norm and less influenced by the out-group norm. However, the
in-group norm did not influence measures of in-group bias and
favoritism in the same way as in Study 1; if anything, the outgroup norm had more of an effect on in-group bias. Anticipation of fairness from the out-group led to slightly more fairness
and anticipation of discrimination, to slightly more in-group
bias, although this effect was severely qualified by the higher
SPECIAL ISSUE: GROUP NORMS AND INTERGROUP RELATIONS
order interactions between in-group and out-group norms on
both in-group bias and favoritism measures. These interactions
reflect the predicted tendency for participants to differentiate
more when the group norms were similar than when they
differed, especially when they reflected discrimination. Furthermore, similarity in group norms led to more positive evaluations of the in-group, lower levels of out-group identification,
and higher levels of prototypicality compared with the dissimilar-norms conditions. 6 The interaction effect for out-group
identification was similar to the interaction found for the
amount of in-group bias and favoritism when the groups shared
a discrimination norm. Identification with the out-group was
lowest when norms were similar and discriminatory but not significantly so when they were similar and fair. However, prototypicality as a member of the in-group (University of Amsterdam students) was highest when norms were similar and fair
and higher than in the dissimilar-norm conditions. Positive intergroup differentiation on the evaluative measure was also
highest in the condition of similar norms of fairness and higher
than in the dissimilar-norm conditions. This suggests that
threats to group distinctiveness implied by similar fairness
norms also underlay attempts to assert a self-definition in group
terms and to differentiate the group positively. In line with our
earlier prediction, this would have been possible because these
indicators were less constrained by the norm of discrimination
than were in-group bias and favoritism in terms of reward
allocations.
General Discussion
The results from these two studies emphasize the importance
of both conformity to in-group norms and the relationship between in-group and out-group norms as influences on group
members' willingness t o express in-group bias, as predicted by
social identity theory. In-group norms influenced in-group bias
and favoritism for the experimental in-groups and positive
differentiation for the natural in-groups, The scant evidence for
an independent effect of out-group norms suggests that participants were not taking their cue from the out-group, which tends
to argue against an explanation of in-group bias in terms of
equity principles. Both studies also revealed that the relationship between in-group and out-group norms can affect strategies o f discrimination and differentiation, albeit in very different ways. We now consider these different findings and their implications in further detail.
In line with predictions, we found ample evidence, in Study
1 in particular, that group members will conform to in-group
norms prescribing levels of discrimination or fairness. This
concurs with other research on social influence and self-stereotyping that demonstrates the power of the group to define the
self and guide behavior (e.g., Turner, 1991; Turner et al., 1987 ).
However, to our knowledge this is one of few experimental studies investigating the role of in-group norms in determining levels of intergroup discrimination. The clearer tendency for our
experimental groups to conform to in-group norms than our
natural groups (i.e., in terms of the net difference between ingroup and out-group rewards) provides an interesting point of
comparison between the two studies. This is even more intriguing given the lower levels of group identification in the more
1231
minimal group setting, because enhanced group identification
might have been predicted to result in greater conformity to ingroup norms. On the other hand, when group identification is
relatively low or the group is not yet well formed, group members may be precisely concerned with establishing their identity.
Worchel et al. (1992) have suggested that in its formative stages
a group places strong emphasis on conformity, and members
are motivated to emphasize their similarity. More secure or better established groups, on the other hand, may allow greater
deviation.
However, another relevant factor is probably the "minimal"
nature of the situation and participants' attempts to make sense
o f the instruction to allocate rewards without any obvious a priori basis for doing so. If membership in an established group
makes the task more meaningful, participants' behavior in the
minimal situation may be more susceptible to the influence of
group norms. Sense making and "uncertainty reduction" may
play a bigger role in minimal intergroup settings than in natural
group settings, leading to a greater influence of prescriptive
norms ( Hogg & Abrams, 1993). In this respect, it is worth noting that the in-group norm for the minimal groups represents
the only, and thus the defining, information about that group.
Normative feedback for an established group, on the other
hand, has to compete with prior knowledge of norms associated
with the group, much of which may undermine these imposed
norms.
Evidence of conformity to in-group norms should not be allowed to obscure the basic evidence for group members' underlying tendency to display in-group bias in reward allocations.
The absolute levels of in-group bias in all conditions of both
studies demonstrate the power of social categorization to elicit
in-group bias, even when it may run counter to group norms
(Tajfel, 1970; Tajfel et al., 1971; for reviews see Brewer, 1979;
Messick & Mackie, 1989; Turner, 1981 ). However, a norm of
fairness was clearly capable of greatly reducing in-group bias
under certain conditions, notably in an experimental in-group
when both it and the out-group were fair and in a natural ingroup when it was discriminatory but the out-group fair.
The present studies also show that the intergroup context,
namely, the combination of in-group and out-group norms and
not just the in-group (or out-group) norm in isolation, is an
important determinant of positive differentiation strategies in
reward allocation and also evaluate differentiation and prototypicality (self-stereotyping). The patterns of in-group bias and
positive differentiation support the notion that similarities and
differences in in-group and out-group norms were perceived
differently and underlay opposing processes in the two studies.
In experimental groups (Study 1 ), group distinctiveness implied by dissimilar group norms predicted positive differentiation. In contrast, when group identity was relatively well formed
(Study 2), similarity of group norms led to enhanced in-group
favoritism and bias in general and most clearly in the most
"competitive" condition, when both group norms were defined
as discriminatory. However, for measures less constrained by
the norm of discrimination, when both norms were fair, greater
6 We also measured group distinctiveness in Study 2. In contrast to
Study 1, we found no significant effects on the group distinctiveness
scale (a = .86) in natural groups.
1232
JETTEN, SPEARS, AND MANSTEAD
evaluative differentiation and prototypicality compared with
the dissimilar-norm conditions was also demonstrated. In Study
2 we were thus able to show that similar norms produced more
"group" type effects than did dissimilar norms regardless of
whether these norms were discriminatory or fair. That this occurred on different measures is explained by the extent to which
the dependent variable was constrained by conformity to the ingroup norm of discrimination.
Group distinctiveness therefore seemed to have a different explanatory status in the two different intergroup contexts we investigated. In the experimental groups, positive differentiation
was a result of group distinctiveness, whereas in the natural
groups, effects on a number of indicators (in-group bias and
favoritism, intergroup evaluation, out-group identification, and
prototypicality) indicated an attempt to assert or defend group
distinctiveness, particularly in the competitive and threatening
context of mutual discrimination. The main variable we used
to interpret differences between our two studies was the degree
of group identification, which was likely to be higher for the
natural groups examined in Study 2 than for the experimental
groups examined in Study 1. Indeed, our measure of identification across the two studies showed that the absolute level of
identification was higher for the natural groups. However, it is
important to be cautious in interpreting this to be the critical
variable accounting for these differences, because experimental
and natural groups are likely to differ on a number of other
dimensions. Obviously, further research must examine the basis
for such differences by experimentally manipulating proposed
moderating variables in order to be more certain that they can
explain the differences in patterns, notably those relating to the
similarity versus dissimilarity of group norms. Although manipulating group identification would seem to be problematic,
we have recently conducted further unpublished research in
which we achieved this by using a bogus pipeline procedure
(Doosje et al., 1995 ) within the same basic minimal-group paradigm used in Study 1. Space limitations prevent a detailed description of the results, but the first indications are promising
and confirm that the responses of high group identifiers become
similar in important respects to those of the natural groups in
Study 2, whereas those of low group identifiers resemble more
the patterns found in Study 1 (Jetten, Spears & Manstead,
1996). Specifically, principal-components analysis of pull
scores showed that whereas high group identifiers displayed the
greatest tendency to differentiate positively when group norms
were similar, low group identifiers positively differentiate most
when group norms were dissimilar. Other measures also confirmed that high group identifiers evaluated their group more
positively, identified most highly with it (i.e., as a dependent
measure), and saw it as more cohesive when the norms were
similar. For low group identifiers, the reverse tended to be true:
They evaluated the in-group more positively when group norms
were dissimilar. These preliminary results give us more confidence in suggesting that degree of identification is at least one
factor that may explain the different patterns we obtained across
our two studies.
This difference o f patterns across studies reinforces the argument that researchers should be wary of generalizing effects
found in artificial groups to natural groups (Mullen et al.,
1992). Particular features characteristic of artificial groups,
such as lower levels of identification, suggest that different principles derived from social identity and self-categorization theory may be applicable in these two domains. Specifically, the
ability of contrasting group norms to result in intergroup
differentiation, in line with the self-categorization principle of
comparative fit, may be strongest when little else is known about
these groups and when the degree of"groupness" and identification with the group are relatively unestablished. The effect of
intergroup similarity, as predicted by social identity theory, may
be strongest for groups that are already sufficiently well defined
that comparative fit and groupness are long since guaranteed.
To summarize, these two studies provide support for conformity to in-group norms and efforts to reflect or reassert group
distinctiveness by means of intergroup differentiation, in addition to the tendency to show in-group bias. One popular interpretation of the social identity explanation of discrimination in
minimal groups is that in-group bias is a fairly universal feature
of intergroup relations. However, even cursory analysis of relations between real-world groups reveals that this is far from
the case (van Knippenberg & Ellemers, 1990). The evidence
of the impact of group norms revealed in the present research
provides one powerful reason why this may be so and suggests
that group norms may play an important role in moderating ingroup bias. On the other hand, the relation between group
norms can also stimulate positive differentiation by eliciting
both contrastive and competitive social comparisons. In this respect, it is possible to argue that the minimal-group paradigm
has deflected the social identity tradition from a fuller consideration of both the ideological content and the social context of
intergroup relations, which were always an important part of its
theoretical heritage (Tajfel, 1972, 1978 ). Although self-categorization theory has more fully embraced these factors, it has
moved away from the motivational bases of intergroup discrimination to consider the cognitive and perceptual dimensions of
intergroup judgments. In elucidating both the perceptual and
motivational aspects of group distinctiveness, the present research demonstrates the distinctive but complementary contributions of self-categorization and social identity principles in
explaining intergroup perceptions and behavior.
References
Allen, V. L., & Wilder, D. A. ( 1975 ). Categorization, belief similarity
and intergroup discrimination. Journal of Personality and Social
Psychology, 32, 971-977.
Asch, S. E. (1956). Studies of independence and conformity: A minority of one against a unanimous majority. Psychological Monographs:
General and Applied, 70, 1-70.
Billig, M. ( 1973 ). Normative communication in a minimal intergroup
situation. European Journal of Social Psychology, 3, 339-343.
Billig, M., & Tajfel, H. ( 1973 ). Social categorization and similarity in
intergroup behaviour. European Journal of Social Psychology, 1, 2752.
Bourhis, R. Y., Sachdev, I., & Gagnon, A. (1994). Intergroup research
with the Tajfel matrices: Methodological notes. In M. P. Zanna &
J. M. Olson (Eds.), The psychology of prejudice: The Ontario Symposium (Vol. 7, pp. 209-232). Hillsdale, N J: Erlbaum.
Bornstein, G., Crum, L., Wittenbraker, J., Harring, K., Insko, C., &
Thibaut, J. (1983a). On the measurement of social orientations in
the minimal group paradigm. European Journal of Social Psychology,
13, 321-350.
SPECIAL ISSUE: GROUP NORMS AND INTERGROUP RELATIONS
Bornstein, G., Crum, L., Wittenbraker, J., Harring, K., Insko, C., &
Thibaut, J. (1983b). Reply to Turner's comment. European Journal
of Social Psychology, 13, 369-381.
Brewer, M. B. (1979). In-group bias in the minimal intergroup situation: A cognitive-motivational analysis. Psychological Bulletin, 86,
307-324.
Brown, R. J. (1984a). The effects of intergroup similarity and cooperative versus 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 (pp. 603-623). Cambridge, England: Cambridge
University Press.
Brown, R. (1988). Group processes: Dynamics within and between
groups. Oxford, England: Basil BiackweU.
Brown, R., & Abrams, D. (1986). The effects of intergroup similarity
and goal interdependence on intergroup attitudes and task performance. Journal of Experimental Social Psychology, 22, 78-92.
I)eutsch, M., & Gerard, H. B. (1955). A study of normative and informational influences upon individual judgement. Journal of Abnorreal and Social Psychology, 51, 629-636.
Diehl, M. (1988). Social identity and minimal groups: The effects of
interpersonal and intergroup attitudinal similarity on intergroup discrimination. British Journal of Social Psychology, 27, 289-300.
Diehl, M. (1989). Justice and discrimination between minimal groups:
The limits of equity. British Journal of Social Psychology, 28, 227238.
Doosje, B., Ellemers, N., & Spears, R. (1995). Perceived intragroup
variability as a function of group status and identification. Journal of
Experimental and Social Psychology, 31, 410-436.
Gerard, H. B., & Hoyt, M. E (1974). Distinctiveness of social categorization and attitude toward ingroup members. Journal of Personafity
and Social Psychology, 29, 836-842.
Hogs, M. A., & Abrams, D. (1993). Towards a single-process uncertainty-reduction model of social motivation. In M. A. Hogs & D. Abrams (Eds.), Group motivation (pp. 173-190). New York: Harvester
Wheatsheaf.
Jetten, J., Spears, R., & Manstead, A. S. R. ( 1996 ). Intergroup discrim-
ination as a function of similarity of intergroup norms and identification. Manuscript in preparation.
Lott, A. J., & Lott, B. E. (1965). Group cohesiveness as interpersonal
attraction. Psychological Bulletin, 64, 259-309.
Messick, D. M., & Mackie, D. M. (1989). Intergroup relations. Annual
Review of Psychology, 40, 45-81.
Mlicki, P. (1993). Us and them: The effects of categorization and inter-
dependence on identification within, and differentiation between, categories and groups. Unpublished doctoral dissertation, University of
Utrecht, Amsterdam, the Netherlands.
Mullen, B., Brown, R., & Smith, C. t 1992). Ingroup bias as a function
of salience, relevance, and status: An integration. European Journal
of Social Psychology,, 22, 103-122.
Mummendey, A., & Schreiber, H. (1984). Social comparison, similarity and ingroup favouritism: A replication. European Journal of
Social Psychology, 14, 231-233.
Newcomb, T. M. (1943). Personality and social change. New York:
Dryden Press.
Pettigrew, T. E (1958). Personality and sociocultural factors in intergroup attitudes: A cross-cultural comparison. Journal of Conflict
Resolution, 2, 29-42.
Roccas, S., & Schwartz, S. H. ( 1993 ). Effects of intergroup similarity
on intergroup relations. European Journal of Social Psychology, 23,
581-595.
1233
Sherif, M. (1936). The psychology of social norms. New York: Harper
& Brothers.
Spears, R., Doosje, B., & Ellemers, N. (in press). Self-stereotyping in
the face of threats to group status and distinctiveness: The role of
group identification. Personality and Social Psychology Bulletin.
Tajfel, H. (1970). Experiments in intergroup discrimination. Scientific
American, 223, 96-102.
Tajfel, H. (1972). Experiments in a vacuum. In J. Israel & H. Tajfel
(Eds.), The context of social psychology: A critical assessment (pp.
69-119). London: Academic Press.
Tajfel, H. (1978). Interindividual behaviour and intergroup behaviour.
In H. Tajfel (Ed.), Differentiation between social groups: Studies in
the social psychology of intergroup relations (pp. 27-60). London:
Academic Press.
Taj fel, H. ( 1982 ). Social psychology of intergroup relations. Annual Review of Psychology, 33, 1-39.
Tajfel, H., Billig, M. G., Bundy, R. P., & Flament, C. (1971). Social
categorization and intergroup behavior. European Journal of Social
Psychology, 1, 149-178.
Tajfel, H., & Turner J. C. (1986). The social identity theory of intergroup behavior. In S. Worchel & W. G Austin (Eds.), Psychology of
intergroup relations (pp. 7-24 ). Chicago: Nelson Hall.
Turner, J. C. ( 1975 ). Social comparison and social identity. European
Journal of Social Psychology, 5, 5-34.
Turner, J. C. (1978). Social comparison, similarity and ingroup-favoritism. In H. Tajfel (Ed.), Differentiation between socialgroups: Studies in the social psychology of intergroup relations (pp. 235-250).
London: Academic Press.
Turner, J. C. ( 1981 ). The experimental social psychology of intergroup
behaviour. In J. C. Turner & H. Giles (Eds.), Intergroup behaviour
(pp. 66-101 ). Oxford, England: Blackwell.
Turner, J. C. (1983a). A second reply to Bornstein, Crum, Wittenbraker, Harring, Insko and Thibaut on the measurement of social
orientations. European Journal of Social Psychology,, 13, 383-387.
Turner, J. C. ( i 983b). Some comments on "The Measurement of Social
Orientations in the Minimal Group Paradigm?' European Journal of
Social Psychology, 13, 351-367.
Turner, J. C. ( 1991 ). Social influence. Milton Keynes, England: Open
University Press.
Turner, J. C., Brown, R. J., & Tajfel, H. (1979). Social comparison and
group interest in ingroup favoritism. European Journal of Social Psychology, 9, 187-204.
Turner, J. C., Hogs, M. A., Oakes, P.J., Reicher, S. D., & Wetherell,
M. S. (1987). Rediscovering the social group: A self-categorization
theory, Oxford, England: Basil Blackwell.
van Knippenber~ A., & Ellemers, N. (1990). Social identity and intergroup differentiation processes. In W. Stroebe & M. Hewstone
(Eds.), European review ofsocialpsychology (Vol. 1, pp. 137-169).
Chichester, England: Wiley.
Vivian, J. E., & Berkowitz, N. H. (1992). Anticipated bias from an outgroup: An attributional analysis. European Journal of Social Psychology, 22, 415-424.
Vivian, J. E., & Berkowitz, N. H. ( 1993 ). Anticipated outgroup evaluations and intergroup bias. European Journal of Social Psychology,
23, 513-524.
Worchel, S., Coutant-Sassic, D., & Grossman, M. (1992). A developmental approach to group dynamics: A model and illustrative research. In S. Worchel, W. Wood, & J. A. Simpson (Eds.), Groupprocess and productivity ( pp. 181-202 ). Thousand Oaks, CA: Sage.
Received J u n e 13, 1995
Revision received July 15, 1996
Accepted July 30, 1996 •