Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Copyright 2001 by the American Psychological Association. Inc.

Journal of Personality and Social Psychology OO22-35I4/O1/$5.OO DOI: 10.1037//O022-3514.81.6.1O90


2001. Vol. 81. No. 6. 1090-1102

Is There Any Scapegoat Around?


Determinants of Intergroup Conflicts at Different Categorization Levels

Thomas Kessler and Amelie Mummendey


Friedrich-Schiller-Universitat

Recategorization at a higher level reduces tensions between groups. However, recategorization may
cause conflicts between the common in-group and a new out-group. Additionally, determinants of
conflict between subgroups may enhance conflict at the higher categorization level. In the context of
German unification, the authors explored these suggestions with an East German 3-wave longitudinal
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

study and a West German control group. Results show that a salient East German versus West German
This document is copyrighted by the American Psychological Association or one of its allied publishers.

categorization enhances conflict between subgroups, whereas categorization as German enhances conflict
at the common in-group level. Determinants of subgroup conflict also influence conflict at the inclusive
level (Germans and foreigners). Thus, recategorization is a 2-edged instrument: Although it reduces
conflict at the subgroup level, it may initiate conflict at the common in-group level.

The explanation of intergroup conflict and attempts to reduce isting groups, but may also simultaneously introduce a new inter-
conflict between groups has been central to social psychology. group conflict between the common in-group and a new out-group
Common expressions of conflict between groups include prejudice by enhancing the salience of categorization at this superordinate
(e.g., hostile attitudes against foreigners or xenophobia), social level. Moreover, another complication may be that both levels of
discrimination (e.g., unjust treatment of others on the basis of their categorization and their associated intergroup conflict may not be
group membership), and collective action (e.g., public protest). completely independent from each other. Thus, determinants of
According to the theoretical frameworks of social identity theory subgroup conflict may also enhance the conflict at the inclusive
(Tajfel & Turner, 1986) and self-categorization theory (Turner, level of the common in-group and the new out-group.
Hogg, Oakes, Reicher, & Wetherell, 1987), one crucial determi- In the current research, we wanted to test these two possible
nant for prejudice and discrimination is the salient level of self- negative side effects of recategorization by exploring recategori-
categorization that specifies which individuals are seen as similar zation in a wider natural context. We examined the two sugges-
to the self ("us") and which are excluded as being others ("them"). tions by exploring intergroup conflicts on different levels of self-
Consistent with this line of reasoning, several models predict that categorization. More specifically, we wanted to focus on two
intergroup conflict can be resolved by reducing the relative sa- different levels of categorization, one subgroup level and one
lience of the categorization level at which the specific intergroup common in-group level, and their influence on intergroup conflict.
conflict occurs (e.g., Brewer & Miller, 1984; Hewstone & Brown, In addition, within a longitudinal design, we wanted to examine
1986; S. L. Gaertner, Dovidio, Anastasio, Bachman, & Rust, 1993; whether the differences between levels of categorization have
Wilder, 1986). The decategorization approach (Brewer & Miller, causal effects on intergroup conflict. The different levels of cate-
1984) focuses on reducing conflict through personalized interac- gorization are placed in the field context of the German unifica-
tion that undermines the salience of group categorization. In con- tion. Within this merger situation (e.g., Terry & Callan, 1998) two
trast, the common in-group identity model (S. L. Gaertner et al., initially separate groups, East and West Germans, are on one hand
1993) stresses the virtues of the recategorization of subgroups into in conflict with each other and yet on the other hand supposed to
one common in-group. Recategorization of former in-group and form one common group. Here, the participants of the merger
out-groups into one common in-group may reduce conflict by could categorize themselves either as members of one of the initial
reducing the salience of the subgroup categorization. However, as groups or as members of the new all-inclusive group. Finally, to
we suggest, recategorization may reduce conflict between preex- replicate and generalize the results, we compared East German (as
a low-status group) and West German (as a high-status group)
samples. In sum, we do not want to challenge the common in-
Thomas Kessler and Amelie Mummendey, Lehrstuhl fur Sozialpsy- group identity model (S. L. Gaertner et al., 1993), rather we want
chologie, Friedrich-Schiller-Universitat, Jena, Germany. to point to possible sources of intergroup conflicts that may arise
The studies reported here were conducted within a research project by recategorization.
supported by Deutsche Forschungsgemeinschaft Grant MU-551-13-3. We
are grateful to David Kenny, Thorsten Meiser, Sabine Otten, Kai Sassen-
berg, and Sven Waldzus for their helpful comments and suggestions on a Common In-Group Identity Model
draft of this article.
Correspondence concerning this article should be addressed to Thomas According to social identity theory (Tajfel & Turner, 1986; see
Kessler, Friedrich-Schiller-Universitat, Lehrstuhl fur Sozialpsychologie, also S. L. Gaertner et al., 1993; S. L. Gaertner, Dovidio, Nier,
Humboldtstrasse 26, D-07743 Jena, Germany. Electronic mail may be sent Ward, & Banker, 1999), intergroup conflict develops by salient
to s7keth@uni-jena.de. categorization into in-group and out-groups (but see L. Gaertner &

1090
RECATEGORIZATION AND INTERGROUP CONFLICT 1091

Insko, 2000). Salient categorization leads to favoritism of in-group bias to the whole group, members of a new out-group may be at a
members by reward allocations (Tajfel, Billig, Bundy, & Flament, disadvantage by becoming salient. To our knowledge, this possi-
1971) and in evaluations (Mummendey & Schreiber, 1983) and bility is rarely mentioned and has not been the subject of empirical
promotes prosocial behavior to and cooperation with in-group research until now. Second, there remains the possibility that the
members (Piliavin, Dovidio, Gaertner, & Clark, 1981; Kramer & original conflict between subgroups may subtly transform itself or
Brewer, 1984) as well as expectations of generalized reciprocity carry over into a new conflict between the common in-group and
(Yamagishi & Kiyonari, 2000). The in-group bias leads to a a new out-group. Generally there are two ways in which different
favorable evaluation of in-group members, which does not, how- levels of intergroup conflict can influence each other: On one
ever, necessarily imply out-group derogation (Brewer, 1979, hand, determinants of intergroup conflict at one level may reduce
1999). Recategorization of previous in-group and out-groups into intergroup conflict on another level of self-categorization. This is
one common in-group reduces tensions (e.g., striving for positive in line with the common in-group identity model, which suggests
in-group distinctiveness; Tajfel & Turner, 1986) between these that a common in-group identity reduces conflict between former
previously distinct groups because recategorizing both groups into subgroups. On the other hand, determinants of an intergroup
one common in-group reduces the salience of the initial social conflict may also foster the intergroup conflict associated with
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

another level of self-categorization. For instance, determinants of


This document is copyrighted by the American Psychological Association or one of its allied publishers.

categories. The process of recategorization redirects some of the


cognitive and motivational processes that originally contributed to the conflict between subgroups may enhance the conflict between
the intergroup conflict. In particular, former out-group members the common in-group and a new out-group. This carryover of
are evaluated more positively because, after recategorization, they conflict between subgroups to the common in-group level may
are now seen as members of the common in-group (e.g., Hogg & occur for two reasons: On one hand, the less powerful subgroup
Hains, 1996; Hogg, Hardi, & Reynolds, 1995). Moreover, recat- within the common in-group may search for an even weaker group
egorization enhances relational justice concerns that foster attach- to which the conflict may be redirected (scapegoating; Allport,
ment and commitment to the common in-group (e.g., Huo, Smith, 1954). On the other hand, group members may recategorize (as a
Tyler, & Lind, 1996). The potential of recategorization to reduce means of social creativity; e.g., Blanz, Mummendey, Mielke, &
intergroup conflict is supported by several empirical findings: Klink, 1998) to evade the original intergroup conflict. The unin-
Previous results show that as salience of a common identity tended side effect may be the emergence of a new out-group and,
increases, there is less conflict between groups and more positive thereby, the development of a new intergroup conflict. Here, the
attitudes about the former out-group (Gaertner, Mann, Dovidio, emergence of a new out-group may be the triggering event for
Murrell, & Pomare, 1990; Gaertner, Mann, Murrell, & Dovidio, intergroup conflict (e.g., Pedersen, Gonzales, & Miller, 2000). In
1989). Favorable conditions of contact such as equal status, coop- this case, recategorization is not sufficient for reducing an inter-
eration, institutional support, and opportunities for personal ac- group conflict, at least under certain conditions (e.g., if some
quaintance (see Allport, 1954) promote the perception of both criteria for favorable intergroup contact such as equal status are
groups as being one common in-group. Concerning the cognitive violated). On the contrary, recategorization may possibly lead to
processes that reduce intergroup conflict, the effect of these favor- an alternative route of conflict—between, for example, the in-
able contact conditions on conflict reduction is mediated by a group and a new out-group which can be competed against more
common in-group identity (S. L. Gaertner, Rust, Dovidio, Bach- easily by the in-group.
man, & Anastasio, 1996). Thus, the representation of a common
in-group is the process that is triggered by favorable conditions of
Status Differences
contact, thereby leading to more positive emotional reactions and
attitudes toward the former out-group (Dovidio, Gaertner, Isen, & Status differences between groups may be one factor that de-
Lowrance, 1995; Dovidio, Gaertner, & Validzic, 1998; S. L. termines whether recategorization may be a viable method for
Gaertner, Dovidio, Rust, Nier, Banker, Ward, Mottola, & Houlette, reducing an intergroup conflict and whether an intergroup conflict
1999). Within the common in-group identity model, it is supposed may carry over to other levels of categorization. In intergroup
that the subgroup identities need not necessarily be abandoned. relations with status differences, members of the superior group
Both levels of categorization, common in-group identity and sub- evaluate intergroup contact more favorably than do members of
group identities, could be simultaneously salient. However, results the lower status group because they have more control over the
for the benefits of maintaining dual identities were mixed (e.g., intergroup-contact conditions (Islam & Hewstone, 1993). Equal
S. L. Gaertner, Dovidio, Nier, et al., 1999; Hornsey & Hogg, status enhances recategorization, however, only if the distinctive-
2000a, 2000b). ness of one's own group is not threatened (see also Brown &
Rather than being a prime tool to reduce intergroup conflict, Wade, 1987; Hornsey & Hogg, 2000b). Perceived low status
recategorization at an inclusive level may also present several impedes the recategorization into one common in-group (Dovidio
factors that could actually enhance conflict and prejudice: First, it et al., 1998; Seta, Seta, & Culver, 2000), thereby reducing the
is not obvious what kind of new intergroup context will emerge possibility of an improvement in the intergroup relationship. In
after recategorization at an inclusive level. Does recategorization accordance with this line of reasoning, S. L. Gaertner et al. (1996)
lead to a more positive view of others and initiate in-group favor- found that high-status group members have more favorable im-
itism to members of the now inclusive group, or does it perhaps at pressions of conditions of contact relative to low-status group
the same time induce prejudice, hostility, and discrimination to- members. Moreover, low status enhances the salience of members'
ward members of a newly emerging out-group? We suggest that social identity compared with members of a superior group. Thus,
whereas members of the common in-group (i.e., the former sub- with low status, it may be more difficult to forge a common
group members) may benefit from the transfer of a pro-in-group in-group identity that rises above the salience of their subgroup
1092 KESSLER AND MUMMENDEY

social identity. Members of low-status groups show higher degrees central constructs in previous research has been resentment about
of bias in affective reactions than members of high-status groups relations between East and West Germans. Resentment is con-
do. In particular, this bias is mainly the result of less favorable ceived as the emotional component of fraternal relative deprivation
affective responses toward the out-group (S. L. Gaertner et al., (Cook, Crosby, & Hennigan, 1977; Runciman, 1966) that also
1996). Generally, perceived low status enhances attempts to over- parallels the notion of prejudice as a negative group-based emotion
come this dissatisfactory situation and may thereby reduce the (Smith, 1993). Thus, this variable is the strongest single predictor
effectiveness of efforts to reduce tensions between groups. Thus, for collective strategies such as realistic and social competition.
status differences between groups may lead to either less recat- A second construct, group efficacy, denotes a collective belief in
egorization or less positive effects of recategorization, particularly the "effectiveness" of a group. Group efficacy implies high moti-
for groups that are regarded as having lower status. vation to work for the benefit of the group, to persist in pursuing
group goals in the face of obstacles and collective difficulties, and
to be willing to accept difficult challenges for the group (Zaccaro,
Intergroup Conflict Blair, Peterson, & Zazanis, 1995). Consistent with this line of
reasoning, previous studies have shown that group efficacy is a
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Expression of Intergroup Conflict


second predictor for collective strategies (Mummendey, Kessler, et
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Several identity strategies have been proposed by social identity al., 1999). Finally, a third variable, in-group identification (Tajfel
theory and relative deprivation theory as a means to cope with less & Turner, 1986) has been shown to predict individual strategies
favorable group status (Blanz et al., 1998; Lalonde & Cameron, (e.g., social mobility). Although identification with one's group
1994). Generally, individual and collective strategies represent two was correlated with collective strategies, this relation was com-
different types of strategies to cope with negative social identity pletely mediated by resentment and group efficacy.
and relative deprivation (Mummendey, Kessler, Klink, & Mielke, To summarize, resentment, group efficacy, and in-group iden-
1999): Individual strategies (e.g., social mobility) attempt to im- tification were predictors for collective strategies (i.e., realistic and
prove a less favorable situation by changing the individual posi- social competition). Whereas resentment has the strongest influ-
tion, thereby leaving the intergroup relations intact. In contrast, ence on collective strategies, the relation between in-group iden-
collective strategies (e.g., public protest and competition) aim at tification and strategy preference was mediated by resentment and
changing a presently existing intergroup situation. Thus, they are group efficacy.
clear expressions of conflict between groups. An obvious conflict
strategy is public protest that mobilizes group members in collec-
Aims of the Present Study
tive action (e.g., demonstrations, sit-ins, and strikes). Other fea-
tures of intergroup conflict are hostile attitudes against foreigners Recategorization as a means to reduce conflict between groups
and xenophobia. Xenophobia and prejudice can take both blatant raises two rarely mentioned issues: First, recategorization at a
and subtle forms (Pettigrew & Meertens, 1995). They imply neg- higher level may constitute a new intergroup setting and, thus, may
ative evaluations of out-group members as well as strong negative introduce a new line of intergroup conflict. Second, both levels of
affect (Smith, 1993). self-categorization with their associated conflict could be interde-
The present study focused on public protest and xenophobia as pendent. Thus, determinants of one intergroup conflict may also
two different expressions of intergroup conflict because both are influence an intergroup conflict at the other categorization level.
expected to be related to different levels of categorization within For instance, recategorization at a higher level may be a strategic
the German merger. Though xenophobia and public protest are not means to evade (not to solve—as assumed in the common in-group
the same type of intergroup conflict (see Struch & Schwartz, identity model) a present intergroup conflict. Therefore, determi-
1989), they are both fostered by salient categorization of an in- nants of conflict between two subgroups may contribute to the
group and out-group. Public protest may be related to the subgroup conflict on a higher level of categorization.
categorization (i.e., East German and West German). In contrast, These suggestions can be summarized in one model (see Figure
xenophobia may be more strongly associated with the inclusive 1). To examine this model in our present study, we assessed the
categorization level (i.e., German). We hoped that the assessment relation between levels of categorization (i.e., categorization as
of two levels of categorization together with their associated East German vs. categorization as German) and the two indicators
intergroup conflicts would enable us to examine the interdepen- of intergroup conflict (i.e., public protest and xenophobia). We
dence between both levels as well as whether there are specific or expected that categorization as East German enhances public pro-
common determinants of intergroup conflicts on different levels of test, whereas categorization as German intensifies xenophobia.
inclusiveness. The relation between higher level of categorization on subgroup
conflict and lower level categorization on intergroup conflict at the
Prediction of Conflict by Social Identity and Relative inclusive level of categorization were also assessed. Moreover, we
assessed whether the predictors of conflict between subgroups
Deprivation
(resentment, group efficacy, and in-group identification) have a
On the basis of social identity theory (Tajfel & Turner, 1986) causal impact on intergroup conflicts. In particular, these predic-
and relative deprivation theory (e.g., Crosby, 1982; Folger, 1987), tors are expected to be causally related to the conflict between
previous studies in the field context of the German merger have subgroups. More challenging is the question of whether determi-
identified several variables that differentially predict individual nants of subgroup conflict also have a causal impact on conflict at
and collective strategies (Mummendey, Kessler, et al., 1999; a higher level of categorization. This influence could either be
Mummendey, Klink, Mielke, Wenzel, & Blanz, 1999). One of the direct (i.e., the classical scapegoat; see Allport, 1954) or be indi-
RECATEGORIZAT1ON AND INTERGROUP CONFLICT 1093

conflict between
common ingroup
common ingroup
categorization
and new outgroup

determinants of
conflict between
subgroups
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

subgroup conflict between


categorization subgroups

Figure 1. Levels of categorization with their related conflict and displacement of intergroup conflict between
subgroups to the categorization level of common in-group.

rect (i.e., mediated by recategorization; see Figure 1). Moreover, 4. High-status group: For either the low- or high-status group, both
the longitudinal design of the present study allows analyzing levels of categorization are related to different intergroup conflicts.
causal relations of variables. Thus, the assumption of a significant However, for high-status groups (i.e., West Germans) the conflicts are
causal influence of levels of categorization on intergroup conflict less intense.
(or vice versa) could be examined.
Finally, we assessed the impact of status differences on the Method
proposed relations between concepts. To this end, we compared an
East German sample as a group that was perceived as having lower Field Situation
status with a West German sample as a group that was perceived
as having superior status. For the superior group, the same rela- The unification of Germany offers an exemplary situation in which to
tions between levels of categorization and intergroup conflict were study the interplay between different levels of self-categorization and
expected. West Germans were expected to feel less resentment their associated intergroup conflict. The unification of Germany is a
concerning the relation between East and West Germans. More- merger that attempts to integrate the two subgroups of East and West
over, the subgroup categorization as West German is less salient Germans into one common and inclusive group of Germans. Members
and, hence, public protest is less strong compared with categori- of each subgroup were able to categorize themselves as either members
of the subgroup (East or West German) or as members of the inclusive
zation as East German. For West Germans, we assumed that both
group (German).
levels of categorization are closely related because they do not
differentiate between West German and German. Hence, there will Within this merging situation some favorable conditions of inter-
group contact (Allport, 1954) are approximately fulfilled: There is
be less or no displacement of intergroup conflict from the sub-
institutional support for the common goal to become one new society.
group to the common in-group level of categorization.
As members of both groups have to be mobile to keep their jobs
or to get new jobs there is a rich acquaintance potential. However,
Hypotheses one of the most obvious problems within the German merger is the
perceived lower status of East Germans. Thus, there is a strong con-
Our hypotheses can be summarized as follows: sensus that East Germans are worse off compared with West Germans.
For East Germans compared with West Germans, several social com-
1. Levels of intergroup conflict: Salient self-categorization at the parisons result in negative outcomes because West Germans are often
subgroup level (e.g., East German or West German) promotes public taken as the standard that East Germans are expected to attain. For
protest, whereas a salient self-categorization at the common in-group example, opinion polls in 1991 and 1995 revealed that most East
level (e.g., German) enhances xenophobia. Germans (above 70%) saw themselves as "second-class citizens"
2. Prediction of intergroup conflict: Variables at the subgroup level (Harenberg, 1991; Spiegel-Redaktion, 1995). Even today, because of
(resentment, group efficacy, or in-group identification) influence in- the perception of the lower status of the East Germans, there continues
tergroup conflict not only at the subgroup level but also at the to be considerable tension between East and West Germans. However,
inclusive level of categorization. the official political guideline is that of harmonizing the tension be-
tween both groups by accentuating the inclusive group membership.
3. Causal structure: There is evidence for causal influence of cate-
gorization levels on intergroup conflict as well as of determinants of Inasmuch as the merger is increasingly accepted, it seems no longer
subgroup conflict on the intergroup conflict between the common appropriate to place the intergroup conflict at the level of East Germans
in-group and a new out-group. versus West Germans.
KESSLER AND MUMMENDEY
1094

Participants and Procedure The response format for all items was a 5-point scale (with participants
indicating how much they agreed or disagreed with the statements pre-
Over 3 years, from 1995 to 1997, 109 participants (65 women and 44 sented, i.e., 1 = do not agree at all to 5 = agree very much).
men) took part in the longitudinal study.1 All were born and were still Levels of categorization. Each level of categorization was assessed
living in East Germany. In the first wave, the mean age was 40 years with one item. Self-categorization as an East German was measured with
(ranging from 19 to 85 years). Educational level of respondents included the item "I see myself as an East German"; categorization as German was
secondary school (n = 31), grammar school (n = 22), technical college measured with the item "I see myself as a German." The response format
(n = 25), and university (n = 30). Fifty-seven percent of participants were for these two items was a 5-point scale that ranged from never to always.
employed (n = 62), 23% were students (/i = 25), and 6% were unem-
ployed persons (n = 6), in addition to 14% who were homemakers and
pensioners (« = 15). Results
Assessed as a control group in 1996, the West German sample included
183 participants (91 women, 90 men; 2 did not indicate their gender). The Levels of Self-Categorization and Predictors of Intergroup
mean age was 31.4 years (range = 18-65 years), and educational levels Conflict
comprised secondary school (n = 40), grammar school (n = 104), tech-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

nical college (n = 11), and university (n = 24). Participants' principal First, we examined the mean change of variables over the three
This document is copyrighted by the American Psychological Association or one of its allied publishers.

occupations were employees (34%; n = 63), students (52%; n = 95), repeated measures from 1995 to 1997. To analyze the observed
unemployed persons (2%; n = 4), and pensioners or homemakers (8%; n = change in the mean of variables, we computed an analysis of
15). variance with repeated measurements for each variable separately.
Each participant completed a questionnaire containing all items relevant There was no significant difference between the measurement
to the present concerns in addition to other questions not related to the occasions in predictor variables (resentment about relations be-
present study. Coworkers in East Germany and West Germany distributed
tween East and West Germans, East German group efficacy, and
the questionnaires to their acquaintances. Participants took part voluntarily;
identification as East German) or in public protest (see Table 1).
they were told that the investigation was related to their "general life-
satisfaction." Participants completed their questionnaires individually ei- Moreover, categorization as East German or German did not
ther at their workplace or at home and returned them by mail. Each significantly change over the period of 3 years. As the only
participant had a code label; we could then collate data from the East significant change, xenophobia increased over time, F(2,
German sample from Time 1, Time 2, and Time 3 without jeopardizing 216) = 17.77, p < .001, with a continuous increase, linear trend,
anonymity. F(l, 108) = 39.72.
In our hypotheses, we suggested that self-categorizations at
different levels are causally related to their specific intergroup
Assessment of Variables
conflicts. In particular, we hypothesized that categorization as East
First, participants rated their perceived status relative to the other Ger- German strengthens public protest, whereas categorization as Ger-
man out-group (East Germans or West Germans). Participants indicated man increases xenophobia. Moreover, we proposed that determi-
whether they perceived their group's relative status as lower, equal, or nants of the conflict at the subgroup level of categorization (i.e.,
superior. In a second section of the questionnaire, participants rated their East German) might also influence the intergroup conflict at the
resentment about relations between East and West Germany, perceived higher level of categorization. Thus, determinants of the conflict at
group efficacy, their degree of in-group identification, and levels of self- the subgroup level also might be causally related to the conflict at
categorization. Finally, participants rated their inclination to public protest the inclusive level of categorization. We set up a panel model with
and whether they agreed with items indicating xenophobia and hostile
structural equation modeling procedures (LISREL 8; Joreskog &
attitudes against foreigners in Germany.
Sorbom, 1993). This panel model included categorization as East
Public protest and xenophobia. Inclinations to public protest were
German and German, public protest and xenophobia as indicators
assessed with four items such as 'The time is ripe that we East Germans
should demonstrate for our rights." The reliability for public protest was of intergroup conflict, and predictors of subgroup conflict (resent-
alpha (Time 1) = .72, alpha (Time 2) = .78, alpha (Time 3) = .84, and for ment, group efficacy, and in-group identification). In addition, the
the West German sample, alpha = .84. Xenophobia was measured with model involved these variables as measured at three times. To find
four items such as "Foreigners and Germans can never be really comfort- evidence for causal influence in a panel analysis, we computed
able with each other, even if they are close friends." The reliability for this cross-lagged regressions (Cook & Campbell, 1979; Rogosa, 1980).
variable was alpha (Time 1) = .75, alpha (Time 2) = .80, alpha (Time 3) = In a three-wave longitudinal design, for each hypothesized causal
.76, and for the West German sample, alpha = .59. relation there are three possible cross-lagged regressions that could
Predictors for xenophobia and public protest. Resentment about the indicate a causal influence. In particular, there was one cross-
relationship between East and West Germans was assessed with three items lagged regression of Variable A at Time 1 on Variable B at Time 2
such as "Being faced daily with the situation of the East Germans one can
(Lag for Times 1-2 regression), and there was a cross-lagged
only become annoyed" with a reliability of alpha (Time 1) = .79, alpha
regression of Variable A at Time 2 on Variable B at Time 3 (Lag
(Time 2) = .80, alpha (Time 3) = .82, and for the West German sample,
alpha = .66. Group efficacy was measured with three items such as "We for Times 2-3 regression), finally, there was a cross-lagged re-
East Germans are not able to manage our fate by ourselves" with a gression that related Variable A at Time 1 to Variable B at Time 3
reliability of alpha (Time 1) = .64, alpha (Time 2) = .64, alpha (Time 3) =
.59, and for the West German sample alpha = .64. Finally, identification
as an East German was captured with three items such as "I consider 1
This longitudinal study was part of a large survey project on the
myself to be East German." Reliability of identification for the three unification of Germany. Within this project, we planned to investigate
measures was alpha (Time 1) = .81, alpha (Time 2) = .85, alpha (Time different research questions concerning intergroup behavior with several
3) = .84, and for the West German sample alpha = .86. Overall, the independent samples (longitudinal and cross-sectional) in East Germany as
reliability of constructs is acceptable, but in some cases low. well as in West Germany.
RECATEGORIZATION AND INTERGROUP CONFLICT 1095

Table 1 model revealed that the assumed equality constraints of the au-
Mean Changes Over Time in the Variables of the Model toregressive coefficients did not hold for categorization as Ger-
man, Ax^l, N = 109) = 4.06, p < .05, and xenophobia, Ax^l,
1995 1996 1997 N = 109) = 9.36, p < .01. Hence, the equality constraints for the
M SD M SD M SD F(2, 216)
autoregressions for categorization as German and xenophobia
Variable
were removed. Moreover, for each variable the second-order au-
Resentment 2.91 0.93 2.87 0.82 2.96 0.92 0.76 toregressive coefficient reached significance even when the first-
Group efficacy 3.53 0.77 3.42 0.74 3.43 0.68 1.73 order autoregressions were controlled for (r values ranged be-
Identification 4.13 0.78 4.09 0.83 4.09 0.88 0.17
tween 2.05 and 5.02). Concerning the hypothesized causal
German categorization 3.72 1.29 3.77 1.16 3.80 1.16 0.30
East German relations between variables, our results indicate that the significant
categorization 3.82 1.10 3.91 1.05 3.91 1.10 0.55 cross-lagged regressions correspond generally to our suggested
Public protest 2.59 0.83 2.52 0.80 2.65 0.88 1.71 hypotheses (see Table 2). There are no other significant cross-
Xenophobia 2.49 0.82 2.69 0.90 2.85 0.86 17.77***
lagged regressions that indicate causality than the coefficients
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

***p < .001. listed in Table 2. Moreover, the equality constraints hold for all
This document is copyrighted by the American Psychological Association or one of its allied publishers.

significant cross-lagged regressions, A ^ ( l , N = 109) < 3.84, p >


.05. Finally, all second-order cross-lagged regressions (lag for
(Lag for Times 1-3 regression). We placed equality constraints on Times 1-3 regressions) were nonsignificant, all t values < 1.25.
the Lag for Times 1-2 and Lag 2-3 regressions. Thus, a causal We removed all paths that were not significant from the model
influence between two variables would be concluded only if the (i.e., all these paths were set 0) and tested the accuracy of the
equality constraint holds and both cross-lagged regressions are
model. The model showed a satisfying fit as is indicated by the
significant. The second-order cross-lagged regression (Lag for
^ ( 5 5 , N = 109) = 56.11, p = .43, as well as other fit indicators,
Times 1-3 regression) was analyzed separately. Moreover, the
root mean square error of approximation = .014, Akaike's infor-
first-order autoregressive coefficients (i.e., stability of variables)
mation criterion (AIC) = 186.11 (saturated model = 240), Tucker-
were set as equal for each variable. The second-order autoregres-
sive coefficients (i.e., the regression of one variable measured at Lewis index (nonnormed-fit index) = 1.00. To evaluate this model
Time 1 on the same variable measured at Time 3) were analyzed further, we compared it to other models. First, including all cross-
as well. lagged regressions in the model did not improve the model fit
A preliminary analysis shows that neither group efficacy (t significantly, ^ ( 4 0 , N = 109) = 39.99, p = .47, with a A^2(15,
values < 1.29) nor in-group identification (f values < 1.31) had N = 109) = 16.12, p > .05. In addition, the full model revealed
significant cross-lagged regression on indicators of intergroup a higher AIC of 199.99 compared with the AIC of our hypothe-
conflict. Consequently, neither variable was a causal antecedent of sized model (AIC = 186.11). This indicates that the hypothesized
intergroup conflict and, hence, they were not further considered. model is more parsimonious than the model in which all cross-
The final panel model included resentment as a predictor of lagged regressions are allowed (see Bozdogan, 1987). Second,
conflict between subgroups (i.e., East Germans and West Ger- removing all cross-lagged regressions decreased the model fit
mans), the subgroup level of categorization (East German), the significantly, ^ ( 6 0 , N = 109) = 98.19, p = .001, with a A^(15,
inclusive level of categorization (German), and both indicators of A' = 109) = 42.08, p < .001. Thus, the nonsignificant cross-lagged
intergroup conflict (public protest and xenophobia). Tests of the regressions add nothing significant to the model, whereas remov-

Table 2
Results of the Three-Wave Panel Model for Levels of Categorization (East German and
German), Intergroup Conflict (Public Protest and Xenophobia), and Resentment

Predictor

Criterion Resentment East German German Public protest Xenophobia

Resentment .54***
East German .21*** .45***
German .11* .72***/.42***
Public protest .14* .12* -.02 40***
Xenophobia .16*** -.15*** .13* .67***7.25**

Note. The coefficients in each cell represent the regression weights for the lags between Time 1 and 2 and
Time 2 and 3. As the regression weights for both time lags were set equal, only their common solution is
indicated. In two cases (i.e., autoregression for German categorization and xenophobia) the equality constraint
did not hold; here, the first number denotes the Time 1 to 2 lag and the second number denotes the Time 2 to 3
lag. The diagonal cells indicate autoregressions between variables over time. The cells under the diagonal
indicate the cross-lagged regression weights. The blank cells represent cross-lagged regression coefficients that
do not differ significantly from zero and, therefore, were set null. Model fit is ^ ( 5 5 , N = 109) = 56.11, p =
.43; root mean square error of approximation = .014; Akaike's information criteria = 186.11(240); nonnormed
fit index = 1.00.
*p<.05. **p<.0l. ***p< .001.
KESSLER AND MUMMENDEY
1096

ing all cross-lagged regressions obviously leads to an inferior Consequently, for an unambiguous assessment of a causal relation
model. between two variables, measured at one time point, unidirectional
The significant cross-lagged regressions reveal a clear picture. causation has to be demonstrated.
Resentment about the relation between East and West Germans is To assess whether the causal influence was unidirectional (i.e.,
a powerful causal antecedent for the other variables in the model: A influences B), we compared four different models for each
High resentment increases East German categorization (/3 = .21) causal path: In the first model bidirectional causation was specified
and German categorization ()3 = .11). Moreover, resentment fos- (A influences B and B influences A), in a second model the
ters inclination for public protest (/3 = .14) and xenophobic atti- expected direction of causal influence was allowed (A influences
tudes (/3 = .16). Consistent with our hypotheses, the German level B), in a third model only the reverse direction was allowed (B
of categorization positively influenced xenophobia (/3 = . 13) as influences A), and in the fourth model no causal path was allowed
the intergroup conflict that applies to the categorization level (no causal influence).2 To assess the unidirectionality of the causal
Germans versus foreigners. Moreover, categorization as East Ger- influence, we anticipated the comparison between the first and the
man positively influences public protest (/3 = .12) as the conflict second model to be nonsignificant. Thus, the addition of the
reverse causal influence adds nothing significant to the expected
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

that is specific for the categorization at the subgroup level (East


causal path. The comparison between the first model and the third
This document is copyrighted by the American Psychological Association or one of its allied publishers.

and West Germans). Interestingly, categorization as East German


negatively influences xenophobic attitudes (/3 = -.15). That is, a model, as well as the comparison between the second and the
stronger East German categorization reduces xenophobic attitudes fourth model, are expected to be significant. The first model
within a time lag of 1 year. A parallel result for the German level comparison revealed that the bidirectional model fit the data sig-
of categorization did not apply: There was only a very small nificantly better than the model that included only the reverse
nonsignificant negative cross lag for categorization as German on causal direction, whereas the second comparison revealed that
public protest (/3 = -.02). Thus, categorization as German has no assuming no causal influence shows a significantly inferior model
causal influence on public protest. fit than assuming the expected causal influence. Finally, compar-
According to our hypotheses, levels of categorization are ex- ing the third and the fourth model revealed whether the reverse
pected to differentially influence both intergroup conflicts. To causal direction adds anything significant to the model that as-
analyze this possibility, we set the influences of each level of sumes no causal influence at all. To summarize, we assumed
categorization on both intergroup conflicts as equal. If the model unidirectional causal influence if the comparison between the first
fit decreases significantly then this equality assumption does not fit and the third model as well as the comparison between the second
the data and, hence, differential prediction of intergroup conflict and the fourth model was significant. In addition, the comparison
by each level of categorization is established. For the categoriza- between the first and the second model as well as the comparison
tion as German, the results revealed a significant decrease in the between the third and the fourth model had to be nonsignificant.
model fit, A ^ ( l , N = 109) = 3.86, p < .05. Introducing this We computed all these model comparisons for all significant
equality constraint for categorization as East German, the model fit paths of the model (see Table 3). The results confirmed that all
decreased significantly as well, A x ^ l , N = 109) = 11.84, p < significant causal paths were unidirectional, except the relation
.001. Consequently, each level of categorization had, as predicted, between resentment and public protest. For this path, the possibil-
a significantly different influence on both intergroup conflicts. ity of bidirectional influence could not be ruled out, as the model
specifying the reverse causation compared with the model without
causality was significantly different. Thus, the reverse causal in-
Evaluation of Causality
fluence (i.e., the influence of public protest on resentment) may
Although the results of the panel model are clear, there are other apply to some extent. However, the path itself did not reach
criteria for the evaluation of causal direction between variables. significance in the bidirectional model (t = 1.90, p > .05). In
According to Kenny's cross-lagged panel approach (e.g., Kenny, general, the assumption of unidirectional causal influence applied
1975, 1979), a relation between two variables could be interpreted to our model. From this we can conclude that the transference of
as causal rather than spurious if the cross lags between two knowledge about causality from the panel model to a cross-
variables differ significantly. However, as it is often criticized sectional structural equation modeling procedure may generally be
(e.g., Cook & Campbell, 1979; Rogosa, 1980), this criterion is not justified. However, in some cases it may lead to an overestimation
an unequivocal criterion for causality, because both cross-lagged of the real size of the causal relation between variables because the
regressions could be significant and at the same time significantly reverse causal direction could not completely be ruled out.
different. Thus, we could have clear bidirectional causation, but,
according to the cross-lagged panel analysis, we have to favor one
over the other. Here, we would' like to elaborate on one point Effectiveness of Recategorization
related to this issue. If a causal influence between two variables is
The levels of categorization causally influence intergroup con-
established in a panel analysis, and this knowledge about causality
flict. East German categorization enhances public protest and
is transferred to a second analysis that involves structural equation
German categorization enhances xenophobia. In addition, East
modeling with cross-sectional data, then it is necessary to have
German categorization extenuates xenophobic attitudes. However,
unidirectional causation. As bidirectional causation could not be
German categorization is not causally related to public protest.
ruled out a priori (Smith, 1982), a given relation between two
variables measured at one time point may be the outcome of
bidirectional causation. Here, modeling only one causal direction 2
We thank David Kenny for suggesting this test for unidirectionality of
may overestimate the real size of the assumed causal influence. causal influence.
RECATEGORIZATION AND INTERGROUP CONFLICT 1097

Table 3 their disadvantaged status position relative to West Germans. A


Comparison of Bidirectional Causal Influence, Initial Causal salient German categorization enhances xenophobia. In contrast, a
Influence, Reverse Causal Influence, and Noncausal Influence salient East German categorization is negatively related to the
intergroup conflict at the inclusive level of categorization. That is,
Model comparison. A*2 a a stronger East German categorization reduces xenophobic atti-
tudes. However, there is no significant causal influence from
Outcome 1-2 1-3 2-4 3^1
categorization as German on public protest. Thus, a salient Ger-
Resentment man categorization does not reduce the intergroup conflict be-
German cat. 1.72 4.78* 4.85* 1.79 tween the subgroups. Nonetheless, as the analyses within each
East German cat. 2.94 12.55*** 12.78*** 3.17 measurement show, levels of categorization have a relative impact
Xenophobia 0.42 11.37*** 11.66*** 0.71 on both intergroup conflicts: Consistent with expectations of the
Public protest 3.28 4.18* 5.89* 3.99*
German cat.
common in-group identity model, a stronger German categoriza-
Xenophobia 1.86 5.55* 5.62* 1.93 tion relative to the East German categorization is negatively re-
Public protest 2.62 0.08 0.08 2.62 lated to public protest. Hence, recategorization reduces the conflict
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

East German cat. between subgroups. However, a stronger German categorization


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Xenophobia 0.44 11.17*** 11.18*** 0.04


relative to the East German categorization is positively correlated
Public protest 0.05 4.48* 4.47* 0.04
with xenophobia. Thus, recategorization enhances the intergroup
Note. 1 = bi-directional causality; 2 = expected causal direction; 3 = conflict at the inclusive level of self-categorization.
reverse causal direction; 4 = no causal influence; cat. = category. Resentment about the relation between East and West Germans
>df= 1.
*p < .05. ***p < .001. is a powerful causal agent: Resentment enhances the salience of
the East German categorization as well as more active public
protest. Moreover, resentment enhances the salience of the Ger-
man categorization. This influence holds even if the salience of the
Thus, there is no causal evidence for the effectiveness of recat-
East German categorization and the readiness for intergroup con-
egorization as a means to reduce conflict between subgroups.
flict (public protest and xenophobia) are controlled for. Thus,
However, for each measurement the relative impact of each cate-
strong resentment may lead to German categorization as a means
gorization level on intergroup conflict was assessed. As both
to evade the negative intergroup relation between East and West
categorization levels are not exclusive, that is, participants could
Germans. Resentment influences xenophobic attitudes through
be high in East German and German categorization as well as low
two different ways: First, resentment enhances the East German
on both, the levels of categorization are increasingly correlated
categorization and thereby fosters xenophobic attitudes. Second,
over time, r (Time 1) = .15, p < .10; r (Time 2) = .22, p < .05;
resentment directly enhances xenophobia. The latter represents the
and r (Time 3) = .35, p < .001. Correspondingly, the correlation
classical scapegoat argument, whereas the first is a more sophis-
between public protest and xenophobia increased nonsignificantly
ticated form of displaced intergroup conflict.
over time, r (Time 1) = -.06, p > .10; r (Time 2) = M,p > .10;
and r (Time 3) = .17, p > .10. To assess the relative impact of The paths discussed above are causal paths, as the panel analysis
different categorization levels on intergroup conflict, we related reveals. Thus, they show significant cross-lagged regression coef-
both measures for the levels of categorization by subtracting the ficients that are consistent over time (i.e., cross-lagged regression
degree of the East German level of categorization from the degree of Time 1 on Time 2 is equal to the cross-lagged regression of
of the German level of categorization. This variable indicates the Time 2 on Time 3). We found no evidence for more distal causal
relative strength of both categorization levels. Here, values above influences that involve cross-lagged regressions of Time 1 on
zero indicate a preference for German categorization over the East Time 3. As an additional criterion of causality, the unidirection-
German categorization and values below zero denote higher pref- ality of causal influences was assessed. Here, causal paths between
erence of East German over the German categorization. Results two variables have to be significant in one direction and zero in the
show that this variable is positively related to xenophobia, r (Time other direction. The results revealed that the causal paths are
1) = .32, p < .001; /-(Time 2) = .34, p < .001; r (Time 3) = .29, unidirectional paths, with the exception of the influence of resent-
p < .01, and negatively to public protest, r (Time 1) = - . 2 8 , p < ment on public protest, for which the reverse influence of public
.01; r (Time 2) = - . 0 6 , ns; r (Time 3) = - . 2 0 , p < .05. The more protest on resentment could not be ruled out. According to this
the participants perceived themselves as German, rather than East evidence, we could specify structural equation models for cross-
German, the more they leaned toward xenophobia, and less toward sectional data that incorporate causal paths without overestimating
public protest; conversely, the more the participants saw them- size of path coefficients.
selves as East German rather than German, the stronger the incli-
nation to public protest was, and the less it was to xenophobia. Comparison of East Germans and West Germans
The results of the East German sample may not be generally
Preliminary Discussion
applicable because East Germans are regarded as having lower
The results concerning resentment and levels of categorization status than West Germans within the German merger. According
as causal antecedents of intergroup conflict correspond closely to to previous results concerning the common in-group identity
our expectations. Each level of categorization differentially influ- model, lower status may reduce attempts to view the intergroup
enced both types of intergroup conflict: East German categoriza- situation as a common in-group (Dovidio et al., 1998). Moreover,
tion promotes more active public protest by East Germans against a minority status enhances the salience of a particular intergroup
KESSLER AND MUMMENDEY
1098

situation, thereby enhancing affective intergroup bias as well as Table 5


negative attitudes toward the majority (S. L. Gaertner et al., 1996). Comparison Between East Germans and West Germans on
By comparing a West German sample with an East German Resentment, Levels of Categorization (East Versus West German
sample, we could test whether this pattern in the relations of and German), and Intergroup Conflict (Public Protest and
variables is basically the same or whether differences between Xenophobia)
high- and low-status groups could be observed. As the West
East West
German sample was assessed in 1996, we compared it with the Germans Germans
East German sample that was also measured in 1996.
First, we tested the validity of the assumed status difference Variable M SD M SD t P<
between East Germans and West Germans. As can be seen from
Resentment 2.87 0.82 2.21 0.72 7.13 .001
Table 4, most East Germans perceived themselves as having a
German categorization 3.77 1.16 3.61 1.16 1.17 .20
lower status than West Germans (81%-86%), some perceived East vs. West German
equal status between East Germans and West Germans (13%- categorization 3.91 1.05 2.69 1.26 8.90a .001
19%) and virtually no one perceived a superior status of East
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Change in categorization
level" -0.14 1.38 0.92 1.21 6.82 .001
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Germans compared with West Germans. The West German sample


Public protest 2.52 0.80 1.69 0.64 9.29a .001
mirrors the results found in the East German sample: Most West 2.69 0.90 2.22 0.63 4.77a .001
Xenophobia
Germans perceived East Germans as having lower status (79%),
some perceived status equality (19%), and nearly no one perceived Note. East German n = 109; West German n = 183.
a
East Germans as superior to West Germans (2%). These results Standard deviations were significantly different for these variables.
b
confirm the assumption of a perceived status difference between Difference value of German and East German categorization or German
and West German categorization, respectively.
East Germans and West Germans that represents a perception
shared by both East and West Germans.
In a second step, we compared the mean structure of variables
between the East German sample and the West German sample. Germans did, the difference was significantly stronger for public
There are several meaningful differences between East Germans protest. This is because public protest does not seem to be a
and West Germans (see Table 5): As expected, East Germans had reasonable option for West Germans.
significantly more resentment compared with West Germans. Al- In a last step, we compared the correlations of variables of East
though East Germans and West Germans did not differ in the Germans and West Germans. For this comparison, we presumed
salience of the German categorization, there is a huge difference in the causal structure revealed in the panel analysis and applied it in
the salience of their subgroup level of categorization (as East both cross-sectional samples. We set up a structural equation
German or West German, respectively). Thus, there was a signif- model that assumed resentment as a predictor for both levels of
icant interaction between levels of categorization (common in- categorization and for both kinds of intergroup conflict. Moreover,
group vs. subgroup) and status (East Germans vs. West Germans), self-categorizations at different levels were predictors for inter-
F(l, 290) = 46.51, p < .001. For West Germans, the salience of group conflict. Self-categorizations at both levels were allowed to
the subgroup categorization was significantly lower than the in- correlate freely as well as both indicators of intergroup conflict,
clusive level of categorization (t = 10.24, p < .001). For East because there was no causal relationship between them. With a
Germans, in contrast, the salience of the subgroup categorization multiple sample analysis, we compared the correlations between
was descriptively stronger than the inclusive level of categoriza- variables for the East German and the West German sample. There
tion, however, this difference did not reach significance (t = 1.04, were several meaningful differences as well as similarities be-
p > .05). Finally, East Germans indicated stronger tendencies to tween both samples. First, the multiple-group analysis replicated
public protest as well as xenophobia compared with West Ger- the differences in the variances of subgroup categorization, public
mans. We found a significant interaction between both kinds of protest, and xenophobia (see Table 6). Resentment was, in both
intergroup conflict and East and West Germans, F(l, samples, equally strongly related to the subgroup categorization
290) = 12.24, p < .001: Whereas East Germans generally indi- (East German and West German), to public protest, and to xeno-
cated stronger inclinations for both intergroup conflicts than West phobia.3 However, only the East German sample showed a signif-
icant relation between resentment and the inclusive categorization
level (j3 = .21, p < .05), whereas in the West German sample the
relation was nonsignificantly negative (/3 = - . 0 8 , p < .10). In
Table 4 both samples, the German level of categorization was significantly
Perceived Status Differences Between East Germans and West
Germans
3
East Germans West For the relation between resentment and public protest the possibility
Germans of bidirectional causation could not be ruled out. Thus, size of this path
East German status 1995 1996 1997 (1996) may be overestimated to some degree. Following Smith (1982), we applied
vs. a nonhierarchical regression analysis to the relation between public protest
West German status and resentment. That is, both causal directions were specified in one model.
Results showed that the influence of resentment on public protest was
Lower 91 86 88 81 91 83 144 79 attenuated but remained significant ((3 = .39, t = 2.03), whereas the
Equal 14 13 21 19 16 15 35 19 influence of public protest on resentment did not reach significance (£S =
Superior 1 1 2 2 4 2
.22, f = .93).
RECATEGORIZATION AND INTERGROUP CONFLICT 1099

Table 6 of categorization, as well as a carry over of intergroup conflict


Comparison of East Germans and West Germans Concerning between subgroups to a conflict between the common in-group and
Determinants of Intergroup Conflict at Different Levels of a new out-group.
Categorization (Subgroup or Inclusive Level)
Levels of Intergroup Conflict
Variable Resentment German Subgroup8 Public protest
There is clear evidence that self-categorizations at each level of
Resentment inclusiveness (i.e., as East German or as German) are causally
German .21*/-.08
Subgroup8 .16** .22*1.50*** related to a specific line of intergroup conflict: Categorization as
Public protest .55*** .00 .09t East German enhances public protest against the status advantage
Xenophobia .24*** .59***7.22*** .08 .171.39*** of West Germans. At the same time, categorization as German
fosters xenophobia. In addition, East German categorization neg-
Note. The coefficient in each cell represents the common solution for the
East German and the West German sample. In the cases of significant
atively influences xenophobia, whereas no such influence could be
differences between samples, the first number indicates the solution for the observed from the level of common in-group (i.e., German) to
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

East German sample and the second number the solution for the West public protest. This confirms our suggestion that different levels of
This document is copyrighted by the American Psychological Association or one of its allied publishers.

German sample. Coefficients in italics represent simple correlations. categorization are distinct causal factors for intergroup conflict.
Model fit is ^ ( 8 , N = 109, 183) = 5.87, p = .66; root mean square error Moreover, a salient self-categorization at the subgroup level re-
of approximation = .00; Akaike's information criteria = 49.87(60); non-
normed fit index = 1.00. duces the intergroup conflict at the common in-group level of
a
The subgroup categorization is either East German or West German. categorization. However, the reverse relation between levels of
t p < . 1 0 . * p < . 0 5 . * * p < . 0 1 . ***/><.001. categorization and intergroup conflict was not found, because the
German level of categorization does not causally relate to public
protest. This may be interpreted as evidence that contradicts the
related to xenophobia and nonsignificantly to public protest (/3 = suggestion that recategorization of intergroup relations at the sub-
.00). However, the positive relation between German categoriza- group level of categorization is beneficial. Nonetheless, comparing
tion and xenophobia was significantly different for the East Ger- the relative impact of categorization levels at each measurement
man sample (/3 = .59, p < .001) compared with the West German point reveals that a stronger relative preference of German cate-
sample (/3 = .22, p < .001). East German categorization was gorization is positively related to xenophobia whereas the relative
marginally significantly related to public protest (/3 = .09, p < .10) preference for East German categorization correlates with public
and nonsignificantly to xenophobia (/3 = .08).4 Both samples protest. Thus, stronger salience of subgroup categorization com-
differed in the correlation between both levels of categorization as pared with the common in-group level leads to an intergroup
well as in the correlation between intergroup conflicts: For West conflict on the subgroup level (i.e., public protest), whereas stron-
Germans, inclusive categorization (i.e., German) and subgroup ger salience of the common in-group compared with subgroup
categorization (as West German) was significantly more strongly categorization enhances intergroup conflict at the inclusive level
correlated (r = .50, p < .001) than the correlation of levels of (i.e., xenophobia). In conclusion, recategorization as a means to
categorization for East Germans (r = .22, p < .05). Finally, for reduce conflict may be a two-edged method, under certain condi-
West Germans public protest and xenophobia were significantly tions: while reducing conflict at the subgroup level, it may foster
more strongly correlated (r = .39, p < .001) than for East Germans conflict between the common in-group level and a new out-group.
(r = .17, ns).
Resentment is causally related to the categorization as German
and xenophobia. A categorization of German influences xenopho- Prediction of Intergroup Conflict
bia. According to Baron and Kenny (1986), these two conditions The predictors of identity-management strategies highlighted by
establish the possibility for a mediation of some effects of resent- relative deprivation theory and social identity theory (i.e., resent-
ment on xenophobia by categorization as German. Results of the ment, group efficacy, and in-group identification) show mixed
multiple-group analyses reveal that resentment has a significantly results concerning their causal relation to intergroup conflict.
indirect effect on xenophobia in the East German sample, /3 Whereas in-group identification and group efficacy are not caus-
(indirect) = . 1 1 , / = 2.25. However, in the West German sample, ally related to indicators of intergroup conflict, resentment is a
no such indirect effect could be observed, /3 (indirect) = —.01, particularly strong causal agent. Resentment enhances both levels
/ = 0.19, because the relation between resentment and German of categorization (i.e., East German and German). One may spec-
categorization was not significant. ulate that different processes may strengthen both levels of cate-
gorization: The East German level of categorization may be caus-
Discussion ally influenced by resentment because the perceived difference
between East and West Germans makes clear the common fate
This article focused on two issues of recategorization not ad-
shared by East Germans. The German level of categorization may
dressed in the common in-group identity model (S. L. Gaertner,
Dovidio, Nier, et al., 1999): First, recategorization may establish a
new line of conflict between the common in-group and a new 4
The relation between East German categorization and public protest
out-group. Second, the intergroup relation on the novel superordi- was low. This is because of the close relation between resentment and East
nate level may be influenced by determinants of the intergroup German categorization as well as between resentment and public protest.
conflict between the subgroups. Results of the present study show Removing resentment from the analyses revealed a much higher path
that specific intergroup conflicts could be found at different levels coefficient for both samples (0 = . 18, f = 2.92).
1100 KESSLER AND MUMMENDEY

be enhanced as a strategic means that conforms to the political differ between East and West Germans, except in the strength of
goals of a German merger, yet makes the unjust treatment of East the German categorization on xenophobia. This relation is signif-
Germans more obvious because if all East Germans are Germans icant for East Germans and West Germans; however, it is signif-
as well, then their unequal status is not justified (Wenzel, 2000). icantly stronger for East Germans. Moreover, the influence of
In addition, resentment enhances both intergroup conflicts. resentment on the subgroup level of categorization as well as on
Thus, strong resentment about the relationship between East and both kinds of intergroup conflict is equal. For West Germans,
West Germans fosters inclinations of public protest (i.e., demon- though, there is a nonsignificant negative relation between resent-
strations on the streets). This is in line with suggestions made by ment and German categorization; recategorization at the common
theories explaining collective actions (e.g., Klandermans, 1997). In identity level does not mediate the influence of resentment on
addition, resentment also strengthens xenophobic attitudes. Here, xenophobia.
the question emerges about the influence of resentment on the Remarkably, the positive correlation between subgroup level
conflict at a higher level of categorization. First, there is a direct and superordinate level of categorization is much stronger for
influence between resentment and xenophobia. This resembles the West Germans relative to East Germans. This indicates that for
scapegoat argument: A conflict with a superior party (individual or West Germans the subgroup level of categorization is basically
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

group) is resolved by searching for some weaker target that could identical to the level of common identity (e.g., being a West
This document is copyrighted by the American Psychological Association or one of its allied publishers.

more easily be overpowered (e.g., Allport, 1954; Staub, 1985). German is being a German). As a consequence, both levels of
Within the present field study, some East Germans may see West intergroup conflict are also less different for West Germans com-
Germans as strong opponents who cannot be easily vanquished. pared with East Germans.
Consequently, they consider another target that seems to be
weaker, like the minority of foreigners in Germany. Second, as the Concluding Remarks
mediation analysis reveals, scapegoating is not the whole story:
The influence of resentment on xenophobia is mediated by the It would be a misunderstanding to interpret this article as re-
categorization as German. This result closely fits the explanation jecting or jeopardizing the common in-group identity model; in-
of triggered displaced aggression (Marcus-Newhall, Pedersen, stead, our results suggest an extension of this model by broadening
Carlson, & Miller, 2000; Miller & Marcus-Newhall, 1997; Ped- the perspective of the common in-group. The new intergroup
ersen, et al., 2000): A perceived frustration that induces enduring context that emerges after successful recategorization may give
anger (e.g., ruminative thoughts; Martin & Tesser, 1989) may act rise to a new intergroup conflict. Moreover, recategorization may
as a primer for the perception of further provocations. Introducing not, under every circumstance, be a sign of reduced intergroup
a new line of categorization may function as a triggering event that conflict. A moderator hypothesis concerning the virtues of recat-
leads individuals to perceive this new out-group as an additional egorization could be suggested: On one hand, as Gaertner and
provocation, thereby enhancing conflict with this new out-group. colleagues (e.g., Gaertner, Dovidio, Rust, et al, 1999) showed,
Possibly in this case the same mechanism can be seen in intergroup favorable conditions of contact may be mediated by a common
contexts as well as in interpersonal relations. Recategorization is in-group identity, leading to a reduction of conflict. On the other
not always a sign of improving intergroup harmony (on different hand, emotional reactions to unfavorable conditions of contact
levels of categorization). On the contrary, recategorization at a (e.g., resentment about perceived lower status or perceived threat)
higher level may foster new forms of conflict and may transpose may be mediated by a common in-group identity leading to an-
one conflict to another level of categorization, at least under other level of categorization, thus enhancing and shifting inter-
certain conditions. We suggest that the most problematic feature in group conflict to other groups.
the contact between East and West Germans is the consensually
perceived lower status of East Germans. Although some favorable References
conditions of intergroup contact have been met, such as institu-
tional support, a common goal, and acquaintance potential, the Allport, G. W. (1954). The nature of prejudice. Reading, MA: Addison-
crucial variable, equal status between the groups, has not yet been Wesley.
established. The unequal status of two groups within a merger may Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable
be interpreted as a threat from the perspective of the lower status distinction in social psychological research: Conceptual, strategic, and
statistical considerations. Journal of Personality and Social Psychology,
group (in the present case, the East Germans). Within intergroup
51, 1173-1182.
situations in which an intergroup threat is perceived, recategori- Blanz, M., Mummendey, A., Mielke, R., & Klink, A. (1998). Strategic
zation as a common in-group may enhance rather than reduce responses to negative social identity: An empirical systematizau'on of
intergroup conflict by reopening a new line of conflict and shifting field data. European Journal of Social Psychology, 28, 697-729.
the original conflict to the new level of categorization. If one Bozdogan, H. (1987). Model selection and Akaike's Information Criterion
cannot be sure whether favorable conditions for contact are met, (AIC): The general theory and its analytical extensions. Psychometrica,
introducing a common in-group identity may lead to mixed 52, 345-370.
outcomes. Brewer, M. B. (1979). In-group bias in the minimal intergroup situation: A
cognitive motivational analysis. Psychological Bulletin, 86, 307-324.
Brewer, M. B. (1999). The psychology of prejudice: In-group love or
Comparison Between East Germans and West Germans out-group hate? Journal of Social Issues, 55, 429-444.
Brewer, M. B., & Miller, N. (1984). Beyond the contact hypothesis:
The comparison between the East German and the West German Theoretical perspectives on desegregation. In N. Miller & M. B. Brewer
samples reveals several similarities and differences. The relations (Eds.), Groups in contact: The psychology of desegregation (pp. 281—
between levels of categorization and intergroup conflict do not 302). Orlando, FL: Academic Press.
RECATEGORIZATION AND INTERGROUP CONFLICT
1101
Brown, R., & Wade, G. (1987). Superordinate goals and intergroup be- on group cohesiveness. European Journal of Social Psychology, 25,
haviour: The effect of role ambiguity and status on intergroup attitudes 159-177.
and task performance. European Journal of Social Psychology, 17, Hornsey, M. J., & Hogg, M. A. (2000a). Assimilation and diversity: An
131-142 integrative model of subgroup relations. Personality and Social Psychol-
Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation: Design ogy Review, 4, 143-156.
and analysis issues for field settings. Boston: Houghton Mifflin. Hornsey, M. J., & Hogg, M. A. (2000b). Subgroup relations: A comparison
Cook, T. D., Crosby, F., & Hennigan, K. M. (1977). The construct validity of mutual intergroup differentiation and common in-group identity mod-
of relative deprivation. In J. Sulls & R. Miller (Eds.), Social comparison els of prejudice reduction. Personality and Social Psychology Bulletin,
processes: Theoretical and empirical perspectives (pp. 307-333). Wash- 26, 242-256.
ington, DC: Hemisphere. Huo, Y. J., Smith, H. I., Tyler, T. R., & Lind, E. A. (1996). Superordinate
Crosby, F. (1982). Relative deprivation and working women. New York: identification, subgroup identification, and justice concerns: Is separat-
Oxford University Press. ism the problem; is assimilation the answer? Psychological Science, 7,
Dovidio, J. F., Gaertner, S. L., Isen, A. M., & Lowrance, R. (1995). Group 40-45.
representations and intergroup bias: Positive affect, similarity, and group Islam, M. R., & Hewstone, M. (1993). Intergroup attributions and affective
consequences in majority and minority groups. Journal of Personality
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

size. Personality and Social Psychology Bulletin, 21, 856-865.


This document is copyrighted by the American Psychological Association or one of its allied publishers.

Dovidio, J. F., Gaertner, S. L., & Validzic, A. (1998). Intergroup bias: and Social Psychology, 64, 936-950
Status, differentiation, and a common in-group identity. Journal of Joreskog, K. G., & Sorbom, D. (1993). USREL 8: Structural equation
Personality and Social Psychology, 75, 109-120. modeling with the SIMPL1S command language. Hillsdale, NJ: Erlbaum.
Folger, R. (1987). Reformulating the preconditions of resentment: A ref- Kenny, D. A. (1975). Cross-lagged panel correlation: A test for spurious-
erent cognition model. In J. C. Masters & W. P. Smith (Eds.), Social ness. Psychological Bulletin, 82, 887-903
comparison, social justice, and relative deprivation (pp. 183-215). Kenny, D. A. (1979). Correlation and causality. New York: Wiley.
London: Erlbaum. Klandermans, B. (1997). The social psychology of protest. Oxford, En-
Gaertner, L., & Insko, C. A. (2000). Intergroup discrimination in the gland: Blackwell.
minimal group paradigm: Categorization, reciprocation, or fear? Journal Kramer, R. M., & Brewer, M. B. (1984). Effects of group identity on
of Personality and Social Psychology, 79, 77—94. resource utilization in a simulated commons dilemma. Journal of Per-
sonality and Social Psychology, 46, 1044-1057.
Gaertner, S. L., Dovidio, J. F., Anastasio, P. A., Bachman, B. A., & Rust,
Lalonde, R. N., & Cameron, J. E. (1994). Behavioral responses to discrim-
M. C. (1993). The common in-group identity model: Recategorization
ination: A focus on action. In M. P. Zanna & J. M. Olson (Eds.), The
and the reduction of in-group bias. In W. Stroebe & M. Hewstone (Eds.),
psychology of prejudice: The Ontario symposium (Vol. 7, pp. 257-288).
European review of social psychology (Vol. 4, pp. 1-26). London:
Hillsdale, NJ: Erlbaum.
Wiley.
Marcus-Newhall, A., Pedersen, W. C , Carlson, M., & Miller, N. (2000).
Gaertner, S. L., Dovidio, J. F., Nier, J. A., Ward, C. M., & Banker, B.
Displaced aggression is alive and well: A meta-analytic review. Journal
(1999). Across cultural divides: The value of a superordinate identity. In
of Personality and Social Psychology, 78, 670-689.
D. A. Prentice & D. T. Miller (Eds.), Cultural divides: Understanding
Martin, L. L., & Tesser, A. (1989). Toward a motivational and structural
and overcoming group conflict (pp. 173-212). New York: Russell Sage
theory of ruminative thought. In J. S. Uleman & J. A. Bargh (Eds.),
Foundation.
Unintended thought (pp. 306-326). New York: Guilford Press.
Gaertner, S. L., Dovidio, J. F., Rust, M. C , Nier, J. A., Banker, B., Ward,
Miller, N., & Marcus-Newhall, A. (1997). A conceptual analysis of dis-
C. M., Mottola, G. R., & Houlette, M. (1999). Reducing intergroup bias:
placed aggression. In R. Ben-An & Y. Rich (Eds.), Enhancing education
Elements of intergroup cooperation. Journal of Personality and Social in heterogeneous schools: Theory and application (pp. 69-108). Ramat-
Psychology, 76, 388-402. Gan, Israel: Bar-Han University Press.
Gaertner, S. L., Mann, J. A., Dovidio, J. F., Murrell, A. J., & Pomare, M. Mummendey, A., Kessler, T., Klink, A., & Mielke, R. (1999). Strategies to
(1990). How does cooperation reduce intergroup bias? Journal of Per- cope with negative social identity: Predictions by social identity theory
sonality and Social Psychology, 59, 692-704. and relative deprivation theory. Journal of Personality and Social Psy-
Gaertner, S. L., Mann, J. A., Murrell, A., & Dovidio, J. F. (1989). Reducing chology, 76, 229-245.
intergroup bias: The benefits of recategorization. Journal of Personality Mummendey, A., Klink, A., Mielke, R., Wenzel, M., & Blanz, M. (1999).
and Social Psychology, 57, 239-249. Socio-structural characteristics of intergroup relations and identity man-
Gaertner, S. L., Rust, M. C , Dovidio, J. F., Bachman, B. A., & Anastasio, agement strategies: A field study in East Germany. European Journal of
P. A. (1996). The contact hypothesis: The role of a common in-group Social Psychology, 29, 259-285.
identity on reducing bias among majority and minority members. In J. L. Mummendey, A., & Schreiber, H. -J. (1983). Better or just different?
Nye & A. M. Brower (Eds.), What's social about social cognition? Positive social identity by discrimination against, or by differentiation
Research on socially shared cognition in small groups (pp. 230-260). from out-groups. European Journal of Social Psychology, 13, 389-397.
London: Sage. Pedersen, W. C , Gonzales, C , & Miller, N. (2000). The moderating effect
Harenberg, W. (1991). Vereint und verschieden [Unified and different]. of trivial triggering provocation on displaced aggression. Journal of
Spiegel Spezial, 1, 10-23. Personality and Social Psychology, 78, 913-927.
Hewstone, M., & Brown, R. (1986). Contact is not enough: An intergroup Pettigrew, T. F., & Meertens, R. W. (1995). Subtle and blatant prejudice in
perspective on the contact hypothesis. In M. Hewstone & R. Brown western Europe. European Journal of Social Psychology, 25, 57-75.
(Eds.), Contact and conflict in intergroup encounters (pp. 1-44). Ox- Piliavin, J. A., Dovidio, J. F., Gaertner, S. L., & Clark, R. D. (1981).
ford, England: Blackwell. Emergency intervention. New York: Academic Press.
Hogg, M. A., & Hains, S. C. (1996). Intergroup relations and group Rogosa, D. (1980). A critique of cross-lagged correlation. Psychological
solidarity: Effects of group identification and social beliefs on deper- Bulletin, 88, 245-258.
sonalized attraction. Journal of Personality and Social Psychology, 70, Runciman, W. G. (1966). Relative deprivation and social justice: A study
295-309. of attitudes to social inequality in twentieth-century England. Berkeley,
Hogg, M. A., Hardi, S. C , & Reynolds, K. J. (1995). Prototypical simi- CA: University of California Press.
larity, self-categorization and depersonalized attraction: A perspective Seta, C. E., Seta, ;. J., & Culver, J. (2000). Recategorization as a method
1102 KESSLER AND MUMMENDEY

for promoting intergroup cooperation: Group status matters. Social Cog- Terry, D. J., & Callan, V. J. (1998). In-group bias in response to an
nition, 18, 354-376. organizational merger. Group Dynamics, 2, 67-81.
Smith, E. R. (1982). Beliefs, attributions, and evaluations: Nonhierarchical Turner, J. C , Hogg, M. A., Oakes, P. J., Reicher, S., & Wetherell, M.
models of mediation in social cognition. Journal of Personality and (Eds.) (1987). Rediscovering the social group: A self-categorization
Social Psychology, 43, 248-259. theory. Oxford, England: Blackwell.
Smith, E. R. (1993). Social identity and social emotions: Toward new Wenzel, M. (2000). Justice and identity: The significance of inclusion for
conceptualizations of prejudice. In D. M. Mackie & D. L. Hamilton perceptions of entitlement and the justice motive. Personality and Social
(Eds.), Affect, cognition, and stereotyping: Interactive processes in Psychology Bulletin, 26, 157-176.
group perception, (pp. 297-315). San Diego, CA: Academic Press. Wilder, D. A. (1986). Social categorization: Implications for creation and
Spiegel-Redaktion. (1995). Stolz aufs eigene Leben. [Proud of our own reduction of intergroup bias. In L. Berkowitz (Ed.), Advances in exper-
life]. Spiegel Nr., 27/95, 40-52. imental social psychology (Vol. 1, pp. 291-355). Orlando, FL: Aca-
Staub, E. (1985). The psychology of perpetrators and bystanders. Political demic Press.
Psychology, 6, 61-85. Yamagishi, T., & Kiyonari, T. (2000). The group as the container of
Struch, N., & Schwartz, S. H. (1989). Intergroup aggression: Its predictors generalized reciprocity. Social Psychology Quarterly, 63, 116-132.
Zaccaro, S. J., Blair, V., Peterson, C , & Zazanis, M. (1995). Collective
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

and distinctness from in-group bias. Journal of Personality and Social


efficacy. In J. E. Maddux (Ed.), Self-efficacy, adaptation, and adjust-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Psychology, 56, 364-373.


Tajfel, H., Billig, M. G., Bundy, R. F., & Flament, C. (1971). Social ment: Theory, research, and application, (pp. 305-328). New York:
categorization and intergroup behavior. European Journal of Social Plenum Press.
Psychology, I, 149-177.
Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup Received October 16, 2000
behavior. In S. Worchel & W. G. Austin (Eds.), Psychology of inter- Revision received June 27, 2001
group relations (2nd ed., pp. 7-24). Chicago: Nelson-Hall. Accepted July 5, 2001 •

Wanted: Your Old Issues!


As APA continues its efforts to digitize journal issues for the PsycARTICLES
database, we are finding that older issues are increasingly unavailable in our inventory.
We are turning to our long-time subscribers for assistance. If you would like to donate
any back issues toward this effort (preceding 1982), please get in touch with us at
journals@apa.org and specify the journal titles, volumes, and issue numbers that you
would like us to take off your hands. (Your donation is of course tax deductible.)

You might also like