Nuestra Culpa: Collective Guilt and Shame as Predictors of Reparation for Historical Wrongdoing

Abstract

Three studies examined the hypothesis that collective guilt and shame have different consequences for reparation. In 2 longitudinal studies, the ingroup was nonindigenous Chileans (Study 1: N = 124/120, lag = 8 weeks; Study 2: N = 247/137, lag = 6 months), and the outgroup was Chile's largest indigenous group, the Mapuche. In both studies, it was found that collective guilt predicted reparation attitudes longitudinally. Collective shame had only cross-sectional associations with reparation and no direct longitudinal effects. In Study 2, collective shame moderated the longitudinal effects of collective guilt such that the effects of guilt were stronger for low-shame respondents. In Study 3 (N = 193 nonindigenous Chileans), the cross-sectional relationships among guilt, shame, and reparation attitudes were replicated. The relationship between shame and reparation attitudes was mediated by a desire to improve the ingroup's reputation.

Full text

Nuestra Culpa: Collective Guilt and Shame as Predictors of Reparation for
Historical Wrongdoing
Rupert Brown
University of Sussex
Roberto Gonza´lez
Pontificia Universidad Cato´lica de Chile
Hanna Zagefka
University of London
Jorge Manzi
Pontificia Universidad Cato´lica de Chile
Sabina C
ˇ
ehajic´
University of Sussex
Three studies examined the hypothesis that collective guilt and shame have different consequences for
reparation. In 2 longitudinal studies, the ingroup was nonindigenous Chileans (Study 1: N ⫽ 124/120,
lag ⫽ 8 weeks; Study 2: N ⫽ 247/137, lag ⫽ 6 months), and the outgroup was Chile’s largest indigenous
group, the Mapuche. In both studies, it was found that collective guilt predicted reparation attitudes
longitudinally. Collective shame had only cross-sectional associations with reparation and no direct
longitudinal effects. In Study 2, collective shame moderated the longitudinal effects of collective guilt
such that the effects of guilt were stronger for low-shame respondents. In Study 3 (N ⫽ 193 nonindig-
enous Chileans), the cross-sectional relationships among guilt, shame, and reparation attitudes were
replicated. The relationship between shame and reparation attitudes was mediated by a desire to improve
the ingroup’s reputation.
Keywords: collective guilt, collective shame, reparation
Historians of the second half of the 20th century will doubtless
come to record that it was marked by several violent intergroup
conflicts around the globe. The wars in Southeast Asia, the long
struggle against apartheid in South Africa, the genocides in
Rwanda and the former Yugoslavia, and the continuing struggles
of indigenous peoples around the world for the restitution of their
homelands and the preservation of their cultures are just a few of
the many social conflicts that have cost millions of lives over the
past 50 years. Although such a record of bloodshed is regrettably
hardly a novel phenomenon, one feature of at least some of these
conflicts is new and is attracting increasing attention from the
social scientific community. That new aspect is the emergence of
political debate about people’s felt culpability for injustices per-
petrated by their group in those conflicts in the past and the extent
to which some restitution should be made to the victims of those
injustices in the form of public apologies, memorials, or material
reparations (Barkan, 2000; Buruma, 1994; Steele, 1990). This
debate has stimulated theorizing and research in social psychology
into the phenomenon of self-conscious emotions and their conse-
quences for intergroup relations (Branscombe & Doosje, 2004;
Leach, Snider, & Iyer, 2002). In this article, we seek to make a
contribution to this developing research area by examining the
effects of experienced collective ingroup guilt and shame on the
desire to make reparations to a “victim” outgroup. We do this in
the naturalistic context of relations between the nonindigenous
majority and one of the indigenous minority groups in Chile.
Reactions to Wrongdoing: The Distinction Between Guilt
and Shame
In the domain of interpersonal relations, there has been much
research into individuals’ emotional reactions to the realization
that they have done something in contravention of some relevant
personal or normative standards (Tangney & Fischer, 1995). An
important point of departure for this work is Lewis’s (1971)
distinction between the self-conscious emotions of guilt and of
shame. For Lewis, both of these reactions involve negative affect,
but the focus of the experience differs: In guilt, the main emphasis
is on the wrongdoing and its consequences for the other (“I did this
bad thing to X, who suffered as a result”), whereas shame is
marked more by a focus on the negative implications of that
Rupert Brown and Sabina C
ˇ
ehajic´, Department of Psychology, Univer-
sity of Sussex, Brighton, United Kingdom; Roberto Gonza´lez and Jorge
Manzi, Escuela de Psicologı´a, Facultad de Ciencias Sociales, Pontificia
Universidad Cato´lica de Chile, Santiago, Chile; Hanna Zagefka, Psychol-
ogy Department, Royal Holloway, University of London, London, United
Kingdom.
This research was supported by Grant 1020954 from the Chilean Na-
tional Funding for Scientific and Technological Research program and a
grant from the British Council allocated to Roberto Gonza´lez. We grate-
fully acknowledge the perceptive comments of Colin Leach and Anja
Zimmermann on an earlier version of this article.
Correspondence concerning this article should be addressed to Rupert
Brown, Department of Psychology, Pevensey 1, University of Sussex, Falmer,
Brighton, BN1 9QH United Kingdom. E-mail: r.brown@sussex.ac.uk
Journal of Personality and Social Psychology Copyright 2008 by the American Psychological Association
2008, Vol. 94, No. 1, 75–90 0022-3514/08/$12.00 DOI: 10.1037/0022-3514.94.1.75
75
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

wrongdoing for one’s self-concept (“I did this bad thing to X, and
therefore I am [seen to be] a bad person”). Because both emotions
are somewhat aversive, people are motivated to alleviate them.
Lewis speculated that this alleviation would take different forms.
Guilt, with its focus on the misdeed, should be more likely to lead
to some form of restitution to the victim (e.g., apology, reparation);
shame, with its focus on the self, should lead to withdrawal from
or avoidance of the situation that gave rise to it.
These ideas provoked considerable empirical research into the
role of guilt and shame in interpersonal relations (e.g., Baumeister,
Stillwell, & Heatherton, 1994; Tangney & Fischer, 1995). Early
on, it was recognized that lay usage of the words guilt and shame
often treats them as synonyms (Shaver, Schwartz, Kirson, &
O’Connor, 1987). Nevertheless, even if they are interchangeable in
ordinary language use, there is growing consensus that, though
positively correlated, the two emotions have a different underlying
psychology and may sometimes lead to different outcomes. Shame
generally seems to be a more intensely and aversively experienced
emotion—associated with anger, a desire to hide, and self-oriented
counterfactuals—whereas guilt has been observed to correlate
more with empathy and action-oriented counterfactuals
(Niedenthal, Tangney, & Gavanski, 1994; Tangney, 1991; Tang-
ney, Miller, Flicker, & Barlow, 1996; Tangney, Wagner, Fletcher,
& Gramzow, 1992). Notably, though, in view of Lewis’s (1971)
prediction, there is not always a greater tendency to make repara-
tions associated with guilt than there is associated with shame
(Roseman, West, & Schwartz, 1994; Tangney et al., 1996).
One debated issue has been how to conceptualize shame. Some
have followed Lewis’s (1971) lead in regarding shame as the
emotion that follows from a negative self-perception, the sense that
one’s character is flawed in some respect (e.g., Tangney, 1991).
Others, though, have linked shame more to the damage to one’s
reputation that may be caused by the public exposure of one’s
misdeeds (R. H. Smith, Webster, Parrott, & Eyre, 2002). Although
these are subtly different conceptualizations of shame, it is likely
that the two components are often closely intertwined. As Mead
(1934) noted many years ago, one’s reputation in the eyes of others
is a major determinant of one’s self-concept. However, though
both accounts predict that shame should still lead to avoidance, it
is possible to imagine circumstances in which a temporary coping
strategy for dealing with the “reputational” aspect of shame could
be to make some kind of public form of restitution if, in so doing,
one’s public image could thereby be enhanced. Later in this article,
we explore this idea empirically.
One other issue concerns the possibility of whether both guilt
and shame can be experienced simultaneously and whether they
may interact. Lewis (1971, p. 42) suggested that both emotions can
be felt at the same time, and others have concurred with this
(Tangney, 1991; Tangney et al., 1992). Tangney et al. (1992) also
introduced the concept of shame-free guilt and guilt-free shame,
the idea that some individuals may be prone to experience guilt but
little shame, or vice versa. Tangney et al. (1992) investigated this
by using partial correlations among their personality measures of
guilt and shame (e.g., guilt-proneness controlling for shame-
proneness); an alternative approach could have been to explore
whether they interact. In other words, does guilt have stronger
associations with reparative tendencies for those who show little
shame? If shame has avoidance consequences on social relations
generally, it seems plausible to suppose that at high levels it might
thereby inhibit the predicted prosocial consequences of guilt, thus
suppressing the positive link between guilt and reparation. We
return to this issue later.
In summary, then, there is some evidence that when people
perceive themselves to have behaved illegitimately, they can ex-
perience guilt or shame (or both) and that, depending on which
predominates, rather different outcomes can occur. Although the
evidence is not unequivocal, guilt seems to be more closely con-
nected to prosocial orientations and shame to stronger negative
self-evaluations, reputational concerns, and various kinds of avoid-
ance behavior.
Collective Guilt and Collective Shame: Conceptual Issues
The discussion above is concerned with the reactions of indi-
viduals to the knowledge that they themselves have transgressed in
some way, usually toward another individual. However, emotions
are not restricted to such interpersonal situations. They can also be
felt in response to other people’s misdeeds, particularly if there is
some psychological connection with those others—for example, if
they belong to the same group (E. R. Smith, 1993). Furthermore,
the victims of those misdeeds may be whole categories of people
and not just isolated individuals. In short, guilt and shame can have
a collective component.
The first social psychological study of collective guilt was by
Doosje, Branscombe, Spears, and Manstead (1998). Drawing on
social identity and self-categorization theories (Tajfel & Turner,
1986; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987), they
proposed that, because group memberships and their associated
category attributes can become internalized into an individual’s
self-concept, it is plausible to assume that the actions of other
ingroup members would have affective implications for that indi-
vidual. Just as the glorious exploits of some members of our
ingroup can lead others of us to bask in reflected glory (Cialdini et
al., 1976), so too can the immoral actions of other ingroup mem-
bers generate feelings in us of “vicarious” remorse or regret
(Lickel, Schmader, Curtis, Scarnier, & Ames, 2005), leaving us
cringing in shared blame, as it were.
Subsequent theorizing has sought to explicate these self-
conscious collective emotions. Following Weiner (1995), Leach et
al. (2002) and Branscombe, Slugoski, and Kappen (2004) argued
that collective guilt arises mainly when group members perceive
that they have some responsibility for their ingroup’s misdeeds or
the subsequent repercussions of those misdeeds. In common with
the work reviewed in the previous section, most commentators
have agreed that collective guilt should lead ingroup members to
want to make reparations to the outgroup (Branscombe, Slugoski,
& Kappen, 2004; Lickel, Schmader, & Barquissau, 2004).
In contrast, collective shame is thought more likely to be in-
voked when people do not feel in control of their (ingroup’s)
actions and when the ingroup is exposed as being weak or incom-
petent (Branscombe, Slugoski, & Kappen, 2004; Lickel et al.,
2004). Conceptualizations of collective shame contain the same
duality that we noted in discussing individual shame. Branscombe,
Slugoski, and Kappen (2004) emphasized the reputational aspect:
“[C]ollective shame involves being publicly exposed as incompe-
tent, not being in control, weak and potentially even disgusting in
the eyes of others” (p. 29). Lickel et al. (2004) added the idea of
shame being associated with some negative ingroup “essence”
76
BROWN, GONZA
´
LEZ, ZAGEFKA, MANZI, AND C
ˇ
EHAJIC
´
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

(Haslam, Rothschild, & Ernst, 2000; Yzerbyt, Rocher, & Schad-
ron, 1997): “[C]ollective shame stems from perceiving that the
actions of the ingroup confirm or reveal a flawed aspect of one’s
social identity . . . (and) . . . implicate something about the very
nature of who they are” (pp. 42–43). As in the case of individual
emotions, it seems likely that the negative essence and the repu-
tational components of shame will often elide each other. More-
over, both viewpoints predict that collective shame, because of the
implied threat to the ingroup’s image, should lead to avoidance of
the events that gave rise to the feelings in the first place or even to
hostility toward the outgroup. In sum, there is some consensus that
the primary antecedent of action tendencies toward reparation
should be collective guilt rather than collective shame.
We do not dissent from this hypothesis, especially as it concerns
durable consequences of these two dysphoric emotions. However,
we believe that in the short term, both collective guilt and shame
can have rather similar and positive associations with reparation
attitudes. One reason for this lies in the “reputational” aspect of
shame. Insofar as ingroup members can see an immediate possi-
bility for improving their group’s image in the eyes of others,
endorsing restitutive policies could also be seen to be an expedient
strategy for alleviating shame. Over time, though, we suspect that
there are likely to be more opportunities to deny feelings and
cognitions about the shame-inducing situation, with a consequent
reduction in tendencies to repair.
What has also not been investigated hitherto, either theoretically
or empirically, is whether the emotions of collective guilt and
shame might interact. As noted earlier, it is plausible to suppose
that group members could feel both emotions simultaneously—
certainly, they are usually correlated positively in the interpersonal
domain (Shaver et al., 1987; Tangney, 1991). If this assumption is
correct, what might be the consequences of feeling guilt with or
without shame? Given the above argument, it is possible to predict
that a relatively high level of shame, with its likely “avoidance”
implications, might inhibit the “normal” prosocial consequences of
guilt. Thus, an optimal situation, at least from the perspective of
promoting positive intergroup outcomes, should be a combination
of relatively high levels of guilt coupled with relatively low levels
of shame.
Collective Guilt and Collective Shame: Prior Research
What empirical research has investigated collective guilt and
shame? We focus first on attempts to measure the two concepts.
Then we discuss research that has examined the consequences of
collective guilt and shame for intergroup attitudes generally and
for a desire to make restitution to the outgroup in particular.
One widely used measure of collective guilt was published by
Branscombe et al. (2004). This consists of five items, four of
which refer to expressions of regret or guilt over the ingroup’s
negative actions toward other groups, and one of which refers to a
desire to make reparations for any damage caused (see also Roc-
cas, Klar, & Liviatan, 2004). Collective shame is not assessed by
this measure. Although variants of this measure have proven useful
in several different intergroup contexts (e.g., Doosje et al., 1998;
Pedersen, Beven, Walker, & Griffiths, 2004), we believe it is
preferable not to include desire to make reparations in the measure
of guilt itself because this is hypothesized to be a consequence of
guilt rather than an integral component of it. It also precludes a
study of the circumstances, if any, under which shame might
predict reparation. Swim and Miller (1999) kept separate their
collective guilt scale from their measure of reparation, but again,
their scale did not measure collective shame. One study that did
attempt to measure both emotions was by Lickel et al. (2005).
Participants were asked to recall an event in which they felt guilty
or ashamed for the actions of someone else (e.g., family member,
ethnic ingroup member) and then to record the emotions that this
event evoked. The emotion words guilty, regret, and remorse
tended to load together on the same factor and distinctly from the
words ashamed, embarrassed, disgraced, and humiliated. Iyer,
Schmader, and Lickel (2007) used a similar technique. However,
Iyer, Leach, and Crosby (2003) found that ratings of emotions felt
when thinking about racial discrimination did not separate so
clearly into guilt and shame factors. Instead, emotions like guilty,
ashamed, regretful, and blameworthy tended to load together into
what Iyer et al. (2003) labeled a guilt factor, whereas sympathetic,
compassionate, and empathetic loaded together into a sympathy
factor (see also Leach, Iyer, & Pederson, 2006).
This inconsistency across studies probably reflects the ambigu-
ity in lay usage of the terms guilt and shame, and more theoreti-
cally grounded items are needed to distinguish the two concepts. In
any event, most existing measures have either not attempted or not
been able to clearly differentiate collective shame from collective
guilt, and some measures of collective guilt conflate guilt and
reparation tendencies. In the studies presented in this article, we
report on our efforts to develop distinct and reliable measures for
collective guilt and shame and then assess the unique ability of
each to predict desire to make reparations to an outgroup.
Leaving aside the question of measures used, research has
consistently found reliable associations between collective guilt
and indicators of intergroup reparation. Doosje et al. (1998, Study
2) presented Dutch student participants with brief historical ac-
counts of the Dutch colonial treatment of Indonesia. Despite the
fact that this sample of student participants could have had no
direct involvement with their country’s past imperialist misadven-
tures, they still reported moderate levels of guilt and desire to make
compensation, and these two measures were positively correlated.
Most other work in this domain has relied on cross-sectional
correlational designs and has investigated only collective guilt.
Swim and Miller (1999) found that White guilt consistently pre-
dicted reparations to African Americans in the form of favorable
attitudes toward affirmative action policies and less prejudice
toward African Americans. This was confirmed by Iyer et al.
(2003), who also found that guilt was only reliably associated with
what they called compensatory forms of affirmative action (e.g.,
special entrance quotas to university, positive job discrimination)
and not with equal opportunities policies (e.g., increased efforts to
attract more minority applicants to job openings). Elsewhere,
McGarty et al. (2005) found that collective guilt felt by nonindig-
enous Australians about the treatment of indigenous Australians
was associated with support for official government apologies to
the indigenous community.
To date, little work has attempted to disentangle the potentially
divergent intergroup consequences of collective guilt and collec-
tive shame. Harvey and Oswald (2000) attempted to induce guilt
and shame experimentally in White Americans. However, their
manipulation had similar effects on both emotions, and the latter
showed similar relationships to the main dependent measure, sup-
77
COLLECTIVE GUILT AS A PREDICTOR OF REPARATION
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

port for Black programs. Lickel et al. (2005) studied the vicarious
emotions aroused by a recollection of the misdeeds of friends,
family members, or ethnic groups. In line with Lickel at al.’s
(2004) model, although shame and guilt were positively correlated
with each other, the former emotion was correlated with motives to
distance one’s self from the perpetrator or the situation, and the
latter was correlated with motives to apologize and repair. More
recently, however, Brown and C
ˇ
ehajic´ (in press) reported two
cross-sectional studies in the former Yugoslavia in which both
collective guilt and shame were positively related to reparation
attitudes.
In some recent research set in the context of the current Iraq war,
Iyer et al. (2007) examined correlates of action intentions to
compensate the Iraqi people for damage caused by the invasion.
They found that neither of their measures of general guilt or shame
(about the situation in Iraq) predicted compensatory intentions
once anger (about the same situation) was controlled. In fact, in
both of their studies, Iyer et al. (2007) found the latter emotion to
be the most potent correlate of compensation intentions. Leach et
al. (2006) also examined the roles of guilt and anger in the context
of political action in Australia over the plight of Aboriginal people
there. Although they found that collective guilt was associated
with compensation attitudes even when controlling for anger,
anger proved a stronger predictor of actual willingness to act to
effect that compensation. In the present Study 3, we examined this
possible role of anger in guilt- and shame-arousing contexts.
In summary, then, several studies have shown that collective
guilt is associated with tendencies to apologize and make restitu-
tion to the outgroup. With three exceptions (Doosje et al., 1998;
Harvey & Oswald, 2000; Iyer et al., 2003, Study 2), these have
relied on cross-sectional correlational data, with all the usual
interpretative difficulties about causation that that implies. Little
research has sought to investigate the consequences of both col-
lective guilt and collective shame in the same study. Such an
omission is surprising in view of the clearly divergent predictions
about the effects of guilt and shame that have been made. In the
first two studies presented here, we looked to fill these lacunae by
conducting longitudinal research into the effects of collective guilt
and collective shame on reparation. Measuring both independent
and dependent variables at two points in time makes some infer-
ences of causality possible (Bijleveld & Van der Kamp, 1998;
Cook & Campbell, 1979; Finkel, 1995). The research was set in
the context of the relationship between indigenous and nonindig-
enous groups in Chile, a hitherto underresearched locale in social
psychology and one that is especially appropriate for studying the
effects of the particular group-based emotions with which we are
concerned in this article.
Chilean Research Context
Chile consists of a majority of nonindigenous Chileans (around
16 million) and several indigenous groups, of which the largest and
most culturally significant is the Mapuche (around 0.8 million).
The Mapuche have fought against invasions of their territory for
over 3 centuries and were finally defeated only in the late 19th
century. Since then, the Mapuche have suffered further infringe-
ments of their land rights, suppression of their culture (e.g., their
language was outlawed under Pinochet’s military regime, 1973–
1989), and severe economic and social deprivation. Recently, they
have become more active in protesting their conditions, which has
sometimes culminated in violent clashes with agents of the state or
private employers. Nonindigenous Chileans have rather ambiva-
lent feelings toward the Mapuche (Saiz, 2002). On the one hand,
the Mapuche are characterized (e.g., in school books) as brave and
fearless warriors, a part of the “founding myth” of the Chilean
nation, and, as such, a source of pride for nonindigenous Chileans.
On the other hand, Mapuche unemployment and alcoholism rates
are disproportionately high in comparison with those of other
groups in Chile, which makes them a ready target of negative
attitudes from the majority. Furthermore, in recent years, there has
been a public debate about nonindigenous people’s group-based
culpability and responsibility for treatment of the Mapuche in the
past and the possible need for reparations. This debate has been
translated into official state action, with the Chilean government
establishing a body for the improvement of the Mapuche’s situa-
tion (Instituto Nacionalde Estadı´stica, 2002; Ministerio de Plani-
ficacio´n y Cooperacio´n, 2003). Thus, issues of group-based guilt,
shame, responsibility, and reparations are likely to be quite salient
with regard to the Mapuche in nonindigenous people’s minds.
Hypotheses
Here, we develop our principal hypotheses. The first hypothesis
(H1) is that feelings of collective guilt held by nonindigenous
Chileans over the historical mistreatment of the Mapuche will be
causally related to attitudes in favor of making reparations to the
Mapuche.
1
This implies that prior feelings of guilt should longi
-
tudinally predict reparation attitudes.
We now turn to the possible effects of collective shame: The-
oretically, as we have shown, shame should be less strongly and
less durably related to reparation. It is possible that in the short
term, there may be a positive association with reparation attitudes,
but we hypothesize that this will not translate into a longitudinal
causal relationship because of its essential ingroup focus. Instead,
as argued earlier, we believe that shame could interact with col-
lective guilt to inhibit the latter’s usual prosocial consequences.
Thus, the second hypothesis (H2) proposes a moderation of the
guilt–reparation relationship by shame such that high-shame re-
spondents will show a weaker relationship between guilt and
reparation than will low-shame respondents.
Study 1
Method
Participants
One hundred twenty-four nonindigenous secondary school stu-
dents from the Temuco region in Chile participated in the survey
(55 boys, 69 girls; age: M ⫽ 15.47 years). For 120 of these
participants, data were collected at two points in time with a time
lag of approximately 8 weeks.
1
The use of causal terminology here must, of course, be understood in
the context of the correlational designs of our studies. Although the
longitudinal designs of Studies 1 and 2 permit stronger causal inferences
than do cross-sectional designs, definitive statements about causality
should be made cautiously.
78
BROWN, GONZA
´
LEZ, ZAGEFKA, MANZI, AND C
ˇ
EHAJIC
´
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Procedure and Measures
Data were collected in Temuco, a city several hundred kilome-
ters to the south of Santiago, in an area in which the proportion of
the Mapuche population is very large. During school class time, all
participants filled out a questionnaire in Spanish that contained the
measures of the independent and dependent variables, as translated
below. The questionnaires distributed at both points in time were
virtually identical. Participation was voluntary and took place with
parental and student written consent. At the conclusion of the
study, all participants were debriefed.
Collective guilt and shame. Collective guilt was measured by
asking participants how much they agreed (or disagreed) with four
statements concerning the nonindigenous people’s current or his-
torical treatment of the Mapuche (see Table 1). Collective shame
was measured by inviting agreement (or disagreement) with three
statements that attributed the cause of the nonindigenous group’s
mistreatment of the Mapuche to some internal factor (see Table 1).
Preliminary factor analysis of these seven items (with the principal
axis method of extraction) with a larger (N ⫽ 359) but equivalent
sample of nonindigenous students at Time 1 (T1) confirmed that
the guilt and shame items did indeed load on separate, if correlated
(.41), factors.
2
In the obtained solution with oblimin rotation, all
four guilt items loaded on the first factor (loadings: .57–.91) but
not on the second factor (loadings: ⬍ .30), and the three shame
items loaded on the second factor (loadings: .47–.89) but not on
the first factor (loadings: ⬍ .30). The two resulting scales had
adequate internal reliabilities (␣s ⫽ .77 and .67 for guilt and
shame, respectively).
Reparation. Desire to make reparation to the Mapuche was
measured with five items that addressed issues of providing resti-
tution to the Mapuche or of reducing their social exclusion (see
Table 1). This scale had satisfactory internal reliability (␣⫽.75).
Prejudice toward the Mapuche. We felt it was important to
control for participants’ initial attitudes toward the Mapuche in
view of previous research indicating correlation between guilt and
prejudice (Swim & Miller, 1999) and between prejudice and
compensatory attitudes (Leach et al., 2006). Therefore, a prejudice
measure was also included. This measure consisted of nine items
taken or adapted from racism measures used in other contexts (e.g.,
Lepore & Brown, 1997; Pettigrew & Meertens, 1995; see Table 1).
This scale had good reliability (␣⫽.81). One reviewer suggested
that this prejudice scale might be empirically indistinguishable
from our measure of collective shame. In fact, this proved not to be
the case. A factor analysis of the shame and prejudice items
(principal axis method of extraction with oblimin rotation) con-
firmed that the shame items loaded together on the same factor
(loadings: .49–.70) and not on either of the two factors that
comprised the prejudice scale (all cross-loadings: ⬍ .30). And the
shame factor was correlated only weakly with the two prejudice
factors (⫺.20 to ⫺.28), which were themselves moderately corre-
lated with each other (.47). Moreover, in the matched sample that
was used for the main analyses, shame and prejudice were again
only weakly correlated (see Table 2). It seems clear, then, that
shame and prejudice were indeed clearly separable constructs.
All of the above items used a 5-point response format (1 ⫽
completely agree,5⫽ completely disagree) and were recoded so
that a high score indicated a high value of the construct in question.
In addition, various demographic details (e.g., age, gender) of
participants were recorded.
Results
We present the findings in three sections. First, we present the
results from cross-sectional analyses of the T1 and Time (T2)
samples, considered separately. This shows the pattern of contem-
poraneous associations among our primary variables. In the second
section, we present the results of the longitudinal analyses, in
which we sought to establish the power of collective guilt (and
shame) to predict reparation attitudes at T2, as predicted by H1,
and also to test the moderation hypothesis (H2). Finally, as in any
cross-lagged panel design, we explored the possible existence of
“reverse” or circular causal processes—namely, that prior repara-
tion attitudes might influence subsequent feelings of collective
guilt and shame.
Cross-Sectional Analyses
The means and intercorrelations of the principal measures are
presented in Table 2. From that table, it can be seen that respon-
dents showed moderate levels of reparation, guilt, and shame (all
around the midpoint of the scale) and rather low levels of prejudice
toward the Mapuche. From the upper right quadrant of Table 2, it
can be seen that, as expected, collective guilt and shame are
positively correlated with each other, and both are positively
associated with reparation attitudes.
To examine their independent associations at T1, we regressed
reparation attitudes onto guilt and shame, and, in a subsequent
step, the interaction term between these predictors. The main
predictors were centered prior to analysis. To be sure that these
were relatively “clean” estimates of the links between collective
guilt, collective shame, and reparation, we first controlled respon-
dents’ age and level of prejudice. This multiple regression ex-
plained a reasonable amount of the variance (R
2
⫽ .48), F(5,
118) ⫽ 21.74, p ⬍ .001, in which the only reliable predictors were
guilt (␤⫽.34, p ⬍ .001) and shame (␤⫽.40, p ⬍ .001). Thus,
both guilt and shame were positively associated with reparation
attitudes in this cross-sectional analysis. Contrary to H2, the in-
teraction term was not reliable (␤⫽⫺.05, p ⬍ .50).
At T2, the same regression analysis produced a very similar
outcome: (R
2
⫽ .55), F(5, 115) ⫽ 28.12, p ⬍ .001. Once again,
both guilt and shame were reliable and positive predictors of
reparation attitudes (␤⫽.47 and .29, respectively; both ps ⬍
.001). At this time point, prejudice was also a reliable correlate of
reparation attitudes (␤⫽⫺.20, p ⬍ .01). The Guilt ⫻ Shame
interaction was again nonsignificant (␤⫽⫺.03, p ⬍ .70).
Longitudinal Analysis
Inspection of the left-hand columns of Table 2 reveals very little
change in the mean levels of the variables. In fact, only shame
showed a significant reduction over time, F(1, 120) ⫽ 5.21, p ⬍
2
This sample comprised a different group of nonindigenous school
students who were taking part in a study closely related to the one reported
here. The larger sample size permitted a statistically robust analysis of the
factorial structure of the guilt and shame items.
79
COLLECTIVE GUILT AS A PREDICTOR OF REPARATION
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Table 1
Items of Principal Measures
Measure Studies 1 and 2 Study 3
Collective guilt I feel guilty for what we, the Non-Indigenous Chileans,
have done to the Mapuche in the past
When I think about the racism that exists towards the
Mapuche, I feel guilty to be a “Non-Indigenous”
Chilean
To think how we “Non-Indigenous” Chileans have stolen
the Mapuche lands makes me guilty
I feel guilty when I realize that we “Non-Indigenous”
Chileans have contributed to the loss of Mapuche
language and customs
I feel guilty for the manner in which the Mapuche have been
treated in the past by “Non-Indigenous” Chileans
When I think how “Non-Indigenous” have stolen the
Mapuche lands, I feel guilty
I feel very bad when I realize what we the “Non-Indigenous”
Chileans have contributed to the loss of Mapuche language
and customs
Sometimes I feel guilty for the things that “Non-Indigenous”
Chileans have done to the Mapuche
When I think what “Non-Indigenous” Chileans have done to
the Mapuche, I feel guilty
Even if I have done nothing bad, I feel guilty for the
behaviour of “Non-Indigenous” Chileans toward the
Mapuche
I feel guilty for the bad living conditions of the Mapuche
To think how we “Non-Indigenous” Chileans show
intolerance, by refusing to offer job contracts to Mapuche
people, makes me feel guilty
Collective shame Due to the long history of discrimination against the
Mapuche, I think that we “Non-Indigenous” Chileans
are predisposed to be racist
When I think of the manner in which the Mapuche have
been treated, I sometimes think that we “Non-
Indigenous” Chileans are racist and mean
Even though I do not discriminate against the Mapuche, I
feel bad when I realize that other “Non-Indigenous”
Chileans do
I feel bad because the behavior of “Non-Indigenous” Chileans
towards the Mapuche people has created a bad image in
the eyes of the world
I feel bad when I see an international report on the treatment
received by the Mapuche on the part of “Non-Indigenous”
Chileans
Sometimes it shames me how others can think for the manner
in which we have harmed the Mapuche
To think how Chile is seen for its treatment of the Mapuche
makes me feel ashamed
I feel humiliated when I think of the negative manner that
Chile is seen by the rest of the world for how it has treated
the Mapuche
I feel shame when I think how “Non-Indigenous” Chileans
have behaved towards the Mapuche
I feel ashamed to be a “Non-Indigenous” Chilean for the way
we have treated the Mapuche
I feel ashamed for the damage done to the Mapuche by
“Non-Indigenous” Chileans
I feel ashamed for the racist tendency of “Non-Indigenous”
Chileans
It shames me when I realize that “Non-Indigenous” Chileans
could be intolerant by nature
Reparation attitude I feel that the Mapuche should have economic benefits as
a reparation for the damage that we’ve caused them
I feel that the Mapuche should have economic benefits as a
reparation for the damage that we’ve caused them
Our government should apologize for all the maltreatment
and deprivation that we’ve caused to the Mapuche
Our government should apologize for all the maltreatment
and deprivation that we’ve caused to the Mapuche
I would like my school to have more materials (books,
magazines, musical instruments, etc.) that would allow
us to have a better understanding of the history and
culture of the Mapuche
I would like my school to have more class materials (books,
magazines, musical instruments, etc.) that would allow us
to have a better understanding of the history and culture of
the Mapuche
I would like our country to be more tolerant and to have a
good relationship between the Mapuche and “Non-
Indigenous” Chileans
I would like our country to be more tolerant and to have a
good relationship between the Mapuche and “Non-
Indigenous” Chileans
I think that in recent times the media have devoted too
much space to speak ill of the Mapuche
I think that in recent times the media have devoted too much
space to speak ill of the Mapuche
Chilean universities should have special scholarships for
Mapuche students
When they offer work, private companies should guarantee
positions to Mapuche people
80
BROWN, GONZA
´
LEZ, ZAGEFKA, MANZI, AND C
ˇ
EHAJIC
´
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

.05. Still, changes in mean levels are not very informative for
testing our hypotheses about the longitudinal influence of guilt and
shame on reparation. To examine this, we regressed T2 reparation
on T1 guilt, shame, and their interaction term, controlling for
initial levels of reparation (Bijleveld & Van der Kamp, 1998;
Finkel, 1995). As before, age and initial prejudice levels were
added as controls.
This longitudinal analysis also explained a respectable propor-
tion of the variance in the criterion measure (R
2
⫽ .37), F(6,
114) ⫽ 11.19, p ⬍ .001. There were just two significant effects: As
expected, the test–retest association for reparation was significant
(␤⫽.43, p ⬍ .001). The only other effect was for guilt, which
was, as predicted by H1, a positive predictor of T2 reparation
attitudes (␤⫽.20, p ⬍ .05). In contrast, the main effect for shame
was far from being significant (␤⫽.03, p ⬍ .80). Contrary to H2,
there was no significant Guilt ⫻ Shame interaction (␤⫽.12, p ⬍
.20).
Reverse Causal Direction
As we have just shown, there is evidence that collective guilt
does indeed have a causal relationship in predicting reparation
attitudes. However, it is possible that a circular relationship exists
such that prior reparation attitudes might have an influence on
subsequent levels of guilt. To examine this possibility, we reversed
the logic of the longitudinal analysis in the previous section. This
time, we regressed T2 guilt on T1 reparation attitudes and shame,
controlling for T1 values of guilt. As before, age and prejudice
Table 1 (continued)
Measure Studies 1 and 2 Study 3
Prejudice I would be bothered if most of my classmates were
Mapuche
I would be concerned if my teacher or boss was Mapuche
I would feel uncomfortable sitting next to a Mapuche
person on a bus
Mapuche people should be marginalized in Chilean
society
If I would meet a Mapuche person in the street, I would
feel tense and nervous
Sometimes I think that this country would be better off
with fewer Mapuche
The Mapuche exaggerate their problems to get help
The Mapuche’s problems are due to themselves
The Mapuche receive more help from the government
than they really deserve
Anger Sometimes I feel angry when I think what my group has done
to Mapuche people in the past
Thinking about how my group has treated Mapuche people
makes me feel angry
Talking about the past and the issues regarding the treatment
of Mapuche people by my group makes me angry
Reputation
management
I would like to improve the image of Chile in the rest of the
world in respect of how we treat indigenous peoples
I believe we should restore the international reputation of
Chile associated with the treatment of indigenous peoples
I would like other people to have a better impression of Chile
as a country in relation to how we have treated indigenous
peoples
I want other countries to respect us again for the way we
treat indigenous people
I believe that it is important that Chile is seen in a more
favorable manner by the rest of the world in relation to
indigenous issues
If we do not resolve the outstanding issues about the
indigenous people in Chile, it will damage us as a country
If we do not change our attitude towards indigenous people,
the international image of Chile will become negative
In order for Chile to improve, it should address the
indigenous issues
The reputation that Chile has at an international level depends
to a large extent on the treatment that we give to the
indigenous peoples
Image threat appraisal I consider that our image as Chileans has been negatively
affected by the way we have addressed Mapuche issues
Sometimes I believe that Chile has lost respect for the way it
has dealt with Mapuche issues
Due to the way we have addressed the Mapuche issues, I
believe that now people judge Chileans negatively
81
COLLECTIVE GUILT AS A PREDICTOR OF REPARATION
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

level were added as controls. This analysis also accounted for
respectable amounts of variance in the dependent measure (R
2
⫽
.42), F(5, 115) ⫽ 16.91, p ⬍ .001. There were three significant
effects: Trivially, T1 guilt was a strong predictor (␤⫽.58, p ⬍
.001), T1 reparation predicted T2 guilt (␤⫽.35, p ⬍ .001), and T1
shame also predicted T2 guilt (␤⫽⫺.24, p ⬍ .02). A parallel
analysis regressing T2 shame on T1 values of the other variables
also yielded a good fit (R
2
⫽ .33), F(5, 115) ⫽ 11.26, p ⬍ .001.
In this analysis, the test–retest association of shame was reliable
(␤⫽.40, p ⬍ .001), as were the associations with T1 reparation
(␤⫽.27, p ⬍ .02) and prejudice (␤⫽⫺.22, p ⬍ .01).
Discussion
There are several noteworthy features of this study. First, as
predicted by H1, collective guilt did indeed have a longitudinal
effect on reparation attitudes. Because our analysis controlled for
initial levels of reparation (Finkel, 1995), there is some basis for
inferring a causal relationship between guilt and reparation. To our
knowledge, this is the first time that this has been demonstrated
longitudinally in a field setting, although, of course, several other
cross-sectional studies have found similar results (Iyer et al., 2003;
McGarty et al., 2005; Swim & Miller, 1999). Second, collective
shame appeared to have no direct causal link to reparation attitudes
because the beta value in the longitudinal analysis was effectively
zero. This stands in stark contrast to the longitudinal effect of guilt.
However, the collective shame measure cannot be dismissed as
simply an unreliable or impotent variable, because it was a signif-
icant predictor of reparation attitudes in both the T1 and T2
cross-sectional analyses. Thus, in the short term, it appears that
collective shame can be “alleviated” in the same way as collective
guilt, by increasing endorsement of reparation attitudes. Its longer
term effects are rather different however.
That said, there were some unexpected findings. First, it was
interesting that there seemed to be some “circular causality” at
work because initial reparation attitudes also predicted subsequent
guilt and shame. In retrospect, this may not be so surprising. This
study may have represented one of the first formal instances in
which these adolescent nonindigenous Chileans had been ques-
tioned about both their group’s potential culpability for the histor-
ical treatment of the Mapuche and what might be done about that.
It is not implausible to imagine that having to answer questions
about whether the Chilean government should apologize and make
restitution to the Mapuche might have instigated individual and
social rumination about the ingroup’s collective guilt. Given the
relatively short time lag of this study (8 weeks), the more the
participants thought about and endorsed reparation attitudes ini-
tially, the greater could have been their feelings of collective guilt
and shame later on. The question of whether such circular causal
effects would still be visible over a longer time period is interest-
ing.
A second issue concerns the absence of any moderation of the
longitudinal effects of guilt on reparation by shame, as predicted
by H2. Apart from the inherent statistical difficulty in detecting
interactions in correlational designs (McClelland & Judd, 1993),
the substantive reason for this is not clear, although, again, we
speculate that it may have to do with the relatively short time span
of this longitudinal design. As we showed, contemporaneously,
shame had a positive association with reparation. It is possible that
this positive association persisted long enough for the predicted
inhibition effect on guilt not to manifest itself but not long enough
to produce any reliable longitudinal effect for shame per se. In a
second study, we examined this possibility by introducing a much
longer time lag.
Study 2
Study 2 was a close replication of Study 1, with one critical
difference. Instead of the relatively short time lag of 8 weeks, we
wanted to determine whether the longitudinal effects of guilt on
reparation would persist for a much longer period—6 months. If
this was the case, this would have important theoretical implica-
tions because it would provide evidence for the robustness of the
consequences of collective guilt. In addition, we sought to reex-
amine the hypothesized inhibitory consequences of shame over
this same longer time period.
Method
Participants
Two hundred and forty-seven nonindigenous secondary school
students from the Temuco area participated in the T1 data collec-
tion (119 boys, 127 girls, 1 unspecified; age: M ⫽ 16.02 years). Of
these, 137 also participated at T2, approximately 6 months later.
This relatively heavy attrition rate was due in part to the more
ambitious time lag, but it was also because the two data-collection
points straddled the end and the beginning of the academic year in
Chile, with consequent losses of students for administrative rea-
sons (e.g., some students changing schools or repeating the year).
Nevertheless, as we report below, the complete panel sample did
not seem to differ much from the respondents who dropped out
Table 2
Study 1: Means and Intercorrelations Among Principal Variables
Variable M
T1
SD M
T2
SD 12 3 4
1. Reparation 3.52 0.86 3.48 0.83 .57 .59/.66 .62/.60 ⫺.24/⫺.34
2. Guilt 2.96 1.03 2.89 1.04 .46 .59 .58/.57 ⫺.15/⫺.15
3. Shame 3.62 0.91 3.40 1.01 .41 .28 .45 ⫺.25/⫺.26
4. Prejudice 2.31 0.78 2.35 0.88 ⴚ.20 ⴚ.17 ⴚ.32 .51
Note. For all correlations of r ⱖ 20, coefficients on or below the diagonal in bold type are Time 1 (T1) and Time 2 (T2) correlations (e.g., from T1 values
of variables in column 1 to T2 values of variables in top row). Above the diagonal are cross-sectional correlations at T1/T2 (r ⱖ .20, p ⬍ .05; r ⱖ .25,
p ⬍ .01; r ⱖ .32, p ⬍ .00).
82
BROWN, GONZA
´
LEZ, ZAGEFKA, MANZI, AND C
ˇ
EHAJIC
´
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

after T1, thus giving us some confidence that it was reasonably
representative of the original sample.
Procedure and Measures
The procedure and measures were identical to those of Study 1.
The measures all had adequate or good internal reliabilities: col-
lective guilt (␣⫽.79), collective shame (␣⫽.64), reparation (␣⫽
.72), and prejudice (␣⫽.76). As in Study 1, preliminary factor
analysis (principal axis method of extraction) with another larger
(N ⫽ 376) but equivalent sample of nonindigenous participants
confirmed the distinctiveness of our collective guilt and shame
measures. Factor analysis with oblimin rotation produced the same
two-factor solution as in Study 1, with guilt items loading on the
first factor (loadings: .58–.89) but not on the second factor (load-
ings: ⬍ .30), whereas shame items loaded on the second factor
(loadings: .69–.86) but not on the first factor (loadings: ⬍ .30).
The correlation between the factors was .45. We again checked on
the separability of the shame and prejudice measures. A factor
analysis of the shame and prejudice items (principal axis method
of extraction with oblimin rotation) showed that the shame items
loaded together on the same factor (loadings: .52–.77) and not on
either of the two factors that comprised the prejudice scale (all
cross-loadings: ⬍ .30). The shame factor was correlated only
weakly with the two prejudice factors (⫺.23–⫺.23), which were
themselves moderately intercorrelated (.44). In the matched sam-
ple that was used for the main analyses, shame and prejudice were
only weakly associated (see Table 3). Once again, it is apparent
that shame and prejudice were clearly separable constructs.
Results
The results are presented in four sections. First, we report on the
effects of the panel attrition. Then we report cross-sectional anal-
yses from T1 and T2. Then we report on the main longitudinal
analysis in which H1 and H2 were tested. Finally, we check on any
reverse causal links.
Panel Attrition
To assess the representativeness of our full panel sample (with
data at both time points), we compared this sample with those for
whom we had data only at T1. One-way analyses of variance on all
the measures revealed no significant differences between the full
longitudinal sample and those who dropped out after T1. In only
one respect did the two samples differ: In the longitudinal sample,
there was a slightly lower proportion of boys than there were in the
T1-only sample (43% vs. 55%), ␹
2
(1) ⫽ 3.99, p ⬍ .05. This minor
difference aside, it seems safe to assume that the full panel was
reasonably representative of the original sample.
Cross-Sectional Analyses at T1 and T2
Means and intercorrelations of the main variables are shown in
Table 3. From there, it can be seen that the mean levels of
reparation, guilt, shame, and prejudice were all very comparable to
those in Study 1. The upper right quadrant of Table 3 shows that,
as in Study 1, collective guilt and shame were positively correlated
with each other, and both were positively associated with repara-
tion attitudes.
To examine their independent associations at T1, we once again
regressed reparation attitudes onto guilt and shame and, in a further
step, the two-way interaction between them. As before, we first
controlled respondents’ age and level of prejudice and centered the
main predictors. The final regression model accounted for a rea-
sonable proportion of the observed variance (R
2
⫽ .43), F(5,
231) ⫽ 35.22, p ⬍ .001. Mirroring the results from Study 1, the
only reliable predictors were guilt (␤⫽.41, p ⬍ .001), shame
(␤⫽.24, p ⬍ .001), and the control variable prejudice level (␤⫽
⫺.35, p ⬍ .001). Thus, both guilt and shame were positively
associated with reparation attitudes in this cross-sectional analysis.
The interaction term was not reliable (␤⫽⫺.02, p ⬍ .70).
The same regressions at T2 produced a similar outcome (R
2
⫽
.40), F(5, 126) ⫽ 16.97, p ⬍ .001. As before, both guilt and shame
were reliable predictors of reparation attitudes (␤s ⫽ .40 and .23,
respectively; ps ⬍ .001 and .01, respectively). Prejudice was also
a reliable correlate of reparation attitudes (␤⫽⫺.34, p ⬍ .001).
The Guilt ⫻ Shame interaction was again nonsignificant (␤⫽
⫺.07, p ⬍ .40).
Longitudinal Analyses
Inspection of the T1 and T2 means in Table 3 reveals that there
was little change in mean levels of the variables over time. The
only significant change was for shame, F(1, 136) ⫽ 4.14, p ⬍ .05.
To test our hypotheses, we regressed reparation attitudes at T2
onto T1 values of guilt, shame, their interaction term, and age and
prejudice level, controlling for initial reparation attitudes. This
regression model accounted for an adequate proportion of the
variance (R
2
⫽ .24), F(6, 126) ⫽ 6.68, p ⬍ .001. Apart from the
test–retest value for reparation (␤⫽.38, p ⬍ .001), the only
significant effects were the hypothesized (H1) main effect for guilt
(␤⫽.22, p ⬍ .04) and the hypothesized (H2) Guilt ⫻ Shame
Table 3
Study 2: Means and Intercorrelations Among Principal Variables
Variable M
T1
SD M
T2
SD 12 3 4
1. Reparation 3.44 0.99 3.44 0.99 .40 .58/.55 .50/.48 ⫺.35/⫺.43
2. Guilt 2.98 1.00 2.91 1.01 .39 .46 .52/.49 ⫺.21/⫺.20
3. Shame 3.69 0.98 3.53 0.97 .30 .24 .47 ⫺.23/⫺.29
4. Prejudice 2.32 0.82 2.40 0.82 ⴚ.19 .01 ⴚ.20 .46
Note. For T2 and T1-T2 correlations, r ⱖ 19; coefficients on or below the diagonal in bold type are Time 1 (T1) and Time 2 (T2) correlations (e.g., from
T1 values of variables in column 1 to T2 values of variables in top row). Above the diagonal are cross-sectional correlations at T1/T2. For correlationsin
bold type, r ⱖ .19, p ⬍ .05; r ⱖ .24, p ⬍ .01; and r ⱖ .30, p ⬍ .001. For T1 correlations (with larger degrees of freedom), all are significant at p ⬍ .001.
83
COLLECTIVE GUILT AS A PREDICTOR OF REPARATION
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

interaction (␤⫽⫺.20, p ⬍ .02). The coefficient for shame was far
from significant (␤⫽.07, p ⬍ .50). Following Aiken and West’s
(1991) suggestion, we plotted the Guilt ⫻ Shame interaction by
applying the regression equation to a combination of two values of
guilt and shame (1 standard deviation below the mean and 1
standard deviation above the mean, labeled low and high, respec-
tively). The results, which are presented in Figure 1, indicate that
the longitudinal effect of guilt on reparation increased at lower
levels of shame, as predicted by H2.
Reverse Causal Direction
Confirming Study 1 and supporting H1, collective guilt ap-
peared to be causally related to reparation attitudes. What about the
opposite pathway over this longer time period? As before, we
regressed T2 guilt on T1 reparation attitudes and shame, control-
ling for T1 values of guilt (with age and prejudice level again
included as controls). This analysis also accounted for variance in
the dependent measure (R
2
⫽ .23), F(5, 127) ⫽ 7.50, p ⬍ .001.
However, there was only one reliable beta coefficient, the unsur-
prising test–retest value for guilt (␤⫽.48, p ⬍ .001). All other
coefficients were far from significant (all ps ⬎ .10). A parallel
analysis regressing T2 shame on T1 values of the other variables
also yielded a significant overall regression equation (R
2
⫽ .27),
F(5, 127) ⫽ 9.29, p ⬍ .001, and, again, only the test–retest
association of shame was reliable (␤⫽.45, p ⬍ .001). All other
coefficients were nonsignificant (all ps ⬎ .11). Thus, in this study,
there was little evidence of circular causality from initial repara-
tion to later feelings of guilt or shame.
Discussion
This study provided a striking confirmation of our hypotheses.
First, despite the threefold increase in time lag, guilt was still
significantly predictive of later reparation attitudes, even when
controlling for initial reparation attitudes. This supports H1. Sec-
ond, and supporting H2, this direct link was moderated by collec-
tive shame such that it held reliably only for those respondents
reporting low initial levels of collective shame. For high-shame
participants, the otherwise positive consequences of guilt seem to
have been suppressed, suggesting, as we had suspected from the
outset, that collective guilt and shame have rather different con-
sequences for people’s desire to make restitution for their in-
group’s past misdeeds.
The effects of collective shame observed in Study 1 were also
mirrored here. As before, it proved to have little direct long-term
effect on reparation attitudes. Its only reliable and independent
association was in the cross-sectional analyses. A possible expla-
2.50
2.75
3.00
3.25
3.50
3.75
4.00
Low Guilt High Guilt
Reparation
Low Shame High shame
Figure 1. Shame moderating the longitudinal relationship between guilt and reparation in Study 2.
84
BROWN, GONZA
´
LEZ, ZAGEFKA, MANZI, AND C
ˇ
EHAJIC
´
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

nation for this correlation is that it reflects the genuine, if short-
term, consequence of feeling collectively shameful about the in-
group’s misdeeds. As we speculated earlier, it is possible that
shame might be temporarily alleviated by attempting to make
reparations, but this lessening is only temporary because of the
underlying negative attribution that shame implies. Such an attri-
bution means that longer term relief is more likely to be gained
from avoidance strategies; hence, the inhibition of the longitudinal
guilt–reparation relationship for high-shame people.
Though the findings of Study 2 largely confirmed those of Study
1, there was one important difference in the pattern of results
observed. This concerns the absence of any reverse path between
reparation and guilt as compared with that found in Study 1. Here,
the difference in time lags between the two studies also seems a
plausible explanation for the inconsistency. The argument ad-
vanced earlier for the reparation–guilt causal link was that being
asked to reflect on reparation issues at T1 might have instigated
increased feelings of guilt in the minds of the nonindigenous
participants. Such a reflexive process may have been less likely
over the 6-month duration of the second study, especially as that
time gap also involved changes in school classes and likely other
significant events in the lives of our adolescent participants. If the
apparently unidirectional causal link that we observed in Study 2
is confirmed in further research over longer periods, it will add
strength to the reasoning behind the hypothesis that predicted it.
Finally, we acknowledge the undesirably high attrition rate and
the consequent loss of statistical power in our longitudinal analy-
ses. Although we had designed the study to be substantially larger,
for administrative and other reasons outside of our control, we did
lose many more participants than we had wished. Still, despite the
attrition, it seems that the full panel sample did not differ substan-
tially from the originally conceived one, and the findings it yielded
were reasonably clear-cut.
Study 3
In Studies 1 and 2, we showed that guilt, but not shame, has
reliable longitudinal main effects on reparation attitudes. This was
entirely consistent with theoretical expectations. Somewhat less
expected, though, were the equally reliable cross-sectional positive
correlations between collective shame and reparation attitudes. In
Study 3, we investigated what might be underlying those associ-
ations. While doing so, we also took the opportunity to refine our
key measures in order to improve their validity and internal reli-
ability.
Earlier, it was noted how conceptualizations of collective shame
have included two aspects. One stresses the perception of some
negative ingroup “essence” that might be to blame for the immoral
actions of ingroup members (Lickel et al., 2004), and the other
focuses more on how the public reputation of the ingroup might
have been damaged by those same reprehensible deeds of its
members (Branscombe et al., 2004). In practice, as we noted
earlier, we suspect that these two components will often be closely
associated because the awareness of a besmirched ingroup repu-
tation in the eyes of others may well lead to a similar negative
perception of the ingroup by its members. Still, insofar as collec-
tive shame does involve such reputational aspects, this does imply
that one strategy for coping with it is to manage the ingroup’s
reputation by appearing to endorse some restitution to the out-
group. A contemporary illustration of this process at work was
provided by Ken Livingston, mayor of London. Commenting on
the 200th anniversary of the abolition of slavery in Britain, Mayor
Livingston said, “A British state that refuses to apologise for a
crime on such a gigantic scale as the slave trade merely lowers our
country in the opinion of the world” (“Marching to London to Hear
a Single Word,” 2007).
Research support for this image-management consequence of
shame was found by Schmader and Lickel (2006). They asked
Hispanic participants to identify an event in which a member of
their group had done something to confirm a negative stereotype
about Hispanics. Feelings of shame about that event were corre-
lated with a desire to repair the image of the ingroup. Such a
strategy may be particularly efficacious in the short term—that is,
when confronted with investigators presenting questionnaire items
concerning reparation attitudes—because it so easily serves what
we believe to be its self- (and ingroup-) presentational needs.
However, such a strategy may be more costly, and hence less
likely to be adopted, over a longer period of time. A more likely
longer term strategy for reducing shame feelings might be denial
or other forms of social or cognitive avoidance. If this analysis is
correct, then it would explain why it is possible to observe positive
shame–reparation associations cross-sectionally (the present Stud-
ies 1 and 2; see also Brown & C
ˇ
ehajic´, in press; Harvey & Oswald,
2000) but not longitudinally. One further implication is that such
a shame–reparation link might be mediated by a desire to protect
or improve the ingroup’s image in the eyes of others. Study 3 was
designed to explore this hypothesis of mediation of the shame–
reparation cross-sectional association.
As noted earlier, Iyer et al. (2007) and Leach et al. (2006) have
suggested that guilt might be a less important predictor of com-
pensatory intentions than is anger. Indeed, Iyer et al. (2007) found
that the effects of both guilt and shame disappeared once anger
was controlled. To determine whether anger could also account for
the guilt–reparation and shame–reparation associations we have
observed, we included a measure of anger in Study 3.
In Studies 1 and 2, the measure of collective shame tapped
mainly the negative ingroup essence component of shame. More-
over, the items comprising it (deliberately) eschewed the words
shame or ashamed because we wanted to avoid semantic overlap
with the items comprising the guilt scale. Although we were
successful in the latter aim, it has to be conceded that the reliability
of the three-item shame measure was suboptimal and may also
have lacked some face validity. Thus, a second goal of Study 3 was
to enlarge and improve the measure of collective shame. We
sought to do this in four ways: First, we added items that captured
more directly the reputational aspect of the emotion. Second, we
incorporated the key emotion words shame, ashamed, and humil-
iated and tied them specifically to the “negative essence” and
reputational concepts. Third, we focused more directly on felt
emotions. In the earlier version of the scale, it might be argued,
there was some conflation of the appraisal aspect of emotion—
“When I think of the manner in which the Mapuche have been
treated, I think that we . . . are predisposed to be racist”—with the
emotion of shame itself. In this improved version, therefore, we
made feelings of shame (and associated emotions) absolutely
explicit and distinct from mere appraisals of threats to the in-
group’s image. Fourth, we increased the range and number of
items dealing with the historical mistreatment of indigenous peo-
85
COLLECTIVE GUILT AS A PREDICTOR OF REPARATION
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ples in Chile. At the same time, we also aimed to improve the
internal reliabilities of the guilt and reparation measures, and we
included some additional measures to check for alternative inter-
pretations of the observed effects.
The hypotheses of this cross-sectional study were as follows:
1. On the basis of Studies 1 and 2 and other empirical work
(e.g., Harvey & Oswald, 2000), we predicted that both
collective guilt and shame would be independently and
positively associated with reparation attitudes.
2. We predicted that the shame–reparation link would be
mediated by a desire to improve the ingroup’s public
reputation.
Method
Participants
Participants were 193 students (undergraduate and preunder-
graduate induction) at a university in Santiago, Chile (91 male, 93
female, 9 unspecified; age: M ⫽ 16.89 years, range ⫽ 14–34
years), who agreed to take part on a voluntary basis.
Procedure
Participants filled out the questionnaire in Spanish during class
time. The questionnaire was described as being a study of social
attitudes and began with a brief introductory paragraph that re-
ferred to the historical conflicts between the Mapuche and the
nonindigenous groups over land, culture, and language issues.
Participants then filled out the questionnaire. Following comple-
tion of the questionnaire, participants were debriefed.
Measures
Collective guilt. This measure comprised eight items. Three of
these items were from Studies 1 and 2 (or were close approxima-
tions thereof), and five were new (see Table 1). Together, these
eight items produced a highly reliable scale (␣⫽.93).
Collective shame. This measure comprised 10 items, all of
which explicitly tapped emotions related to the different facets of
shame (see Table 1). These 10 items also made a highly reliable
scale (␣⫽.93).
Appraisal of image threat. A separate scale specifically mea-
suring appraisals of the threat to the ingroup’s image rather than
the actual emotions caused by those appraisals was devised from
three items (see Table 1; ␣⫽.75).
Reparation attitudes. This scale comprised the five original
items together with two new ones (see Table 1). It also had a
satisfactory internal reliability (␣⫽.86).
Reputation management. This was a new scale that tapped a
concern with and a desire to improve the reputation of the ingroup
(nonindigenous Chileans). It consisted of nine items (see Table 1).
This scale had very good internal reliability (␣⫽.91).
Anger. Three items tapped participants’ anger at the past mis-
treatment of the Mapuche (see Table 1). These formed a reliable
scale (␣⫽.86).
All items used a 7-point response format (1 ⫽ completely
disagree,7⫽ completely agree) so that a high score indicated a
high value of the construct. Various demographic details (e.g., age,
gender) of participants were also recorded.
Results
The results from this study are presented in two sections. In the
first section, we present factor analyses of the new collective guilt
and shame scales to demonstrate their empirical distinctiveness. In
the second section, we present findings from multiple regression
analyses that tested our two hypotheses.
Collective Guilt and Shame Measures
All 18 items comprising the guilt and shame scales were factor
analyzed with the principal axis extraction method with oblimin
rotation. Inspection of the eigenvalues and the scree slope clearly
indicated a two-factor solution (59.4% of variance accounted for).
All of the guilt items loaded on the same factor (loadings: .58–.90;
cross-loadings on the other factor: ⬍ .12). All of the shame items
loaded on the second factor (loadings .65–.85; cross-loadings on
the other factor: ⬍ .13). This analysis thus provided excellent
evidence for the distinctiveness of the new guilt and shame scales,
and, as noted earlier, both had excellent internal reliabilities
(⬎ .90). As usual, these overall guilt and shame scales were
moderately intercorrelated, r(186) ⫽ .68, p ⬍ .001.
Regression Analyses
Table 4 presents the means and standard deviations of the
principal variables used in the multiple regression analyses. To test
our mediation hypothesis for the shame–reparation link, we re-
gressed reparation attitudes onto guilt and shame in a first step and
Table 4
Study 3: Means and Intercorrelations Among Principal Variables
Variable MSDGuilt Shame Reputation Appraisal Anger
Reparation 4.28 1.48 .57 .56 .57 .38 .49
Guilt 3.15 1.47 .68 .48 .33 .45
Shame 3.46 1.51 .61 .53 .44
Reputation 4.04 1.41 .59 .38
Appraisal 3.86 1.47 .29
Anger 3.93 1.56
Note. For all correlations, p ⬍ .001.
86
BROWN, GONZA
´
LEZ, ZAGEFKA, MANZI, AND C
ˇ
EHAJIC
´
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

then added our measure of reputation management in a second step
(Baron & Kenny, 1986). The variance explained in the first step
of this model was respectable (R
2
⫽ .37), F(2, 183) ⫽ 54.36,
p ⬍ .001. In this first step, both guilt and shame were reli-
able predictors of reparation attitudes (␤s ⫽ .34 and .32, respec-
tively; both ps ⬍ .001).
3
Adding reputation management in the
second step produced a reliable increase in variance explained
(⌬R
2
⫽ .065), F
change
⫽ 21.04, p ⬍ .001. The regression coeffi
-
cient for guilt dropped only slightly in this step (␤⫽.30, p ⬍
.001), whereas that for shame dropped markedly (␤⫽.15) to
become nonsignificant (p ⬍ .07). And, as expected, the coefficient
for the mediator was also highly reliable (␤⫽.32, p ⬍ .001). The
Sobel test for the predicted mediation of Shame ⫻ Reputation
management was reliable (z ⫽ 3.76, p ⬍ .001), indicating that
mediation had occurred.
4
A test for possible mediation of the
Guilt ⫻ Reparation effect proved nonsignificant (z ⫽ 1.49, p ⬍
.14). Finally, only shame predicted reputation management (␤⫽
.53, p ⬍ .001); the coefficient for guilt was small and unreliable (␤
⫽.12, p ⬍ .12).
Next, we checked for various alternative explanations of the
above finding. In a further regression analysis, we added appraisal
of image threat in a first step before entering guilt and shame. The
logic here was to demonstrate that the associations between guilt
and shame held even when controlling for threat appraisals. It
seemed that they did. Though appraisal of image threat had a
reliable association with reparation (␤⫽.40, p ⬍ .001), the
relationships among guilt, shame, and reparation were still highly
significant in the second step of the analysis (␤⫽.34, p ⬍ .001,
and ␤⫽.24, p ⬍ .005, respectively). Moreover, adding the
reputation mediator in the final step still resulted in a nonsignifi-
cant effect for shame (␤⫽.14, p ⬍ .11), a reduction that a Sobel
test indicated was reliable (z ⫽ 2.85, p ⬍ .005). In contrast, the
beta for guilt was little affected by the addition of the mediator
(␤⫽.30, p ⬍ .001).
In a further analysis, we examined whether our effects for
guilt and shame held even when controlling for anger about the
plight of the Mapuche (Iyer et al., 2007). Accordingly, we
added anger in the first step (␤⫽.49, p ⬍ .001). However, in
the second step, guilt and shame still proved to be reliable
predictors of reparation attitude (␤s ⫽ .27 and .26, respectively;
both ps ⬍ .001). Moreover, the hypothesized mediation of the
shame effect was still clearly visible in the third step: The
shame effect dwindled to nonsignificance (␤⫽.12, p ⬍ .16), a
reliable mediation effect according to a Sobel test (z ⫽ 3.47,
p ⬍ .001); the guilt effect was little changed (␤⫽.24, p ⬍
.001). Thus, it seems that our effects pertain above and beyond
any role played by anger.
Finally, we added both appraisal of image threat and anger as
controls in the first step of the regression. Both proved to be inde-
pendent predictors of reparation attitude (␤s ⫽ .28 and .41, respec-
tively; both ps ⬍ .001). Nevertheless, the guilt and shame effects were
still significant in the third step, even with both controls added (guilt:
␤⫽.28, p ⬍ .001; shame: ␤⫽.20, p ⬍ .03). And the predicted
mediation of the shame–reparation link was still observable in the
final step, whereas the guilt–reparation link was little affected (guilt:
␤⫽.25, p ⬍ .001; shame: ␤⫽.11, p ⬍ .20). Again, a Sobel test
indicated that mediation had occurred (z ⫽ 2.64, p ⬍ .01).
Discussion
The results from this third study support our hypothesis about
one of the possible underlying reasons for the cross-sectional
association between shame and reparation attitudes. As we had
surmised, that association is mediated by a desire to improve or
maintain the ingroup’s reputation. Moreover, the shame–
reparation link, and its mediation, held even when controlling for
appraisals of threat to the ingroup’s image and anger over the
treatment of the Mapuche.
Three other features of this study deserve comment. One con-
cerns the development of improved measures of guilt and shame.
The new measures now had high internal reliabilities; had good
face validity in the sense that they now clearly tapped intergroup
emotions and not appraisals; and, as we have seen, were both
predictive of reparation attitudes. In the absence of an alternative
contextualized measure of shame elsewhere, we look forward to
other investigators exploring its potential in other fields.
The second noteworthy point concerns the consistency of the
results from the cross-sectional analyses across all three studies,
despite the use of different scales and university instead of school
students in this third study. As we have shown, both collective
guilt and collective shame are reliably associated with reparation
attitudes contemporaneously. Although existing theory conven-
tionally does not predict such prosocial effects for shame, there
seems little doubt about the empirical robustness of the finding
(see also Brown & C
ˇ
ehajic´, in press; Harvey & Oswald, 2000).
Third, it is interesting that these associations between guilt,
shame, and reparation held even when controlling for anger over
the plight of the outgroup. This finding contrasts with that of Iyer
et al. (2007), who found that neither guilt nor shame were predic-
tive of reparation when anger was controlled (but cf. Leach et al.,
2006). There would seem to be two possible explanations for this
disparity. One lies in the measures of guilt, shame, and anger used
in the two studies. Iyer et al. (2007) used a technique developed by
Lickel et al. (2005) in which respondents are asked to indicate how
much of each of several emotions (e.g., guilty, remorseful,
ashamed, disgraced, furious, angry) they feel (in this case, about
the situation in Iraq). As the authors acknowledged, this method
introduces some ambiguity about whether the emotions are per-
sonal or group-based. In contrast, our guilt, shame, and anger items
are explicit in their reference to emotions felt about what the
ingroup has (or has not) done to the outgroup. In contextualizing
the questions in this way, we believe that our scales are more
3
The Guilt ⫻ Shame interaction was nonsignificant in these analyses
and so is not considered further. Likewise, in Studies 1 and 2, there were
no interactions between guilt and shame at a cross-sectional level.
4
It is also worth noting that we obtained very similar results in another
study (N ⫽ 192 Chilean university students). In that study, we used
measures of guilt and shame rather similar to those used in Studies 1 and
2—in fact, they correlated at .79 and .73, respectively, with the original
measures—and measures of reparation attitude and reputation management
identical to those used in Study 3. Regressing reparation on guilt and shame
yielded significant effects for both (␤s ⫽ .28 and .39, respectively; both
ps ⬍ .001). Adding the mediator in a second step affected the guilt
relationship little (␤⫽.25, p ⬍ .001) but reduced the shame effect
noticeably (␤⫽.30, p ⬍ .01), a significant drop according to a Sobel test
(z ⫽ 2.54, p ⬍ .02).
87
COLLECTIVE GUILT AS A PREDICTOR OF REPARATION
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

clearly focused on intergroup emotions. A second explanation for
the difference between the studies is that Iyer et al. (2007) focused
on a contemporary and highly charged political situation (the
ongoing war in Iraq), whereas our Chilean indigenous–
nonindigenous relationship comprises a mixture of some contem-
porary but mainly historical mistreatment of the outgroup. Con-
ceivably, with the passage of time, self-conscious emotions like
guilt and shame regain some of their potency to predict reparative
tendencies independently of anger, whereas with ongoing ingroup
transgressions, the latter emotion may have more immediate action
potential (Leach et al., 2006).
General Discussion
In drawing general conclusions from these three studies, we
make the following brief remarks: First, we believe that our
findings help to substantiate the theoretical distinction between
guilt and shame at a collective level. Although several commen-
tators have argued that guilt and shame have different underlying
psychologies and should have different social consequences
(Branscombe et al., 2004; Lickel et al., 2004; Tangney & Fischer,
1995), until now there has been little direct evidence to demon-
strate this at an intergroup level. Apart from Lickel et al. (2005),
who included friends together with groups proper (e.g., family,
ethnicity) as potential sources of vicarious shame and guilt, and
Iyer et al.’s (2007) recent study of reactions to the war in Iraq,
there has been little research showing that the two collective
emotions can be distinguished empirically and can then lead to
different outcomes. In the studies reported here, we have shown
not only that they can be reliably measured but also that they have
different longitudinal effects: Collective guilt leads to a subsequent
increase in reparative attitudes toward the outgroup, albeit espe-
cially for low-shame people, whereas collective shame appears to
have only short-term effects on reparation and no independent
longitudinal effects. Parenthetically, we can also note that collec-
tive shame has been observed to be positively correlated with
reparation attitudes in cross-sectional studies conducted in very
different contexts, such as postconflict Bosnia (Brown & C
ˇ
ehajic´,
in press) and interethnic relations in the United States (Harvey &
Oswald, 2000).
Second, our findings shed further light on the consequences of
collective shame. As we have shown, shame seems to be a re-
sponse to both a perception that the ingroup is flawed in some way
and to a concern about how others see the ingroup. This latter
reputational aspect of shame is what can lead to short-term proso-
cial effects as ingroup members seek to present their group in a
better light (Schmader & Lickel, 2006). The mediation results
from Study 3 are certainly consistent with this analysis. One
interesting further implication of this argument could be to exam-
ine the consequences of making reparation attempts public as
opposed to allowing them to remain anonymous. If collective
shame is primarily concerned with the potential damage to the
image of the ingroup, then one might expect it to be more sensitive
to such anonymity manipulations than would be collective guilt.
There is some preliminary evidence to support such a contention
(Coen & Brown, 2005). However, in noting that shame can have
short-term prosocial effects, we do not wish to imply that it cannot
also have the kinds of negative consequences that traditional
theorizing predicts. Indeed, we suspect that the fundamentally
aversive quality of shame leads people to “take to the path of least
resistance” in dealing with it. In questionnaire studies, as here, the
easiest course may be one of reputation management, endorsing or
claiming to endorse restitution to the outgroup. In other contexts,
blaming the victim group or avoiding it may be a less effortful
option. A recent experimental study in which collective shame was
manipulated independently from collective guilt found evidence of
just such negative reactions (Brown & Chatfield, 2006).
Third, though we believe that our findings have clarified some
fundamental issues in this burgeoning research domain, there
obviously remains much to be done. For example, it would be
important to investigate the antecedents of collective guilt and
shame. At the start of this article, we noted that there are some
trends in international politics toward identifying, and then encour-
aging acceptance of responsibility for, illegitimate collective ac-
tions. However, history is replete with the failure of groups to
acknowledge their misdeeds (e.g., Cohen, 2001). So, one question
is to identify the conditions that give rise to people’s experience of
their ingroup’s culpability and what form that self-conscious emo-
tion takes (shame or guilt). In two recent studies in Bosnia, we
have found that a key antecedent of collective guilt is some
acknowledgement of and acceptance of responsibility for the in-
group’s misdeeds, whereas an appraisal of the extent to which the
misdeeds threaten the ingroup’s image in the eyes of others seems
to stimulate collective shame (C
ˇ
ehajic´ & Brown, 2006). Last, it is
obviously important to extend the longitudinal work we have
initiated here to include more than two measuring points and a
longer time lag. Such a design would give a more complete picture
of the temporal dynamics involved in the relationships between
collective guilt, shame, and various outcome variables.
We believe that our findings on the beneficial consequences of
collective guilt have practical implications. Given that they were
obtained from adolescent students, we are encouraged to believe
that incorporation of material that highlights the part of dominant
groups in perpetrating injustices into educational curricula could
play a useful role in raising awareness and changing attitudes. But,
to end on a cautionary note, we would not want to argue that
stimulating collective guilt can act as a universal panacea for
rectifying intergroup inequalities. For one thing, we have evidence
from elsewhere in our research program that it can be (adversely)
implicated in mediating among outgroup contact, knowledge, and
intergroup anxiety (Zagefka, Gonza´lez, Brown, & Manzi, 2005).
And it is possible that too frequent and repeated reminders of
ingroup misdeeds could eventually transform the socially progres-
sive emotion of collective guilt into the more introspective and
potentially harmful emotion of collective shame.
References
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and
interpreting interactions. London: Sage.
Barkan, E. (2000). The guilt of nations. London: Johns Hopkins University
Press.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable
distinction in social psychological research: Conceptual, strategic, and
statistical considerations. Journal of Personality and Social Psychology,
51, 1173–1182.
Baumeister, R. F., Stillwell, A. M., & Heatherton, T. F. (1994). Guilt: An
interpersonal approach. Psychological Bulletin, 115, 243–267.
88
BROWN, GONZA
´
LEZ, ZAGEFKA, MANZI, AND C
ˇ
EHAJIC
´
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Bijleveld, C. J. H., & van der Kamp, L. J. T. (1998). Longitudinal data
analysis: Designs, models and methods. London: Sage.
Branscombe, N. R., & Doosje, B. (2004). Collective guilt: International
perspectives. New York: Cambridge University Press.
Branscombe, N. R., Slugoski, B., & Kappen, D. M. (2004). The measure-
ment of collective guilt: What it is and what it is not. In N. R.
Branscombe & B. Doosje (Eds.), Collective guilt: International perspec-
tives (pp. 16–34). New York: Cambridge University Press.
Brown, R., & C
ˇ
ehajic´, S. (in press). Dealing with the past and facing the
future: Mediators of the effects of collective guilt and shame in Bosnia.
European Journal of Social Psychology.
Brown, R., & Chatfield, J. (2006). Shame on us: Social consequences of a
group-based dysphoric emotion. Unpublished manuscript.
Buruma, I. (1994). The wages of guilt: Memories of war in Germany and
Japan. London: Jonathan Cape.
C
ˇ
ehajic´, S., & Brown, R. (2006). The burden of our times: Antecedents of
group-based guilt and shame. Unpublished manuscript.
Cialdini, R. B., Borden, R. J., Thorne, A., Walker, M. R., Freeman, S., &
Sloan, L. R. (1976). Basking in reflected glory: Three (football) field
studies. Journal of Personality and Social Psychology, 34, 366–374.
Coen, S., & Brown, R. (2005, August 30). It depends on your point of view:
Perspective taking, anonymity and the experience of group-based guilt
and shame. Paper presented at the British Psychological Society Social
Psychology Section Annual Conference, Edinburgh, Scotland.
Cohen, S. (2001). States of denial: Knowing about atrocities and suffering.
Oxford, England: Polity Press.
Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation: Design
and analysis issues for field settings. London: Houghton Mifflin.
Doosje, B., Branscombe, N. R., Spears, R., & Manstead, A. S. R. (1998).
Guilty by association: When one’s group has a negative history. Journal
of Personality and Social Psychology, 75, 872–886.
Finkel, S. E. (1995). Causal analysis with panel data. Thousand Oaks, CA:
Sage.
Harvey, R. D., & Oswald, D. L. (2000). Collective guilt and shame as
motivation for White support for Black programs. Journal of Applied
Social Psychology, 30, 1790–1811.
Haslam, N., Rothschild, L., & Ernst, D. (2000). Essentialist beliefs about
social categories. British Journal of Social Psychology, 39, 113–127.
Instituto Nacionalde Estadı´stica. (2002). Estadı´sticas sociales de los pueb-
los indı´genas en Chile [Social statistics of the indigenous peoples in
Chile]. Santiago, Chile: Gobierno de Chile.
Iyer, A., Leach, C. W., & Crosby, F. (2003). White guilt and racial
compensation: The benefits and limits of self-focus. Personality &
Social Psychology Bulletin, 29, 117–129.
Iyer, A., Schmader, T., & Lickel, B. (2007). Why individuals protest the
perceived transgressions of their country: The role of anger, shame, and
guilt. Personality & Social Psychology Bulletin, 33, 572–587.
Leach, C. W., Iyer, A., & Pederson, A. (2006). Anger and guilt about
ingroup advantage explain the willingness for political action. Person-
ality & Social Psychology Bulletin, 32, 1232–1245.
Leach, C. W., Snider, N., & Iyer, A. (2002). Poisoning the consciences of
the fortunate: The experience of relative advantage and support for
social equality. In I. Walker & H. Smith (Eds.), Relative deprivation:
Specification, development, and integration (pp. 136–163). New York:
Cambridge University Press.
Lepore, L., & Brown, R. (1997). Category activation and stereotype
accessibility: Is prejudice inevitable? Journal of Personality and Social
Psychology, 72, 275–287.
Lewis, H. B. (1971). Shame and guilt in neurosis. New York: International
Universities Press.
Lickel, B., Schmader, T., & Barquissau, M. (2004). The evocation of moral
emotions in intergroup contexts: The distinction between collective guilt
and collective shame. In N. R. Branscombe & B. Doosje (Eds.), Col-
lective guilt: International perspectives (pp. 35–55). New York: Cam-
bridge University Press.
Lickel, B., Schmader, T., Curtis, M., Scarnier, M., & Ames, D. R. (2005).
Vicarious shame and guilt. Group Processes and Intergroup Relations,
8, 145–157.
McClelland, G. H., & Judd, C. M. (1993). The statistical difficulties of
detecting interactions and moderator effects. Psychological Bulletin,
114, 376–390.
McGarty, C., Pederson, A., Leach, C. W., Mansell, T., Waller, J., & Bliuc,
A.-M. (2005). Group-based guilt as a predictor of commitment to apol-
ogy. British Journal of Social Psychology, 44, 659–680.
Mead, G. H. (1934). On social psychology. Chicago: University of Chicago
Press.
Ministerio de Planificacio´n y Cooperacio´n. (2003). Informe comisio´n de
verdad histo´rica y nuevo trato de los pueblos indı´genas [Commission of
historical truth and new treatment of the indigenous peoples]. Santiago,
Chile: Gobierno de Chile.
Niedenthal, P. M., Tangney, J. P., & Gavanski, I. (1994). “If only I
weren’t” versus “If only I hadn’t”: Distinguishing shame and guilt in
counterfactual thinking. Journal of Personality and Social Psychology,
67, 585–595.
Pederson, A., Beven, J., Walker, I., & Griffiths, B. (2004). Attitudes
toward indigenous Australians: The role of empathy and guilt. Journal
of Community and Applied Social Psychology, 14, 233–249.
Pettigrew, T. F., & Meertens, R. W. (1995). Subtle and blatant prejudice in
Western Europe. European Journal of Social Psychology, 25, 57–75.
Roccas, S., Klar, Y., & Liviatan, I. (2004). Exonerating cognitions, group
identification, and personal values as predictors of collective guilt
among Jewish-Israelis. In N. R. Branscombe & B. Doosje (Eds.), Col-
lective guilt: International perspectives (pp. 130–147). New York: Cam-
bridge University Press.
Roseman, I. J., West, C., & Schwartz, T. S. (1994). Phenomenology,
behaviors, and goals differentiate discrete emotions. Journal of Person-
ality and Social Psychology, 67, 206–221.
Saiz, J.-L. (2002). Atribucio´n de estereotipos: Los indı´genas Mapuches que
perciben los Chilenos [Attribution of stereotypes: the indigenous Mapu-
che as perceived by Chileans]. In J. F. Morales, D. Pa´ez, A. L. Kornblit,
&D.Asu´n (Eds.), Psicologı´a social (pp. 145–151). Buenos Aires,
Argentina: Pearson Education.
Schmader, T., & Lickel, B. (2006). Stigma and shame: Emotional re-
sponses to the stereotypic actions of one’s ethnic ingroup. In S. Levin &
C. van Laar (Eds.), Stigma and group inequality: Social psychological
approaches (pp. 261–286). Mahwah, NJ: Erlbaum.
Shaver, P., Schwartz, J., Kirson, D., & O’Connor, C. (1987). Emotion
knowledge: Further exploration of a prototype approach. Journal of
Personality and Social Psychology, 52, 1061–1086.
Smith, E. R. (1993). Social identity and social emotions: Toward new
conceptualizations of prejudice. In D. M. Mackie & D. L. Hamilton
(Eds.), Affect, cognition and stereotyping (pp. 297–315). San Diego, CA:
Academic Press.
Smith, R. H., Webster, J. M., Parrott, W. G., & Eyre, H. L. (2002). The role
of public exposure in moral and nonmoral shame and guilt. Journal of
Personality and Social Psychology, 83, 138–159.
Steele, S. (1990). The content of our character: A new vision of race in
America. New York: St. Martin’s Press.
Swim, J. K., & Miller, D. L. (1999). White guilt: Its antecedents and
consequences for attitudes toward affirmative action. Personality &
Social Psychology Bulletin, 25, 500–514.
Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup
behavior. In S. Worchel & W. G. Austin (Eds.), Psychology of inter-
group relations (pp. 7–24). Chicago: Nelson-Hall.
Tangney, J. P. (1991). Moral affect: The good, the bad, and the ugly.
Journal of Personality and Social Psychology, 61, 598–607.
Tangney, J. P., & Fischer, K. W. (1995). Self-conscious emotions: The
89
COLLECTIVE GUILT AS A PREDICTOR OF REPARATION
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

psychology of shame, guilt, embarrassment and pride. New York: Guil-
ford Press.
Tangney, J. P., Miller, R. S., Flicker, L., & Barlow, D. H. (1996). Are
shame, guilt, and embarrassment distinct emotions? Journal of Person-
ality and Social Psychology, 70, 1256–1269.
Tangney, J. P., Wagner, P., Fletcher, C., & Gramzow, R. (1992). Shamed
into anger? The relation of shame and guilt to anger and self-reported
aggression. Journal of Personality and Social Psychology, 62, 669–675.
Turner, J. C., Hogg, M. A., Oakes, P. J., Reicher, S. D., & Wetherell, M. S.
(1987). Rediscovering the social group: A self-categorization theory.
Oxford, MA: Blackwell.
Weiner, B. (1995). Judgments of responsibility: A foundation for a theory
of social conduct. New York: Guilford Press.
Yzerbyt, V. Y., Rocher, S., & Schadron, G. (1997). Stereotypes as expla-
nations: A subjective essentialistic view of group perception. In R.
Spears, P. J. Oakes, N. Ellemers, & A. Haslam (Eds.), The social
psychology of stereotyping and group life (pp. 20–50). Oxford, MA:
Blackwell.
Zagefka, H., Gonza´lez, R., Brown, R., & Manzi, J. (2005). To know you is
to love you? Differential longitudinal effects of intergroup contact and
knowledge on intergroup anxiety and prejudice among indigenous and
nonindigenous Chileans. Unpublished manuscript.
Received September 23, 2005
Revision received May 21, 2007
Accepted May 22, 2007 䡲
New Editors Appointed, 2009–2014
The Publications and Communications Board of the American Psychological Association an-
nounces the appointment of six new editors for 6-year terms beginning in 2009. As of January 1,
2008, manuscripts should be directed as follows:
● Journal of Applied Psychology (http://www.apa.org/journals/apl), Steve W. J. Kozlowski,
PhD, Department of Psychology, Michigan State University, East Lansing, MI 48824.
● Journal of Educational Psychology (http://www.apa.org/journals/edu), Arthur C. Graesser,
PhD, Department of Psychology, University of Memphis, 202 Psychology Building, Memphis,
TN 38152.
● Journal of Personality and Social Psychology: Interpersonal Relations and Group Processes
(http://www.apa.org/journals/psp), Jeffry A. Simpson, PhD, Department of Psychology,
University of Minnesota, 75 East River Road, N394 Elliott Hall, Minneapolis, MN 55455.
● Psychology of Addictive Behaviors (http://www.apa.org/journals/adb), Stephen A. Maisto,
PhD, Department of Psychology, Syracuse University, Syracuse, NY 13244.
● Behavioral Neuroscience (http://www.apa.org/journals/bne), Mark S. Blumberg, PhD, De-
partment of Psychology, University of Iowa, E11 Seashore Hall, Iowa City, IA 52242.
● Psychological Bulletin (http://www.apa.org/journals/bul), Stephen P. Hinshaw, PhD, Depart-
ment of Psychology, University of California, Tolman Hall #1650, Berkeley, CA 94720.
(Manuscripts will not be directed to Dr. Hinshaw until July 1, 2008, as Harris Cooper will
continue as editor until June 30, 2008.)
Electronic manuscript submission: As of January 1, 2008, manuscripts should be submitted
electronically via the journal’s Manuscript Submission Portal (see the website listed above with
each journal title).
Manuscript submission patterns make the precise date of completion of the 2008 volumes
uncertain. Current editors, Sheldon Zedeck, PhD, Karen R. Harris, EdD, John F. Dovidio, PhD,
Howard J. Shaffer, PhD, and John F. Disterhoft, PhD, will receive and consider manuscripts through
December 31, 2007. Harris Cooper, PhD, will continue to receive manuscripts until June 30, 2008.
Should 2008 volumes be completed before that date, manuscripts will be redirected to the new
editors for consideration in 2009 volumes.
90
BROWN, GONZA
´
LEZ, ZAGEFKA, MANZI, AND C
ˇ
EHAJIC
´
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.