Running Head: GUILT IN SOCIAL CHANGE CONTEXT Group-based guilt and reparation in the context of social change

Publisher's copyright statement: This is the peer reviewed version of the following article: Dumont, K. & Waldzus, S. (2014). Groupbased guilt and reparation in the context of social change. Journal of Applied Social Psychology. 44 (4), 331-341, which has been published in final form at https://dx.doi.org/10.1111/jasp.12224. This article may be used for non-commercial purposes in accordance with the Publisher's Terms and Conditions for self-archiving.


Introduction
Probably no country in history has so openly and thoroughly confronted its past as South Africa has done (Gibson, 2004). After the first democratic elections in 1994, Nelson Mandela as the then president of the African National Congress (ANC) government, appointed the Truth and Reconciliation Commission (TRC) in 1995 to expose and document human rights abuses perpetrated under apartheid. The approach of the TRC aimed at healing communities by truth-telling, forgiveness, acceptance and trust, and thus played a key role in facilitating transitional justice and social equality. Since race reflects the tortured history of apartheid, it is not surprising that public debates on guilt as a response to apartheid were racialist. The debate between the two South African Nobel Prize laureates, Desmond Tutu and FW de Klerk, illustrates this point. Tutu, as chair of the TRC, on various occasions expressed his regret that " [...] by and large, the white community failed to take advantage of the Truth and Reconciliation process [...]" and his belief that "[...] many of them carry a burden of guilt" (Tutu, 2003, p. 1). FW de Klerk rejected any form of assigning collective guilt to white South Africans as it would label them as "morally inferior" and because labels of this kind would be the cause of ethnic tension, prejudice and discrimination (de Klerk, 2000). Research has demonstrated that the perspectives -particularly of white and black "born-free" South Africans -on guilt and reparation as responses to apartheid are as polarized as the perspectives of these two Nobel Prize laureates (Gibson, 2004;Klandermans, Werner & van Doorn, 2008).
South Africa is not unique with regard to its atrocious history and its way of dealing with it but it seems quite distinct in experiencing one of the most radical social change processes at the same time. The ANC government did not only address the legacy of apartheid by appointing the TRC but also by developing and implementing racial transformation policies, which are far more extensive than those adopted elsewhere (Habib & 4 Bentley, 2008). Consequently, the government's racial transformation policies include not only affirmative action in education, employment and sport, but also in programmes such as black economic empowerment that address transformation of medium-and large-business enterprises to achieve black ownership, management, and skills development. Since the transition to black majority rule, many white South Africans increasingly feel marginalized as a small minority with much to lose from these race-targeted policies (Durheim et al., 2009).
Thus, the question arises as to which impact the social change processes have on white South Africans' sense of guilt and reparation intentions as a response to apartheid. Two studies conducted with white adolescents will be reported which, to our knowledge, are the first to test the relationship between group-based guilt and reparation by considering changes in the relations of status between former perpetrator and victim groups within a real intergroup setting.
Since Doosje et al. (1998) in-group identification has been defined as a determinant in the experience of group-based guilt and its consequences for reparation. The general assumption proposed by these authors is that people are motivated to hold a positive view of their group, particularly when they strongly identify with the group. Consequently, people who attach great importance and commitment to a group are less likely to accept negative information about their group (Branscombe, Ellemers, Spears & Doosje, 1999). With regard to group-based guilt, the authors therefore hypothesised and found that under certain conditions high in-group identification leads people to search for a means of avoiding this feeling and, in turn, to a lower likelihood of support for reparations to the out-group. The conditions that either reduce or enhance the impact of in-group identification are ambiguous valence of information about transgression (see Doosje et al., 1998), information about transgression provided by the out-group (see Doosje et al., 2004), previously offered financial reparation (see Doosje et al., 2004Doosje et al., , 2006; and taking the perspective of the disadvantaged group (see Zebel, Doosje & Spears, 2004).
Whilst Doosje et al. (1998Doosje et al. ( , 2004Doosje et al. ( , 2006 focused on contextual conditions moderating the relationship between in-group identification, group-based guilt, and reparation intentions, McGarthy (2002), andBranscombe (2004), proposed in-group identification as a distal antecedent that is mediated via proximal constructs in predicting group-based guilt and, consequently, reparation tendencies. According to Branscombe (2004), in-group identification and its correlates, such as group salience and group norms, 6 play an indirect role via proximal predictors, such as the perception of responsibility of the in-group's actions, perception of illegitimacy/immorality of the in-group's actions, and the cost and difficulty of achieving justice (Branscombe et al., 2002;Lickel et al., 2005;Wohl & Branscombe, 2008;and Cehajic et al., 2009).
As social context has been demonstrated in numerous studies to determine the when and how in-group identification predicts group-based guilt, one can assume that social context is equally important in determining the when and how group-based guilt results in reparation intentions. The present study will particularly focus on the latter by studying the impact of social change processes on the relationship between group-based guilt and reparation intentions.

Social change and group-based guilt
Research has demonstrated that experienced social change or the perceived possibility of social change determines attitudes and behaviour towards out-groups. Riek, Mania and Gaertner (2006) found in their meta-analysis on intergroup threat that realistic threat, which includes perceptions of threats to physical and economic well-being of the in-group, is a strong predictor of negative out-group attitudes towards low-status rather than high-status groups. The authors reason that this effect "[...] may be due to high status group members' perception that low status out-groups are trying to usurp their resources" (Riek et al., 2006, p. 345). In line with the conclusion of Riek et al. (2006) are the findings of Leach, Iyer and Pedersen (2006, Study 3) and Harth, Kessler and Leach (2008); that demonstrated that groupbased guilt was predictive for reparation that is aimed at compensation but not if it is aimed at promoting equality.
A similar pattern has been identified in studies which focus on intergroup help (see Jackson & Essen, 2000;Nadler & Halabi, 2006;and Cunningham & Platow, 2007). Jackson and Essen (2000) demonstrated, for instance, that members of a materially high status group 7 were willing to provide empowerment to a low status out-group on condition that it did not threaten the in-group's material advantage. Arguing from a social identity perspective, Nadler (2002) and Nadler and Halabi (2006) showed that any threat by the out-group to the status of the in-group, whether material or related to other comparison dimensions, results in withholding empowerment help.
We would like to argue that as out-group attitudes and intergroup help are affected by social change, so probably are group-based guilt and reparation. Compensation toward the group of victims as a response to group-based guilt felt by the perpetrators not only unburdens them from an unpleasant negative emotion, but also regulates and eventually changes the intergroup relationship. Therefore, as group-based guilt and reparation intentions stand in a functional relationship with social change, we propose that changes in the intergroup relations impact on group-based guilt and reparation intentions. Whether the relationship between group-based guilt and reparation intentions is moderated by the perception of social change was tested in two studies that were conducted with white South Africans who are too young to have been personally involved in apartheid-related transgressions (the "born-free generation").

Study 1
To examine the role of group-based guilt and its relationship to reparation intentions in the context of South Africa, we first explored the salience of apartheid in white adolescents. This was achieved by asking the participants to name three events in the history of South Africa that came spontaneously to mind. Second, we explored whether white participants perceive social change among white and black South Africans as a loss of status for their in-group and a gain of status for the out-group. Last, the following hypotheses were tested: As participants were not provided with either positive or negative information about apartheid and previous research had indicated that the perspectives on apartheid vary 8 (Gibson, 2004), we assumed to replicate the findings of Doosje et al. (1998Doosje et al. ( , 2006) on ambiguous valance of information and expected that identification with the in-group (white South Africans) would negatively predict group-based guilt. Further, in line with previous theories and research demonstrating that the relationship between in-group identification and group-based guilt is mediated by proximal cognitive factors (Branscombe, 2004), the study aimed to demonstrate that the negative relationship between in-group identification and group-based guilt is mediated by perceived in-group responsibility (H1). Most important and consistent with our reasoning on group-based guilt and its consequences within the context of social change, it was predicted that the relationship between group-based guilt and reparation would be moderated by perceptions of status change of the in-group. More precisely, participants should be less ready to react to group-based guilt with reparation to the group of victims the more they perceived that the in-group has been losing status (H2).
Change in the in-group's status on its own does not capture the complexity of social change. It is relative, that is to say, perceived in-group status change is either the outcome of comparison processes with relevant out-groups and/or the outcome of temporal comparison processes (Pettigrew & Meertens, 1995). In line with Social Identity Theory (Tajfel & Turner, 1986) the study therefore also controlled for perceived changes of the status of the out-group as well as the average status of the in-group and out-group spanning from the past to the future. As readiness for reparation is probably constrained by participants' perception of the social status of their own family and its change, it was decided to control for both variables, too. Further, the relative ideal status of the in-group was treated as covariate, because the attitude toward actual status change might depend on such a point of reference.

Participants
Participants were white pupils from a middle-class, multiracial school based in East London, South Africa. The average age of the students was 16 years (with a range from 15 to 17 years of age). Altogether 52 participants submitted completed questionnaires. Twentyseven females and 25 males participated. Ethical clearance for the study was granted by the University of Fort Hare (East London), and consent was obtained from the headmaster of the school and the pupils.

Procedure and measures
Participants were informed that the questions formed part of a project that addresses learners' ideas on whether history is important for people's current lives or not. The participants were first asked to name three historical events in the history of South Africa that came to mind spontaneously. They were then given a range of questions by which the principle variables were measured. Participants completed the questionnaires in their classrooms, in the presence of a teacher and one research assistant. They were informed that the study was anonymous. On completion of the questionnaire, all participants were debriefed.
If not otherwise stated, participants were asked to indicate their agreement or disagreement with each statement on a 5-point Likert scale, ranging from 1 (totally disagree) to 5 (totally agree).
In-group identification was measured by four items selected and adapted from the 10 item self-investment scale proposed by Leach et al. (2008): 'I feel committed to white South Africans'; 'I am glad to be a white South African'; 'I think that white South Africans have a lot to be proud of'; and 'Being a white South African is an important part of how I see myself' (alpha = .65).
Perceived in-group responsibility was measured by three items adapted from the scale developed by Cehajic and Brown (2006): 'White South Africans should feel responsible for the things that happened during apartheid'; 'I think that white South Africans are responsible for what happened during apartheid'; and 'I consider white South Africans as responsible for the atrocities committed during apartheid' (alpha = .66).
Group-based guilt was measured by four items adapted from the scale developed by Brown et al. (2008) and Brown and Cehajic (2008)  Social change at the group level was operationalized as perceived change in the economic status of white and black South Africans. It was measured by applying the intergroup perception ladder representing an adaptation of Cantril's Self-Anchoring Scale (Finchilescu & de la Rey, 1991), which assesses intergroup comparison by including the temporal dimension. Participants were presented with a drawing of a ladder with 12 rungs (labelled from 0 to 11) and asked to imagine that this ladder represented economic status in 11 South Africa. The top step represented the best economic status one could imagine, whilst the bottom step represented the worst. The task of the participants was to indicate on which step the in-group (that is, white South Africans) and the comparison group (namely black South Africans) had stood in the past (25 years ago); the step on which they stand today; on which step they will stand in 15 years' time; in 50 years' time; and on which they should ideally stand.
Family status was operationalized as change in the economic status of the individual's family in comparison to that of the out-group (that is, black South Africans). The measurement was identical to the one used to assess perceived social change at a group level, except that participants were asked to indicate their opinion about on which step their family and the comparison group stood economically in the past (25 years ago), on which step they stood today; on which they will stand in 15' and in 50 years' time. Last, the information of the ideal in-group status was used as a measure.

Preliminary analysis
First, we assessed whether apartheid is a salient historical event for the participants.
Two independent raters coded the three events listed by the participants. Two types of 12 historical event were distinguished: apartheid-related events versus events unrelated to apartheid.
Apartheid-related events include apartheid as an historical period (usually named respectively. The two raters discussed each ambiguous case until agreement was reached.
Thirty-four participants named an apartheid-related event as the first historical event that came to mind. Nine mentioned it as a second event, and four named it in third place.
Consequently, only five participants did not mention an apartheid-related event at all.
Although, an overwhelming majority of our participants named apartheid-related events, they were on average rather reluctant in accepting responsibility for apartheid and for supporting reparation; and indifferent in accepting guilt (see mean values in Table 1).
<Insert Table 1 here> 13 Furthermore, the study explored how participants perceived changes in the economic status relations between white and black South Africans. Figure 1 depicts the perceptions of status loss for white and status gain for black South Africans. The perceived group statuses of white and black South Africans differed significantly for the past, t(49) = 13.04, p < .001; present, t(49) = -6.93, p < .001; the future in 15 years, t (49) = -7.07, p < .001; and in 50 years, t(48) = -4.3, p < .001.

Hypotheses testing
As the first step, we tested the hypothesis that in-group identification negatively predicts group-based guilt and that this relationship is mediated by perceived in-group responsibility (H1). We tested H1 with regression analysis by following the suggestions of Preacher and Hayes (2008)  group-based guilt, in that the more participants identify with white South Africans, the less they perceive in-group responsibility for apartheid and the less they feel group-based guilt.
<Insert Table 2 here> As the second step, the effect of group-based guilt on reparation intentions moderated by perceived in-group status change (H2) was estimated by using Ordinary Least Squares (OLS) regression (Hayes & Matthes, 2009). In-group identification and all remaining status and status change variables were included as statistical controls ( Table 2). The results indicate the main effects of guilt, in-group status change, family status change and out-group status change, R 2 = .73, F (10, 37) = 9.79, p < .001, n = 48. In line with H2, the interaction term between guilt and in-group status change was statistically significant, and increased significantly the explained variance of reparation intentions, ∆R 2 = .04, F (1, 47) = 5.17, p < .05. 1 The unstandardized simple slopes as depicted in Table 2 indicate, as predicted, that the relationship between guilt and reparation was strongest for in-group status stability/gain (at 1 SD above the mean level), but declined the more a loss of status was perceived (at mean level and at 1 SD below mean level).

Discussion
The results of Study 1 confirmed both hypotheses: First, that identification represents a distal predictor for group-based guilt that is mediated by the proximal predictor perceived in-group responsibility for the wrongdoing (Branscombe, 2004). Second, the results clearly support the prediction that the more participants perceived that the in-group has been losing status, the weaker the relationship between group-based guilt and reparation was.
Given that this is, to our knowledge, the first study considering the perceptions of changes in status relations between former groups of perpetrators and of victims in a real intergroup context and its impact on the relationship between group-based guilt and reparation, we decided to replicate the study, using a larger sample.

Study 2
The aim of Study 2 was to replicate the finding that the relationship between groupbased guilt and reparation intentions is moderated by in-group status change (H1). It was decided to provide a less ambiguous setting, in that apartheid was portrayed as an atrocity and white South Africans were portrayed as perpetrators, and black South Africans were portrayed as victims. In accordance with Doosje et al. (1998Doosje et al. ( , 2006 and Iyer and Leach (2009), we assumed that such unambiguous contextual information might make it impossible even for high identifiers to deny the in-group's responsibility for apartheid. Hence, we expected that the importance of pre-existing differences in in-group identification as predictor of group-level emotions will be reduced. The latter would result in no direct association between in-group identification and group-based guilt.

Participants
The chosen participants were whites from a lower middle-class, multiracial high school in the area of East London, South Africa. Altogether 108 participants (aged from 13 -20, M = 15.5, 60 female, 48 male) submitted completed questionnaires. Ethical clearance for the study was approved by the University of Fort Hare (East London), and consent was obtained from the headmaster of the school and the pupils.

Procedure and measures
Study 2 differed from Study 1 in that participants were provided with a short introduction in which apartheid was portrayed as an atrocity and white South Africans were portrayed as perpetrators and black South Africans as victims (e.g. negative valence of information, Doosje et al., 1998Doosje et al., , 2006. After reading the information, participants were asked to answer a range of questions. If not otherwise stated, participants were asked to indicate their agreement or disagreement with each statement on a 6-point Likert scale ranging from 1 (totally disagree) to 6 (totally agree).
In-group identification was measured by the four items used in Study 1. Two additional items of the 10 item self-investment scale  were included to redress the rather low Cronbach's alpha as found in Study 1: 'I feel solidarity with white South Africans' and 'I feel a bond with white South Africans' (alpha = .73).
Group-based guilt, reparation and social change. The measures for group-based guilt (alpha = .85) and reparation (alpha = .84) were identical to the measures used in Study 1, as was the instrument to assess and compute the social change variables (in-group status change, in-group status, out-group status change, out-group status, family status change, family status and relative ideal status of the in-group).

Preliminary analysis
In line with Doosje et al. (1998Doosje et al. ( , 2006 and Iyer and Leach (2009), but differing from Study 1, in-group identification did not correlate with any of the assessed variables (Table 3).
We therefore decided not to include this variable in any further analysis. 2 The less ambiguous setting, by portraying apartheid as an atrocity, might also have contributed to the fact that participants accepted guilt on average. Moreover, the results replicate the perceptions of the loss of status for white and the gain of status for black South Africans (see Figure 2). The perceived status positions of white and black South Africans in the past, t(101) = 14.64, p < .001; at present, t(102) = -6.82, p < .001; in 15 years, t(102) = -7.85, p < .001; and in 50 years, t(102) = -7.53, p < .001, differed significantly.

Hypothesis testing
The hypothesis (H1), that the effect of group-based guilt on reparation tendencies is moderated by perceived in-group status change was tested as in Study 1. All remaining status and status change variables were included as statistical controls, R 2 = .33, F(9, 86) = 4.81, p < .001, n = 96. A significant main effect was found for guilt, whilst a marginal direct effect was found for the in-group's ideal status (Table 4). All other variables did not have any direct impact on the intentions of reparation. As predicted by H1, the interaction term between guilt and in-group status change was significant, and increased significantly the explained variance of reparation intentions, ∆R 2 = .03, F (1, 95) = 4.12, p < .05. As in Study 1, the effect of group-based guilt on reparation was strongest for in-group status stability/gain (at 1 SD above the mean level), and declined to non-existent the more participants perceived in-group status loss (at mean level and at 1 SD below mean level).

Discussion
The results clearly replicated the finding of Study 1 that the relationship between group-based guilt and reparation is moderated by in-group status change, that is to say, the more white participants perceive the loss of status for their in-group, the less group-based guilt results in the behavioural consequence of supporting reparation.

General discussion
Group-based guilt and reparation have been studied extensively in recent years. The findings of this research have improved our understanding of the psychological and contextual factors relevant for this intergroup emotion and its consequences. However, little research has examined such emotion and its consequences in contexts in which it is probably most important, namely in societal contexts that experience radical social change processes.
We argued and hypothesised that group-based guilt and reparation intentions toward the group of victims as its consequence stand in a functional relationship with social change. The 18 results of two correlative studies conducted within the context of South Africa supported the hypothesis by consistently showing that the more white participants perceive a loss of status for white South Africans, the weaker is the relationship between group-based guilt and reparation intentions.
The present results contribute to at least three research domains. First, they complement research on group-based guilt and reparation tendencies that have been conducted mainly with dominant groups in relatively stable intergroup relations. Whilst replicating the importance of in-group identification in predicting group-based guilt in ambiguous (Study 1), but not in un-ambiguous (Study 2) contexts (Doosje et al., 1998, the current results go beyond this research as they speak to the contingencies of guilt-related behaviour at times when the status of the in-group is at stake. Second, by providing evidence that people take into account the implications of group-based guilt and reparation for current intergroup dynamics, the current findings also complement previous research on consequences of social change for intergroup attitudes and intergroup support (Riek et al., 2006;Jackson & Essen, 2000;Nadler, 2002;Nadler & Halabi, 2006;and Cunningham & Platow, 2007). Last, although we found convincing evidence for the moderating role of perceived status loss on the relation between group-based guilt and readiness for reparation, our data do not allow us to conclude whether this moderation is driven by the promotion of the change of status among those participants who see a lack of it or by a more reactive prevention of a further loss of status among those participants who see too much of it. One could argue that, similar to the different functions of in-group bias (Scheepers, Spears, Doosje & Manstead 2006), both processes might play a role. Which one is more important depends on the situation. For example, the ideal status for white South Africans expressed by the participants only had a negative impact on the readiness for reparation when apartheid was unequivocally framed as an atrocity in the instructions (that is, in Study 2 but not in Study 1). The clear, negative framing in Study 2 might have given more weight to differences in support of social change compared to those in Study 1. However, the current data do not allow for more than speculation on this issue. Further experimental research is necessary to distinguish between different functions of guilt-related behavioural tendencies.
Additional limitations of the current studies should be addressed in future research.
First, the hypotheses in the present research were tested by using correlational studies. The replication of the findings using a more controlled methodological approach is pending.
Second, the present study did not include possible proximal factors such as prejudice which might mediate the moderation function of the change of status as found in the present studies.
Third, the present research was limited to the perspective of one group (white participants).
As the context of the change of status is, however, characterized by numerous intergroup interactions and discourses within the society that involves the perspectives of all groups, one can assume that the shared belief systems which impact on group-based emotions and intergroup behaviour also depend on relevant expectations, beliefs and goals held by members of the former group of victims. Future research should explore whether or not the perceptions of social change have a similar impact on the assignment of group-based guilt and demand for reparations by members of the former groups of victims (Wohl & Branscombe, 2008). Fourth, the present study did not incorporate alternative responses to apartheid such as moral outrage and anger which can be either directed toward the in-group or the out-group and which are considered to be powerful motivators toward justice and equality (Iyer, Leach & Petersen, 2004). Since both group-based emotions are considered to result in efforts that go beyond apologies and compensation, their prevalence among young white South Africans as well as their impact within a social change context should be studied in future research.

20
Finally, besides all outlined limitations, the results of the present studies point toward a trend among young white South Africans that needs to be taken seriously. Study 1 revealed that an overwhelming majority of our participants considers apartheid as an important experience in the history of South Africa. This consideration might be a result of political correctness but can also be seen as an indicator of how omnipresent apartheid is in the minds of young white South Africans. Although apartheid is in their minds, they seem to distance themselves from it (see Study 1). Even when feeling guilt as a response to apartheid, they hesitate in opting for reparations when the status of their in-group is perceived as already declining (Studies 1 and 2). The latter implies that losing economic status might be perceived as "already paying for the past". Yet, reducing apartheid to a legacy of economic inequality between white and black South Africans bears the risk of denying other important dimensions such as the human-rights violations and racism against black communities during apartheid and at present. Such denial would not only reinforce the polarized perspectives on apartheid as expressed by the two Nobel Prize laureates, His Excellency the Archbishop Desmond Tutu and former president FW de Klerk but it can also hinder the development of a shared culture of social, political and economical inclusion that is so important for the new South Africa to uphold the civil and human rights of its constitution. Footnotes 1. Analyses without these covariates in this and the second study showed the assumed direction of the interaction term between guilt and in-group status change but did not reach statistical significance. The results can be obtained on request from the first author.
2. The inclusion of the in-group identification in the model in a separate OLS regression (Hayes & Matthes, 2009) did not change the pattern of the results and identification did not have any direct impact on reparation tendencies, as found in Study 1. Note: One SD below the mean indicates the perception of in-group status loss, while one SD above the mean indicates the perception of in-group status gain/stability. Note: † p < .10, * p < .05, ** p < .01, *** p < .001; means*** are significant (p < .001) different from 0 for the social change variables and from 3.5 for the remaining variables.