This is an open access article distributed under the terms of the Creative Commons
Attribution License (After the genocide committed against the Tutsis by extremist Hutus in 1994, the Rwandese government has promoted national identity as a tool of reconciliation, as a replacement of the subgroup (sometimes referred to as ethnic) identities. At a conference held in the office of the president in 1995, one of the recommendations was the prohibition of the use of ethnic references in schools, universities and public and private sectors (Présidence de la République Rwandaise, 1995). A special clause in the constitution of the Republic of Rwanda of 4th June 2003, Chapter II, Title I, article 9 concerns “eradication of ethnic, regional and other divisions and promotion of national unity.” Following this, the mention of ethnicity in identity cards has been removed, and any reference to ethnic groups and identities has been banned in the public sphere.
The Rwandese government highlights that both the Tutsis and the Hutus come from one ancestor, Gihanga Ngomijana (de Heusch, 1966; Kagame, 1959; Muzungu, 2006). The Rwandese government’s version of national history, which stresses harmonious relationships between Hutus and Tutsis during pre-colonial times, is also enforced through Ingando camps, in which former combatants, released prisoners, or future members of the country’s elite are sent for civic education and training (see also Bilali, 2014).
The initiative for the promotion of Rwandese nationality seems politically astute given the excesses associated with ethnicity, which led to one of the worst genocides in human history. As observed by the Rwandese political scientist Shyaka (2003, p. 191):
The foundation of the Rwandan identity can never be reduced to the ethnic particularities of Hutu, Tutsi and Twa. The latter would even weaken it. The Rwandan identity on the other hand, offers a vital structure that lessens conflicts and counteracts the genocidal ideology. Through the promotion of confidence among citizens, and by unifying the collective identity perceptions of Rwandans, the Rwandan identity brings to light an ontological essence of the true roots of Rwandans from a shared “organic nation” where commonalities and convergences take precedence over differences and divergences.
However, the problem that arises is to know whether this promotion of Rwandese national identity is effective in reducing the antagonism between the genocide survivors (members of the victim group) and the non-victims (members of the perpetrator group).
The Common In-Group Identity Model
In social psychological terms, the recommendation that Rwandans should identify with the Rwandan nation rather than ethnic groups evokes the common in-group identity model (CIIM). According to this model, identification with a group at a superordinate level tends to improve relationships between members of subordinate groups (Gaertner & Dovidio, 2000). This model is itself inspired by social identity (Tajfel & Turner, 1986) and self-categorization (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987) theories. It is based on the idea that people have multiple social identities that are organized hierarchically, from the more specific (the individual self) to the more abstract (humanity), with group identity being situated at an intermediate level. According to the CIIM, recategorizing sub-groups (e.g., ethnic groups) into a superordinate category (e.g., Rwanda), increases the perceived similarity between members of different subgroups, therefore leading to more harmonious intergroup attitudes and relations. One aim of our study is to test, in the Rwandese context, whether identifying with the superordinate national group is associated with more reconciliation sentiment among members of the victim and of the perpetrator groups. In addition, we will examine the mediating role of perceived similarity.
These positive effects of identification with a superordinate group have been demonstrated in several empirical studies in both laboratory experiments with minimal groups (Dovidio, Gaertner, Validzic, Matoka, Johnson, & Frazier, 1997) and among members of real life groups, including nationality, sport teams, and shared sexual orientation (Levine, Prosser, Evans, & Reicher, 2005; Levine & Thompson, 2004; Stürmer, Snyder, & Omoto, 2005). Closer to our purpose, studies conducted by Wohl and Branscombe (2005) examined whether the inclusive categorization (that is to say, as humans) leads to forgiveness towards, and less collective guilt assigned to the group that is held responsible for past atrocities or injustices against the in-group. The participants were North American Jews and Aboriginal Canadians. The results show that when the categorization as human beings was made salient, members of the victim group (North American Jews or Aboriginal Canadians) saw themselves as similar to members of the perpetrator group (Germans and European Canadians), assigned less collective guilt, and were more willing to forgive them for the harm caused. This was not the case when subordinate identities were made salient. Recent studies conducted in Central Africa (Democratic Republic of the Congo, Burundi, and Rwanda) by Vollhardt and Bilali (2014) also showed that holding an inclusive victimhood orientation – which amounts to include both in-group and out-group victims of a past conflict into a common victimized in-group – was associated with more positive intergroup attitudes, in contrast to negative intergroup attitudes that were associated with holding an exclusive victimhood orientation. These studies suggest that negative emotions or attitudes towards perpetrator groups may actually be reduced through inclusive categorization.
In the study by Wohl and Branscombe (2005), the temporal and generational distances between past hostilities and their effects were large. Whether the same effects would be observed in people who lived at the time of these hostilities and in close proximity to the perpetrators of harmful acts is one of the questions we will address in the present study. A recent study by Licata, Klein, Saade, Azzi, and Branscombe (2012) showed that identification with Lebanon had a positive effect on intergroup attitudes of young Christian Maronite participants toward Muslims after the Lebanese civil war (1975-1990), whereas identification with their religious group had the opposite effect. But the situation in Rwanda is even more critical, as the scale and severity of the violence was higher, and the events are more recent.
However, the CIIM might not equally apply to all social groups in all socio-historical contexts. In the context of asymmetric power relations, the norms reflected by the superordinate category are often perceived as those defined by the dominant group (Mummendey & Wenzel, 1999). As a consequence, minority group members might resist the inclusive categorization, which in return might backlash and raise distinctiveness threats. For example, Noor, Brown, and Prentice (2008a) examined the role of psychosocial variables involved in intergroup forgiveness. In the study conducted in Northern Ireland, a positive association between superordinate (Northern Irish) identification and intergroup forgiveness was obtained only among Irish Catholics; no association was observed among Protestants. Yet, they also identified with the superordinate level. One reason that has been advanced for this pattern is that Protestants may not have seen the superordinate identity as an inclusive identity but one that served to distinguish Protestants (a majority in Northern Ireland but a minority in Ireland) from Catholics (Noor, Brown, & Prentice, 2008b). In other words, it was not really viewed as an inclusive identity for all Northern Irish people. These results draw attention to the fact that the CIIM is not equally relevant for all groups.
Numerical Status and Inclusive Identification
Previous studies such as those by Noor et al. (2008b) indicate that the CIIM does not take into account the possibility that some groups may identify more than others at the superordinate level. Other studies (Binder et al., 2009; Dixon, Tropp, Durrheim, & Tredoux, 2010; Tropp & Pettigrew, 2005) point to the influence of one particular variable, the relative size of the group, on the effectiveness of policies aimed at reducing intergroup bias. They found that members of disadvantaged minority groups tend to be less influenced by interventions aiming to reduce intergroup bias than majority group members. In general, majority group members prefer policies that promote inclusive identification (Verkuyten, 2006), in order to maintain the social values that favour the status quo, and benefit the majority group (Dovidio, Gaertner, & Saguy, 2009; Jost & Banaji, 1994). In contrast, minority group members prefer policies that promote dual identity, and recognize and respect subgroup identities, while at the same time fostering a strong identification with the inclusive group (Dovidio, Gaertner, & Saguy, 2007; Hornsey & Hogg, 2000; Simon & Ruhs, 2008).
One reason why minority group members may prefer a dual identity strategy is that an exclusive focus on inclusive identity may actually be perceived as a threat to their distinct ethnic identity (Tajfel & Turner, 1986). Members of many racial and ethnic minorities, such as Blacks and Latinos in the United States, are particularly concerned about the loss of their distinct ethnic or cultural identities that is implied by a strong superordinate identity that corresponds, in reality, to the majority's identity (Bergsieker, Shelton, & Richeson, 2010). Dual identity becomes the choice strategy for minority group members because it provides them with respect for their identity while expressing, at the same time, shared values with the majority group.
In Rwanda, the genocide was perpetrated against members of a minority, the Tutsis, presented as a distinct ethnic group, by members of the ethnic majority, the Hutus. It might then be expected that members of the victim group will tend to be more reluctant than members of the non-victim group to identify with the superordinate group, Rwanda. This reluctance to give up their ethnic identity might be strengthened by the fact that this identity is associated with the recognition of their victim status. As a consequence, one might expect that superordinate group identification is more effective in reducing intergroup animosity among majority group members than among minority group members.
However, besides the groups’ numerical status, other factors might intervene in determining intergroup attitudes in this post-genocide situation. The socio-emotional needs of former victims and perpetrators could play a major role.
Needs of Victims and Perpetrators
According to the socioemotional needs-based model, victims of hostilities suffer a threat to their identity as meaningful and powerful actors. They generally experience a sense of loss of control, status and honour. At the same time, members of the perpetrator group experience threat to their identity as moral actors. They usually suffer from a sense of moral inferiority and are concerned about the rejection of the moral community to which they belong (Shnabel, Nadler, Ullrich, Dovidio, & Carmi, 2009). Thus, according to these authors, members of the victim group are likely to experience a threat of status and power loss, while members of the perpetrator group are likely to experience a threat to their moral image and social acceptance.
These threats felt by victims and perpetrators also produce different motivational consequences (Nadler & Shnabel, 2008). Victim groups could either seek revenge to restore their group's power, or prosecute their perpetrators to recognize the wrongs and injustices caused to the in-group. It is usually up to the victims to judge whether to cancel the moral debt of the perpetrators (Shnabel et al., 2009). These strategies, when adopted by the victim group, tend to increase the distancing with out-group members. At the same time, perpetrator group members could either deny the painful consequences of their actions and/or responsibilities for the harm caused to victims (Schönbach, 2010), or adopt positive attitudes towards members of the victim group, show empathy, ask for forgiveness (Nadler & Liviatan, 2006), and seek to forge ties of friendship with the victims (Shnabel et al., 2009).
Two studies conducted by Shnabel et al. (2009) with Israelis Jews presented either as perpetrators towards Arabs (victims) in the Israeli-Palestinian conflict or as victims of the Germans (perpetrators) during World War II showed that, when taking the victim's perspective, Israeli Jews expressed more conciliatory attitudes if they received empowering messages from the perpetrators whereas, when taking the perpetrator's perspective, they expressed more conciliatory attitudes if they received messages of acceptance from the victims.
In the Rwandese post-genocide situation, commemorations of the genocide are particularly sensitive moments for intergroup relations. Memorials are being created; ceremonies of exhumation and inhumation of corpses are organized; and, every year, one week of national commemoration takes place, starting on April 7th, and it is extended to a period of 100 days by some groups of survivors (Ibreck, 2010; Vidal, 2004). These commemorations are instrumental in recognizing the victims and are aimed at fostering reconciliation. However, the commemoration period might also revive the ethnic identities, as well as tarnish the moral image of the non-victim group members, therefore lowering their self-esteem. In this context, identifying with the national in-group, rather than with the perpetrator group, should allow members of the non-victim group to protect their self-esteem during commemorations. We might then expect that identification with Rwanda could also lead to more reconciliation sentiment because it increases self-esteem, in contrast with the moral stigma associated with the ethnic subgroup. However, this effect should only be obtained among members of the former perpetrator group.
Finally, as suggested by the social psychological literature on group-based emotions (Iyer & Leach, 2009; Wohl, Branscombe, & Klar, 2006), perceiving a common identity with the victims of one’s in-group’s misdeeds tends to increase negative self-focused emotions such as collective guilt, group-based shame, or remorse, whereas identifying with the subordinate in-group has the opposite effect. In turn, these emotions generally tend to be associated with more willingness to reconcile, or more positive behavioural intentions towards members of the victim group (Doosje, Branscombe, Spears, & Manstead, 1998). We will then also examine the potential mediating role played by group-based remorse.
Hypotheses
The first aim of this study is to test the common in-group identity model in the post-genocide situation in Rwanda, where members of the victim and of the non-victim groups lived through the genocide and continue cohabiting after it. Rwandans are currently subject to substantial legal and normative pressure that delegitimizes ethnic identities and imposes a single and inclusive national identity.
We will also check the mediating role of perception of similarity that derives from the common ingroup identity model, as well as that of self-esteem during genocide commemorations, inspired by the needs-based model of reconciliation, and that of group-based remorse, deriving from the literature on group-based emotions. We predict that the effect of Rwandan identification on reconciliation will be mediated by perceptions of similarity in the two groups, and that self-esteem during commemorations and group-based remorse will also mediate this link, but only among non-victims.
Indeed, albeit the commemorations of the genocide are emotionally disturbing for members of both groups (Ibreck, 2010), these events are particularly threatening for the moral image of the perpetrator group members, thus potentially undermining their self-esteem. In this context, identifying with the Rwandan nation could give members of the perpetrator group an opportunity to restore the negative identity and self-esteem that were bestowed on their ethnic group identity by their role in the genocide. National identity, which has been infused with a positive value since the political reconciliation process was put in place by the government, allows members of the perpetrator group to overcome their negative image and self-esteem. In turn, the satisfaction of their socio-emotional need for self-esteem during these commemorations should be associated with more reconciliation sentiment towards members of the victim group. Such an effect should not be expected among victims, whose identification with the Rwandan nation would not confer the same benefit, and could be potentially threatening, because it does not recognize their ethnic identity and/or it may be seen as a means of diluting their history of victimization. Thus, we would predict that self-esteem would mediate the effect of national identification on intergroup reconciliation, but only for members of the perpetrator group.
Finally, identifying with the superordinate national group should be associated with more group-based remorse among members of the perpetrator group, which should also mediate the positive association between Rwandan identification and reconciliation. Reconciliation should not be associated with group-based remorse among members of the victim group.
Method
Participants
The study was conducted in 2009 at the Institute of Agriculture, Technology and Education of Kibungo (INATEK) located in the Eastern Province of Rwanda. Participants were students at the Faculty of Psychology and Education in their 3rd and 4th years of bachelor degrees. The only inclusion criterion was to be Rwandese.
The sample comprised 247 students (203 women), including 113 genocide survivors (referred to as survivors), and 134 people who were not victims (referred to as non-victims). 49 persons who were abroad during the genocide also completed the questionnaire, but their answers were not included in the present study. Participants were between 20 and 55 years old (M = 30.74, SD = 8.41) at the time of data collection, in 2009. Most of them (62.3%) were at least 10 years old in 1994, and 81.5% of them declared they had lost at least one relative in this genocide, 99.1% among the survivors and 73.9% among the non-victims.
The relatively high number of survivors in the sample, compared with their demographic representation in Rwanda, could be attributed to two factors: genocide survivors might be more represented in psychology departments, which could be linked to their personal motivation to learn more about trauma and counseling in order to help their relatives and colleagues dealing with genocide related trauma. This cannot be ascertained given the absence of statistics about ethnicity in post-genocide Rwanda. The other possible explanation is that genocide survivors could have been more interested in participating in a study about reconciliation and commemoration of the genocide than non-victims. However, these are just tentative explanations that could not be tested in the present study.
Procedure
The Dean of the Faculty of Psychology at INATEK agreed to give access to the students. He also informed the lecturers about the study, and requested them to facilitate in the recruitment. Explanations about the study were provided to the students at the beginning of a class, and those who were willing to participate remained at the end of the lecture for completing the questionnaire. The principal investigator introduced herself and specified that participation was voluntary. Participants then signed the consent form and completed the questionnaire.
For obvious reasons linked with the post-genocide context, it was not possible to administer different questionnaires to members of the survivor and non-victim groups, or to include questions about ethnicity in the questionnaire. In order to distinguish members of the two groups, participants were asked to state if they were in Rwanda during the genocide and to self-identify as either survivor or non-survivor. For the purpose of this study, a genocide survivor was defined by reference to the definition proposed by the Ibuka association and approved by the government: "a genocide survivor is anyone who was chased from 1/10/1990 until 31/12/1994 because of his/her ethnicity or ideology to fight against the genocide of the Tutsi". Therefore, although we can expect that most participants who self-identified as survivors belonged to the Tutsi ethnicized group, this category could also comprise members of moderate Hutu families who were also persecuted during the genocide period.
Once they had identified themselves as a survivor or non-survivor, and after completing a series of demographic questions, they were directed to the appropriate following questions depending on their group (“if you identified yourself as a survivor, go to question number x, and if you identified yourself as a non-survivor, go to question y”). The term “non-survivor” is highly ambiguous in English, but the meaning was clear in Kinyarwanda (obviously referring to the participant status as “not a survivor” rather than “did not survive”). To avoid confusion we have referred to those who were not survivors as “non-victims” in this article, except where referring to the actual wording of items in the study.
Materials
The questionnaire comprised two sections; one concerned the level of group identification and the other measured reconciliation sentiment, perception of intergroup similarity, self-esteem during the commemorations, group-based remorse, and other questions that will not be presented in this article. All variables were measured on 5-point Likert scales (from 1 – “disagree completely” to 5 – “completely agree”). The questionnaire was written in Kinyarwanda.
Group Identification
The first part of the questionnaire was designed to assess the level of identification with the Rwandese nation and the two subordinate groups, namely, the group of survivors and the group of non-victims. As the mention of ethnic group names is forbidden in Rwanda, these measures served as proxy for ethnic identification. The seven items measuring national identification, created by Verkuyten and Hagendoorn (1998), were adapted at the two levels.
We created a summary score by averaging responses to the seven items to assess the level of identification with the Rwandese nation: I feel good about being Rwandese; I often regret that I am Rwandese (reversed); In general, I am glad to be Rwandese; If I could be born again, I would want to be Rwandese; Being Rwandese is important for me; If someone says something bad about the Rwandese they say something bad about me; Overall, I often do not like being Rwandese (reverse-coded). The alpha values corresponding to these scores were α = .75 for the non-victim group and α = .63 for the victim group.
Unfortunately, it proved impossible to obtain a reliable measure of subgroup identity, because of the terminology used in the items: “survivor” and “non-survivor”. The use of this indirect language proved inefficient for measuring ethnic identification, leading to seemingly absurd or double-meaning items such as “I feel good about being a survivor” or “Being a non-survivor is important for me”. As a consequence, the subordinate group identification scales proved unreliable in both groups (α = .44 in the survivor group; .34 in the non-victim group). This variable will therefore not be used in the following analyses.
The second part of the questionnaire was designed to assess reconciliatory sentiment, perception of intergroup similarity, self-esteem, and group-based remorse.
Reconciliation Sentiment
The reconciliation sentiment measure was adapted from the questionnaire of Mukashema and Mullet (2010). These authors distinguished two reconciliation factors: the intra- and the inter-personal components of reconciliation sentiment. Only the latter was used in the present study. It was composed of 14 items referring to trust and collaboration with out-group members: I think I can again share pleasurable activities with members of the survivor (non-survivor) group; I think I can make matrimonial alliances with members of the survivor (non-survivor) group; I think I can now make alliances with members of the survivor (non-survivor) group; I think I can now have serious discussions with members of the survivor (non-) group; I think I can trust members of the survivor (non-survivor) group; I feel good when I am in the company of members of the survivor (non-survivor) group; I think I can ask for the help of members of the survivor (non-survivor) group in case it is needed; I think I can tell my difficulties to the members of the survivor (non-survivor) group; I think I can participate in the celebrations and parties of the survivor (non-survivor) group; I think I can count on them in any circumstances; I think I can now be liked by the members of the survivor (non-survivor) group; I think I can accept the fact that I am not always in agreement with the members of the survivor (non-survivor) group; I think I can now work with the members of the survivor (non-survivor) group; I think I don’t feel a strong desire for revenge toward the members of the survivor (non-survivor) group. The alpha values corresponding to these scores were α = .78 for survivors and α = .91 for non-victims.
Perception of Intergroup Similarity
The perception of intergroup similarity measure was composed of five items referring to major components for the construction of Rwandese identity. In order to avoid comparisons with Rwandese people who were in the diaspora (outside Rwanda) during the genocide, each question started with “I think members of the survivor (non-survivor) group who were in Rwanda during the Tutsi genocide:” … have the same vision of the future as the members of my group; … share the same culture as my group members; … do not share physical traits with members of my group (reverse-coded); … share the same myths, values and rituals with members of my group; … have the same interests as members of my group. The Cronbach alpha values corresponding to these scores were .65 for survivors and .70 for non-victims.
Self-Esteem During Commemorations
The self-esteem measure was composed of 10 items based on the Rosenberg Self-Esteem Scale (Rosenberg, 1965) and adapted to measure self-esteem during the commemorations of the genocide against the Tutsi. This is a very sensitive period (every year in April) during which the identities of survivor or non-victim group members are salient and in which the socio-emotional needs linked with these statuses are probably felt more strongly. Each item started with “During the commemoration of the genocide against the Tutsi:” …on the whole, I am satisfied with myself; … I think I am no good at all (reverse-coded); … I feel that I have a number of good qualities; … I am able to do things as well as most other people; … I feel I do not have much to be proud of (reverse-coded); … I certainly feel useless at times (reverse-coded); … I feel that I'm a person of worth; … I wish I could have more respect for myself (reverse-coded); … I am inclined to think that I am a failure (reverse-coded); … I take a positive attitude toward myself. The alpha values corresponding to these scores were α = .79 for survivors and α = .63 for non-victims.
It should be noted that we did not measure victimization and collective guilt, which are included in the needs-based model. We only measured the part of the model that refers to self-esteem. In fact, the questionnaire was designed in 2008, and these data were collected in 2009, before the model was published. Yet, as the model is relevant to this context and since we had included the measure of self-esteem during commemoration, we decided to consider a partial test of the needs-based model.
Group-Based Remorse
Group-based remorse was measured through a single item: “I think I have remorse regarding what happened during the genocide perpetrated against the Tutsi.”
Results
Before testing our hypotheses, we compared means on all variables of interest between the two groups and we examined correlations. There was no significant difference in strength of identification with Rwanda: survivors tended to identify as much (M = 4.30, SD = 0.57) as non-victims (M = 4.40, SD = 0.66), t(245) = 1.29, p = .20, d = .16. But non-victims (M = 4.02, SD = .78) were more willing to reconcile than survivors (M = 3.77, SD = 0.60), t(231) = 2.60, p = .01, d = 0.35. Non-victims (M = 3.93, SD = 1.22) felt more remorse than did survivors (M = 3.55, SD = 1.33), t(220) = 2.19, p = .03, d = 0.38, and they tended to perceive more intergroup similarity (M = 3.51, SD = 0.96) than did survivors (M = 3.30, SD = 0.93), t(237) = 1.71, p = .09, d = 0.22, although this trend was only marginally significant. Finally, there was no difference in reported self-esteem during commemorations of the genocide between the two groups (M = 3.70, SD = 0.81 for non-victims; M = 3.79, SD = 0.64 for survivors); t(242) = 0.96, p = .34, d = 0.12.
We then examined the correlations between these variables separately for the two groups (see Table 1). Identification with Rwanda was positively correlated with reconciliation sentiment, and with perception of intergroup similarity in the two groups. Perception of intergroup similarity correlated positively with reconciliation sentiment in both groups, but this correlation tended to be stronger among non-victims (z = 1.88; p = .06). Self-esteem during commemorations was positively correlated with Rwandan identification and perception of intergroup similarity in the two groups (though only marginally among survivors), and also with reconciliation sentiment, but only among non-victims. This suggests that self-esteem during commemorations played a role in determining non-victims’ attitudes towards reconciliation, but that it did not among members of the survivor group. Finally, group-based remorse was positively correlated with reconciliation sentiment in both groups but not with Rwandan identification. This latter result discards this variable as a potential mediator, therefore not supporting our hypothesis. It will not be included in subsequent analyses. Interestingly, remorse was positively correlated with perception of similarity only among the non-victims, and negatively with self-esteem during commemorations only among the survivors.
Table 1
Pearson Bivariate Correlations between Main Variables of Interest among Survivors (Above Diagonal) and Non-Victims (Below Diagonal)
| 1 | 2 | 3 | 4 | 5 | |
|---|---|---|---|---|---|
| 1. Rwandan Identification | .22* | .39*** | .18† | -.13 | |
| N | 104 | 106 | 111 | 98 | |
| 2. Reconciliation | .29*** | .31*** | .14 | .20* | |
| N | 129 | 104 | 102 | 98 | |
| 3. Similarity | .28*** | .58*** | .19† | .002 | |
| N | 133 | 129 | 104 | 98 | |
| 4. Self-esteem | .17* | .28*** | .20* | -.20* | |
| N | 133 | 128 | 132 | 96 | |
| 5. Remorse | .13 | .38*** | .32*** | .02 | |
| N | 124 | 124 | 124 | 124 |
†p < .1. *p < .05. **p < .01. ***p < .001.
In order to explain through which processes identification with the Rwandan nation leads to more reconciliation sentiments in the two groups, we then carried out moderated mediation analyses using the PROCESS macro for SPSS (Hayes, 2013; Model 59), which applies Preacher and Hayes’s (2008) bias-corrected bootstrapping procedure. We examined if perception of intergroup similarity and reported self-esteem during commemorations accounted for the relation between Rwandan identification and reconciliation sentiment and if these mediations were similar for the two groups.
Results, presented in Figure 1 and Table 2, show that, for the two groups, the effect of Rwandan identification on reconciliation sentiment was significantly mediated by perception of intergroup similarity (the bias-corrected 95% confidence interval did not include 0). But group membership moderated the link between perception of similarity and reconciliation sentiment: this effect was stronger in the non-victim group than in the survivor group. In addition, although there was no significant interaction between Rwandan identification and group membership, an analysis of conditional indirect effects showed that the effect of Rwandan identification on reconciliation sentiment was also significantly mediated by self-esteem during commemorations, but only in the non-victim group. In addition, group membership also independently predicted reconciliation sentiment: Non-victims expressed more reconciliation sentiment than did survivors.
Figure 1
Moderated mediation analysis of the effect of Rwandan identification on reconciliation sentiment.
Table 2
Moderated mediation analysis of the effect of Rwandan identification on reconciliation sentiment
| b | S.E. | 95% C.I. | |
|---|---|---|---|
| Total effects | |||
| Rwandan identification to Reconciliation sentiment (c1) | .29*** | .07 | [.15, .43] |
| Group to Reconciliation sentiment (c2) | .23** | .09 | [.05, .40] |
| Direct effects | |||
| Rwandan identification to Reconciliation sentiment (c’1) | .12 | .07 | [-.02, .26] |
| Group to Reconciliation sentiment (c’2) | .17* | .08 | [.01, .33] |
| Paths from IVs to mediators | |||
| Rwandan identification to Perception of intergroup similarity (a11) | .52*** | .09 | [.34, .71] |
| Rwandan identification to Self-esteem during commemorations (a12) | .21** | .08 | [.05, .36] |
| Paths from mediators to DV | |||
| Perception of intergroup similarity to Reconciliation sentiment (b11) | .31*** | .05 | [.22, .40] |
| Self-esteem during commemorations to Reconciliation sentiment (b12) | .13* | .06 | [.01, .24] |
| Interaction | |||
| Group x Perception of intergroup similarity to Reconciliation sentiment (b21) | .24* | .09 | [.06, .43] |
| Conditional direct effect of Rwandan Identification to Reconciliation sentiment | |||
| Victim group | .13 | .12 | [-.10, .36] |
| Non-Victim group | .11 | .09 | [-.06, .28] |
| Conditional indirect effects of Rwandan Identification to Reconciliation sentiment | |||
| Mediation through Perception of intergroup similarity – Victim group | .11 | .06 | [.02, .24] |
| Mediation through Perception of intergroup similarity – Non-Victim group | .18 | .08 | [.07, .38] |
| Mediation through Self-esteem during commemoration – Victim group | .01 | .02 | [-.01, .10] |
| Mediation through Self-esteem during commemoration – Non-Victim group | .03 | .02 | [0, .10] |
Note. Only significant indirect paths are presented. All non-significant p values > .15.
Total R2 for the model is .30, F(7,222) = 13.74, p < .001.
Unstandardized b coefficients. *p < .05. **p < .01. ***p < .001.
In brief, these results suggest that identifying with Rwanda tended to increase reconciliation sentiments among members of both groups, in part because it led them to perceive their in-group and the out-group as being more similar. However, this effect of perception of similarity on reconciliation sentiment was stronger among members of the non-victim group. In addition, this positive effect of national identification was also partly explained through the fact that it tended to bolster self-esteem during genocide commemorations, but only among non-victims.
Discussion
As anticipated, some of our results are in line with the common in-group identity model while the other results are more coherent with the needs-based model of intergroup reconciliation, or show that the majority/minority status must be taken into account. It is not our intention to present these models as alternative approaches. Instead, we would like to consider them as complementary, in particular when dealing with real groups and real conflicts.
In accordance with the common in-group identity model, our results produce two important observations. First, the effect of national identification on intergroup reconciliation attitudes is positive and significant for both groups. Second, the mediating effect of perceived intergroup similarity is significant for both groups. These results are clearly supportive of the common in-group identity model’s assumption that a) identification with an inclusive superordinate group improves intergroup relations, and b) this effect is due to the increase in perceived similarity between the groups. It should go without saying that these results are also consistent with self-categorization theory.
Yet, our results also show the limitations of this approach, especially given that it does not take into consideration possible differences between majorities and minorities, or between survivors and non-victims of real intergroup conflicts, in their motivation to identify with the inclusive common identity, in their willingness to reconcile, and in the mediating effects of perceived intergroup similarity.
Minority group members have been shown to be more reluctant to identify with a superordinate group that includes a majority (Bergsieker, Shelton, & Richeson, 2010). While our results do not show differences between the Rwandese groups in the level of their national identification, they show that survivors perceived less intergroup similarity than non-victims, and that this variable played a weaker role in mediating the link between national identification and reconciliation sentiment. However, this effect could also be due to the fact that survivors might resist the loss of their ethnic identity because it would undermine the recognition of their victimhood. Our data do not allow us to disentangle these two possible explanations. The absence of difference in national identification might be due to the government policy, which exerts strong normative pressure in that direction.
Unfortunately, our test of the differences in social identifications between ethnic groups is limited, given our inability to directly measure the participants’ ethnic identification. This was due, as mentioned previously, to the legal restrictions on the public use of ethnic group names. Consequently, we were not able to examine the moderating effects of this important variable on the processes described by the common in-group identity model. In particular, this prevents us from determining whether, in the Rwandan situation, the re-categorization strategy adopted by the Rwandan government proves more efficient than a dual identity strategy (see Dovidio et al., 2009; Paluck & Green, 2009).
According to the needs-based model of intergroup reconciliation approach, victims and perpetrators have different underlying motivations to seek or accept reconciliation, which can determine their propensity to accept or not to give priority to the inclusive common identity as opposed to subgroup identities. On this basis, we have proposed that perpetrators are more likely to benefit from a common superordinate identity, in part because such an identity may improve a subgroup image that was tarnished by their group’s participation in past violence against the minority. Two results are in line with this argument. First, non-victim participants were more willing to reconcile than survivors; second, the level of self-esteem during commemorations mediated the effect of national identification on intergroup reconciliation attitudes only for non-victims.
As negative self-centered group-based emotions such as group-based guilt, shame, or remorse have been shown to affect intergroup reconciliation processes, we also examined the role played by group-based remorse. Results showed that this emotion was, as expected, more strongly reported among non-victims than among survivors, though this difference was relatively weak given this post-genocidal context. This variable was not significantly correlated with national identification in the two groups. As expected, group-based remorse was linked with more reconciliatory sentiments in the non-victim group. However, we also found this correlation in the survivor group. These unexpected results might be linked with “survivor guilt,” which was described as a guilty feeling experienced by survivors of mass violence, who may wonder why they survived whilst so many others died (Lifton, 1980). However, this emotion has been described as self-centered rather than group-centered, and thus it should not be linked with intergroup reconciliation sentiments as we observed in the present study.
Globally, our results suggest that the policies of strengthening national identification, whilst also suppressing ethnic identities, endorsed by the Rwandese government after the 1994 genocide, might have been efficient in promoting intergroup reconciliation (notwithstanding the limitation reported above). In our – non-representative – sample, levels of national identification were at comparable levels in the two groups, and this variable was associated with more reconciliatory attitudes in the two groups. However, this strategy seems to be more efficient among non-victims than among survivors. Our study provides partial evidence that the different needs felt by members of victim and perpetrator groups after violent intergroup conflicts are at stake. Yet, this study also lacked a proper measure of the need to recover a sense of agency that, as postulated by this model, could mediate the link between national (and ethnic) identification and reconciliation sentiment among victims. Further studies are needed to test this hypothesis.
These results indicate that, in real-life situations, when survivors and non-victims of past hostilities have to live together, explaining the complex social psychological processes involved in intergroup reconciliation requires the complementary use of more than one theory. In the case of our investigation of intergroup reconciliation in Rwanda in the context of genocide commemorations, combining the common in-group identity model with the needs-based model of intergroup reconciliation proved relevant.