A questionnaire survey (N = 247) investigated the influence of identification with the Rwandan nation on reconciliation sentiments between members of the survivor and of the non-victim groups of the 1994 genocide in Rwanda. Results showed that, whereas the two groups did not differ in their level of identification with the nation, members of the non-victim group were more willing to reconcile than members of the survivor group. Perceived intergroup similarity mediated the effect of national identification on reconciliation sentiment for both groups, but this effect was stronger among non-victims. Finally, self-esteem during commemorations also mediated this effect, but only among non-victims. We discuss the importance of people’s motivation to reconcile with out-group members in post-genocidal contexts in light of the common in-group identity model (Gaertner & Dovidio, 2000) as well as the needs-based model of intergroup reconciliation (Nadler & Schnabel, 2008).
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 (
The Rwandese government highlights that both the Tutsis and the Hutus come from one ancestor, Gihanga Ngomijana (
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
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).
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 (
These positive effects of identification with a superordinate group have been demonstrated in several empirical studies in both laboratory experiments with minimal groups (
In the study by
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 (
Previous studies such as those by
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 (
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.
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 (
These threats felt by victims and perpetrators also produce different motivational consequences (
Two studies conducted by
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 (
Finally, as suggested by the social psychological literature on group-based emotions (
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 (
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.
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 (
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.
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.
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.
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
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.
The reconciliation sentiment measure was adapted from the questionnaire of
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.
The self-esteem measure was composed of 10 items based on the Rosenberg Self-Esteem Scale (
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 was measured through a single item: “I think I have remorse regarding what happened during the genocide perpetrated against the Tutsi.”
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 (
We then examined the correlations between these variables separately for the two groups (see
1 | 2 | 3 | 4 | 5 | |
---|---|---|---|---|---|
1. Rwandan Identification | .22* | .39*** | .18† | -.13 | |
104 | 106 | 111 | 98 | ||
2. Reconciliation | .29*** | .31*** | .14 | .20* | |
129 | 104 | 102 | 98 | ||
3. Similarity | .28*** | .58*** | .19† | .002 | |
133 | 129 | 104 | 98 | ||
4. Self-esteem | .17* | .28*** | .20* | -.20* | |
133 | 128 | 132 | 96 | ||
5. Remorse | .13 | .38*** | .32*** | .02 | |
124 | 124 | 124 | 124 |
†
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 (
Results, presented in
Moderated mediation analysis of the effect of Rwandan identification on reconciliation sentiment.
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] |
Total R2 for the model is .30,
Unstandardized b coefficients. *
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.
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 (
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
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 (
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.
This study was financially supported by the Belgian Technical Cooperation (BTC), Université libre de Bruxelles (ULB), and the de Meurs-François and Van Buuren fund. The funding of BTC supported the research design, the data collection, analysis, interpretation and the reporting of the findings. The de Meurs-François and Van Buuren fund supported the reporting process. ULB and BTC provided material support such as laptops and SPSS licences.
The authors have declared that no competing interests exist.
We want to acknowledge the help of Université Libre de Bruxelles and INATEK, in particular the clinical psychology department, administrators of this department and students who kindly accepted to participate in the present study. We also appreciate the assistance received from Benoite Umubyeyi, Darius Gishoma, and helpful comments from the editor and three anonymous reviewers of the manuscript.