This is an open access article distributed under the terms of the Creative Commons
Attribution License (The way we remember the past of our ethnic group can play an important role in shaping present-day intergroup relations (Jetten & Wohl, 2012; Liu & Hilton, 2005; Páez & Liu, 2010). Particularly in post-conflict societies that have undergone large political transformations and where ethnicity has been repeatedly constructed and deconstructed, collective memories of the past can stand in the way of harmonious intergroup relations in the present. Importantly, these conflicts and the related memories can also get “exported” to a diaspora, as many people seek refuge abroad during a war. It is not unlikely that members of conflicted groups find themselves living in the same host country, as is the case with, for instance, Turks and Kurds in Western Europe. If the relations between formerly conflicted immigrant groups are not amicable, this can undermine social cohesion in the host society.
In this paper we examine interethnic relations between groups originating from the same conflict area (former Yugoslavia) and currently living in a third country to which they have migrated (Australia). Former Yugoslavia is an exemplary case of a post-conflict society that has set in motion substantial waves of emigration. After the dissolution of the Socialist Federative Republic of Yugoslavia in the early 90s and the establishment of new independent states, three ethnic groups – Croats, Serbs and Bosniaks – engaged in a violent conflict, and many emigrated overseas to countries that had opened their doors to humanitarian migrants. Australia is one of the countries in which relatively large numbers of migrants of Bosniak, Croatian and Serbian ethnicity have found refuge in the early 90s (Colic-Peisker & Walker, 2003; Forrest, Hermes, Johnston, & Poulsen, 2013) but also in earlier decades of the 20th century in the aftermath of World War II (Colic-Peisker, 2005; Hay, 1998).
Rather than studying intergroup relations in terms of group stereotypes and attitudes, we focus on voluntary contacts with ethnic out-groups as a (self-reported) behavioral measure of intergroup relations. Compared to perceptions and feelings, developing relationships with ethnic out-group members is a more direct and informative indicator of actual intergroup relations. Furthermore, we examine contacts with ethnic out-group members relative to in-group contacts, or what we will call relative out-group contacts (ROC). We do so because we are interested in the difference in frequency of contact with in-group and out-group members and not in the actual level of out-group contact. In this way, we account for differences in sociability, as some people have fewer relationships in general, including with their ethnic in-group. Focusing on three ethnic groups in the Australian diaspora gives us a unique possibility to examine interethnic contacts, which would possibly be more difficult in the region of origin. Given that ethnic segregation is still rather strong and often institutionalized in the republics that were formed after the dissolution of Yugoslavia (Ajduković & Čorkalo Biruški, 2008), the opportunities to come into contact with out-group members are still limited (Žeželj, Ioannou, Franc, Psaltis, & Martinovic, 2017). In Australia, these groups have primarily settled in several larger cities (Melbourne, Sydney, Perth, Brisbane) and can therefore easily seek out each other’s company.
Given the long history of inter-ethnic tensions in the Balkans, we turn to group identities and collective memories of the past as relevant predictors of relative out-group contacts in the present. More specifically, we examine the role of a superordinate Yugoslavian identification and subgroup ethnic identification in predicting relative out-group contacts, and we argue that these associations are further explained by two dominant and emotionally laden representations of history: nostalgia about the former Yugoslavia (i.e., Yugonostalgia) and collective guilt assignment for past wrongdoings.
Ethnic and Yugoslavian Identification
Group identities play an important role in shaping intergroup relations. According to Self-Categorization Theory (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987) and Social Identity Theory (Tajfel & Turner, 1979), individuals categorize themselves and others into social groups, and they attach value and importance to the groups they belong to. Furthermore, people can have multiple group belongings, and subgroup identities can be nested in superordinate ones (Gaertner, Dovidio, Anastasio, Bachman, & Rust, 1993).
Nested identities are relevant in the case of the former Yugoslavia, which was a federation of republics inhabited by different ethnic groups of Southern Slavs. Thus, people could identify to different degrees with their ethnic (or subgroup) identities and with the superordinate Yugoslavian identity. The latter was salient in society (Sekulic, Massey, & Hodson, 1994), and was understood predominantly in a civic sense (Sekulic, 2004).
In countries that have undergone significant political transformations, identities from a previous era do not immediately wither away (see Poppe & Hagendoorn, 2001, for the example of the former Soviet Union). Even though there was a decrease in the percentage of people declaring themselves as primarily Yugoslavian after the dissolution of Yugoslavia between 1981 and 1991 (Sekulic, 2004), Yugoslavian identity could still be a meaningful identity for people originating from this area, and they might still identify to various degrees with Yugoslavia in the present. Research on other diaspora communities indicates that the nation of origin often remains an important and meaningful source of identity, as among Chinese, Greek and Italian immigrants (e.g., Ang, 2003; Cohen, 2008).
We propose that relationships between Serbs, Bosniaks and Croats in Australia depend on both the degree of ethnic and Yugoslavian identification. For higher ethnic identifiers other sub-groups from former Yugoslavia represent a clear outgroup, and according to Social Identity Theory (Tajfel & Turner, 1979), strong ‘us’ and ‘them’ dynamics are likely to fuel favouring the in-group over the out-group (ingroup bias). To clarify, even though group identification does not necessarily translate into outgroup derogation (Brewer, 1999), negative responses to outgroups are likely among those who identify more strongly with the group and when the group finds itself in competitive and threatening contexts. The war between ex-Yugoslavian ethnic groups was a “territorially-based quests for self-determination” (Hasani, 2003, p. 201; see also Meier, 1999). While physical safety and territorial disputes are no longer a source of direct threat to immigrants in Australia, members of these groups might still bear a grudge against other immigrant groups from former Yugoslavia whom their group fought against during the war. The history of threat, competition and conflict can leave its marks on the present. We therefore expected that those immigrants from former Yugoslavia who care strongly about their ethnic in-group will turn more to their in-group and away from the rival out-group. Thus, higher ethnic identifiers should have relatively fewer contacts with ex-Yugoslavian ethnic out-groups in Australia.
The role of the superordinate Yugoslavian identity could be expected to be different. According to the Common In-group Identity Model (Gaertner et al., 1993), identifying with a superordinate category can improve attitudes towards out-groups that fall under this same overarching category. Strong superordinate identification enhances shared identity perceptions with other subgroups, and because a superordinate category is inclusive towards the out-group, intergroup trust is enhanced. Evidence for the beneficial effects of identification with a superordinate category has been found in various studies (e.g., Dovidio et al., 1997; Nier et al., 2001), including research on the conflict between Turks and Kurds (Çelebi, Verkuyten, Köse, & Maliepaard, 2014). Furthermore, longitudinal research has shown that identification with a superordinate group is conducive to more intergroup friendships over time (West, Pearson, Dovidio, Shelton, & Trail, 2009). At the same time, given that a superordinate national identity also includes ethnic in-group members, a strong Yugoslavian identity would not necessarily be linked with fewer contacts with co-ethnics. Altogether, as we expected the discrepancy between out-group and in-group contacts to be smaller for people who identify more strongly with the Yugoslavian category, we predicted Yugoslavian identification to be related to relatively more contacts with out-groups from former Yugoslavia living in Australia.
Remembering the Past
Research has shown that the way in which history is told and taught, with an emphasis on specific historical elements and the omission or even denial of others, can have a strong effect on intergroup attitudes and behaviours (Jetten & Wohl, 2012; Sibley, Liu, Duckitt, & Khan, 2008; Smeekes, Verkuyten, & Poppe, 2011). The war that took place after the dissolution of Yugoslavia has left its marks in different ways, and it has shaped collective memories and related emotions about the past. In this particular geopolitical context, two important and controversial emotional memories about the past include Yugonostalgia (Kolstø, 2014; Lindstrom, 2005; Volčič, 2007) and collective guilt (Brown & Čehajić, 2008; Jelić, Čorkalo Biruški, & Ajduković, 2013).
Yugonostalgia
Research on past-oriented group-based nostalgia indicates that while being nostalgic about the national past can bring in-group members closer together (Wildschut, Bruder, Robertson, van Tilburg, & Sedikides, 2014), this same sentiment can also have detrimental effects on intergroup relations (Smeekes, 2015; Smeekes, Verkuyten, & Martinovic, 2015). This research has primarily focused on European immigrant-receiving societies such as the Netherlands, and it shows that the more nostalgic majority members are about their culturally homogeneous national past, the less willing they are to accept immigrants. Thus, group-based nostalgia can be a barrier for harmonious intergroup relations in some contexts.
In contrast, we argue that, in the context of groups from former Yugoslavia, nostalgia could improve intergroup relations. Unlike being nostalgic for the national past of a country that continues to exist but that is disrupted by the arrival of immigrants, Yugonostalgia refers to a longing for the past of a superordinate (federal) category that has fully ceased to exist. We expected that individuals who were more Yugonostalgic would also more readily seek contacts with ethnic out-group members from former Yugoslavia, because through such contacts they could psychologically restore the past that they longed for. This is in line with research showing that nostalgia functions as a coping mechanism to deal with identity discontinuity (Sedikides, Wildschut, Routledge, & Arndt, 2015), and with research that indicates that nostalgia about an encounter with an out-group member is associated with a sense of having a shared identity and more positive out-group attitudes (Turner, Wildschut, & Sedikides, 2012). At the same time, there is no reason to assume that Yugonostalgia would be a barrier for contacts with co-ethnics, and more Yugonostalgic individuals might even have more contact with co-ethnics because with them they can also remember “the good old days”. Yet our assumption is that the past of a multicultural Yugoslavia can be more easily restored in people's minds through contact with diverse groups from former Yugoslavia, and that Yugonostalgic people will therefore have relatively more contact with out-group members than those who are less Yugonostalgic.
According to intergroup emotions theory (Mackie, Devos, & Smith, 2000), people need to identify with their group in order to experience group-based emotions, and higher identifiers experience stronger emotions. In a series of studies, Wildschut and colleagues (2014) have shown that group-based nostalgia also qualifies as a group based emotion, and Smeekes (2015) has confirmed that it correlates positively with in-group identification. Therefore, stronger Yugoslavian identification can be expected to be associated with a stronger feeling of Yugonostalgia. In contrast, stronger ethnic identification could be a barrier to Yugonostalgic feelings. Given that during Yugoslavian times ethnic differences were suppressed and ethnic identification was discouraged (Godina, 1998), strong ethnic identifiers might have felt that not enough distinctiveness was granted to their ethnic subgroup (Brewer, 1991). It could be argued that high ethnic identifiers felt threatened by Yugoslavia and therefore do not long for the “good old Yugoslavian days”. Altogether, we expected identification with the overarching Yugoslavian identity to be related to more Yugonostalgia, whereas ethnic identification will be related to less Yugonostalgia. Yugonostalgia, in turn, is expected to be conducive to higher relative out-group contact with other groups from former Yugoslavia.
Collective Guilt Assignment
People do not only have nostalgic memories about former Yugoslavia but can also feeling guilty on behalf of their in-group or blame out-groups for the dissolution of the country, the outbreak of the war, and the atrocities committed. Collective guilt occurs when there is a discrepancy between how one thinks one’s in-group should have behaved and how it actually behaved. This can refer to the present but also to the past when people feel guilty about their ancestors’ past wrongdoings and the harm done to another group (Allpress, Barlow, Brown, & Louis, 2010; Doosje, Branscombe, Spears, & Manstead, 1998; Imhoff, Wohl, & Erb, 2013; Rees, Allpress, & Brown, 2013). Collective guilt has been studied mostly in contexts where the distinction between victims and perpetrators is clear (e.g., colonialism, genocide, the Holocaust). However, in other conflict situations it is more difficult to argue that one group is to be fully blamed and the other is an innocent victim. Even though one group might have initiated the conflict and caused most damage, atrocities eventually are committed on both sides. Depending on the historical narratives about the conflict that these groups put forward, all conflicted parties can be simultaneously presented as perpetrators and victims to some degree (Wohl, Branscombe, & Klar, 2006).
Research from a context in which violence was committed on both sides (Northern Ireland) shows that members of both groups can feel moderately guilty. However, feelings of in-group guilt tend to be lower among individuals who have directly suffered at the hands of the outgroup (Hewstone et al., 2004). Wohl and colleagues (2006, p. 7) argue that in such conflict situations collective guilt on behalf of one’s in-group is “inhibited by righteous indignation that stems from having experienced harm at the hands of the outgroup”. Thus, rather than examining feelings of collective guilt on behalf of one’s in-group, it is in such contested contexts particularly interesting to focus on guilt assignment to out-group versus the in-group.
In the case of former Yugoslavia, ethnic groups can blame each other for the dissolution of Yugoslavia, the outbreak of the war, and the violence. We argue that people who see out-groups as relatively guiltier than their in-group will more strongly avoid contacts with these ‘perpetrator’ groups. Our reasoning is that group members who attribute relatively more guilt to the out-group have stronger feelings of resentment and find it harder to forgive the rival group, and this lack of forgiveness stands in the way of reconciliation and positive interethnic relations (Noor, Brown, & Prentice, 2008). Also, in the aftermath of a conflict, people who blame the out-group for mistreatment of their in-group members should be more likely to turn to their in-group for comfort and support, and by extension, have more contacts with in-group members. This is in line with the theoretical reasoning and empirical support of the Rejection-Identification Model (Branscombe, Schmitt, & Harvey, 1999), according to which rejection by out-group members strengthens identification with one's ethnic in-group. In contrast with out-group guilt, collective guilt on behalf of one’s in-group has been found to evoke empathy and active responses such as endorsing reparations for historical wrongdoings (Brown & Čehajić, 2008). Recognizing in-group responsibility for the past deeds also has been found to correlate with more favourable intergroup relations (Čehajić & Brown, 2010). Based on all of the above, we expected that higher assignment of guilt to out-groups relative to the in-group relates to lower levels of relative out-group contact.
The way group members attribute group-based guilt depends on group identification. Some degree of identification is necessary to experience collective guilt, which is again in line with the reasoning from intergroup emotions theory (Mackie, Devos, & Smith, 2000). Yet, given that guilt is a negative emotion that undermines a positive social identity (Doosje et al., 1998; but see Doosje, Branscombe, Spears, & Manstead, 2004), high in-group identifiers are particularly motivated to suppress feelings of collective guilt. The recognition of collective guilt is especially problematic for group members who derive a positive sense of self from their glorified in-group (Roccas, Klar, & Liviatan, 2006). Reducing feelings of guilt for the in-group and shifting the blame to out-group members is a strategy for improving the moral status of one’s group (Doosje et al., 1998). In the context of groups from former Yugoslavia, whose ethnic identities were glorified during the war by the political elites' nationalist rhetoric (Wilmer, 1997), we expected higher ethnic identifiers to more readily attribute guilt to rival ethnic out-groups, as well as to deny their in-group's guilt. Thus, ethnic identification was expected to be related to more relative guilt assignment to out-groups for the outbreak of the war and atrocities committed.
In contrast, identification with the overarching Yugoslavian identity should reduce the discrepancy between in-group and out-group guilt assignment. Identifying with an overarching category increases mutual understanding and brings sub-groups closer together (Gaertner et al., 1993). This blurring of distinctions between in-group and out-group might mean that high common ingroup identifiers hold more nuanced views on in-group and out-group guilt than low common in-group identifiers. The former should be more likely than the latter to recognize some actions of their in-group as wrong and to show comparatively more understanding for the actions of the out-group and thus attribute relatively less blame to them. Experimental research has confirmed that identification with a more inclusive category leads to decreased collective guilt assignment to out-groups for their harmdoing (Wohl & Branscombe, 2005). We therefore expect a negative relationship between Yugoslavian identification and relative out-group guilt assignment. Note that this does not mean that we expect high Yugoslavian identifiers to blame their ethnic in-group more than ethnic out-groups in absolute terms, but we expect the difference between assignment of guilt to the in-group and out-group to be smaller for those who feel more Yugoslavian.
In summary, we hypothesize that higher Yugoslavian identification will be related to relatively more ethnic out-group contact (ROC), which can be explained through higher Yugonostalgia (H1) and lower relative assignment of guilt to out-group (H2). Further, we expected higher ethnic identification to be related to less relative out-group contact, via lower Yugonostalgia (H3) and via higher relative assignment of guilt to the out-group (H4).
Method
Data and Participants
To test our hypotheses, we collected unique survey data in 2013 among 87 participants of Croatian, Serbian and Bosniak ethnicity residing in Australia. Collecting these data proved to be a serious challenge and many participants who were approached declined given the sensitive nature of the survey. This suggests that the ethnic group tensions from former Yugoslavia are still salient among these immigrants in present-day Australia.
The data were collected using both online questionnaires and paper and pencil versions that were sent out to participants by mail and returned by mail to the researchers.i We contacted sports clubs, cultural clubs and religious associations in Brisbane, Sydney, Melbourne and Perth, and via these channels we targeted a broad range of participants in terms of their socio-demographic background. The questionnaire was prepared in English and translated to Croatian by a native speaker. It was then back-translated to English by another native speaker. A native Serbian speaker translated the Croatian version to Serbian. In Bosnia and Herzegovina both the Serbian and Croatian languages are spoken. As Croatian and Serbian are very similar, and the differences are mainly grammatical and rarely semantic, it did not seem necessary to back-translate this version to English. Participants could choose which language they wanted to complete the questionnaire in, and 44 per cent chose Croatian, 25 per cent Serbian and 25 per cent English. Out of 87 surveys, 44 (51.6%) were completed online and 43 (49.4%) used paper and pencil.
The three ethnic groups were equally represented in the data (NCroats = 28, NSerbs = 30, NBosniaks = 29). Gender distribution was balanced, with females making up 52% of the sample. Age ranged from 22 to 74 (M = 47, SD = 13.8). Regarding the level of education, 8% of the participants had only followed primary education, 29% secondary education, 24% vocational training, and 39% had attended university. Most participants (79%) were born in former Yugoslavia and only 13 participants (15%) were born in Australia to Bosniak, Croatian or Serbian parents. 6% of the participants did not answer the question about country of birth.
Measures
The independent variable Yugoslavian identification was measured by four items adapted from commonly used scales (e.g., Verkuyten, 2005): “I feel Yugoslavian”, “I identify strongly with the Yugoslavian people”, “My Yugoslavian identity is an important part of me”, and “I am proud to be Yugoslavian”. The answers ranged from (1), completely disagree, to (7), completely agree. The items formed a reliable scale and they were averaged (Cronbach’s α = .96). The same four items were used to measure ethnic identification, but this time with reference to one’s ethnic group, e.g., “I am proud to be Serbian” (α = .90).
The first mediator, Yugonostalgia, was measured with five items (7-point scales ranging from completely disagree to completely agree) adapted from a scale previously used to measure national nostalgia (Smeekes et al., 2015). The items were: “It is a pity that Yugoslavia ceased to exist”, “I get nostalgic when I think back of Yugoslavia in the past times”, “I often long for Yugoslavia from the past”, “I experience nostalgic feelings when I hear old Yugoslav music”, and “I wish I could turn back the time”. The scale was reliable and the items were averaged (α = .93).
The second mediator, relative out-group guilt, was measured using questions about the extent to which participants agreed that their in-group (e.g., Bosniaks) and the two out-groups (Croats and Serbs) were to be blamed for the following three events: the dissolution of Yugoslavia, the outbreak of the war, and the atrocities committed. The answer scale ranged from (1), not at all guilty, to (7), very much guilty. For each ethnic group we computed a composite score for out-group guilt across the three items and two outgroups (α = .70 for Bosniaks, .97 Serbs, .58 Croats), and subtracted in-group guilt from it (α =.91 for Bosniaks, .94 Serbs, .86 Croats) to obtain a measure of relative assignment of guilt to out-group.
The dependent variable relative out-group contacts with subgroups from former Yugoslavia was measured by asking participants to respond to the question (1 = never, to 7 = daily): “How often do you hang out with people from the following ethnic groups living in Australia? This can be at your work or study/school, as well as in your neighborhood and in your free time.” Multiple groups were listed, including Australians, Italians, Greeks and Asians, but for the relative out-group contact measure we focused only on contacts with Serbs, Croats, and Bosniaks. We computed an average out-group contact score and then subtracted in-group contact from out-group contact, which yielded a score for relative out-group contact.
To ensure that the three conceptually close measures – ethnic identification, Yugoslavian identification, and Yugonostalgia – represent different constructs, we ran a factor analysis in SPSS. A principle component analysis with oblique rotation yielded three factors with Eigenvalues above 1. Each item loaded highly on the construct it was supposed to measure, with loadings ranging between .813 and .956. Cross-loadings were all below .200, which confirms that ethnic identification, Yugoslavian identification, and Yugonostalgia are empirically distinct constructs.ii
In the analyses we controlled for ethnic group membership using dummy variables for Serbian (1) and Croatian (1) ethnicity, whereas Bosniak ethnicity (0) served as a reference group.
Results
Descriptive Findings
Table 1 shows the descriptives and correlations between the main variables for the sample as a whole, and broken down by ethnic group. For the total sample, the correlations for the hypothesized associations were mostly significant and all were in the expected direction. Yugoslavian identification and Yugonostalgia were associated with more, and ethnic identification and relative assignment of guilt to out-group with less, relative out-group contact. Further, Yugoslavian identification was related to more and ethnic identification to less Yugonostalgia. Lastly, ethnic identification correlated positively and Yugoslavian identification negatively (though not significantly) with relative assignment of guilt to out-group. The patterns were largely similar for each of the three groups, but they sometimes failed to reach significance due to small sample sizes. Particularly the Croatian and Bosniak samples resembled each other and replicated the findings from the overall sample. Among Serbian participants, however, we did not find the expected correlations of Yugoslavian identification and Yugonostalgia with relative out-group contact.
Table 1
Descriptive Statistics of the Core Constructs: Total Sample and by Ethnic Group
| Total sample | 1 | 2 | 3 | 4 | Range | M | SD |
|---|---|---|---|---|---|---|---|
| 1. Relative out-group contact | - | -6.00 – 2.50 | -2.17 | 1.84 | |||
| 2. Yugoslavian identification | .29** | - | 1.00 – 7.00 | 3.33 | 2.29 | ||
| 3. Ethnic identification | -.42*** | -.31** | - | 1.00 – 7.00 | 5.61 | 1.62 | |
| 4. Yugonostalgia | .30** | .63*** | -.29** | - | 1.00 – 7.00 | 4.34 | 2.10 |
| 5. Relative out-group guilt | -.52*** | -.15 | .39*** | .08 | -2.33 – 6.00 | 2.13 | 2.32 |
| Croats | |||||||
| 1. Relative out-group contact | - | -5.50 – 2.50 | -2.30 | 1.90 | |||
| 2. Yugoslavian identification | .46* | - | 1.00 – 5.75 | 1.93 | 1.50 | ||
| 3. Ethnic identification | -.45* | -.19 | - | 3.00 – 7.00 | 5.92 | 1.30 | |
| 4. Yugonostalgia | .49* | .64*** | -.36† | - | 1.00 – 6.00 | 2.57 | 1.72 |
| 5. Relative out-group guilt | -.54** | -.57** | .40* | -.47* | -2.33 – 5.00 | 1.24 | 1.62 |
| Serbs | |||||||
| 1. Relative out-group contact | - | -5.00 – 0.00 | -1.33 | 1.57 | |||
| 2. Yugoslavian identification | -.02 | - | 1.00 – 7.00 | 4.59 | 2.11 | ||
| 3. Ethnic identification | -.25 | -.09 | - | 1.75 – 7.00 | 5.28 | 1.70 | |
| 4. Yugonostalgia | .05 | .53** | -.12 | - | 2.20 – 7.00 | 5.11 | 1.55 |
| 5. Relative out-group guilt | -.33† | .09 | .56** | .10 | -0.67 – 5.67 | 0.97 | 1.96 |
| Bosniaks | |||||||
| 1. Relative out-group contact | - | -6.00 – 0.00 | -2.91 | 1.73 | |||
| 2. Yugoslavian identification | .27 | - | 1.00 – 7.00 | 3.44 | 2.38 | ||
| 3. Ethnic identification | -.55** | -.45* | - | 1.00 – 7.00 | 5.66 | 1.79 | |
| 4. Yugonostalgia | .50** | .53** | -.33† | - | 1.00 – 7.00 | 5.29 | 1.85 |
| 5. Relative out-group guilt | -.56** | -.38† | .58** | -.27 | 0.00 – 6.00 | 4.26 | 1.79 |
†p <.10. *p < .05. **p < .01. ***p <.001.
Analysis of variance (ANOVA) indicated that Serbs, Croats and Bosniaks did not differ in their levels of ethnic identification, F(2, 83) = 1.15, p = .322 (see Table 1). However, Croats scored significantly lower on Yugoslavian identification than Serbs and Bosniaks, F(2, 82) = 12.23, p < .001, η2 = 0.23. Regarding Yugonostalgia, the group differences were also significant and substantial, F(2, 80) = 21.62, p < .001, η2 = 0.35, with Croats being significantly less nostalgic about Yugoslavia than Serbs and Bosniaks. The latter two groups scored significantly higher than the mid-point of the scale, t(27) = 3.81, p = .001 and t(27) = 3.69, p = .001, whereas Croats scored significantly below the mid-point, t(26) = -4.31, p < .001. In relation to relative out-group guilt, Bosniaks blamed the out-groups substantially more compared to Croats and Serbs blaming the out-groups, F(2, 77) = 27.13, p < .001, η2 = 0.41. However, all groups blamed the in-group less than the out-groups (Croats: Mingroup = 3.88 vs. Moutgroup = 5.13; Serbs: Mingroup = 4.50 vs. Moutgroup = 5.47; Bosniaks: Mingroup = 2.22 vs. Moutgroup = 6.47). Lastly, Bosniaks had the least relative out-group contact, followed by Croats and then Serbs, F(2, 83) = 6.23, p = .003, η2 = 0.13. For all groups, intragroup contact was more frequent than intergroup contact (Croats: Mintragroup = 5.18 vs. Mintergroup = 2.85; Serbs: Mintragroup = 4.70 vs. Mintergroup = 3.37; Bosniaks: Mintragroup = 6.03 vs. Mintergroup = 3.12).iii, iv
Group Identification, Collective Emotions About History, and Intergroup Contact
To determine whether ethnic and Yugoslavian identifications were significantly and positively related to relative out-group contact, and whether these relationships could be explained by Yugonostalgia and relative assignment of guilt to out-group, we estimated a mediation model in Mplus (version 7). Because of the small sample size we used manifest variables (averaged scores) rather than latent constructs. We regressed relative out-group contact on Yugonostalgia, relative out-group guilt, Yugoslavian identification and ethnic identification. Further, we regressed Yugonostalgia and relative out-group guilt on the two types of identification. Thereby we specified four mediation paths, two from Yugoslavian identification to relative out-group contact, and two from ethnic identification to relative out-group contact. The covariance between ethnic and Yugoslavian identification was accounted for in the model, and so was the covariance between the two mediators. In addition, we controlled for ethnic group differences in relation to the two mediators. As there were no significant differences between ethnic groups with respect to relative out-group contact in this model, we did not include these paths. We used bootstrapping with 10,000 replacement samples to obtain confidence intervals for the indirect paths (Preacher & Hayes, 2008).
Figure 1 shows the path coefficients for the hypothesized model. Yugoslavian identification was significantly and positively related to Yugonostalgia but it was unrelated to relative out-group guilt. In contrast, ethnic identification was significantly and positively related to attributing relatively more guilt to out-groups, but there was no association between ethnic identification and Yugonostalgia. In turn, the more Yugonostalgic the participants were, and the less they blamed the out-groups for the dissolution of Yugoslavia relative to the in-group, the higher their relative out-group contact was.v
Figure 1
Results from a structural path model predicting relative out-group contacts among Croats, Serbs and Bosniaks in Australia (N = 87).
Note. Standardized coefficients presented, total effects in the brackets; the covariance of the two mediators (ß = -.008, p = .935) and the two independent predictors (ß = -.311, p = .003) was accounted for. We controlled for significant ethnic group differences in relation to the mediators.
*p < .05. **p < .01. ***p < .001.
The indirect effect of Yugoslavian identification on relative out-group contact through Yugonostalgia was significant, β = .145, p = .016, and the confidence interval did not include zero [low CI = .026, high CI = .263]. This confirmed Hypothesis 1. In addition, and in accordance with Hypothesis 4, ethnic identification was indirectly related to less ROC via higher relative assignment of guilt to out-group, β = -.181, p = .001 [low CI = -.288, high CI = -.074]. The remaining two indirect paths were not significant: Yugoslavian identification was not indirectly linked to relative out-group contact via relative out-group guilt, β = .053, p = .277 [low CI = -.043, high CI = .149], and ethnic identification was not indirectly linked to relative out-group contact via Yugonostalgia, β = -.035, p = .227 [low CI = -.091, high CI = .022]. Hypotheses 2 and 3 were thus rejected.vi, vii
Additional Analyses
To allow for a better understanding of the specific processes involved, we re-estimated two separate models including only attributions of guilt to the in-group followed by attributions of guilt to the out-group.viii In-group guilt was related to significantly more relative out-group contact (β = .391, S.E. = .101, p < .001), whereas out-group guilt was related to marginally less relative out-group contact (β = -.193, S.E. = .101, p = .055). Further, those participants higher in Yugoslavian identification reported greater awareness of in-group guilt (β = .214, S.E. = .092, p = .019) whereas higher ethnic identification was related to lower perceptions of in-group guilt (β = -.400, S.E. = .119, p < .001). Yugoslavian and also ethnic identification were not significantly related to out-group guilt, β = .160, S.E. = .118, p = .174 and β = -.029, S.E. = .105, p = .784, respectively. We can therefore conclude that group identities, and particularly ethnic identification, are more strongly associated with assignment of guilt to the in-group than to the out-group. Furthermore, in-group guilt was a better predictor of ROC than outgroup guilt. The finding that out-group guilt matters less could be due to a ceiling effect: the scores on out-group guilt in this sample were rather high (M = 5.69, SD = 1.44), reducing the ability of this measure to meaningfully account for variance.
Next, we re-ran the model replacing relative out-group contact first with in-group contact, and afterwards, with out-group contact. We found that higher ethnic identifiers had more contact with in-group members, β = .390, S.E. = .119, p = .001, but they did not have significantly less contact with out-group members, β = -.008, S.E. = .128, p = .948. In contrast, those higher in Yugoslavian identification reported more contact with out-group members, β = .251, S.E. = .109, p = .021, but not less contact with in-group members, β = .078, S.E. = .106, p = .465. Further, relative out-group guilt was associated with more in-group contact (β = .372, S.E. = .074, p < .001) and less out-group contact, though the latter path was not significant (β = -.171, S.E. = .090, p = .183). Yugonostalgia was, in contrast, related to less in-group contact (β = -.126, S.E. = .099, p = .284) and more out-group contact (β = .204, S.E. = .108, p = .106), yet both paths were not significantly different from zero. Similar to the results for guilt, the separate measures of contact seem to point in the expected directions, but the effects are weaker than the effects on the combined relative score.
Discussion
In this paper we examined the role of group identities and collective emotional memories of the past in shaping present-day intergroup relations between previously conflicted ethnic groups from former Yugoslavia that now live in Australia. Our study demonstrates that violent conflicts “there and then” can have implications for intergroup relations in the “here and now”. What happened in the past in one’s country of origin can still be meaningful for how ethnic group members relate to each other in the country of settlement. The past violent conflict leaves its mark not only on people still living in the area but also on those who have migrated to a far-away place.
We focused on intergroup contacts as a behavioral measure of intergroup relations. While intergroup contacts were not absent in our sample, Bosniak, Croatian and Serbian participants did not gravitate much towards each other but clearly hung out more with people from their own ethnic group. This finding suggests that people from these groups might have partially transferred the conflict to this new place, at least to the extent that they did not regularly seek each other’s company. Also the fact that it was difficult to find people who were willing to take part in a survey about groups from former Yugoslavia illustrates that intergroup boundaries are still very clear in the new country of settlement. This general pattern corresponds to the findings from research conducted in the countries that formed after the dissolution of Yugoslavia. Namely, the relationships between members of these groups are, generally speaking, still not very amicable (Ajduković & Čorkalo Biruški, 2008; Rogić & Šakić, 1997).
Empirically, we examined factors that might stimulate or hinder the development of interethnic contacts, and as such, are conducive to, or stand in the way of, improving intergroup relations among these diaspora groups from the same conflict area. Our findings demonstrated that relative out-group contact was associated with collective emotional memories and with Yugoslavian and ethnic identification. Higher Yugoslavian identifiers had more contacts with ethnic out-groups from former Yugoslavia (relative to in-group contacts), and this was due to their being more nostalgic about the Yugoslavian past. Higher ethnic identifiers, on the other hand, had fewer relative out-group contacts, and this was due to attributing more guilt for the conflict to out-groups as compared to the in-group.
Identification with former Yugoslavia was associated with stronger nostalgia about a federal country that ceased to exist, and via Yugonostalgia, to relatively more outgroup contacts. This is in line with research that demonstrates that nostalgia can function as a coping mechanism for dealing with identity discontinuity, and that nostalgic feelings can bring category members closer together (Turner et al., 2012; Wildschut et al., 2014). Our findings thus extend the literature on group-based nostalgia that has shown that being nostalgic about an in-group’s past can lead to negative intergroup relations between natives and immigrant minorities (Smeekes, 2015; Smeekes et al., 2015). We have shown that intergroup relations depend on the specific type of in-group past that one is nostalgic about, and in particular that feeling nostalgic about a former shared category can have positive intergroup implications.
Yugonostalgia in our study was not associated with ethnic identification, which suggests that a stronger sense of belonging to one’s ethnic group does not necessarily undermine nostalgic feelings about the superordinate category. However, looking at the correlations per ethnic group (Table 1), we see that there is a marginally significant negative relationship between Yugonostalgia and feeling Croatian as well as feeling Bosniak, whereas this relationship is much weaker and far from being significant in the Serbian sample. These findings suggest that we should be careful when interpreting the link between ethnic identification and Yugonostalgia, as this might vary across ethnic groups. Particularly for Croats, who wanted independence and self-determination (Antonić, 1997), feeling strongly Croatian could be related to wanting to distance oneself from the Yugoslav past. Nostalgia implies longing for the “good old days”, and high Croatian identifiers might not remember those old days as good, but rather as the days when their group did not have a sovereign state.
Ethnic identification, however, was associated with relative assignment of guilt to out-group. All three ethnic groups showed strong in-group bias in the sense that they considered their in-group as less guilty for the war compared to the out-groups (e.g., Çelebi et al., 2014; Doosje et al., 1998; Jelić et al., 2013). And because higher identifiers are more concerned about creating a positive intergroup distinctiveness (Brewer, 1991), particularly when their identification includes glorification (Roccas et al., 2006), strong ethnic identifiers denied in-group guilt and thus attributed relatively more blame to the out-groups. In turn, perceiving the out-group as relatively guiltier than the in-group was associated with relatively fewer contacts with out-group members. This indicates that collective guilt attributions for past violence are meaningful for current intergroup relations. It should be noted that recognition of in-group guilt seems to be a more relevant explanation for the relationship between ethnic identification and relative out-group contacts than the assignment of guilt to out-group. Future research with larger samples could help to better disentangle the roles of the two types of guilt attributions.
Interestingly, Yugoslavian identification was not associated with relative assignment of guilt to out-group. There was no evidence in the overall sample that identifying with an overarching (former) category, by increasing mutual understanding (Gaertner et al., 1993), decreases relative guilt assignment to out-groups for the harm doing (Wohl & Branscombe, 2005). However, the associations for the separate ethnic groups (see correlations in Table 1) show that for Croats and Bosniaks, Yugoslavian identification was associated with lower relative assignment of guilt to the out-group, which would be in line with our theorizing. This was not found for Serbs, for whom the correlation between Yugoslavian identification and relative out-group guilt was even positive, though not significant. It should be noted that Serbs were more fervent supporters of the Yugoslavian federation than other ethnic groups in former Yugoslavia (Antonić, 1997). Perhaps Serbs who identify strongly as Yugoslavian reproach other groups, and particularly Croats, for proclaiming independence and causing the dissolution of the country. This would explain why for Serbs we did not find the expected negative association between Yugoslavian identification and relative out-group guilt, which we expected based on the common in-group identity model (Gaertner et al., 1993).
Limitations
There are three main limitations of this research that should be addressed in future research. First, our study was cross-sectional, which means that we cannot draw any conclusions about causality. In particular, reversed causal paths cannot be ruled out. It could, for instance, be the case that intergroup contacts help people to take the perspective of the out-group and therefore assume more responsibility for their in-group’s actions and attribute relatively less guilt to out-groups (Čehajić & Brown, 2010). There is also research showing that remembering an encounter with an out-group member in a nostalgic way can increase the sense of shared identity (Turner et al., 2012). Future research should focus on confirming the proposed directionality of paths longitudinally.
Another limitation is that the sample size was relatively small and participants were not randomly selected. Despite that, in terms of gender, age, and education levels, the sample was well balanced suggesting that we managed to reach a demographically diverse group of participants. The reason for the small sample size is the fact that there was a lot of opposition to take part in a survey on the topic of former Yugoslavia. This enhances the likelihood that our sample is biased (e.g., consisting mainly of people who were more open to engaging with these topics). Therefore, we suggest that the mean scores be interpreted with caution. For example, the mean score for Yugoslavian identification is probably higher than what we would have obtained from a representative sample, and the mean score on ethnic identification is probably lower. However, the differences in mean scores that we have detected for the different ethnic groups seem plausible in light of the history of the conflict in the former Yugoslavia. For instance, the finding that the Croatian participants identified much less with Yugoslavia and felt less nostalgic about it, is meaningful in this context because Croats (together with Slovenians) were the first to proclaim independence in 1991 and had moreover done so once before, during World War II (Glenny, 1999). Furthermore, it is also understandable that for Bosniak participants the discrepancy between out-group and in-group guilt was the highest given that in terms of the numbers of displaced persons and casualties (e.g., the Srebrenica massacre), Bosniaks as a group ended up suffering the most in the war (Brunborg, Lyngstad, & Urdal, 2003; Friedman, 2013).
While sample selectivity might have affected the mean scores, there is no reason to believe it affected the relationships between the constructs. Importantly, the numbers of Serbs, Bosniaks and Croats were comparable, and the associations between the different constructs were quite similar for the three ethnic groups. Nevertheless, future studies, based on larger samples, should test a multi-group model and examine whether the proposed theoretical paths apply equally to all three ethnic groups. Moreover, such studies could also examine guilt attributions and contact frequencies separately in relation to each out-group instead of looking at the combined out-group scores.
A third limitation is that we could not systematically examine and compare the processes across migration generations because there were only thirteen second generation participants in the data set. We are confident, based on a robustness check, that the findings from the overall model hold equally when tested only among first generation immigrants. However, to test whether the processes are the same among the children of immigrants, future studies should use larger samples of second-generation participants.
Indeed, generational differences may be important for the representations of the past. Collective emotions about past events, such as guilt or Yugonostalgia, can be experienced not only by first generation immigrants from former Yugoslavia but also their Australian born children. On the one hand, second generation immigrants might experience these emotions to a lesser extent because they were not directly part of that specific past. On the other hand, precisely because of temporal and spatial distance, the second generation might attribute more guilt to out-groups than first generation migrants who would have experienced both the positive and negative aspects of living in former Yugoslavia both in peace and in war times. Looking at the mean comparisons in our sample, it seems that there is some support for this reasoning, given that first generation migrants were more Yugonostalgic than the second generation and attributed (marginally) more guilt to out-groups compared to their in-group. A more systematic comparison of first and second generation migrants would thus be an interesting avenue for future research on collective memories and intergroup relations.
Conclusion
Our study is one of the first to examine interethnic contacts among diaspora groups originating from the same conflict area. Using unique survey data collected among members of three conflicted groups (Croats, Bosniaks and Serbs) from a dissolved country (former Yugoslavia) that have emigrated to the same host country (Australia), we have shown how relevant group identities and collective memories of the past can be in shaping present-day interethnic relations.