Introduction
Traumatic events affect how people behave. Exposure to wartime violence can hamper political participation (Lyall Reference Lyall2009) or increase it (Bateson Reference Bateson2012; Bellows and Miguel Reference Bellows and Miguel2009; Blattman Reference Blattman2009). Experiencing violence can lead to apathy (Wood Reference Wood2006), but also to backlash mobilization (Francisco Reference Francisco2004) and stronger social cohesion (Gilligan et al. Reference Gilligan, Pasquale and Samii2014). These diverging findings suggest that the relationship between traumatic events and behavior is meaningful, but far from simple. In this article, I explore how people who experienced wartime displacement behave toward distant strangers who experienced the same.
In particular, I explore whether drawing attention to someone’s identity as a displaced person affects their altruism toward displaced strangers who are members of an ethnic outgroup. The idea of “altruism born of suffering” (Staub Reference Staub2003, Vollhardt Reference Vollhardt2009) posits that people who have experienced hardship may be more likely to help others. Whether such helping behaviors are limited to ingroup members is less clear; some studies suggest that this is the case (Gilligan et al. Reference Gilligan, Pasquale and Samii2014, Voors et al. Reference Voors, Nillesen, Verwimp, Bulte, Lensink and Van Soest2012), while others do not (Bauer et al. Reference Bauer, Cassar, Chytilová and Henrich2013, Whitt and Wilson Reference Whitt and Wilson2007). I explore the flexibility of ingroup salience by recategorizing strangers ordinarily seen as outgroup members into ingroup members, where the new ingroup is defined by a superordinate identity—an identity that is shared and includes both groups (Gaertner et al. Reference Gaertner, Dovidio, Anastasio, Bachman and Rust1993, Transue Reference Transue2007). I then examine whether such recategorization results in more altruism exhibited toward the new ingroup.
To explore this relationship between shared experience of hardship and altruism, I study formerly displaced ethnic Serbs’ reactions to Syrian refugees traveling along the Balkan route. Following the Yugoslav wars of the 1990s and early 2000s, almost a quarter million refugees and internally displaced people remain in Serbia. Since 2015, hundreds of thousands of other refugees, mostly from Syria and Afghanistan, have traveled through Serbia on their way to northern Europe. Formerly displaced ethnic Serbs provide a valuable opportunity to examine the role of shared hardship because they do not share characteristics that might otherwise be used to elicit a sense of commonality with the new refugees, like common religion or ethnicity. Rather, this context provides a hard test of recategorization as Serbs displaced during the Yugoslav wars may harbor anti-Muslim sentiment, which was relevant in the ethno-religious context of those wars and which may also matter in the context of refugees from Syria and Afghanistan, many of whom are perceived to be Muslim.
Serbian public discourse discussed Serbia’s role in the so-called refugee crisis in two ways (Šelo Šabić Reference Šelo Šabić2017). First, it emphasized empathy that Serbian citizens might feel given their shared wartime displacement experiences. Second, it highlighted Serbian exceptionalism in its humane response to the refugees, particularly when compared to other Balkan route countries. These two narratives present an opportunity to examine how ingroup salience affects altruism: does a reminder of displacement (superordinate group) better motivate altruism than a reminder of Serbian (ingroup) identity?
To explore this question, I recruited participants who experienced wartime displacement and conducted an online survey experiment. Only individuals who experienced displacement due to conflicts in Croatia (1991–1995), Bosnia (1992–1995), or Kosovo (1998–1999) were included. Half of eligible participants received the superordinate “displacement” prime, while the other half received an ingroup “Serbian” prime; the primes reflect the Serbian discourse on the crisis and follow scholarship that compares two contextually salient primes (as opposed to a control and a treatment) (Chong and Druckman Reference Chong and Druckman2007; McClendon and Riedl Reference McClendon and Riedl2015). Participants then played the dictator game in which they had the option of sharing their participation earnings with an anonymous Syrian refugee family; this is a widely used behavioral measure of altruism (Hoffman et al. Reference Hoffman, McCabe, Shachat and Smith1994).
The priming increased the salience of the superordinate identity among treated participants, but had no effect on their altruism. In fact, participants who experienced wartime violence in addition to displacement were more generous in response to a treatment that primed the opposite—their Serbian identity. This challenges a number of findings on superordinate identity salience and altruism (Charnysh et al. Reference Charnysh, Lucas and Singh2015; Gaertner et al. Reference Gaertner, Dovidio, Anastasio, Bachman and Rust1993; Levine and Thompson Reference Levine and Thompson2004; Riek et al. Reference Riek, Mania, Gaertner, McDonald and Lamoreaux2010) and suggests that simply increasing the salience of a superordinate displaced identity does not necessarily increase altruism toward displaced strangers. The findings suggest that host society interventions may even benefit from capitalizing on differences between hosts and refugees (Brewer Reference Brewer1991). More broadly, these findings speak to scholarship on identity salience (Klar Reference Klar2013; Diamond Reference Diamond2020), outgroup derogation (Hewstone et al. Reference Hewstone, Rubin and Willis2002), host society attitudes toward refugees (Bansak et al. Reference Bansak, Hainmueller and Hangartner2016), and immigrant inclusion (Gorodzeisky and Semyonov Reference Gorodzeisky and Semyonov2020; Schenk Reference Schenk2021; Tyrberg Reference Tyrberg2024).
Background
Reactions to Refugees
Host society reactions to refugees and asylum seekers vary, often in response to who the refugees are or how they are perceived. For example, survey participants across 15 European countries express significantly greater support for asylum applicants with high employability, severe vulnerabilities, and consistent asylum testimonies (Bansak et al. Reference Bansak, Hainmueller and Hangartner2016). In contrast, Muslim asylum applicants receive far less support (Bansak et al. Reference Bansak, Hainmueller and Hangartner2016) and often experience discrimination in the destination country (Paz and Kook Reference Paz and Kook2021). Contact and exposure are likewise important, if inconsistent, determinants of host society attitudes toward refugees (Altındağ and Kaushal Reference Altındağ and Kaushal2021, Getmansky et al. Reference Getmansky, Sınmazdemir and Zeitzoff2018, Zorlu Reference Zorlu2017).
Recent studies examine the effectiveness of several different interventions aimed at improving host society attitudes. First, perspective taking—asking the participant to imagine they were a refugee—results in supportive attitudes and behaviors among Americans with no personal experience of displacement (Adida et al. Reference Adida, Lo and Platas2018; Williamson et al. Reference Williamson, Adida, Lo, Platas, Prather and Werfel2021), among Greeks whose ancestors experienced displacement after the Greco-Turkish War of 1919–1922, and among Germans whose ancestors were expelled from the former Eastern territories of the German Reich after the end of World War II (Dinas et al. Reference Dinas, Fouka and Schläpfer2021). Second, emphasizing shared religion between Syrian refugees and Turkish citizens improves certain Turkish attitudes toward refugees and increases respondent donations to a refugee charity (Lazarev and Sharma Reference Lazarev and Sharma2017).
These recent studies reveal that while perspective taking and emphasizing shared religion appear to be effective in fostering refugee inclusion, some interventions fail and may even lead to backlash. For example, providing facts about countries’ commitments to accept refugees does not change attitudes among some and results in negative reactions among others (Adida et al. Reference Adida, Lo and Platas2018; Getmansky et al. Reference Getmansky, Sınmazdemir and Zeitzoff2018). Similarly, a positive reminder that an open-door policy saves innocent lives can reduce support for refugees (Getmansky et al. Reference Getmansky, Sınmazdemir and Zeitzoff2018). Thus, more research is needed not only concerning effective strategies but potentially counterproductive ones as well. This is especially true for understudied communities such as those that have experienced wartime displacement themselves (but see Hall Reference Hall2016, Reference Hall2018 and Corstange and York Reference Corstange and York2018). In addition, more work is needed to explore the role of ethnicity in these dynamics (Bracic Reference Bracic2022).
With this goal in mind, I examine an intervention proven successful in several contexts: recategorization based on a shared superordinate identity (Gaertner et al. Reference Gaertner, Dovidio, Anastasio, Bachman and Rust1993; Transue Reference Transue2007). Like Lazarev and Sharma (Reference Lazarev and Sharma2017), I exploit a common identity between hosts and refugees, but unlike them, I choose an identity acquired by experiencing displacement. Like Dinas et al. (Reference Dinas, Fouka and Schläpfer2021), I leverage a personal experience of the same sort of hardship—displacement—but unlike them, I examine participants with first-hand experience of displacement. I also diverge from Dinas et al. (Reference Dinas, Fouka and Schläpfer2021) in examining the effect of a shared identity alone, without a perspective-taking component that explicitly draws parallels between participants’ personal experiences and a current refugee crisis. In doing so, I conduct a hard test by comparing the effects of a shared acquired identity to those of a highly salient identity dimension that constitutes a difference between hosts and refugees: ethnicity.Footnote 1 Finally, I examine host society reactions to refugees in a relatively underexamined context: Serbia.
Serbia
Serbia, along with other Western Balkan states, often appears in popular imagination as part of a volatile region shaped by deeply rooted ethno-nationalist sentiment. This view is, at best, simplistic (Todorova Reference Todorova1997). The wars that accompanied Yugoslavia’s dissolution in the 1990s and the 2000s were indeed fought along ethno-religious lines and resulted in the partition of a multi-ethnic Yugoslavian federation into states that were far less ethnically diverse (Subotić Reference Subotić2015). The causes of the wars, however, were complex and not rooted in ancient ethnic hatreds (Gagnon Reference Gagnon2004, Jović Reference Jović2001). Ethnic hatred was created anew—by cultural, political, and intellectual elites who engaged in manipulation of the public (Jović Reference Jović2001, Klanjšek and Flere Reference Klanjšek and Flere2011, Sekulić et al. Reference Sekulić, Massey and Hodson2006).
Since before the wars and continuing to present day, Serbian identity has been variously constructed, narrated, and contested. In Yugoslavia, under Tito’s leadership, identity narratives selectively highlighted aspects of unity among the Southern Slavic ethnic communities (Malešević Reference Malešević2002) to facilitate the construction of the supranational Yugoslavian identity (Kalemaj and Lleshi Reference Kalemaj, Lleshi, Meka and Bianchini2020). In the late 1980s, Serbia saw a shift in elite discourse. Serbian national identity became its focus, with discussions eventually coalescing around two opposing but mutually reinforcing constructions: First Serbia and Other Serbia (Russell-Omaljev Reference Russell-Omaljev2016).
The First Serbia discourse was “all-penetrating and dominant” (Russell-Omaljev 86, Reference Russell-Omaljev2016) and placed Serbian ethnicity at its core. After gaining power in the League of Communists of Serbia in 1987, Milošević mobilized supporters through extreme ethno-nationalist rhetoric that relied on a set of myths no longer suppressed for the purpose of unity: the Kosovo myth, the myth of Saint Sava, the myth of victimization in Jasenovac, and myths of Serbian military victories and the sacrifices of its heroes (Malešević Reference Malešević2002). These myths collectively ground Serbian identity in narratives of sacrifice, victimhood, and heroism while presenting Serbian culture as largely unchanged since the 8th century. Their content is “solely ethnic (Serb) and occasionally religious (Orthodox Serb)” (Malešević Reference Malešević2002, 177). Serbian political and cultural elites invoked these myths in the period before and during the Yugoslav wars, strengthening the exclusionary identity by portraying Croats, Kosovar Albanians, the United States, and the European Union as enemies (Malešević Reference Malešević2002, Subotić Reference Subotić2011).
Other Serbia emerged in 1992 from the civic opposition to the Yugoslav wars and the Milošević regime. Witnessing atrocities committed in the name of ethnic superiority, Other Serbia struggled with the idea of a Serbian national identity and embraced a pacifist, civic collective identity grounded in moral and ethical principles rather than ethnicity (Russell-Omaljev Reference Russell-Omaljev2016). Supported by a minority of citizens, Other Serbia explicitly rejected the dominant First Serbia discourse (Dawson Reference Dawson2014) and Serbian ethno-nationalism (Russell-Omaljev Reference Russell-Omaljev2016). It openly fought against war crimes and ethnic cleansing.
In the post-2000 period, Other Serbia splintered when the anti-war stance no longer unified the liberal intelligentsia (Russell-Omaljev Reference Russell-Omaljev2016). First Serbia discourse refocused on Europe in response to Europe’s view that Serbia was responsible for the breakup of Yugoslavia and the war atrocities (Subotić Reference Subotić2011). While ethnic nationalism became less central as a new divide emerged between moderate and fundamentalist nationalists (Russell-Omaljev Reference Russell-Omaljev2016), the loss of Kosovo—which First Serbia considered inherently Serbian—reinforced ethno-nationalist sentiment (Subotić Reference Subotić2011). Though Kosovo’s loss remains a central issue today, narratives of the war appear less frequently in contemporary conversations about identity (Wygnańska Reference Wygnańska2021).
The Balkan Route
During the European refugee crisis between 2014 and 2017, over 3 million people applied for asylum in several European states (Eurostat 2018). Syrians and Afghans comprised the two largest asylum applicant groups; the remainder were predominantly from Iraq, Pakistan, Albania, Eritrea, Kosovo, Somalia, Nigeria, and Bangladesh (Dustmann et al. Reference Dustmann, Fasani, Frattini, Minale and Schönberg2016). Depending on their country of origin, most refugees reached the EU through Mediterranean routes, the route along the border of the easternmost EU member states, routes through the Canary Islands and the Iberian peninsula, or the western Balkan route (Frontex European Border & Coast Guard Agency 2018).
With its history of conflict and its prominent position on the Balkan route, Serbia presents an opportunity to explore the link between experiences of displacement and altruism. Serbia is home to many people experiencing displacement, either as refugees or as persons internally displaced during the Yugoslav wars. In total, Serbia has hosted 537,937 refugees, mostly from conflicts in Croatia (1991–1995) and Bosnia and Herzegovina (1992–1995), and 209,021 internally displaced persons, mostly from the conflict in Kosovo (1998–1999) (KIRS 2016). In June 2014, 43,763 refugees still lived in Serbia, as did 204,049 internally displaced people (KIRS 2016), comprising a total of 3.5% of the Serbian population (KIRS 2016, RZS 2016); the actual proportion may have been even higher as only those formally registered as displaced were included in the official count.
In recent years, Serbs have witnessed tens of thousands of refugees traveling the Balkan route. From October 2015 to February 2016, between 1,000 and 10,000 refugees entered Serbia daily, most continuing north toward Germany and beyond (UNHCR 2016). In early March 2016, Austrian border restrictions aimed at closing the Balkan route triggered a wave of similar policies in the countries to the south. Following these border closures, the number of refugees traveling the Balkan route was dramatically reduced, although substantial numbers still passed through, legally and illegally (Vasovic Reference Vasovic2016). In March 2017, the UNHCR reported that 7,900 refugees entered Serbia legally; 707 of them formally registered their intent to seek asylum there (UNHCR 2017).
Serbian public discourse surrounding the crisis differed substantially from discourse in many other European countries. While some press coverage in Hungary and France persistently promoted hostility and hate speech, and while the media in Czechia, France, and the United Kingdom wrote more about national security than about caring for the newly displaced, the Serbian press focused on helping refugees (Georgiou and Zaborowski Reference Georgiou and Zaborowski2017). Early narratives concentrated on the criminality of smuggling networks, but as the numbers passing through rose, humanitarian narratives prevailed, emphasizing the human dimension of the crisis (Petrović Trifunović and Poleti Ćosić Reference Petrović Trifunović, Poleti Ćosić, Pudar Draško, Pavlović and Meka2018; Šelo Šabić Reference Šelo Šabić2017).
One common narrative favorably compared Serbia’s response to the crisis with those of its neighbors (Galijaš Reference Galijaš2019, Petrović Trifunović and Poleti Ćosić Reference Petrović Trifunović, Poleti Ćosić, Pudar Draško, Pavlović and Meka2018). Serbia was thus “saving Europe’s soul” (Petrović Trifunović and Poleti Ćosić Reference Petrović Trifunović, Poleti Ćosić, Pudar Draško, Pavlović and Meka2018, 223) and “teaching Europe a lesson” (223) while “Bulgaria [set] tanks on refugees” (224), “Croats [forced] women and children into a frozen river” (224), and Hungary’s Victor Orbán, the “fascist from the heart of Europe” (224), built a wall. This framing presented Serbian “hospitality and kindness toward refugees” (Petrović Trifunović and Poleti Ćosić Reference Petrović Trifunović, Poleti Ćosić, Pudar Draško, Pavlović and Meka2018, 221) as a stark departure from how EU members and other states along the Balkan route approached the crisis.
Another common narrative connected Serbs and the refugees, highlighting Serbian empathy stemming from the experience of displacement and wartime violence during the Yugoslav conflicts (Šelo Šabić Reference Šelo Šabić2017). For example, in response to a photograph of a smiling Serbian policeman holding a Syrian child, some Serbs reported experiencing “the moment of salvation, because they themselves have been through the war” (Blic, 12 September 2015, 4 in Petrović Trifunović and Poleti Ćosić Reference Petrović Trifunović, Poleti Ćosić, Pudar Draško, Pavlović and Meka2018, 223).
Scholars link these narratives to a combination of politically salient factors. The first is Serbia’s need to rebuild its image in the aftermath of the atrocities during the Yugoslav wars and its stigma as a “rogue state” (Petrović Trifunović and Poleti Ćosić Reference Petrović Trifunović, Poleti Ćosić, Pudar Draško, Pavlović and Meka2018, 212). The second is Serbia’s status as an EU candidate. The refugee crisis allowed Serbia to advance both agendas by linking the policy of welcoming to memories of Serbia’s own wartime plight and by showcasing the country as a competent and responsible champion of European values (Šelo Šabić Reference Šelo Šabić2017). These efforts were recognized in December 2015 with the opening of the first accession negotiation chapters. Finally, scholars have also noted that Serbian discourse reflects its status as a transit country and not a destination (Galijaš Reference Galijaš2019). Indeed, as countries along the Balkan route progressively closed their borders, the discourse in Serbia became more security-driven (Sardelić Reference Sardelić2017; Šelo Šabić Reference Šelo Šabić2017).
In short, Serbia offers a compelling opportunity to explore how previously displaced people behave toward displaced strangers in a real-life context, given its (1) residents who experienced displacement, (2) prominent position on the Balkan route, and (3) crisis-response narratives of ingroup exceptionalism and shared hardship.
Theoretical Expectations
Humans possess and can activate multiple social identities. Which social identity is activated depends on the environment, as individuals self-categorize based on contextually salient social categories (Turner et al. Reference Turner, Hogg, Oakes, Reicher and Wetherell1987). These social categories then classify the ingroup, to which the individual belongs, and outgroups (Tajfel and Turner Reference Tajfel and Turner1979). Once people are categorized into groups, perceived differences between members of the same group are minimized (Tajfel Reference Tajfel1969), while perceived differences between ingroup and outgroup members are enhanced. When an individual activates a social identity, she begins to behave as a member of that ingroup (Turner and Onorato Reference Turner, Onorato, Tom, Roderick and Oliver1999).
People tend to systematically favor their own group members over those of other groups (Billing and Tajfel Reference Billing and Tajfel1973; Dasgupta Reference Dasgupta2004). This bias can manifest itself in ingroup favoritism, outgroup derogation, or both (Hewstone, Rubin, and Willis Reference Hewstone, Rubin and Willis2002). Prosocial behaviors that benefit others are more commonly exhibited toward an ingroup: people are more likely to help, trust, or share resources with members of their own group (Penner et al. Reference Penner, Dovidio, Piliavin and Schroeder2005). Correspondingly, people are less likely to be prosocial toward outsiders. For example, Greeks donate more money to feed ethnic Greek children than Roma children (Linos et al. Reference Linos, Jakli and Carlson2021), and Germans are more likely to help a German native than a Muslim immigrant after a mishap (Choi et al. Reference Choi, Poertner and Sambanis2019). More generally, people who strongly identify with their national ingroup tend to express more support for exclusionary immigration policies, though this depends on the way people think about national identity (Bracic et al. Reference Bracic, Israel-Trummel and Shortle2023; Breidahl et al. Reference Breidahl, Holtug and Kongshøj2018; Goodman and Alarian Reference Goodman and Alarian2021; Rapp Reference Rapp2022; Schildkraut Reference Schildkraut2011). When the national ingroup is defined in civic terms—a commitment to the country’s laws and institutions—people are more open to refugees, migrants, or asylum seekers. When the ingroup is instead defined in ethnic terms, commitment to the national ingroup results in exclusionary attitudes (Esses et al. Reference Esses, Hamilton and Gaucher2017).
After Tito’s Yugoslavia, views of Serbian national identity have been shaped by the dominant First Serbia discourse, which grounded the identity in ethnicity, and its much smaller opposite, Other Serbia, which grounded it in civic terms (Russell-Omaljev Reference Russell-Omaljev2016). While the two counterparts have since divided into multiple visions of identity that fill out the space between them, ethnocultural narratives of Kosovo belonging to Serbia persist and are quite common (Wygnańska Reference Wygnańska2021). For example, 78 percent of the participants in this study strongly agree that Kosovo and Metohija belong to Serbia.Footnote 2 And Serbs who believe that “every country belongs primarily to its first inhabitants” are more willing to engage in an anti-refugee demonstration (Hasbún López et al. Reference Hasbún López, Martinović, Bobowik, Chryssochoou, Cichocka, Ernst-Vintila, Franc, Fülöp, Ghilani, Kochar, Lamberty, Leone, Licata and Žeželj2019). It is therefore reasonable to expect that ethnic identity remains salient in Serbia and that it’s linked to exclusionary attitudes.
Group categorization is often spontaneous, but it is not unalterable. The characteristics on the basis of which categorization occurs can be modified by manipulating the perceiver’s expectations, goals, motives, past experiences, as well as situational factors (Dovidio and Gaertner Reference Dovidio and Gaertner2000). Ingroups form based on random assignment and clothing color (Wright et al. Reference Wright, Aron, McLaughlin-Volpe and Ropp1997), but also on the basis of ethnicity (Montoya and Pittinsky Reference Montoya and Pittinsky2016), disability (Darling Reference Darling2013), and environmental disaster experience (Vezzali et al. Reference Vezzali, Cadamuro, Versari, Giovannini and Trifiletti2015). Personal experience of displacement should therefore also enable people to form ingroups.
Recategorization, as proposed by the Common Ingroup Identity Model (Gaertner et al. Reference Gaertner, Dovidio, Anastasio, Bachman and Rust1993), seeks to reshape group boundaries by creating an inclusive ingroup in place of two or more outgroups. While recategorization processes don’t do away with the smaller groups, they increase the salience of the superordinate identity, emphasizing the shared bond between the smaller groups. As the salience of a superordinate identity common to two outgroups increases, behaviors between members should improve following the cognitive processes that motivate ingroup favoritism (Gaertner at al. Reference Gaertner, Dovidio, Nier, Banker, Ward, Houlette, Loux, Capozza and Brown2000). Interventions that emphasize superordinate identities, ranging from dress similarity (Dovidio et al. Reference Dovidio, Gaertner, Isen and Lowrance1995) to party identification (Riek et al. Reference Riek, Mania, Gaertner, McDonald and Lamoreaux2010), can reduce intergroup bias and lead to increased helping behaviors (Dovidio et al. Reference Dovidio, Gaertner, Validzic, Matoka, Johnson and Frazier1997; Levine et al. Reference Levine, Prosser, Evans and Reicher2005). For example, British citizens whose identity as Europeans was made salient indicate a substantially higher willingness to donate money to help victims of an environmental disaster in Europe than participants whose British identity was salient instead (Levine and Thompson Reference Levine and Thompson2004). Among Australians, those identifying strongly as humans are substantially more welcoming toward asylum seekers, while those strongly identifying as Australians are not (Nickerson and Louis Reference Nickerson and Louis2008). Indian Hindus whose salience of their Indian national identity was increased express more altruism toward Indian Muslims (Charnysh et al. Reference Charnysh, Lucas and Singh2015).
The relationship between shared hardship and altruism grounded in the common ingroup identity model could result from several mechanisms. People with a universal orientation find it easier to see similarities with others (Phillips and Ziller Reference Phillips and Ziller1997), while people with higher levels of empathy might be more likely to draw parallels between their own experiences and someone else’s (Zaki and Ochsner Reference Zaki and Ochsner2012). Empathy and universal orientation are mechanisms that could lead a person who experienced displacement to form inclusive victim consciousness—a perception that the suffering of their ingroup is similar to the suffering of other groups (Vollhardt et al. Reference Vollhardt, Cohrs, Szabó, Winiewski, Twali, Hadjiandreou and McNeill2021). Inclusive victim consciousness may then promote altruism toward others who experienced displacement.
An alternative mechanism—perceived ingroup superiority—is a perception that fosters a competitive posture toward other groups (Roccas et al. Reference Roccas, Sagiv, Schwartz, Halevy and Eidelson2008) and positions the ingroup as morally superior (Noor et al. Reference Noor, Shnabel, Halabi and Nadler2012). People who see their ingroup as superior may develop exclusive victim consciousness where they perceive their own experiences of suffering as unique (Vollhardt et al. Reference Vollhardt, Nair and Tropp2016). This likely fosters negative attitudes toward other groups who also suffered (Vollhardt et al. Reference Vollhardt, Cohrs, Szabó, Winiewski, Twali, Hadjiandreou and McNeill2021). Finally, intergroup contact could also shape the relationship between shared hardship and altruism, but the direction of the effect may vary (Homola and Tavits Reference Homola and Tavits2018, Vollhardt et al. Reference Vollhardt, Cohrs, Szabó, Winiewski, Twali, Hadjiandreou and McNeill2021, Zorlu Reference Zorlu2017). These mechanisms illuminate the potential relationship between shared hardship and altruism. This study, however, is not designed to distinguish between them.
Serbian public discourse on the refugee crisis appealed to both ingroup and superordinate identities. The first narrative highlights Serbian exceptionalism and could evoke feelings of ingroup favoritism. The second narrative focuses on shared hardship, evoking memories of Serbia’s own wartime plight. Among Serbs who experienced wartime displacement, such a narrative might increase the salience of a superordinate displacement identity. During the refugee crisis, Serbs were likely exposed to these narratives. Given the contextual salience of these narratives, I therefore compare the effectiveness of an ingroup and a superordinate prime, foregoing a control condition (Chong and Druckman Reference Chong and Druckman2007; McClendon and Riedl Reference McClendon and Riedl2015). Following social identity theory, I expect participants receiving the ingroup ethnic identity prime to express less altruism toward refugees traveling the Balkan route. By contrast, I expect that recategorization will lead to higher altruism toward the refugees:
H1: Participants whose displaced identity is primed will, on average, engage in significantly higher levels of altruism than participants whose ingroup identity is primed.
While all those displaced in war suffer, such experiences will differ. A meaningful source of variation is exposure to violence, which can affect attitudes and behaviors, political or otherwise (Bateson Reference Bateson2012, Lyall Reference Lyall2009, Wood Reference Wood2006). In Burundi, for example, conflict victimization at the individual level is positively associated with altruistic behavior (Voors et al. Reference Voors, Nillesen, Verwimp, Bulte, Lensink and Van Soest2012). In Nepal, violence-affected communities also exhibit higher levels of altruism and cooperation (Gilligan et al. Reference Gilligan, Pasquale and Samii2014).
The intensity of violence also matters. In Sierra Leone, people from households that experienced more direct civil war victimization—such as murder or arson—are more likely to vote, join local political and community groups, and attend community meetings (Bellows and Miguel Reference Bellows and Miguel2009). Ex-combatants from northern Uganda, who were victims of rebel abduction and conscription, have substantially higher rates of voting and becoming community leaders (Blattman Reference Blattman2009). Among them, witnessing violence accounts for a substantial portion of this relationship. Still, while numerous studies demonstrate higher altruism among people who experienced wartime violence, this relationship is not universal. In the Republic of Georgia and Sierra Leone, increased altruism is only expressed toward ingroup members (Bauer et al. Reference Bauer, Cassar, Chytilová and Henrich2013), while in Kyrgyzstan, people in victimized neighborhoods express less prosociality toward ingroup and outgroup members alike (Hager et al. Reference Hager, Krakowski and Schaub2019).
Nevertheless, as most research shows that exposure to wartime violence increases prosociality, my second hypothesis is:
H2: Participants who experienced war violence in addition to displacement will, on average, engage in significantly higher levels of altruism than participants who did not.
Empirical Strategy
I conducted an online survey experiment. A local research firm, Ninamedia, administered the survey between June 1 and August 2, 2017. Participants were recruited using targeted Facebook advertisements, a method that has recently been used to recruit immigrant respondents (Tyrberg Reference Tyrberg2024). I chose this method of recruitment after exhausting several other recruiting options, none of which garnered a sufficient number of participants. I discuss the unsuccessful strategies in the Online Appendix. While not ideal, this recruitment strategy is better than snowball sampling, which is often used to recruit hard-to-reach populations but results in participants who are interconnected and often similar to one another (Shaghaghi et al. Reference Shaghaghi, Bhopal and Sheikh2011). In fact, Facebook-recruited convenience samples can provide useful evidence for experimental treatment effects, even though they have lower generalizability (Krupnikov et al. Reference Krupnikov, Nam, Style, James and Donald2021).
Given the sensitive nature of my online survey experiment, recruitment materials and the consenting procedure explicitly identified it as a study of wartime displacement and reactions to the European refugee crisis. A separate statement in the survey itself warned participants about upcoming questions about their wartime experience; if they wished, they could skip these questions without reading them while still completing the survey.
Potential participants first answered four screening questions. They were eligible to participate if they experienced displacement either as refugees or as internally displaced persons during the conflicts in (1) Croatia (1991–1995); (2) Bosnia (1992–1995); or (3) Kosovo (1998–1999); and (4) were at least seven years old when the conflict occurred (Bauer et al. Reference Bauer, Cassar, Chytilová and Henrich2013).Footnote 3 The sample consists of 384 ethnic Serbs who experienced displacement.Footnote 4
Eligible participants randomly received either the ingroup Serbian prime or the superordinate displaced prime, reflecting the common narratives discussed earlier (Chong and Druckman Reference Chong and Druckman2007; McClendon and Riedl Reference McClendon and Riedl2015). To increase the salience of either the ingroup ethnic identity or the superordinate identity, I used a questions-as-treatments framework and asked participants a series of questions listed in the Appendix (Bloom et al. Reference Bloom, Arikan and Courtemanche2015, Transue Reference Transue2007, Tyrberg Reference Tyrberg2024). The final question in both primes was open-ended, asking for a short answer to “What does it mean to be Serbian?” or “What does it mean to be displaced?” The number and selection of priming questions were determined in consultation with local experts.Footnote 5 While the experimental treatments reflect the narratives in Serbian discourse on the crisis, the treatments do not directly mention the refugee crisis. The ingroup identity prime therefore primes only the Serbian ethnic ingroup, without reference to the refugee crisis or Serbia’s role in it. The superordinate prime likewise primes the displaced identity by referring to past experiences of displacement in the Yugoslav wars without referencing the refugee crisis.Footnote 6
I captured altruism by asking participants to play the dictator game (Hoffman et al. Reference Hoffman, McCabe, Shachat and Smith1994):Footnote 7
To thank you for participating in this study, the University of Oklahoma gives you 800 dinars. If you wish, you have the opportunity to divide these 800 dinars between you and a Syrian refugee family in need that would like to seek asylum in Serbia. The family will be chosen with the aid of the Commissariat for Refugees and Migration of the Republic of Serbia and will remain anonymous. After this study is over, this family will receive the bulk sum of whatever the participants in this study send to them. You don’t have to send anything to this family, but if you want to send something, type the amount in the box below.
The survey continued with several questions capturing attitudes and relevant population covariates such as age (Freund and Blanchard-Fields Reference Freund and Blanchard-Fields2014), income (Chowdhury and Jeon Reference Chowdhury and Jeon2014), gender (Croson and Gneezy Reference Croson and Gneezy2009), education, and exposure to war violence (Grossman et al. Reference Grossman, Manekin and Miodownik2015).
A word on generalizability. The sample is not representative of ethnic Serbs who have experienced displacement. Since recruitment materials were explicit about the purpose of the study, it is likely that people who self-selected into the study were willing to contemplate their own displacement. The sample is overly representative of people who are comfortable with technology, and Facebook in particular, and also disproportionately female. While gender effects on altruism are inconsistent and overall null (Balliet et al. Reference Balliet, Li, Macfarlan and Van Vugt2011), estimating the direction of the other biases in relation to altruism is difficult. The results of the analysis that follows should therefore not be generalized beyond this sample of participants.
Results
This section first discusses the effectiveness of priming and then presents the dictator game results. Briefly, the superordinate priming questions increased the comparative salience of the displaced identity, but the increased salience of this identity resulted in neither higher nor lower average contributions to the Syrian family in the dictator game (no support for H1). Participants who experienced wartime violence donated significantly more to the refugee family (support for H2) but did so in response to the ingroup and not the displacement prime.
The Relative Salience of the Superordinate Identity
The manipulation check following the priming questions asked participants to rate how strongly they felt like a Serbian and how strongly they felt like a displaced individual (5-point scale).Footnote 8 The superordinate treatment group reported a significantly higher average salience of the displaced identity compared to the ingroup prime participants (p < 0.05; t-test). The average salience of Serbian identity, however, was the same for both groups (see Table 1). Further, among ingroup prime participants, the average salience of the displaced identity was significantly lower than the average salience of Serbian identity (p < 0.05; t-test). This was not the case for the superordinate prime participants; they felt just as strongly displaced as they felt Serbian (see Table 2). The manipulation check suggests that the superordinate priming questions achieved their intended result. I next examine whether the priming that successfully raised the salience of the superordinate identity also increased altruism.
Table 1. Salience of displaced and ethnic identities, between groups (t-test)
Table 2. Salience of displaced and ethnic identities, within groups (t-test)
Levels of Altruism Exhibited by the Two Groups
Even though participants receiving the superordinate prime reported a significantly higher salience of their identity as displaced individuals, they were no more generous toward the Syrian refugee family than ingroup prime participants. Figure 1 presents the average amount sent to the Syrian refugee family by those receiving the ethnic ingroup and superordinate primes. The difference is not statistically significant: participants sent 29.5% (ethnic ingroup prime) and 24% (superordinate prime) of their 800 dinars to the family.
Figure 1. This figure presents the average amount of dinars sent to the Syrian refugee family in need along with 95% confidence intervals, by prime. Participants had an endowment of 800 dinars.
To control for covariates possibly influencing decision-making, I use a regression. The first column in Table 3 presents OLS regression results with covariates for age, gender, income, education, and personal experience of wartime violence.
Table 3. OLS regression results (dependent variable: amount sent in the dictator game)
Note: *p < 0.05
Regression results confirm the initial null finding.Footnote 9 As the superordinate prime had no discernible bearing on the participants’ levels of generosity, I have no support for H1, which hypothesized that a more salient superordinate identity would result in more altruism.Footnote 10 This result suggests that recategorization, even when it works, does not always lead to prosociality.
Personal Experience of Conflict
Here I examine whether personal experience of conflict among displaced Serbs moderates the effect of the superordinate prime. This was not a pre-registered line of inquiry and is therefore exploratory, but it offers insight into the null finding. To take a closer look at the role of exposure to violence, I divide the sample into two subgroups: (1) participants who experienced displacement and also lost someone close to them in the war and (2) participants who experienced displacement but didn’t lose anyone.
Participants who did not lose a close person during the war contributed the same amount in the dictator game, on average, regardless of the two primes (left panel of Figure 2). Participants who lost someone close to them in the war, however, responded differently to the two primes. Those who received the ingroup ethnic identity prime contributed significantly more to the refugee family. Figure 2 (right panel) shows that this group was more altruistic than participants who experienced both displacement and violence but who instead received the superordinate prime. Table 3 shows the results of two additional OLS models, now split by subgroup. The findings in column 2 are for the displacement-only subgroup and show the null effect of the displacement prime. The findings in column 3 are for the displacement and war violence subgroup and show a negative effect of the displacement prime.Footnote 11
Figure 2. This figure presents the average amount of dinars sent to the Syrian refugee family in need along with 95% confidence intervals, by prime and by personal experience of violence. Participants had an endowment of 800 dinars.
In H2, I expected that participants who experienced wartime violence would exhibit more generosity. They did so, but only when receiving the ingroup ethnic identity prime. This result presents a challenge to research on altruism that finds that people who have experienced violence favor ingroup members (Bauer et al. Reference Bauer, Cassar, Chytilová and Henrich2013). It also underscores what has been found elsewhere: personal experience of violence is a factor that shapes other-regarding behavior (Gilligan et al. Reference Gilligan, Pasquale and Samii2014, Voors et al. Reference Voors, Nillesen, Verwimp, Bulte, Lensink and Van Soest2012).
Additional Exploratory Analyses
Here, I conduct a few additional analyses. Since this study was not designed to test additional explanations, most of the measures that I use in this section are adequate for exploratory purposes, but not ideal. Several are also subject to post-treatment bias (Montgomery et al. Reference Montgomery, Nyhan and Torres2018). In most of the analyses that follow, I therefore examine the two treatment groups separately. To be clear, these analyses do not offer any definitive answers, only suggestive insight for future work. I examine feelings of closeness toward refugees, generalized anti-Muslim sentiment, contact, empathy, ideology, perceived threat, conceptions of Serbian identity, and positive identity attachment.
There are several factors that potentially shape the relationship between shared hardship and altruism. First, if people develop exclusive victim consciousness, shared hardship may not lead to altruism (Vollhardt and Bilali Reference Vollhardt and Bilali2015). Exclusive victim beliefs are based on perceived distinctiveness of ingroup victimization (Brewer Reference Brewer1991, Vollhardt Reference Vollhardt2013). For example, when the Holocaust is invoked in discussions about other rights abuses, some Jewish people welcome the reference (Schulz Reference Schulz2001, Wayne and Zhukov Reference Wayne and Zhukov2022), while others argue that it erodes the unique character and horror of the Holocaust (Rothberg Reference Rothberg2011). While this survey did not probe for feelings of distinctiveness explicitly, I asked participants how close they felt toward their ethnic ingroup, toward people who experienced wartime displacement in general, and toward Syrian and Afghan refugees. The only feelings of closeness that matter are the specific feelings felt toward Syrians and Afghans. The more closeness participants in both ingroup and superordinate subsamples felt toward these particular refugees, the more generous they were toward the family in need (Appendix Tables 15-17).
Second, when animus exists between two subgroups, recategorization is more challenging (Vollhardt Reference Vollhardt2009). The Yugoslav wars, during which this study’s participants experienced displacement, were fought along ethno-religious lines. Generalized anti-Muslim sentiment might be salient, presenting a barrier to recategorization such that ethnic Serbs may be less open to the idea of shared hardship with refugees who are largely perceived to be Muslim. I find, however, that generalized anti-Muslim sentiment was not statistically significant in either treatment group (Appendix Table 15). Exploratory evidence further shows that Serbian participants feel significantly more warmly toward Syrian refugees than they do toward Kosovar Albanians, which indicates that ethnic Serbs distinguish between wartime enemies and the new refugees—even though both groups are perceived to be Muslim (Appendix Table 18). While participants overwhelmingly subscribe to the ethno-nationalist notion that Kosovo and Metohija belong to Serbia and exhibit proportionate levels of animus toward Kosovar Albanians, these attitudes do not generalize to the refugees.
Finally, contact and empathy may shape altruism in light of shared hardship (Szabo et al. Reference Szabó, Vollhardt and Mészáros2020, Vollhardt et al. Reference Vollhardt, Cohrs, Szabó, Winiewski, Twali, Hadjiandreou and McNeill2021). I find that participants who saw refugees traveling the Balkan route were no more and no less altruistic, regardless of the prime they received (Appendix Figure 4). Similarly, I obtained null findings when examining emotion-specific empathy (Olderbak et al. Reference Olderbak, Sassenrath, Keller and Wilhelm2014) as a potential moderator (Appendix Figures 1-3). Altogether, it seems that distinctiveness likely mattered in the context of the superordinate prime, while anti-Muslim sentiment, contact with refugees, and empathy did not.
I next turn to the ingroup prime. As discussed earlier, in the decades following Tito’s death, Serbian identity was mutually constructed and challenged by two opposing viewpoints: a dominant exclusionary narrative, which placed Serbian ethnic identity at its core, and the smaller, civic narrative (Russel-Omaljev Reference Russell-Omaljev2016). Although this stark polarization has softened, the symbolic importance of Kosovo—and with it an exclusionary vision of Serbian identity—remains highly salient (Wygnańska Reference Wygnańska2021). An ingroup identity prime should have led to less altruism (Hasbún López et al. Reference Hasbún López, Martinović, Bobowik, Chryssochoou, Cichocka, Ernst-Vintila, Franc, Fülöp, Ghilani, Kochar, Lamberty, Leone, Licata and Žeželj2019, Penner et al. Reference Penner, Dovidio, Piliavin and Schroeder2005). Why didn’t it?
The ingroup prime did not change how strongly Serbian the participants felt. Still, the way participants think about Serbian identity might matter. Specifically, these findings could be driven by higher altruism among those who favor a civic vision of identity. One of the ingroup treatment questions asked participants “What does it mean to be Serbian?” This question was open-ended and 53% of the ingroup prime participants gave answers that can be categorized as having either an ethnic or a civic understanding of identity. Comparing the levels of altruism between these two groups reveals no statistically significant differences (Appendix Table 22). While participants certainly have divergent conceptions of what it means to be Serbian, these differences are not reflected in their generosity. This is especially puzzling given the two polarizing Serbian identity narratives—one exclusionary (ethnic) and one not (civic).
An alternative explanation stems from the narrative surrounding the refugee crisis in Serbia, which framed Serbs as exceptionally helpful. This narrative might have led participants to exhibit more generosity because they felt good about being Serbian or because they were trying to follow the exemplar of Serbs as helpful toward refugees (Han et al. Reference Han, Workman, May, Scholtens, Dawson, Glenn and Meindl2022). I am unable to examine the effects of this rhetoric directly. Instead, I use the answers to “What does it mean to be Serbian?” to explore this possibility by categorizing them into positive, neutral, and negative attachments to being Serbian. I find no differences in generosity. Serbs who express a positive attachment to their identity are no more and no less altruistic to refugees compared to Serbs who do not (Appendix Table 36). Thus, neither potential explanation for the generosity associated with the ingroup prime bears out.
The literature identifies two other factors that may affect altruism toward refugees, writ large. The first is political ideology (Szabo et al. Reference Szabó, Vollhardt and Mészáros2020, Vollhardt et al. Reference Vollhardt, Cohrs, Szabó, Winiewski, Twali, Hadjiandreou and McNeill2021, but see Vollhard et al Reference Vollhardt, Nair and Tropp2016), for which I find no support in this sample. The second is threat perception (Vollhard et al. Reference Vollhardt, Cohrs, Szabó, Winiewski, Twali, Hadjiandreou and McNeill2021). I find that displaced Serbs who would feel unsafe if Serbia accepted more Syrian and Afghan refugees send significantly less to the refugee family in need. This is the case for both treatment groups (both factors in Appendix Tables 20-21).
The exploratory analyses find several potential explanations unhelpful. They also highlight two promising avenues for future research, which should be tackled systematically: closeness and perceived threat.
Conclusion
While the findings in this study are unexpected, they are not unique. A recent multi-country study of willingness to help a stranger during the COVID-19 pandemic shows that compared to participants in Germany, England, Ireland, and Sweden, participants in Serbia defy the typical expectation of ingroup favoritism based on an ethno-religious dimension (Carol et al. Reference Carol, David, Malešević and Uzelac2024). Religious Serbian participants were no less likely to help a fictitious Muslim neighbor than non-religious Serbs. In fact, Serbs were more willing to help an outgroup member than participants from any other country included in the study. Together with the findings in this study, this suggests that the typical relationship between ingroup favoritism and altruism may not hold in Serbia. As the case of Serbia is relatively underexplored (Carol et al. Reference Carol, David, Malešević and Uzelac2024), scholars may consider further work in order to gain a better understanding of how Serbian identity shapes altruism.
This study has several limitations. First, the lack of a control condition precludes a baseline to which altruism levels associated with the ingroup and superordinate primes could be compared. Second, due to the convenience sample of self-selected participants, I am unable to generalize these findings to the general population of displaced Serbs. Scholars who study shared hardship often contend with this limitation, as recruiting hard-to-reach populations for participation in research of a sensitive nature requires snowball or convenience sampling (Szabo et al. Reference Szabó, Vollhardt and Mészáros2020, Vollhardt et al. Reference Vollhardt, Cohrs, Szabó, Winiewski, Twali, Hadjiandreou and McNeill2021). Future research might therefore consider fielding similar surveys using larger, representative samples, while establishing clear baselines for comparisons.
Altogether, these results carry some implications for displacement interventions. Perhaps a successful intervention might rely on differences between refugees and hosts, in addition to naming commonalities. People may not necessarily reject a superordinate identity altogether; they might identify with a shared, superordinate identity while preferring an ingroup identity to be acknowledged as distinct (Hornsey and Hogg Reference Hornsey and Hogg2000). Such an intervention would be unlikely to dissuade people who already feel closeness toward the new refugees from altruism, but it might encourage those who feel distinct to be generous. Future work might also consider interventions that aim to reduce perceptions of threat or interventions that state the need for cross-group solidarity explicitly.
The need to understand why humanitarian appeals work and why they fail is acute. As the global number of displaced people rises, so does the chance that members of newfound host societies will have experienced conflict displacement themselves (Fisk Reference Fisk2018). Knowing how to narrate a potential commonality between a host society and a refugee population not only increases the chance that refugees will find a safe haven but likely also affects how displaced members of the host population process their own trauma. In the short term, understanding the interaction between superordinate identity and altruism can help policymakers reduce animosity; in the long term, it can aid interethnic integration and coalition-building efforts.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/rep.2025.4
Data availability statement
The data that support the findings of this study are available on request from the author. The data are not publicly available due to the sensitive nature of this research. Participants agreed to the data being made available to approved researchers.
Acknowledgements
I would like to thank W. Nicholson Price II, Mackenzie Israel-Trummel, Allyson Shortle, and the participants of the Comparative Politics Workshop at Michigan State University for invaluable feedback on the manuscript. I would like to thank Lucija Dežan for research assistance in coding open-ended responses.
Funding statement
This work was supported by the University of Oklahoma College of Arts & Sciences Funding Assistance Program, which played no role in the design, execution, analysis and interpretation of data, or writing of the study.
Ethical standards
This project was approved by the University of Oklahoma IRB (IRB: 7020) and pre-registered at EGAP (ID: 20160809AB).
Open access
