The Political Impact of Displacement: Wartime IDPs, Religiosity, and Post-War Politics in Bosnia

Abstract

Following armed conflict, why do some members of ethno-religious groups vote for political parties that use religious appeals while others do not? I argue that internal displacement shapes the relationship between conflict and post-war political outcomes. Specifically, individuals who become internally displaced during armed conflict will use their religious faith to cope with the trauma of displacement, thereby strengthening their religiosity. This heightened religiosity then leads them to prefer religiously oriented parties after conflict. Analyzing survey data from Bosnian Muslims, I show that internally displaced respondents were more likely to vote for the religious nationalist Party of Democratic Action nearly a decade after conflict. Employing matching analysis, I then verify that these internally displaced persons became more religious than other respondents compared to before the war. My findings therefore provide evidence that trauma and religiosity combine to shape post-war voting behavior for members of ethno-religious groups.

1. INTRODUCTION

Armed conflicts centered on an ethnic or religious cleavage are likely to prime the ethnic or religious identities of members of the warring groups. If such a war is followed by a democratic regime, citizens should be primed to vote for political parties making appeals on the basis of ethnicity or religion, particularly because this context may heighten ethnic group members’ desire for expressive public reinforcement of their ethnic identity (Ferree 2006) and/or their need to use ethnicity as a shortcut to gain information about politicians’ intentions in a new, information-poor democracy (Chandra 2007). However, not all citizens support ethno-religious political parties in this situation. In Bosnia and Herzegovina, for example, though post-war elections following the 1995 Dayton Peace Agreement have often functioned as ethnic censuses (Touquet 2011), some Bosnian Serbs, Croats, and Muslims have supported their respective ethno-religious nationalist parties more consistently than others.

Why is this the case? More specifically, following an ethno-religious armed conflict, why do some members of ethno-religious groups vote for political parties that use religious appeals while other members do not? Previous findings in the literature concerning armed conflict's impact on post-war electoral politics do not address this question directly, focusing instead on how experiencing wartime violence heightens post-war political participation (Blattman 2009) or how wartime victimization at the hands of specific actors shapes post-war political attitudes (Balcells 2012).

In this article, I present a new relationship between wartime experiences and post-war political outcomes. I argue that ethno-religious group members who become internally displaced during armed conflict are more likely to vote for religiously oriented political parties after the conflict ends. The argument is premised on the notion that internal displacement is a traumatic experience. As a result, internally displaced persons (IDPs) will turn to religion as a coping mechanism with which to deal with this trauma. In doing so, their religiosity will become stronger. Consequently, compared to non-IDPs, they will demonstrate greater post-war support for political parties which promote policies tied to religious customs and beliefs.

In order to hold contextual variables constant, I focus on the Bosnian Muslims as my ethno-religious group of interest. Using data from a survey carried out by the Norwegian Research Council in 2003–2004, I show that Bosnian Muslim respondents who were internally displaced during the 1990s Bosnian war were more likely to vote for the religious nationalist Party of Democratic Action (SDA) several years after the conflict. I also provide evidence for the mechanisms within the argument by demonstrating that internally displaced Bosnian Muslim respondents attended religious services more frequently before the war and were more likely to report becoming more religious since the war compared to non-displaced Bosnian Muslim respondents. In doing so, I employ a matching analysis to verify the latter mechanism.

I also test several alternative explanations for SDA vote choice, showing that the experience of different types of wartime violence, mistrust of other ethnic groups, lower socio-economic status, and pre-war rural residence are not significantly related to voting for the SDA after accounting for wartime IDP status. As a robustness check, I use an original set of municipal level data from Bosnia to provide an additional test of the relationship between displacement and post-war political behavior. I find that municipalities containing a much higher proportion of Bosnian Muslim IDPs after the war relative to the 1991 municipal population tended to give much greater electoral support to the SDA in the 2004 municipal elections compared the last pre-war election in 1990.

The results of this article make four important contributions to the research literature concerning the relationship between armed conflict and political attitudes and behavior.

First, I present a new empirical relationship between wartime experiences and post-war political outcomes. A number of recent studies have examined the impact of wartime experiences on individuals and groups, with a focus on economic and political behavior. Specifically, the studies have shown that individuals experiencing wartime violence are more altruistic and risk-seeking (Voors et al. 2012) and more likely to vote and become involved in local community politics (Bellows and Miguel 2009; Blattman 2009). By showing that experiencing internal displacement during war increases the likelihood of voting for a specific type of political party, my results suggest a strong link between wartime experiences and post-war political attitudes and behavior outside of victimization by wartime violence.

Second, this article fills a gap in the existing literature by analyzing how wartime experiences shape political attitudes via trauma and religion. Within the growing number of studies examining the link between wartime experiences and post-war political participation, very few have investigated how wartime trauma can impact post-war political attitudes. One important exception to this trend has been the work of Laia Balcells, who found that victims of violence during the Spanish Civil War remained opposed to the political identity of the perpetrators of the violence, even passing this attitude down across generations within their families (Balcells 2012). Another study found more post-war support for left-wing political parties in areas of Italy victimized by Nazi and Fascist violence during World War Two (Costalli and Ruggeri 2015), while an analysis of IDF soldiers found that combat experience led them to prefer military solutions to the Israeli-Palestinian conflict and spurred them to vote for more hawkish political parties (Grossman, Manekin, and Miodownik 2015). More broadly, it also appears that higher levels of violent victimization in a society are associated with lower perceived legitimacy of political institutions and lower levels of political trust (Grosjean 2014).

Though insightful, none of these studies have examined how wartime trauma may shape religious beliefs and how religious views shaped by war may subsequently impact post-war political attitudes and behavior. This is a potentially significant omission, since experimental research indicates that individuals whose religious identity is primed instead of their ethnic identity give higher priority to social issues and moral values in politics and exhibit a preference for candidates who focus on moral issues (McCauley 2014). Furthermore, given the long-term impact of violent trauma on political ideology and political behavior in post-civil war Spain and post-World War Two Italy, heightened religiosity and religious identity resulting from wartime trauma may also have a significant effect on political preferences in post-war democratic politics.

Third, in conjunction with this contribution, one of the mechanisms outlined here bridges academic disciplines by incorporating research on the psychology of trauma and the psychology of religion into the study of armed conflict. By showing the value of this inter-disciplinary approach, the article's findings emphasize the need for a new sub-area of study within political science research on war.

Fourth, this article is one of the first to investigate the political attitudes and behaviors of IDPs. Though it is understudied in social science, internal displacement is a public policy issue that is growing in magnitude across the world, as the total number of IDPs worldwide rose from 23.3 million to 38 million between 2005 and 2014, an increase of 63% (Internal Displacement Monitoring Centre 2015). Consequently, it is important to examine whether the experiences, beliefs, and attitudes of IDP populations have a political impact. This study thus comprises a pioneering effort to trace the form and degree of political participation and agency among IDPs in post-war democratic politics.

The article is structured as follows. Section 2 elaborates the theoretical argument. Section 3 presents background information on the Bosnian conflict, the Bosnian Muslims, and the SDA political party. Section 4 presents the research design. Section 5 presents and discusses the results. Section 6 concludes.

2. THEORY: INTERNAL DISPLACEMENT, RELIGIOSITY, AND POST-WAR POLITICAL OUTCOMES

My argument linking internal displacement to post-war support for political parties making religious appeals consists of three parts. First, conflict-induced internal displacement produces psychological trauma. Second, this trauma heightens the religiosity of IDPs, as they turn to their religious faith to cope with their traumatic experiences. Third, IDPs’ heightened religiosity leads them to support religiously oriented political parties following the end of armed conflict. Below, I define some of my key terms before elaborating my argument.

By “ethno-religious” groups, I refer to groups whose ethnic and religious identity markers overlap, such that individual group members can choose to prioritize either the ethnic or religious component of their group's identity (Ruane and Todd 2010). One example of this type of group identity is Sikhism, where the adjective “Sikh” refers to both a religion and an ethnicity. Other examples include Armenians and Uighers. Prioritizing the ethnic component of the identity signifies that individual members identify with the group's customs and traditions in a secular fashion. This may include observance of holidays and rituals derived from religious sources but separated from any faith in the supernatural aspects of religion or regular attendance at religious services.

On the other hand, prioritizing the religious component of group identity signifies that individual members believe in supernatural aspects of the group's religion, adhere to scriptural tenets and guidelines, and attend religious services regularly. The latter are the set of beliefs and practices that I refer to collectively as “religious identity” in this article. Similarly, when I refer to “religiosity,” I build on prior work in referring to attitudes and actions that principally include attendance at religious services, performance of religious rituals (prayer in particular), and spiritual beliefs and faith (Ringdal and Ringdal 2010).

For “internally displaced persons,” I incorporate the United Nations Office for the Coordination of Humanitarian Affairs’ definition of IDPs, which refers to people who are forced to flee their homes and who have not crossed an internationally recognized state border (United Nations Office for the Coordination of Humanitarian Affairs 2004). However, I limit my use of the concept to individuals displaced by armed conflict. Despite the many public policy challenges posed by a large global population of IDPs, political science research has thus far not devoted much attention to this issue. The few studies that have examined the political role and impact of IDPs have produced mixed results concerning their political agency. Thus, individuals displaced by election violence in Kenya were more likely to reject the use of violence as an acceptable form of political expression (Linke 2013), while displaced persons in Aceh, Indonesia were more likely to vote in post-conflict elections than non-displaced persons (Shewfelt 2008). On the other hand, while IDPs in Georgia successfully mobilized for collective action in both formal and informal organizations, their success was largely predicated on assistance from international actors who advocated on their behalf (Røkke 2012).

In terms of the specific relationship between IDPs and religion, prior theoretical and empirical work suggests that IDPs tend to originate from countries and regions with fewer economic opportunities and that, all else being equal, people with lower incomes are more likely to be displaced (Adhikari 2013). IDPs are also likely to be poorer than refugees, since leaving a country during conflict is costlier and requires more resources than being displaced within a country, to the point that even reaching a neighboring country is becoming more expensive (Mundt and Ferris 2008, 3). In turn, research on the sociology of religion indicates that poorer individuals tend to be more religious (Norris and Inglehart 2011). Taken together, these findings suggest that IDPs will have access to an established religious faith and set of religious practices during armed conflict and will be relatively more religious than non-IDPs prior to conflict.

In turn, given that internal displacement produces significant material and emotional loss, along with a heightened risk of experiencing violence, wartime IDPs are more likely to suffer trauma relative to non-IDPs. Mental health research thus indicates that conflict-induced IDPs have much higher rates of post-traumatic stress disorder, anxiety, and depression than the general population (Thapa and Hauff 2005). Studies in various countries, including Bosnia, also show that wartime IDPs experience significantly more traumatic events relative to refugees and the non-displaced (Mels et al. 2010; Powell et al. 2003; Roberts et al. 2009; Schmidt, Kravic, and Ehlert 2008).

Wartime IDPs’ greater susceptibility to trauma may consequently lead them to become more religious during armed conflict. The reason for this is that in a context of profound loss, as is the case during internal displacement, individuals may turn to mental schemas that promise to alleviate trauma. A prime candidate for this kind of schema would be a readily available religious tradition, which can offer therapy by explaining the meaning and purpose behind seemingly purposeless events while also giving individuals a renewed sense of control over their lives. In addition, continued observance of religious rituals can substitute for lost routine and structure in other areas of life and provide a social bond to withstand the trauma of displacement. Using religion as a coping mechanism to deal with trauma is in turn likely to strengthen individuals’ religiosity.

After experiencing trauma, individuals often wonder whether their suffering had some larger cause, seeking to reconstruct a benign and meaningful view of the world. In principle, this goal aligns with religion's promise to answer questions about the meaning and purpose of life (Peres et al. 2007). In addition, trauma survivors can place a traumatic event within a religious narrative framework, allowing them to process the event more quickly and to perceive a greater meaning or purpose to their suffering (McIntosh 1995). Religion can also provide trauma victims with a supportive social network of people sharing similar values and beliefs (Boehnlein 2006).

Moreover, many studies indicate that religion and religiosity are effective resources for alleviating trauma. Thus, traumatized individuals who adopt a spiritual outlook on their religious faith and frame their religious beliefs in terms of a quest for meaning and significance tend to experience positive growth following trauma (Harris et al. 2008; Calhoun et al. 2000). Furthermore, individuals who use their religiosity to cope with trauma retain a perception of the world as benevolent and meaningful, which helps shield them from some of the psychological consequences of trauma (Zukerman and Korn 2014). From a historical perspective, Rodney Stark has also argued that Christianity spread so quickly and widely in the ancient world partly because of its effectiveness at providing meaning, coherence, and purpose in the wake of vast crises and tragedy (Stark 1996).

By turning to religion in order to cope with their condition, traumatized individuals are likely to enhance their religiosity, particularly if they use religion to understand and make sense of the traumatic event (Tedeschi and Calhoun 1996). Thus, as they use religion to integrate and stabilize their view of the world and to confront death, tragedy, and injustice, trauma survivors’ religious faith and practice will grow stronger (Shaw, Joseph, and Linley 2005). Lab experiments have also demonstrated that being more aware of death, seeing the world as random and uncertain, and perceiving a loss of control all heighten belief in God (Norenzayan and Gervais 2013). Once a religious identity is strengthened, it may also lead people to go to great efforts to defend it, particularly if this identity is tied to one's ego and protection from trauma (Hogg, Adelman, and Blagg 2010).

In terms of historical contexts, studies of Holocaust survivors have shown them to be more religious (Carmil and Breznitz 1991). Furthermore, studies in the United States revealed much stronger short-term religious identity among college students in the aftermath of September 11th, 2001 (Ysseldyk, Matheson, and Anisman 2010). Victims of political violence in the Peruvian highlands were also more likely to convert to evangelical Protestantism during and after the conflict there (Gamarra 2000).

Additional research suggests that IDPs should be particularly likely to turn to religion and have their religiosity increase during armed conflict. Specifically, people with access to religion tend to use religious coping following trauma (ter Kuile and Ehring 2014). Accordingly, even though some research suggests that trauma survivors who hold very strong religious beliefs prior to their traumatic experience may become less religious, traumatized individuals who use religion to cope with their trauma are generally likely to increase their religious activities and beliefs (ter Kuile and Ehring 2014). In turn, trauma victims currently living in a religious environment are highly likely to use religion to cope with their trauma (ter Kuile and Ehring 2014).

Other research indicates that well-developed religious beliefs are also resilient to trauma and become even stronger in the wake of traumatic events (e.g., Overcash et al. 1996). Reconciling these findings, it seems that individuals who experience post-traumatic stress disorder from a single or isolated traumatic event tend to develop weaker religious beliefs, while those experiencing multiple instances of traumatic victimization independent of post-traumatic stress disorder tend to strengthen their religious beliefs (Falsetti, Resick, and Davis 2003). In line with these findings, conflict-induced IDPs’ religiosity may grow stronger because of their greater risk of experiencing multiple traumatic events during war. Depending on the nature of the conflict, an individual IDP may also be surrounded by more devout religious followers, thus placing them in a religious environment and making it more likely that they will turn to religion as a coping mechanism, which will in turn increase their religiosity.

Therefore, with ready access to religious faith and practice, I posit that IDPs will use religion to cope with the trauma of displacement and to find emotional relief and a sense of renewed meaning, order, and control in their lives. As a result, IDPs will come to identify with their religious faith even more fervently as they practice it and reaffirm their beliefs, thus becoming more religious than they were prior to the conflict.

Subsequently, any strengthened religious identity among ethno-religious group members may also have significant repercussions for post-war democracy. In this case, individuals who become more religious as a result of armed conflict may be more likely to identify with and support political leaders and parties whose policies match their religious values. In particular, there may be a strong link between these individuals’ attitudes and the act of voting for religiously oriented political parties.

This link starts from the assumption that individuals who have become more religious are more likely to take an active role in religious institutions. These institutions, in turn, are likely to possess the well-developed social capital (i.e., networks, activists, leaders, monetary resources, infrastructure) needed to register individual members to vote, take them to or host political campaign events, and help them turn out to vote. Assuming that most of the leaders of these religious institutions support more religiously-oriented political parties, individual members of these organizations will not only turn out to vote in relatively greater numbers than individuals who do not belong to the organizations, but will also vote for these types of political parties.

The link between greater religiosity and support for religiously oriented parties may be especially likely in a political environment like post-war Bosnia. Since 1995, a few ethno-religious political parties have consistently won elections and served in government, meaning that the country has seen low to moderate party fragmentation (Hulsey 2010). At the same time, the political system has produced very few new parties, only moderate electoral volatility, and high ethnic polarization (Džankić 2015). This dominance and institutionalization of ethnic parties stems from the Dayton Peace Agreement's constitutional framework for Bosnia, which created a decentralized, consociational system of government based on ethnic power-sharing.

Given this institutional design and the overlap of ethnicity with religion among Bosnian ethnic groups, Bosnian voters face significant pressure to vote for their respective ethno-religious party. Bosnia also lacks the sort of informally institutionalized cross-ethnic ties that have been shown to lower the salience of ethnicity in other countries (Dunning and Harrison 2010). These institutional features arguably make it more likely that more religious Bosnian Muslims will develop a political connection with the SDA as their greater religiosity prompts them to become more involved with religious institutions. The subsequent hypothesis is:

Members of ethno-religious groups who were internally displaced as a result of armed conflict are more likely to vote for religiously oriented political parties following conflict than those who were not internally displaced.

Alternative explanations for post-war vote choice stem from the experience of wartime violence, mistrust of other ethnic groups, lower socio-economic status, and pre-war voting patterns. In terms of violence, individuals who are victimized by intense violence during war experience a context where there does not appear to be a clear order or structure to victimization, leaving them helpless in trying to learn how to avoid violence. Thus, this context is also likely to produce trauma, potentially spurring those experiencing this kind of violence to strengthen their religiosity. The first rival hypothesis is thus:

R1: Members of ethno-religious groups who experience high levels of intense violence during armed conflict are more likely to vote for religiously oriented political parties following conflict than those who do not experience this type of violence.

On the other hand, following an ethnic conflict, it is possible that individual members of an ethno-religious group may have very low levels of trust regarding other ethnic groups in the polity. This may be especially true for IDPs, given that their displacement likely resulted from the actions of members of other ethnic groups. In the presence of lingering feelings of fear and mistrust of other ethnic groups, individual ethno-religious group members may feel more secure by supporting politicians who represent their ethno-religious group. They may also find rhetoric and policies promoting religious nationalism to be comforting and reassuring signals that these politicians will follow through on promises of benefits and protection from future depredation and attacks by other ethnic groups. The second rival hypothesis is thus:

R2: Members of ethno-religious groups who greatly mistrust other ethnic groups in the polity are more likely to vote for religiously oriented political parties following conflict than those who do not greatly mistrust other ethnic groups.

Conversely, since IDPs tend to have few economic resources after armed conflict, they may support political parties which benefit individuals with low socioeconomic status. Thus, if religiously oriented political parties promote policies that benefit lower income voters, then IDPs’ support for these types of parties may be a result of their poverty and relative economic deprivation, not heightened religiosity. Thus:

R3: Members of ethno-religious groups from lower socio-economic backgrounds are more likely to vote for religiously oriented political parties following conflict than those from wealthier socioeconomic backgrounds.

Finally, individuals who become internally displaced during armed conflict may have been more likely to vote for a religiously oriented political party before conflict, such that any tendency to vote for such a party after conflict is a reflection of pre-war attitudes and behavior instead of greater religiosity. This pattern may be particularly applicable to individuals who grow up in rural areas, as they tend to be more religious and give more support to politicians who espouse conservative and traditional ethnic and religious values. The alternative hypothesis is:

R4: Members of ethno-religious groups who grew up in rural areas are more likely to vote for religiously oriented political parties following conflict than those who did not grow up in rural areas.

3. CASE BACKGROUND: BOSNIA, THE BOSNIAN MUSLIMS, AND THE SDA

The empirical tests of my theory focus on the Bosnian Muslim ethno-religious group during and after the 1990s war in Bosnia-Herzegovina. Part of communist Yugoslavia from 1945 to 1990, the official identity of Bosnia and most Bosnian Muslims prior to the war was civic and secular (Toal and Dahlman 2011). At the same time, however, religious identity was resurfacing, as evidenced by increasing religious attendance (Ramet 2006), the building and renovation of mosques (Ramet 2002), and clerics’ quest for a more active public leadership role (Friedman 1996). The war in Bosnia, which began on April 6th, 1992 and did not officially end until the signing of the Dayton agreement on December 14th, 1995, saw 100,000 people killed, of which over 64,000 were Bosnian Muslims (Toal and Dahlman 2011). A total of 1.2 million Bosnian Muslims also fled their homes, of which roughly 300,000 were IDPs. Thus, approximately 15% of the pre-war Bosnian Muslim population became internally displaced as a result of the conflict (Kukić 2001).

Two years before the war, the SDA political party had formed as a religious nationalist party representing the Bosnian Muslims. During the war, the party's top leaders were also the principal leaders of the internationally-recognized Bosnian government. In its outlook and many of its policies, the SDA appealed to Bosnian Muslims for support while promoting a more religious identity for the group. However, its attempts to influence and gain allegiance from all Bosnian Muslims did not result in universal support from members of this group during the war (Maček 2009). In addition, not all Bosnian Muslims voted for it after the war ended, meaning that ethnic ties alone are insufficient to explain variation in the party's performance in post-war elections. In the immediate aftermath of the war, the SDA also attempted to build support from the Bosnian Muslim IDP population, which included promises of significant improvements in their circumstances. They did not deliver on these promises, however, to the point that IDPs realized they were lying (Halilovich 2013). Since the survey data used in this article ask which party respondents voted for in the 2002 elections, evidence of continued strong electoral support for the SDA among Bosnian Muslim IDPs seven years after the war would suggest other sources of affinity besides empty promises of material gain.

4. RESEARCH DESIGN

To test my hypothesis, I analyzed individual-level survey data from Bosnian Muslim respondents following the Bosnian conflict. The data come from the 2003–2004 South East European Social Survey Project, which was funded by the Research Council of Norway (Simkus 2013). Carried out in six different Balkan countries, the survey contained 21,916 respondents in total, including 6,809 from Bosnia, of which 2,559 were Bosnian Muslim. Accordingly, I only analyzed the Bosnian Muslim respondents for purposes of this article.

The survey asked respondents to answer a wide range of questions, including items relating to voting behavior, religious practices and religiosity, and wartime experiences and displacement status. One item in particular asked respondents which political party they voted for in the last election prior to the survey. I recoded this item as a dummy variable, so that my main dependent variable, SDA Vote, takes the value of 1 if the respondent voted for the SDA in the 2002 elections. The primary independent variable, Wartime IDP, takes the value of 1 if the respondent reported being internally displaced during the war and 0 otherwise.

Additional survey items allowed me to code several control variables. For R1, respondents were asked whether they personally saw or witnessed shooting, artillery fire, or airstrikes directed at them, their family, or their community during the war. I thus coded the variable Violence as a dichotomous indicator taking the value of 1 if respondents answered that they had witnessed any one of these three types of violence and 0 if they had not witnessed any of them. Another question asked respondents whether they agreed or disagreed with the statement that “Among nations [i.e., ethnic groups] it is possible to create cooperation, but not full trust.” Ethnic Mistrust is thus a dichotomous indicator taking the value of 1 if respondents either agreed or strongly agreed with this statement, enabling me to test R2.

To test the potential role of socioeconomic status outlined in R3, I included Income, reflecting a respondent's total net income per month, as well as Education, coded in ascending order of the highest degree the respondent obtained. In the case of the Bosnian Muslims, individuals with less education were more likely to have voted for the SDA before the conflict (Bougarel 1997). Moreover, per R3, respondents with less income, which includes many IDPs, may vote for the SDA on class lines.

To test R4, I included the variable Rural Upbringing, taking the value of 1 if respondents said they lived in a rural area when they were fourteen and 0 otherwise. This measure thus accounts for rural residents’ greater likelihood of being internally displaced during war, and, in this case, rural Bosnian Muslim residents’ greater support for the SDA. Additional variables included Gender, taking the value of 2 if respondents were female and 1 if they were male, and Age, with the variable coded according to each respondent's age in years.

5. ANALYSIS

5.1. SDA Vote Choice

To test my hypothesis against the rival explanations outlined above, I used a pair of probit regressions. Model 1 of Table 1 indicates that there is a statistically significant and positive bivariate relationship between having been a wartime IDP and voting for the SDA in 2002. Model 2 presents a multivariate probit model which includes the full set of control variables representing rival explanations for voting for the SDA. Holding these other variables constant, wartime IDP status remains positive and statistically significant at 95% confidence.

Table 1. SDA vote choice, Bosnian Muslim respondents in the 2002 Bosnian elections

Standard errors in parentheses. *p < 0.05, **p < 0.01, ***p < 0.001.

The signs of the coefficients for the control variables are in the theoretically expected direction, as experiencing wartime violence, expressing mistrust of other ethnic groups, and growing up in a rural area are all positively related to voting for the SDA, while education and income are negatively related. None of these variables are statistically significant in the presence of IDP status, however. In contrast, age is negatively related to SDA vote choice at the 99% level of statistical significance. This result suggests that older respondents, having grown up under an officially atheist communist regime, were not as willing to support a political party oriented toward religious and ethnic nationalism.

Overall, the models in Table 1 lend greater support to my hypothesis compared to the rival hypotheses. Having experienced wartime internal displacement is a more powerful predictor that a Bosnian Muslim respondent voted for the SDA seven years after the war than having experienced violence, lacking trust in other ethnic groups, having lower socioeconomic status, or growing up in an area in which most people voted for the SDA before the war.

Substantively, I used the results from Model 2 of Table 1 to calculate the tangible impact that having been a wartime IDP had on the predicted probability that Bosnian Muslim respondents voted for the SDA, holding the control variables at their observed values (Hanmer and Kalkan 2013). I computed both a single calculation of the predicted probability, as well as 1,000 simulations of the calculation, which provided the mean effect and 95% confidence interval for the predicted probability. The results indicate that experiencing internal displacement during the war corresponded, on average, with a twenty percentage point increase in the likelihood of voting for the SDA seven years after the war. This mean effect is also statistically significant at 95% confidence.

In terms of the trauma and religiosity mechanism linking the relationship between IDP status and SDA vote choice, one of the survey questions asked respondents how often they had attended religious services in 1990, while another question asked them whether their religiosity had changed since 1990. Table 2 presents a cross-tab of the relationship between pre-war religious service attendance and wartime IDP status.

Table 2. IDP status vs. pre-war religious attendance, Bosnian Muslims

N = 484.

The cross-tab shows that Bosnian Muslims who became IDPs during the war also went to religious services more frequently before the conflict. In particular, relative to non-IDPs, IDPs were much more likely to have attended religious services at least once a week and, especially, at least once a day. This result confirms the premise that Bosnian Muslim IDPs could access a developed set of religious beliefs and practices in the face of trauma.

Table 3 presents a cross-tab of the relationship between wartime IDP status and changes in respondents’ religiosity compared to before the war.

Table 3. IDP status vs. change in religiosity, Bosnian Muslims

N = 518.

According to these results, wartime IDPs were more likely to have become more religious, while non-IDPs were likely to have become less or much less religious. In the online appendix, I also show that this relationship between wartime internal displacement and greater religiosity holds true for both rural and non-rural Bosnian Muslim respondents. Thus, rural IDPs were much more likely to say they had become more religious compared to rural non-IDPs, while non-rural IDPs were also much more likely to say they had become both more religious and much more religious relative to non-rural non-IDPs.

These outcomes support the notion that the traumatic experience of wartime displacement led Bosnian Muslim IDPs to become more religious than they were prior to the war. Moreover, it appears that wartime IDP status boosted individuals’ religiosity regardless of whether they came from a more rural background before the war. In combination, these results provide strong support for my argument, whereby heightened religiosity resulting from the trauma of wartime internal displacement spurs IDPs to vote for religiously oriented parties in post-war elections.

5.2. Matching — Greater Religiosity and IDP Status

In order to verify that the religiosity of Bosnian Muslim IDPs had increased compared to before the war, I also analyzed this relationship with a matched sample of respondents to account for the fact that conflict-induced internal displacement is not randomly assigned. The dependent variable in this analysis takes the value of 1 if respondents reported being more or much more religious compared to before the war and 0 otherwise. Following suggested practice, the matched sample contained individuals who were highly similar on a set of key covariates, with the main difference being that some were internally displaced during the war while others were not (Ho et al. 2007).

I generated the matched sample via propensity score matching, using the gender, age, education, and rural upbringing variables as my key covariates. I matched on these variables because previous research indicates that women (Miller and Hoffman 1995), people of a certain age (Argue, Johnson, and White 1999), and individuals with less education (Stark 1963) are more religious, and that rural residents adhere to more traditional religious beliefs (Nelsen, Yokley, and Madron 1971). In the context of the Bosnian conflict, these groups were also more likely to have been displaced during the war. Moreover, since the survey was administered between November of 2003 and March of 2004 and the rural background question asked about rural residence at age 14, capturing respondents with rural upbringing before the war started in April of 1992 required including only those who were born, at the latest, in March of 1978. Accordingly, I restricted the sample to respondents who were at least 26 years old.

Matching on these covariates, Table 4 shows that respondents who received the “treatment” of internal displacement during the war were more likely to have become more religious than those who were not internally displaced. Table 4 also indicates that the matched sample is balanced, as the p-values on all t-tests are not statistically significant, meaning that the null hypothesis that the covariates are balanced cannot be rejected. As a robustness check on these findings, I also employed coarsened exact matching to test the relationship. The results, reported in the online appendix, corroborate IDPs’ greater tendency to have become more religious compared with non-IDPs. The matching analysis thereby confirms the religiosity mechanism linking wartime IDP status with post-war voting support for the SDA.

Table 4. Matching results and balance statistics — Greater religiosity vs. IDP status

N = 405.

5.3. Robustness Check — Bosnian Municipal Data

As an additional means of validating my hypothesis, I collected and assembled an original dataset of municipal-level variables in Bosnia. Prior to the war, the data include each municipality's total population, income, and rural population, as well as the number of municipal council seats the SDA won in the 1990 election. I also added data regarding the number of people killed in each municipality during the war. Post-war, the data include the number of IDPs registered in each municipality as of 2005, categorized by ethnicity, as well as the number of votes cast for the SDA in the 2004 municipal elections.

The pre-war demographic variables are from the 1991–1992 Bosnian census conducted and maintained by the Bosnian State Statistical Office and Bosnian Federal Statistical Office (Federalni Zavod Za Statistiku 1998; Republika Bosna I Hercegovina Državni Zavod Za Statistiku 1994). The 1990 election results were published by Suad Arnautović (1996), a professor at the University of Sarajevo. Stefano Costalli and Francesco Moro (2012) made wartime casualty data available in a publication examining the severity of violence in Bosnian municipalities during the war. I located all of the IDP data in a report published by the Bosnian Ministry for Human Rights and Refugees in 2005 (Nenadić et al. 2005). Results for all post-war Bosnian elections are available on the website of the Bosnian Election Commission (Centralna Izborna Komisija Bosne I Hercegovine 2016).

The dependent variable in the municipal-level analysis is SDA Support Increase, coded as the difference between the proportion of the vote won by the SDA in each municipality in the 2004 local elections versus the proportion of municipal council seats it won in the 1990 election. Although the proportion of municipal council seats won in 1990 is not an ideal measure for comparison, it represents the best measure given the limitations of available data. An additional caveat is that post-war elections included votes from both inside and outside the municipality owing to the peace agreement's stipulation that displaced residents and refugees could vote in their pre-war municipality. I coded the proportion of votes cast for the SDA from both inside and outside the municipality, as well as the proportion of total votes. However, I used only the latter measure in constructing my dependent variable.

The principal independent variable in this analysis is 2005 Bosnian Muslim IDPs, coded as the proportion of the number of Bosnian Muslim IDPs registered as living in a municipality in 2005 relative to the total 1991 municipal population. This variable is thus meant to capture the relative influx of internally displaced Bosnian Muslims on a municipal level. Control variables include 1991 Per Capita Income and 1991 Percent Rural Population, also measured at the municipal level. These variables were the best proxies available to measure post-war municipal income and rural residence because Bosnia did not publish any official census results between 1991 and 2016. My other control variable is Percent Total War Victims, which is the proportion of the number of people killed in a municipality during the war relative to its 1991 population. All three of these variables may have been related to the proportion of IDPs originating from a given municipality during the war, as well as the degree of municipal electoral support for the SDA in the 1990 elections. In addition, the post-war political settlement created new municipalities in Bosnia. Where applicable, I matched data on IDPs and election results to the appropriate pre-war municipality.

Table 5 presents regression results using the municipal data. This model demonstrates that, holding all else constant, a statistically significant relationship exists between relatively greater post-war municipal electoral support for the SDA nearly a decade after the war and the number of Bosnian Muslim IDPs residing in a municipality relative to its pre-war population.

Table 5. Municipal SDA electoral support, 1990–2004, vs. influx of Bosnian Muslim IDPs

Standard errors in parentheses. *p < 0.05, **p < 0.01, ***p < 0.001.

Substantively, the relationship between the influx of Bosnian Muslim IDPs and the increase in electoral support for the SDA at the municipal level is also significant. Holding the other variables at their observed values, municipalities with no Bosnian Muslim IDPs experienced a 5% net loss in voting support for the SDA in 2004 relative to 1990, while municipalities experiencing an influx of Bosnian Muslim IDPs equal to 5% of their pre-war population experienced a 5% net gain in support. This outcome demonstrates that the individual-level relationship between conflict-induced internal displacement and post-war political support for religious nationalist politicians may also have a collective impact.

6. CONCLUSION

The findings presented in this article suggest that internal displacement can shape the voting behavior of individual members of ethno-religious groups following armed conflict. The mechanism linking this relationship runs through heightened wartime religiosity. Specifically, internally displaced members of an ethno-religious group are especially likely to increase their religiosity as they utilize their pre-existing religious faith and practices to cope with the trauma of wartime displacement. This increased religiosity in turn leads them to support religiously oriented parties in post-war elections. In contrast, the experience of wartime violence, feelings of mistrust toward other ethnic groups, lower socio-economic status, and pre-war rural residence cannot explain individuals’ support for these types of parties outside the experience of internal displacement.

By showing a link between internal displacement, strengthened religiosity, and post-war political attitudes and behavior, my findings thus indicate the need to undertake a broader examination of how traumatic wartime experiences combined with shifts in religious attitudes can shape individuals’ identities and political preferences following the end of conflict. In addition, these results indicate that IDPs are a politically significant group who merit greater study in political science. Further study of this group may yield broader findings concerning the stability and health of post-war democracy and the types of policies it is likely to produce. Additional surveys and interviews could also help determine which specific wartime experiences stemming from internal displacement are most likely to produce trauma and heightened religiosity. This approach would require more studies focusing on just one or a few ethnic groups, since IDP data at the large N cross-national level is sparse.

Regarding generalizability, this article's findings are limited by the institutional context of political competition in post-war Bosnia. Thus, post-war democracies that avoid instituting ethnic consociationalism may provide religious voters with a greater number of political parties to choose from in elections. Countries with less overlap between religion and ethnicity may also yield different results for the voting behavior of religious individuals, particularly if citizens have access to ties and connections that cut across ethnic and religious lines. Given these limitations, the article's findings should be tested in a broader set of post-war democracies with similar institutions and party structures.

In terms of policy, this article's findings suggest that ethno-religious groups fighting armed conflicts will develop a more religious collective identity if the group suffers high levels of internal displacement. If so, such an identity shift may exacerbate polarization along religious lines within and between ethno-religious groups. Polarization on these lines may then hinder the effective functioning and consolidation of post-war democracy if a greater range of opposing views have to be included and reconciled within the new regime. This scenario suggests that, following armed conflict, foreign states and international organizations engaged in reconstruction and peacebuilding should channel high levels of aid and resources to IDPs, particularly with respect to mental health services. Providing IDPs with non-religious resources for coping with trauma may moderate or diminish their connection to religion in a context where religion has been tied to polarization and violence, thus creating a smoother path to post-war reconciliation and unity.