Youthful victimization in post-war Bosnia and Herzegovina: A multi-level analysis of a national sample

Abstract

Intra-national conflicts with racial or ethnic elements can complicate post-war reconciliation. From 1992 to 1995, much of the former Yugoslavia, a nation largely drawn from three distinct ethnic groups, was embroiled in such a conflict. After the signing of the Dayton Peace Accord, it was feared that schools would become a surrogate battlefield for school-aged children within the newly created nation of Bosnia and Herzegovina (BiH). Group threat theory and the imbalance of power thesis provide differing views on such conflicts. Group threat theory posits that as a population – in this case a school – approaches maximum ethnic diversity, the residents – in this case the students – will feel increasingly threatened, resulting in higher cross-group victimizations. The imbalance of power thesis suggests that a group’s decision to victimize another group depends on the relative lack of ethnic diversity: the extent to which one ethnic group dominates a school, the likelihood of victimization of any smaller groups increases. We explore which of these two theories best explains victimization levels within a sample of 2003 school-aged BiH adolescents born in areas dominated by Muslim Bosnians, Eastern Orthodox Serbians, or Roman Catholic Croatians. We find that there is an ethnic component to victimizations: students born in Serbia face higher levels of victimization than do their Bosnian-born counterparts under conditions that fit better with group threat theory than the imbalance of power thesis. We speculate about the significance of these findings for national ethnic harmony in BiH.

Keywords

Victimization adolescents multi-level analysis post-conflict Bosnia-Herzegovina

Introduction

Schools are social microcosms of their respective communities (Brunson and Miller, 2009; Morrison, 2001). They also act as social control mechanisms, regulating student behavior through rule teaching and enforcement (Gottfredson, 2001). The interdependence between communities and their schools is well documented across a range of topics, including economic conditions, crime, politics, family life, and social support via social network ties (Benabou, 1996; Brunson and Miller, 2009; Christenson and Reschly, 2010; Hubbard et al., 2006; Wood, 2008). Schools can serve as physical safe havens and provide students with a sense of normalcy even during times of conflict (Barakat et al., 2013).

Schools in formerly war-torn areas, including nations enjoying relative peace in a post-war era, may have to confront forces that challenge beliefs about schools and social cohesion. As Buckland (2006: 7) observed: ‘While education does not cause wars, nor end them, every education system has the potential either to exacerbate or to mitigate the conditions that contribute to violent conflict.’ We contend that schools in Bosnia and Herzegovina (BiH) represent a possible battleground and catalyst for continued student-on-student inter-ethnic violence (Nelles, 2006).

Post-war BiH had a decentralized system of school governance vested in local municipalities and regional cantons. These geographic areas are generally split along ethno-national lines. Beginning in the mid-1990s, BiH politicians, through the passage of divisive education laws, promoted latent or even blatant ethnic division among school-age youth (Nelles, 2006). The net result was schools that may ‘entrench intolerance, create or perpetuate inequality and intensify social tensions that can lead to civil conflict and violence’ (Johnson and Van Kalmthout, 2006: 3).

The claim that ethnic identity-driven conflicts in the post-World War II era have caused the deaths of millions and the displacement of tens of millions seems beyond dispute (Horowitz, 2000; Mann, 2005; Reynal-Querol, 2002). As Horowitz (2000: xi) notes in the second edition of Ethnic Groups in Conflict:

More than the secessionist wars in Biafra or Bangladesh, more than the mass killings in Rwanda or Burundi, more than the ethnic brutality of previous decades in more than a score of countries, the violent fragmentation of [the former Soviet Union and the former Yugoslavia] convinced Western opinion that group loyalty still had a murderous aspect in the heartland of Europe a half century after Hitler.

Ethnic conflicts can yield other forms of victimization beyond genocide and population displacement. In fact, our goal in this work is to investigate the linkages between the ethnic origins of school-attending youths in post-war BiH and their self-reported victimizations. Specifically, the main ethnic groups include Roman Catholic Croats, Eastern Orthodox Serbs, and Muslim Bosnians, the latter called Bosniaks. For decades following World War II, the three religious groups had coexisted; however, at the start of the 1992 war, long-simmering ethnic and religious differences began to percolate to the surface and conflict soon enveloped the region.

The current study employs two theories that offer different reasons for the emergence of intergroup conflict. According to group threat theory, conflicts between groups grow out of their respective economic and political power arrangements (Blalock, 1967; Blumer, 1958). Each identifiable group in a geo-political area sees other groups as a threat to its social, political, and economic position in the larger community. This theory has applications to the public school. Using a group threat framework, children in schools with populations that approach maximum diversity will perceive the greatest threat, resulting in the highest cross-group victimization levels (Blalock, 1967). Conversely, the imbalance of power thesis tells us that a group’s relationships with other groups depend on the relative imbalance of physical or social power existing between groups (Agirdag et al., 2011). This second theory is not, strictly speaking, about person-to-person imbalance – as in a very strong person bullying a far weaker-appearing peer on the playground. Rather, power imbalance predicts that victimizations are more likely when one identifiable group is numerically much larger than others in the same geophysical space, whereas relative peace is the result of a balance in relative group size (Graham, 2006).

In sum, we explore the viability of each theory’s arguments as an explanation of victimization experiences reported by school children living in a region torn by war only a decade before. Before addressing this work’s central issue, we first turn to an overview of the 1992–5 war in the former Socialist Federal Republic of Yugoslavia (SFRY) and the role of education in the emergent BiH.

The Balkans war: 1992–5

Beginning in August 1990, following the devolution of the SFRY, the former nation split into several autonomous and semi-autonomous entities. One of the SFRY’s republics, BiH, with its three socially and culturally distinct constituencies, was in a politically awkward position. Some residents of BiH, primarily Bosnian Serbs, desired to stay with a Yugoslavia dominated by the Republic of Serbia, whereas others, primarily Bosniaks and Bosnian Croats, favored decentralization and eventual independence (Budimlić et al., 2010; Hjort and Frisen, 2006; Trbovich, 2008). After several years of political wrangling, a national sovereignty election was held in early 1992, which was boycotted by Bosnian Serbs. The outcome was never in doubt, as 99.7 percent of those voting – roughly two-thirds of all eligible voters – favored complete separation from Yugoslavia and the creation of an independent Bosnia and Herzegovina (Commission on Security and Cooperation in Europe, 1992).

A series of armed conflicts quickly followed, some involving Bosniaks and Croats allied against Serbians and still others pitting Bosniaks against Croats. These lasted from early 1992 through the signing of the Dayton Peace Accord (DPA) in December 1995 (U.S. Department of State, 2015). Within parts of BiH, fighting was especially brutal and violent, including acts of so-called ‘ethnic cleansing.’ A literal translation of a Serbo-Croatian phrase (etnicko ciscenje), ethnic cleansing is ‘the expulsion, imprisonment, or killing of an ethnic minority by a dominant majority in order to achieve ethnic homogeneity’ (Merriam-Webster, n.d.). Mann (2005: 4) reminds us that the ‘essentially modern’ phenomenon of ethnic cleansing can occur as one group seeks to establish its hegemony over another group identified as ‘the enemy.’ As importantly, Mann (2005: 6) notes that, in cases of ethnic cleansing, ethnicity ‘trumps social class as the main form of social stratification,’ or how a society is ordered. The results are often, as was the case in the former Yugoslavia, horrific (Andreopoulos, 1997; Mann, 2005).

After the conflict, it was found that up to 8000 Bosniak men had been killed in the Srebrenica massacre by persons linked to the Bosnian Serb Army, called an act of genocide by the International Criminal Tribunal for the Former Yugoslavia (2004). The best available estimate suggests that upwards of 100,000 soldiers and civilians died in the war, nearly two-thirds being Bosniaks (Ball et al., 2007).

What makes the Bosnian war and its aftermath unique is the DPA’s decision to form a nation made up of three semi-autonomous regions split along ethnic lines and following geographic boundaries formed out of the conflict (Szasz, 1997). Since 1999, BiH has consisted of the Federation of Bosnia and Herzegovina (FBiH), occupied primarily by Bosniaks (52 percent), Bosnian Croats (22 percent) and Bosnian Serbs (18 percent). The Republika Srpska (RS), with over 80 percent Bosnian Serbs, envelopes the FBiH on three sides. Situated between the left and right halves of the RS is the self-governing Brčko District, populated by roughly one-third Bosnian Croats, 30 percent Bosnian Serbs, a quarter Bosniaks, and about 10 percent ethnic Yugoslavs (Central Intelligence Agency, n.d.).

At the time of data collection, the BiH was subject to a different kind of fighting, one for the hearts and minds of the nation’s population. The people of BiH felt the war’s social and psychological after-effects long after active combat (Brown and Cehajic, 2008; Hjort and Frisen, 2006; Klarić et al., 2007). Many critics questioned if reconciliation was possible, as politicians were still ‘rattling their [nationalist] sabers’ to remain in power well into the 21st century (Franovic and Rill, 2009: 31) and memories of neighbor-on-neighbor war-time atrocities committed in multiethnic towns and villages had dimmed little since the war (Parent, 2012).

Education in post-war BiH

A defining issue for the various geo-political groups in BiH has been the nation’s school system (Bush and Saltarelli, 2000; Risher and Kabil, 2010; UNICEF, 2009). The DPA gave authority over public education to three political entities. The FBiH further parceled out some control over education to the 10 local cantons but kept most of it in Sarajevo (Risher and Kabil, 2010). The DPA’s goal was to provide equal access to an educational system that would assist in coalescing the nation. In practice, however, the schools became instruments of segregation and not reconciliation (Risher and Kabil, 2010; Tveit et al., 2014). In this regard, Bush and Saltarelli (2000) found four negative elements commonly observed in post-conflict educational systems were endemic in BiH: (1) manipulating history for political purposes, (2) manipulating textbooks, (3) instilling self-worth and hatred for others, and (4) segregating education to ensure inequality, lowered esteem and stereotyping.

The consensus among international pedagogical experts is that education in BiH could be but was not yet – or not at the point that the data were collected – an effective instrument of community-building and reconciliation (Barbieri et al., 2013; Tortsi, 2009; UNICEF, 2009). As previously discussed, schools are, after families, perhaps the most influential and effective mechanism of community-building, especially when there is ‘local ownership’ within the impacted communities (Donais, 2012; Weinstein et al., 2007). As explained by Donais (2012: 1), ‘in post-conflict contexts, the notion (of local ownership) conveys the commonsense wisdom that any peace process not embraced by those who live with it is likely to fail.’ However, most BiH schools were in effect mono-ethnic, employing a ‘two schools under one roof’ model implemented in 1997 and abandoned in 2012 (Hunt et al., 2013; Tveit et al., 2014). The two schools under one roof policy meant that different ethnic groups taught in a single building experienced very different curriculums, each focusing on their respective ethnic affiliation, history, religion, linguistics, and language. Moreover, teaching about the history of the 1992–5 war varied by the dominant ethnic and religious identity of each village, town, or enclave (that is, neighborhood) (Massey et al., 1999). The history of the war generally took an ‘us against them’ stance when it came to identifying who were victims, defenders, and aggressors (Kreso, 2008). Education appears to have been used for post-war propaganda and ethnic division, further exacerbating and perpetuating ethnic conflicts and social marginalization (Oberschall, 2000).

In summary, the political ethno-histories taught in the decade after the war served as a collective narrative of ‘our people,’ linking post-war generations in BiH to an ethnic identity already shared by war survivors. The repackaging and manipulation of the Balkan war story depended upon the storyteller (Oberschall, 2000; see also Hayden, 1996). The construction of collective narratives that either implicitly or explicitly stereotyped, diminished, and degraded out-groups further exacerbated the divisive tensions that were identified as being partially responsible for the war (Fischer, 2006; Salomon, 2004). Inequality and exclusion – driven by ethnic identity – formed the basis of fear and perceived threat, and they have important consequences for inter-ethnic intolerances in the context of ethnic enclaves (Massey et al., 1999). By the early 2000s, the exploration of ethnicity-based school victimizations through a multi-level process nesting students in schools and schools within neighborhoods was a well-established practice (Benbenishty and Astor, 2005). The likelihood that students in BiH schools escaped pervasive inter-ethnic distrust seems low, and many victimizations are likely to have had an inter-ethnic component. A primary goal of this study, then, is to explore the theoretical ties between students’ ethnic identities, their schools’ ethnic demographic composition, and the self-reported victimization levels of students in those schools.

Understanding ethnic identity and victimization in post-war BiH

Once considered normative in the SFRY, post-war ethnic mixing and integration at the city and neighborhood levels were by the end of the first decade of the 21st century no longer a social reality (Adjudkovi and Biruski, 2008). Most scholars do not consider either pre-war ethnic relations or ancient ethnic hatreds a major cause of conflict in former Yugoslavia (Gagnon, 1994; Sekulić et al., 2006). Instead, nationalist politicians seemed to incite war and violence by politicizing ethnic identity, resulting in segregated ethnic enclaves and severed cross-ethnic ties (Gagnon, 2006; Sekulić et al., 2006). Tragic events during war became part of the cultural narrative (Gagnon, 2006). What was once a muted distrust among Bosniaks, Eastern Orthodox Serbians, and Roman Catholic Croatians evolved into reason enough for heightened intergroup violence (Costalli and Moro, 2011).

By the late 1990 and early 2000s, the nations of the former Yugoslavia were among the most factionalized areas in the world, as measured by ethnicity, linguistics, and religious diversity (Alesina et al., 2003). Divisions along these lines alone may not explain the consequences of civil wars such as experienced by BiH in 1992–5 (Reynal-Querol, 2002). What is needed is a theoretical framework to examine which school children are being victimized due to their ethnicity by what other students.

Clearly, schools are central to understanding ethnic conflict in post-war BiH. We focus on BiH’s various ethnic groups within the schools, each group relative to the others. As previously described, two theories offer possible conceptual lenses through which to view the ties between ethnicity and victimization. The diversity of ethnicities and races reflected in each school, called integration or evenness, is crucial for both models (Graham, 2006; Massey and Denton, 1988). First, group threat theory stipulates that economic and political contexts are foundational to the existence of frustration and negative attitudes between in-groups and out-groups (see Blalock, 1967; Blumer, 1958). At its core, the perspective centers on the in-group’s greater negative attitudes towards the out-group because of a real or perceived threat to in-group interests (see, too, Blumer, 1958; Quillian, 1995; Riek et al., 2006; Stephan et al., 2009).

Group threat theory further suggests that children in schools with populations that approach maximum diversity will perceive the greatest threat, resulting in the highest cross-group victimization levels (Agirdag et al., 2011). Maximum diversity reflects polarization. Reynal-Querol (2002: 32) contends that, under a polarization model, ‘the situation that leads to the point of maximum tension is when there are two social groups with the same size.’ Hence, she also sees polarization or ethnic evenness as leading to increased conflicts and victimizations.

Conversely, the imbalance of power thesis suggests that a group’s decision to victimize or remain peaceful towards one or more other groups depends on the relative imbalance of physical and/or social power existing between various groups (Agirdag et al., 2011; Cederman et al., 2013). A key part of such power is the ability to inflict punishments and distribute rewards (Blau, 1964; Emerson, 1962; French and Raven, 1959). Viewed from this perspective, victimization is more likely to happen when one group is much larger than another, whereas relative peace between groups is the result of a balance in relative group size (Agirdag et al., 2011; Olweus, 1997), a prediction that is the opposite of that modeled by group threat theory. In situations of power imbalance, as diversity levels in each community approach equality (where all significant groups are represented in roughly equal proportions in the community), the groups conflict more, where conflict can be broadly defined as various forms of intolerance towards each other or narrowly defined as subjecting ‘the other’ to heightened levels of personal victimization (Agirdag et al., 2011; Lam, 2013; Semyonov et al., 2012).

Research questions

This article’s central thesis is that school environments interact with individual-level ethnic identifies and can impact individual-level victimization outcomes. A school’s social environment may reflect power arrangements manifested at geo-political levels. Here, the focus is on aggregations of ethnicities between and within schools and how they impact individual-level victimization. In this context, personal identities are often embedded in an ethnic identity that provides a basis for differentiation from others who do not share it.

We employ the following research questions to explore the victimization experiences of public school students in BiH:

Is there a significant ethnic effect on victimization across schools?

Is there a significant ethnic effect on victimization within schools?

Does the effect of school diversity on personal-level victimization depend upon individual-level ethnicity?

With which theoretical perspective do the data best agree – group threat theory or imbalance of power thesis?

The observed relationships between self-reported victimization levels, personal ethnic identity, and level of school ethnic evenness should be largely unaffected by the concurrent examination of personal-biographical measures and crime-related measures drawn from multiple domains.

Research methods

Data sources

In 2007, researchers at the University of Sarajevo’s Faculty of Criminal Sciences used a 12-page self-report questionnaire to gather information about individual and family demographic characteristics, youths’ gang involvement and offending, and ecological risk and protective factors such as victimization and family relationships. The researchers drew items related to youth offending and risk and protective factors from measures with demonstrated psychometric properties (Elliott et al., 1985).

The researchers collected the data from a two-stage cluster sample of 2206 youth between the ages of 13 and 17 studying in 53 primary and secondary schools located in the five largest cities in BiH: Sarajevo, Banja Luka, Mostar, Tuzla, and Zenica. The administrative areas of the city school systems included not only these named cities, but also the surrounding suburban and rural areas. Schools constituted the first cluster and classrooms the second. The final sample reflected the relative size of each area, and within each area the researchers selected the schools at random for participation. In the selected schools, researchers also randomly selected classrooms, and invited all students in the selected classrooms to participate. Complete responses were available for 2003 respondents, because we excluded 203 respondents owing to missing values; this figure represented nearly 90 percent of all available students in the selected schools.

Dependent variable

Four items measured respondents’ self-reported victimization. Each asked how often the youth had experienced a specific form of victimization. Personal victimization questions were whether respondents ‘had been hit by someone’ or ‘attacked by someone.’ Property victimization questions asked about being the victim of a robbery or a victim of theft. All items used a five-point scale to measure victimization during the previous 12 months (1 = never, 2 = once or twice, 3 = 3–5 times, 4 = 6–10 times, 5 = more than 10 times). Scores on the four items were summed to form a victimization index with a range from 4 to 20. The following mean victimization index scores (and standard deviations) were noted for youth indicating a particular country of birth (COB): Bosnia-born = 5.34 (SD 1.76); Croatia-born = 5.76 (SD 2.24); Serbia-born = 5.49 (SD 1.82); and Other COB = 5.75 (SD 1.89).

Independent variables

Our research questions focus on individual ethnic identity and the ethnic composition of each school. We employed COB as a proxy measure of ethnic identity. The Montenegrin responses (n = 20) were collapsed into the Other category. Within our final sample, 1778 (88.8 percent) were by birth Bosnian, 90 (4.49 percent) were Croatian, 55 (2.75 percent) were Serbian, and 80 (3.99 percent) were Others. For analytic purposes, Bosnia-born was the reference category.

In the interests of as complete a test as possible, we employ both Blau’s index and Theil’s H as measures of each school’s ethnic diversity. The measurement of diversity (that is, within unit variation) describing the distribution of individual traits such as ethnicity remains an active area of study (Reardon and Firebaugh, 2002; Van Dick et al., 2008; Van Knippenberg and Schippers, 2007; Williams and O’Reilly, 1998). Conceptualization and related measurement issues are part of this focus, which led Harrison and Klein (2007) to suggest the organization by type of the various diversity indexes: separation, variety, and disparity. Separation and disparity refer to differences in attitude or position and to differences in power status and hierarchy, respectively. For our purposes, we focus on diversity as variety. Diversity as variety assumes that, ‘within units, members differ from one another on a categorical attribute V . . . units differ in the extent to which members are evenly spread across all categories of V . . . differences between units in their relative spread or diversity on V will be associated, usually positively, with vital unit consequences (e.g., problem solving. . .)’ (Harrison and Klein, 2007: 1204).

Blau’s index

Blau’s index (1977) builds upon indexes used in other disciplines such as economics (for example, Herfindahl, 1950; Hirschman, 1964) and ecology (for example, Simpson, 1949).¹ Applied to our school data, a maximum value occurs when students in a school are spread equally or obtain ‘evenness’ within a school and would be indicated by a value of 0.75.

Theil’s H (Theil’s diversity index)

Thiel (1972) proposed a diversity index that combines information on the number of entities or groups in a unit (for example, a school) and the corresponding proportion of each group within the unit (Hao and Fong, 2011).² Zero occurs when all individuals are members of a single group (no diversity) and the maximum score occurs if individuals are evenly distributed among the M groups. The entropy index is, therefore, a measure of ‘evenness’ that captures the degree to which groups are evenly distributed within units (Massey and Denton, 1988). The higher the number, the more diverse the school’s ethnic populations. The maximum score occurs when all groups have equal representation in a school, such that the proportion of each group is the same.

Control variables

We used several control variables drawn from the life-style exposure literature, which suggests that daily routines can impact victimization risk (Hindelang et al., 1978; Maxfield, 1987; Meier and Miethe, 1993). Table 1 contains the descriptive statistics for the individual- and school-level variables. Such things as self-reported delinquency and criminal activity, general rule breaking, drinking and other drug use, and aggressive behavior can increase victimization rates. We include parental monitoring and commitment to school as protective factors. Biological sex, too, is seen as a risk factor of some importance.

Table 1.

Descriptive statistics.

Variable	Mean	SD	Min.	Max.	n
Individual-level variables
Age	14.96	1.32	12	20	2003
Male	0.51	0.50	0	1	2003
Parental Monitoring	15.44	3.10	4	20	2003
School Commitment	14.49	2.95	4	20	2003
Living Arrangements	0.85	0.36	0	1	2003
Gang	0.04	0.21	0	1	2003
Stealing^a	1.98	1.45	1	5	2003
Lied about Age^b	1.29	0.86	1	5	2003
Aggressive Behavior^c	7.70	3.32	4	30	2003
Alcohol Use	0.84	1.11	0	4	2003
Tobacco Use	0.61	1.21	0	4	2003
Bosnian	0.89	0.32	0	1	2003
Croatian	0.04	0.21	0	1	2003
Serbian	0.03	0.16	0	1	2003
Other Ethnicity	0.04	0.20	0	1	2003
School-level variables
Blau’s index	0.19	0.15	0	0.59	53
Theil’s diversity index	0.27	0.19	0	0.72	53

Failed to pay for movies, bus fare, etc.

Lied to gain entrance or make a purchase.

Refers to self-reported instances over the previous 12 months of the following: damaging property, carrying a hidden weapon, hitting someone with the idea of hurting them, using a weapon or force to get money, involvement in group fights, and attacking someone with a weapon.

Two risk/protective measures within the family domain are often linked to victimization experiences (Esbensen et al., 2010; Zimmerman and Rees, 2014). Each subject responded to four items related to parental monitoring measured on a five-point scale (1 = strongly disagree to 5 = strongly agree). The responses captured the extent to which (a) youth informed parents where they were going, (b) parents knew where the youth was when not at home or at school, (c) youth knew how to get in touch with parents, and (d) parents knew whom the youth was with when not at home. Collectively, these items reflected two domains of monitoring – parental knowledge of an adolescent’s whereabouts and adolescents’ disclosure of activities to parents – identified by Hoeve and colleagues (2009) as essential for understanding parental effects on youthful misconduct. Reliability analyses indicated the summated four-item Parental Monitoring scale tapped a single latent variable (alpha = 0.78). Second, we also include a dichotomous variable to capture the family structure of the respondent’s home. Living Arrangement is dummy-coded so that 0 equaled other than two parents, while 1 equaled two parents living at home.

Two items reflected students’ self-reported delinquency and drug use (Gatti et al., 2015). Respondents were asked if, in the past 12 months, they had (a) avoided paying for something such as movies, bus or subway rides (Stealing) and (b) lied about their age to get into some place or to buy something (Lied about Age). Each was measured on a five-point scale (0 = never, 1 = once or twice, 2 = 3–5 times, 3 = 6–10 times, 4 = more than 10 times). Our models included two indicators of drug use in the past 12 months as separate variables (Tobacco Use and Alcohol Use), measured on a five-item scale, ranging from a score of 0 for never to 4 for nearly every day.

Six items asked about aggressive and violent behavior, including ‘damaged property,’ ‘carried a hidden weapon,’ ‘hit someone with the idea of hurting them,’ ‘used a weapon or force to get money,’ ‘was involved in group fights,’ and ‘attacked someone with a weapon’ (Smith-Khuri et al., 2004). For each self-reported behavior we used a five-point response set indicating how often the youth had engaged in Aggressive Behavior over the previous 12 months: 0 = never, 1 = once or twice, 2 = 3–5 times, 3 = 6–10 times, 4 = more than 10 times. The scale’s internal consistency was good (alpha = 0.8).

Criminologists tend to view school as a protective measure (Zimmerman and Rees, 2014). We used four items to create a School Commitment scale: (a) ‘I try hard in school’, (b) ‘In general, I like school’, (c) ‘Grades are very important to me’, and (d) ‘I usually finish my homework.’ Possible answers used a five-point Likert agreement–disagreement response set, which created a reliable scale (alpha = 0.72).

Our analyses also control for age in years (Age) and gender, in this case being male or not (Male).

Modeling strategy

Our strategy for creating multi-level models begins with a null model to establish a baseline for evaluating larger models. After examining the null model with no predictors to assess variability at each level, we estimated two-level regression models (hierarchical linear models – HLM), which simultaneously examined individual-level ethnicity and school-level ethnic diversity indexes as predictors of the victimization index (Raudenbush and Bryk, 2002). The HLM approach generates efficient slope estimates and unbiased standard errors when level-one units (that is, persons) nested within level-two groups (that is, schools) share similar traits (that is, are clustered). It also allows for estimation of cross-level interactions. To answer our question about the personal ethnicity/school ethnic diversity interactions, we modeled the slopes of the school-level ethnicity indexes as a function of the personal ethnicity variables. Lastly, we estimated a full model inclusive of a host of important individual-level control variables related to individual-level victimization outcomes. As none of the individual-level ethnicities varied significantly across schools, we opted not to model these slopes with random coefficients. As a protection against model misspecification, we re-estimated the models allowing for the coefficients of the individual-level ethnicity variables to vary randomly across schools. The results that form the basis of our findings were substantively unchanged when using this alternate modeling strategy.

Findings

Question 1: Is there a significant ethnic effect on victimization across schools?

To examine if there was school-level variation in the victimization outcomes, we first estimated an unconditional model without level-one or level-two covariates. Those analyses (not shown) indicated that there was significant school-level variability in youth self-reported victimization (τ00 = 0.147, p < .001), justifying our multi-level modeling strategy.

Question 2: Is there a significant ethnic effect on victimization within schools?

In Table 2, we add the key independent variables (that is, respondent COB and indexes of diversity) to the unconditional models. Both the Theil’s and Blau’s indexes suggest that school-level diversity has a positive and significant association with respondent victimization. This finding is consistent with the view that relationships among pupils in school may reflect ongoing ethnic and political tensions in BiH (Smith, 2010).

Table 2.

Hierarchical regression models of victimization outcomes regressed on respondent country of birth, Blau’s index, or Theil’s diversity index.

Ethnicity type	Model 1: Theil’s diversity index		Model 2: Blau’s index
Ethnicity type	β	(SE)	β	(SE)
Croatian^a	0.247	(0.21)	0.227	(0.21)
Serbian^a	0.113	(0.24)	0.114	(0.24)
Other^a	0.230	(0.21)	0.223	(0.21)
Blau’s index			1.022*	(0.43)
Theil’s index	0.737*	(0.34)
Intercept	5.175	(0.11)	5.183	(0.10)

Notes: n = 2003; β = Linear Coefficient; SE = Standard Error.

Reference category: Bosnia.

p < .05; **p < .01; ***p < .001 (two–tailed tests).

Question 3: Does the effect of school diversity on personal-level victimization depend upon individual-level ethnicity?

An initial answer to our third question comes from adding cross-level interaction terms between respondent COB and the corresponding diversity index to the models. The interaction effects in Table 3 use the Bosnian COB as a reference category. Once again, the positive and significant cross-level Serbia × the interaction terms for both Theil’s and Blau’s indexes, coupled with the significant negative effect of individual-level Serbian ethnicity, are consistent with our hypothesis that the effect of school-level diversity manifests in different ways depending upon individual-level ethnicity.³

Table 3.

Preliminary interaction analysis for association between individual ethnicity, Blau’s index, and Theil’s diversity index.

Model 3: Theil’s diversity index					Model 4: Blau’s index
	Individual ethnicity		Individual ethnicity × Theil’s index		Individual ethnicity		Individual ethnicity × Blau’s index
	β	(SE)	β	(SE)	β	(SE)	β	(SE)
Croatian^a	0.93	(0.55)	−1.38	(1.04)	0.75	(0.47)	−1.41	(1.13)
Serbian^a	−1.50**	(0.55)	4.65**	(1.44)	−1.38**	(0.52)	6.33**	(1.97)
Other^a	1.06	(0.52)	−1.93	(1.12)	1.03*	(0.45)	−2.60*	(1.30)
Theil’s index	0.78*	(0.35)
Blau’s index					1.14*	(0.45)
Constant	5.16***	(0.11)			5.16***	(0.10)

Notes: n = 2003; β = Linear Coefficient; SE = Standard Error.

Reference category: Bosnia.

p < .05; **p < .01; ***p < .001 (two-tailed tests).

We explored this effect by calculating the average marginal effects of school diversity (that is, simple slopes) for each individual-level ethnicity on victimization levels. In Figure 1, we see the slope of the line for the average marginal effect of school ethnic diversity as having a slightly positive and significant effect on victimization across for Bosnia-born youth (b = 1.14, p < .05). However, the overall effect on increasing levels of diversity for Serbians is much stronger (b = 7.48, p < .000). We tested the difference between these two slopes and found that the effect of increasing diversity on victimization levels for Serbia-born youth is significantly higher compared with that for their Bosnia-born counterparts ( $χ^{2} = 10.31, d f = 1, p < . 01)$ .

Figure 1.

Relationship between individual ethnicity and predicted respondent victimization across values of Blau’s index.

Similar findings are evident in Figure 2. The effect of diversity measured using Theil’s index is positive and significant for Bosnia-born students (b = 0.78, p < .05). For these school children, increasing levels of school diversity lead to increased levels of victimization. The effect of increasing levels of diversity for Serbia-born students is also positive and significant (b = 5.43, p < .001). Again, there is a significant difference between these two effects, with the effect of increasing diversity being stronger for Serbia-born students compared with that for their Bosnian counterparts ( $χ^{2} = 10.32, d f = 1, p < . 01)$ .

Figure 2.

Relationship between individual ethnicity and predicted respondent victimization across values of Theil’s diversity index.

In Table 4, we provide a more conservative test of our third hypothesis by controlling for an array of variables that previous researchers suggest are related to victimization. As expected, the cross-level respondent COB/school diversity interactions for both indexes remain positive and significant for Serbia-born students. These findings, coupled with the significant negative effect reported for both indexes of individual-level Serbian COB (−1.26, p < .05; −1.13, p < .05), are consistent with our suggestion that the effect of school-level diversity manifests in different ways depending upon individual-level ethnicity.

Table 4.

Hierarchical regression models of victimization outcomes regressed on individual ethnicity × Blau’s index or ethnicity × Theil’s index interaction and controls.

Variable	Model 5: Theil’s diversity index		Model 6: Blau’s index
Variable	β	(SE)	β	(SE)
Individual-level variables
Age	−0.065	(0.04)	−0.064	(0.04)
Male	0.123	(0.07)	0.124	(0.77)
Parental Monitoring	−0.007	(0.01)	−0.007	(0.01)
School Commitment	−0.001	(0.01)	−0.001	(0.01)
Living Arrangements	−0.067	(0.09)	−0.066	(0.09)
Gang	−0.416*	(0.19)	−0.419*	(0.19)
Stealing^a	0.050	(0.03)	0.049	(0.03)
Lied about Age^b	0.150**	(0.05)	0.150**	(0.05)
Aggressive Behavior^c	0.201***	(0.01)	0.201***	(0.01)
Alcohol Use	−0.034	(0.04)	−0.033	(0.04)
Tobacco Use	0.101**	(0.04)	0.101***	(0.04)
Croatian^d	0.640	(0.49)	0.499	(0.42)
Serbian^d	−1.260*	(0.49)	−1.130*	(0.46)
Other^d	0.706	(0.46)	0.707	(0.40)
School-level variables
Blau’s index			0.831	(0.42)
Theil’s diversity index	0.579	(0.33)
Cross-level interactions
Croatian × Theil’s index^d	−0.888	(0.94)
Serbian × Theil’s index^d	3.773**	(1.30)
Other × Theil’s index^d	−1.050	(1.01)
Croatian × Blau’s index^d			−0.846	(1.01)
Serbian × Blau’s index^d			4.991**	(1.77)
Other × Blau’s index^d			−1.460	(1.17)
Intercept	4.480***	(0.660)	4.46***	(0.653)

Notes: n = 2003; β = Log-Odds Coefficient; SE = Standard Error.

Failed to pay for movies, bus fare, etc.

Lied to gain entrance or make purchase.

Reference group: Bosnian.

*p < .05; **p < .01; ***p < .001 (two-tailed tests).

We again calculated the average marginal effects of school diversity for the different individual-level ethnicities (data not shown but available on request). Our findings remain consistent with the simpler model presented in Table 3. The effect of increasing levels of school diversity as measured by Blau’s index has a strong positive effect on victimization for Serbian students (b = 5.822, p < .01). This effect for Bosnia-born students is no longer significant (b = 0.83, p > .05). There is a significant difference between these two slope coefficients, indicating a stronger positive effect of increasing levels of diversity on increased levels of victimization for Serbia-born students compared with Bosnians $(χ^{2} = 10.32, d f = 1, p < . 01)$ . The effect of increasing levels of school diversity as measured by Thiel’s index has a strong positive effect on victimization for Serbian students (b = 4.353, p < .01). This effect for Bosnia-born students is no longer significant (b = 0.57, p > .05). There is a significant difference between these two slope coefficients, indicating a stronger positive effect of increasing levels of diversity on increased levels of victimization for Serbia-born students compared with students whose COB is Bosnia $(χ^{2} = 8.40, d f = 1, p < . 01)$ .

Figure 3 also provides additional insight into the most common COB within schools and its relationship to school diversity. The proportions of each COB group within a school and the schools’ corresponding diversity indexes will tend towards convergence at higher levels of diversity. However, Bosnia-born students consistently represent the numerical majority in the school samples. Recall the magnified effect of increasing diversity for either index on Serbia-born students’ victimization levels when compared with all other reported COBs.

Figure 3.

COB proportions in school by diversity index.

Question 4: With which theoretical perspective do the data best agree – group threat theory or imbalance of power thesis?

Overall, our findings are consistent with theoretical ideas of majority–minority power struggles, spatial dominance, and intolerance captured by group threat theory (see Blau, 1977; Blalock, 1967; Massey et al., 1999; Sekulić et al., 2006). Assuming respondent COB provides an adequate proxy measure of ethnic identity, intergroup contact between the study’s three main groups may foster tolerance when there is a clear majority within a school. However, this tolerance may falter as evenness in group representation increases within schools. This observation may be particularly true for post-war BiH, where post-war nationalism constituted a powerful political platform and a student’s own ethnic identity may determine positive or negative experiences during school (Aleksić and Đorđević, 2013).

In both Models 5 and 6 for the individual difference covariates, self-identified gang members in our sample had on average significantly fewer victimizations than non-gang members, a finding that runs counter to extant findings reported by other US and European researchers (see Carson and Esbensen, 2014; Fox et al., 2011; Klein, 2006). Lying about one’s age significantly increased victimization levels in both models, as did aggressive behavior and tobacco use. Drug use, however, failed to perform in similar ways. As importantly, age, gender, parental monitoring, school commitment, and living arrangements failed to play significant roles as covariates.

Summary and conclusions

The present study examined whether ethnicity was related to self-reported victimizations across a wide range of BiH schools, simultaneously considering both the direct and the moderating effects of the levels of evenness of ethnic representation. Our results are consistent with group threat theory: as the level of evenness in ethnic representation increases within schools dominated by Bosnia-born students, the victimization levels of Serbian students increased. Our results also suggest that personal ethnic identity’s impact on victimization within BiH schools depends on school demographics and tends to reflect neighborhood and state-level relations between the constituent ethnicities in BiH. This finding agrees well with the theoretical notion that out-group levels of threat perceptions are partially driven by a history of conflict with the in-group (Shamir and Sagiv-Schifter, 2006). Educational researchers report that, at the time of the data collection, BiH schools essentially endorsed between-group stereotypes and intolerances (Quist and Resendez, 2002; Shamir and Sagiv-Schifter, 2006). Furthermore, the reality or perception of threat from another group can lead to emotional responses such as fear, anxiety, and anger, with behavioral responses inclusive of overt aggression, retaliation, and harassment (Stephan et al., 2009). Such negative responses are precisely what most school administrators would normally try to prevent.

Given the ferocity of the Balkan war, both Roman Catholic Croatians and Bosniaks generally came to view Serbs during and after the conflict as ‘the enemy’ (Costalli and Moro, 2011; Mojzes, 2011; see, also, Ball et al., 2007). Serbian school children have expressed remorse and guilt over the excesses of ‘their side’ in the 1992–5 war (Brown and Cehajic, 2008). Although Roman Catholic Croats and Bosniaks also fought each other, albeit infrequently, these conflicts did not appear to translate into higher post-war victimizations for the Croatia-born youth in our study. Victimizations were highest for Serbia-born youth only in a situation of evenness compared with those born in Bosnia. It was more than just being a minority group outsider or being a non-Bosnian, then, that determined one’s suitability as a target for victimization. We suspect that it is the deep-seated distrust and distain felt by Bosnia-born youths for Serbia-born youths, likely a product of the war, that can help us understand these victimizations. Again, this situation is more probable as the out-group approaches parity or evenness with the group that defines itself as egregiously victimized in the past (Hayden, 1996).

The current study does have several important limitations. First, we do not measure individuals’ ethnic identity directly and instead use country of birth as a proxy. The absence of a direct measure also precludes us from determining how strongly each student identifies with his or her chosen ethnicity. Although COB can serve as a meaningful proxy for ethnic identity (Stronks et al., 2009), an individual’s ethnic identity in post-war BiH represents an exceptionally sensitive and complex personal, social, and political topic. For example, political manipulation continues to derail attempts to conduct the first complete census since the beginning of the war, originally scheduled for completion in 2013. A central concern is the collection of individuals’ self-determined ethnic identity, language, and religion. Nonetheless, our findings using COB tend to agree with theoretical expectations surrounding inter-ethnic relations in BiH.

Although we find a clear connection between evenness in the distribution of ethnicities in BiH schools and victimization, we recognize that some of this association may be due to students’ pre-existing level of dissatisfaction with the level of ethnic segregation found in the city or town in which the school is located. That is, the significant inter-ethnic differences in victimization levels may, at least in part, be driven by the minority ethnicity’s frustration with the segregated housing and political tensions still prevalent in BiH. One way to test this would be to measure segregation at the city level and include city as a third level in our multi-level model. These types of data would normally be included in a city- or country-level census and, as noted, is problematic for the study of segregation in post-war BiH.

Lastly, we recognize that the cross-sectional nature of our data exposes us to issues of causal ordering. Longitudinal data would allow us to control for and parcel out the effects of prior levels of victimization on current levels of victimization and capture a more precise estimate of the effect of individual and school-level measures of ethnic identity.

In conclusion, the relationship between individual ethnicity and victimization in BiH schools is moderated by the ethnic distribution at the school level. Namely, as the representation of Serbia-born youth in a school increases towards numerical equality with Bosnia-born youth, victimization of Serbia-born youth increases. We see this as evidence of school demographics and individual characteristics reflecting inter-ethnic tensions found in BiH neighborhoods and the Balkans war in general. The role of schools in peace-building efforts could be one of reconciliation of new and younger generations of Bosniaks, Croats, and Serbs. However, should ethnicity continue to be a political issue and used as a measure of national loyalty, ethnic tensions will be further entrenched in BiH and be reflected in the victimization levels of BiH students.

Footnotes

Acknowledgements

The following individuals at the University of Sarajevo participated in the data collection phase of this study: Almir Maljevic, Elmedin Muratbegovic, Muhamed Budimlić, and Eldan Mujanović.

Authors’ Note

Points of view are those of the authors and do not necessarily reflect the official position of the U.S. Department of Justice. An earlier version of this work was presented at the 15th Annual Conference of the European Society of Criminology in Porto, Portugal, 2–5 September 2015.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by Research Award 1704-257 from the International Center, National Institute of Justice, Office of Justice Programs, U.S. Department of Justice.

ORCID iD

L Thomas Winfree Jr

Notes

References

Agirdag

Demanet

Van Houtte

Van Avermaet

(2011) Ethnic school composition and peer victimization: A focus on the interethnic school climate. International Journal of Intercultural Relations 35: 465–473.

Aleksić

Đorđević

(2013) Nationality (citizenship) vs. ethnicity: Clash of identities in Bosnia and Herzegovina. National Question in Central Europe 155.

Alesina

Devleeschauwer

Easterly

Kurlat

Wacziarg

(2003) Fractionalization. Journal of Economic Growth 8: 155–194.

Andreopoulos

(1997) Genocide: Conceptual and Historical Dimensions. Philadelphia: University of Pennsylvania Press.

Ball

Tabeau

Verwimp

(2007) The Bosnian Book of the Dead: Assessment of the Database (Full Report). HiCN Research Design Note 5. Institute of Development Studies, University of Sussex. URL (accessed 22 January 2021): https://hrdag.org/wp-content/uploads/2013/02/rdn5.pdf.

Barakat

Connolly

Hardman

Sundaram

(2013) The role of basic education in post-conflict recovery. Comparative Education 49: 124–142.

Barbieri

Vrgova

Bliznakovski

(2013) Overcoming ethnic-based segregation: How to integrate public schools in Macedonia and Bosnia and Herzegovina. Institute for Democracy, Societas Civilis - Skopje. URL (accessed 25 January 2021): https://idscs.org.mk/wp-content/uploads/2013/12/zastoneidscs_2013_final.pdf.

Benabou

(1996) Heterogeneity, stratification, and growth: Macroeconomic implications of community structure of school finance. American Economic Review 86: 584–609.

Benbenishty

Astor

(2005) School Violence in Context: Culture, Neighborhood, Family, School, and Gender. Oxford: Oxford University Press.

10.

Blalock

(1967) Toward a Theory of Minority-Group Relations. New York: Capricorn Books.

11.

Blau

(1964) Exchange and Power in Social Life. New York: Wiley.

12.

Blau

(1977) Inequality and Heterogeneity: A Primitive Theory of Social Structure, Vol. 7. New York: Free Press.

13.

Blumer

(1958) Race prejudice as a sense of group position. The Pacific Sociological Review 1(1): 3–7.

14.

Brown

Cehajic

(2008) Dealing with the past and facing the future: Mediators of the effects of collective guilt and shame in Bosnia and Herzegovina. European Journal of Social Psychology 38(4): 669.

15.

Brunson

Miller

(2009) Schools, neighborhoods, and adolescent conflicts: A situational examination of reciprocal dynamics. Justice Quarterly 26(2): 183–210.

16.

Buckland

(2006) Post-conflict education: Time for a reality check. Forced Migration Review Supplement, July: 7–8.

17.

Budimlić

Maljević

Muratbegović

(2010) Bosnia and Herzegovina. In: Junger-Tas

Marshall

Enzmann

Killias

Steketee

Gruscznska

(eds) Juvenile Delinquency in Europe and Beyond. New York: Springer.

18.

Bush

Saltarelli

(2000) The Two Faces of Education in Ethnic Conflict. Florence, Italy: UNICEF Innocenti Research Centre.

19.

Carson

Esbensen

F-A

(2014) The mediating effects of delinquent attitudes on race, race heterogeneity, and violent offending. Journal of Crime and Justice 37(1): 23–41.

20.

Cederman

L-E

Skrede Gleditsch

Buhaug

(2013) Inequality, Grievances, and Civil War. Cambridge: Cambridge University Press.

21.

Central Intelligence Agency (n.d.) The World Factbook: Europe: Bosnia and Herzegovina. URL (accessed 25 January 2021): https://www.cia.gov/the-world-factbook/countries/bosnia-and-herzegovina/.

22.

Christenson

Reschly

(eds) (2010) Handbook of School–Family Partnerships. New York: Routledge.

23.

Commission on Security and Cooperation in Europe (1992) Report: The Referendum on Independence in Bosnia-Herzegovina. Washington, DC: Commission on Security and Cooperation in Europe.

24.

Costalli

Moro

(2011) The patterns of ethnic settlement of violence: A local level quantitative analysis of the Bosnian war. Ethnic and Racial Studies 34(12): 2096–2114.

25.

Donais

(2012) Peacebuilding and Local Ownership: Post-Conflict Consensus-Building. New York: Routledge.

26.

Elliott

Huizinga

Ageton

(1985) Explaining Delinquency and Drug Use. Beverly Hills, CA: Sage Publications.

27.

Emerson

(1962) Power–dependence relations. American Sociological Review 27(1): 31–41.

28.

Esbensen

F-A

Peterson

Taylor

Freng

(2010) Youth Violence: Sex and Race Differences in Offending, Victimization, and Gang Membership. Philadelphia, PA: Temple University Press.

29.

Fischer

(2006) Integration or segregation? Reforming the education sector. In: Martina

(ed.) Peacebuilding and Civil Society in Bosnia-Hercegovina. Berlin: LIT Verlag.

30.

Fox

Rufino

Kercher

(2011) Crime Victimization and Gang Membership. Huntsville, TX: Sam Houston State University, Crime Victims’ Institute.

31.

Franovic

Rill

(eds) (2009) I Cannot Feel Good If My Neighbor Does Not. Belgrade/Sarajevo: Center for Nonviolent Action.

32.

French

Raven

(1959) The bases of social power. In: Cartwright

(ed.) Studies in Social Power. Ann Arbor, MI: Institute for Social Research.

33.

Gagnon

Jr (1994) Ethnic nationalism and international conflict: The case of Serbia. International Security 19(3): 130–166.

34.

Gagnon

Jr (2006) The Myth of Ethnic War: Serbia and Croatia in the 1990s. New York: Cornell University Press.

35.

Gatti

Soellner

Bräker

A-B

Verde

Rocca

(2015) Delinquency and alcohol use among adolescents in Europe: The role of cultural contexts. European Journal of Criminology 12: 362–377.

36.

Gottfredson

(2001) Schools and Delinquency. Cambridge: Cambridge University Press.

37.

Graham

(2006) Peer victimization in school: Exploring the ethnic context. Current Directions in Psychological Science 15(6): 317–321.

38.

Hao

Fong

(2011) Linking dichotomous segregation with multi-group segregation: Weighted segregation ratios in selected US metropolitan areas. Social Science Research 40(1): 379–391.

39.

Harrison

Klein

(2007) What’s the difference? Diversity constructs as separation, variety, or disparity in organizations. Academy of Management Review 32(4): 1199–1228.

40.

Hayden

(1996) Imagined communities and real victims: Self-determination of ethnic cleansing in Yugoslavia. American Ethnologist 23(4): 783–801.

41.

Herfindahl

(1950) Concentration in the U.S. steel industry. Unpublished doctoral dissertation, Columbia University, New York.

42.

Hindelang

Gottfredson

Garofalo

(1978) Victims of Personal Crime: An Empirical Foundation for a Theory of Personal Victimization. Cambridge, MA: Ballinger.

43.

Hirschman

(1964) The paternity of an index. The American Economic Review 54(5): 761–762.

44.

Hjort

Frisen

(2006) Ethnic identity and reconciliation: Two main tasks for the young in Bosnia-Herzegovina. Adolescence 41(161): 141–163.

45.

Hoeve

Dubas

Eichelsheim

, et al. (2009) The relationship between parenting and delinquency: A meta-analysis. Journal of Abnormal Child Psychology 37: 749–775.

46.

Horowitz

(2000) Ethnic Groups in Conflict. 2nd edn. Berkeley, CA: University of California Press.

47.

Hubbard

Mehan

Stein

(2006) Reform as Learning: School Reform, Organizational Culture, and Community Politics in San Diego. Abingdon: Taylor and Francis.

48.

Hunt

Durakovic

Radeljkovic

(2013) Bosnia today: Despair, hope, and history. Dissent 60(3): 23–26.

49.

International Criminal Tribunal for the Former Yugoslavia (2004) Appeals chamber judgment in the case of the Prosecutor versus Radislav Krstić. URL (accessed 22 January 2021): http://www.icty.org/x/cases/krstic/acjug/en/krs-aj040419e.pdf.

50.

Johnson

van Kalmthout

(2006) Editorial. Forced Migration Review, Supplement, July: 3. URL (accessed 22 January 2021): http://reliefweb.int/sites/reliefweb.int/files/resources/8EA5936A8A4C07ABC12571BE002E99DA-FMR-Jul2006.pdf.

51.

Klarić

Stevanović

, et al. (2007) Psychological consequences of war trauma and postwar social stressors in women in Bosnia and Herzegovina. Croatian Medical Journal 48(2): 167–176.

52.

Klein

(2006) Street gang violence in Europe. European Journal of Criminology 3: 413–437.

53.

Kreso

(2008) The War and Post-War Impact on the Educational System of Bosnia and Herzegovina. Netherlands: Springer.

54.

Lam

(2013) Developing ethnic identities and ethnic attitudes in children: Theory of research in context. In: Agyemang

Airhihenbuwa

de Graft

(eds) Ethnicity: Theories, International Perspectives and Challenges. Hauppauge, NY: Nova Science Publishers.

55.

Mann

(2005) The Dark Side of Democracy: Explaining Ethnic Cleansing. Cambridge: Cambridge University Press.

56.

Massey

Denton

(1988) The dimensions of residential segregation. Social Forces 67(2): 281–315.

57.

Massey

Hodson

Sekulić

(1999) Ethnic enclaves and intolerance: The case of Yugoslavia. Social Forces 78(2): 669–693.

58.

Maxfield

(1987) Lifestyle and routine activities theories of crime: Empirical studies of victimization, delinquency, and offender decision-making. Journal of Quantitative Criminology 3(4): 275–282.

59.

Meier

Miethe

(1993) Understanding theories of criminal victimization. Crime and Justice 17: 459–499.

60.

Merriam-Webster. Ethnic cleansing. URL (accessed 22 January 2021): http://www.merriam-webster.com/dictionary/ethnic%20cleansing.

61.

Mojzes

(2011) Balkan Genocides: Holocaust and Ethnic Cleansing in the Twentieth Century. New York: Rowman & Littlefield.

62.

Morrison

(2001) The school system: Developing its capacity in the regulation of a civil society. In: Strang

Braithwaite

(eds) Restorative Justice and Civil Society. Cambridge: Cambridge University Press.

63.

Nelles

(2006) Bosnian education for security and peacebuilding? International Peacekeeping 13(2): 229–241.

64.

Oberschall

(2000) The manipulation of ethnicity: From ethnic cooperation to violence and war in Yugoslavia. Ethnic and Racial Studies 23(6): 982–1001.

65.

Olweus

(1997) Bully/victim problems among schoolchildren: Facts and intervention. European Journal of Psychology and Education 12(4): 495–510.

66.

Parent

(2012) Identifying factors promoting or obstructing healing and reconciliation: Observations from an exploratory research field in ex-Yugoslavia. International Journal of Peace Studies 17(1): 25–45.

67.

Quillian

(1995) Prejudice as a response to perceived group threat: Population composition of anti-immigrant and racial prejudice in Europe. American Sociological Review 60(4): 586–611.

68.

Quist

Resendez

(2002) Social dominance threat: Examining social dominance theory’s explanation of prejudice as legitimizing myths. Basic and Applied Social Psychology 24(4): 287–293.

69.

Raudenbush

Bryk

(2002) Hierarchical Linear Models: Applications and Data Analysis Methods. New York: Sage Publications.

70.

Reardon

Firebaugh

(2002) Measures of multigroup segregation. Sociological Methodology 32(1): 33–67.

71.

Reynal-Querol

(2002) Ethnicity, political systems and civil wars. Journal of Conflict Resolution 46(1): 29–54.

72.

Riek

Mania

Gaertner

(2006) Intergroup threat and outgroup attitudes: A meta-analytic review. Personality and Social Psychology Review 10: 336–353.

73.

Risher

Kabil

(2010) Child Friendly Schools. Case Study: Bosnia and Herzegovina. New York: UNICEF, Education Section.

74.

Salomon

(2004) Does peace education make a difference in the context of an intractable conflict? Peace and Conflict: Journal of Peace Psychology 10(3): 37–48.

75.

Sekulić

Massey

Hodson

(2006) Ethnic intolerance and ethnic conflict in the dissolution of Yugoslavia. Ethnic and Racial Studies 29(5): 797–827.

76.

Semyonov

Gorodzeisky

Glikman

(2012) Neighborhood ethnic composition and resident perceptions of safety in European countries. Social Problems 59(1): 117–135.

77.

Shamir

Sagiv-Schifter

(2006) Conflict, identity, and tolerance: Israel in the Al-Aqsa Intifada. Political Psychology 27(4): 569–595.

78.

Simpson

(1949) Measurement of diversity. Nature 163(4148): 688.

79.

Smith

(2010) The influence of education on conflict and peace building. Paper commissioned for the Education for All Global Monitoring Report 2011, The hidden crisis: Armed conflict and education. Paris: UNESCO. URL (accessed 22 January 2021): http://unesdoc.unesco.org/images/0019/001913/191341e.pdf.

80.

Smith-Khuri

Iachan

Scheidt

, et al. (2004) A cross-national study of violence-related behavior in adolescents. Archives of Pediatrics and Adolescent Medicine 158: 539–544.

81.

Stephan

Ybarra

Rios Morrison

(2009) Intergroup threat theory. In: Nelson

(ed.) Handbook of Prejudice, Stereotyping, and Discrimination. New York: Psychology Press.

82.

Stronks

Kulu-Glasgow

Agyemang

(2009) The utility of ‘country of birth’ for the classification of ethnic groups in health research: The Dutch experience. Ethnicity & Health 14(3): 255–269.

83.

Szasz

(1997) The Dayton Accord: The Balkan peace agreement. Cornell International Law Journal 30: 759–768.

84.

Theil

(1972) Statistical Decomposition Analysis; With Applications in the Social and Administrative Sciences. Amsterdam: North-Holland Publishers.

85.

Tortsi

(2009) Segregated education and texts: A challenge to peace in Bosnia and Herzegovina. International Journal on World Peace XXVI(2): 65–82.

86.

Trbovich

(2008) A Legal Geography of Yugoslavia’s Disintegration. Oxford: Oxford University Press.

87.

Tveit

Cameron

Kovač

(2014) ‘Two schools under one roof’ in Bosnia and Herzegovina: Exploring the challenges of group identity and deliberative values among Bosniak and Croat students. International Journal of Educational Research 66: 103–112.

88.

UNICEF (2009) Divided Schools in Bosnia and Herzegovina. Sarajevo: UNICEF BiH.

89.

U.S. Department of State (2015) Dayton Accords. URL (accessed 22 January 2021): https://2009-2017.state.gov/p/eur/rls/or/dayton//index.htm.

90.

Van Dick

Van Knippenberg

Hägele

, et al. (2008) Group diversity and group identification: The moderating role of diversity beliefs. Human Relations 61(10): 1463–1492.

91.

Van Knippenberg

Schippers

(2007) Work group diversity. Annual Review of Psychology 58: 515–541.

92.

Weinstein

Freedman

Hughson

(2007) School voices: Challenges facing education after identity-based conflicts. Education, Citizenship and Social Justice 2: 41–71.

93.

Williams

O’Reilly

(1998) Demography and diversity in organizations: A review of 40 years of research. Research in Organizational Behavior 20: 77–140.

94.

Wood

(2008) The social processes of civil war: The wartime transformation of social networks. Annual Review of Political Science 11: 539–561.

95.

Zimmerman

Rees

(2014) Do school disciplinary policies have positive social impacts? Examining the attenuating effects of school policies on the relationship between personal and peer delinquency. Journal of Criminal Justice 42(1): 54–65.