Individual and Contextual Influences of War on Ethnic Prejudice in Croatia

Abstract

This article examines what the determinants of ethnic prejudice in Croatia were in the aftermath of the 1991–1995 war. The analysis is based on a nationwide survey (N = 2,202) conducted in March and April 1996, less than a year after the cessation of war activities in Croatia. The main focus of our analysis is on how war influences the ethnic prejudice of individuals. The influences of individual war-related experiences and the effects of regional differences in the level of war activities are analyzed simultaneously by conducting a multilevel analysis. The main findings are that individual war-related experiences have little impact on prejudice, but that the contextual influence of war is somewhat stronger. Variables that are not directly related to the conflict—such as education, religiosity and size of the place of residence—have the strongest effect on prejudice. Insofar as our results can be generalized to other large-scale ethnic conflicts, they indicate that a recent history of conflict is not in itself a major hindrance to the process of ethnic reconciliation.

NOTES

Notes

¹ Sources: 1991 data: Woodward (1995:33); 2001 data: census data from Croatian Bureau of Statistics.

² Forbes (1997) has argued convincingly that the main focus in the studies of ethnic conflict should be on group-level causes of conflict, and not on individual-level prejudices and antipathies. Nevertheless, we find it quite obvious that high levels of ethnic hatred can increase the likelihood of ethnic conflict.

³ See, for example, the differences between Quillian (1995), Pedersen (1996), and Alexseev (2003).

⁴ For a short overview of early research, see Noel and Pinkney (1964). For more recent results, see Quillian (1995), Evans and Need (2002), and Semyonov et al. (2004).

⁵ There is a certain amount of confusion with regards to the administrative borders of opcine in those articles that are based on the same set of data (Kunovich and Hodson 1999; Massey et al. 2003) The Croatian government introduced reforms to the municipality system, resulting in a considerable increase in the number of municipalities, to around 350. However, in this survey, the old Yugoslavian definition of opcina is used, and the number of municipalities reported in Kunovich and Hodson (1999:651) and Massey et al. (2003:64) is not correct.

⁶ Questions 156 to 189 in the question list in Appendix 1.

⁷ If included in the models as separate categories, both pensioners and housewives have nonsignificant positive effects.

⁸ The full list of coding of municipalities is presented in Appendix 2.

⁹ Because large-scale disturbances in the economic and social situation caused by the war, it is probably best to use prewar data. We do not have the necessary data from the last prewar census in 1991 and rely on data from the 1981 census.

¹⁰ In practice it means being reported as a job seeker at the local labor office.

¹¹ Rasbash et al. (2003) employ the term “Variance Partition Coefficient” instead of ICC.

¹² Testing the significance of variance components by using estimated standard errors and z- or t-tests is somewhat problematic because the sampling distribution of the variance estimates is skewed to an unknown degree (Raudenbush and Bryk 2002:55; Rasbash et al. 2003) recommend the use of likelihood ratio test as a preferred test. We will thus perform the latter test both when discussing the goodness of fit of the models and the significance of the variance components.

¹³ It is also possible that some of the authors of previous studies have not performed a test for the possible nonlinear effects of age. Exclusion of the second-order term of age variable in our models results in a nonsignificant linear effect of age.

¹⁴ Nevertheless, Kunovich, and Hodson (2002) find negative effects of media exposure on ethnic prejudice.

¹⁵ For an interesting and comprehensive study of this topic, see Massey et al. (1999).

¹⁶ The reduction in deviance of 17.11, chi-square distributed with the five degrees of freedom

¹⁷ The variance of level-2 residuals is reduced from around 34 in the empty model to around 24 in model 1.

¹⁸ Real sample size, before weighting.

¹⁹ The exact number of individuals who moved, voluntarily or involuntarily, because of war activities in the areas where they lived is not easy to estimate. Ramet (2002:69) reported that more than half a million Croatian residents had been driven from their homes by the end of 1991.

²⁰ This is not to say that there are no data at all. Scattered surveys do exist, most often focusing on mental health in the aftermath of war (e.g., de Jong et al. 2002; Scholte, Olff, and Ventevogel 2004).