Critical Configurations : Settlement Patterns and Ethnic Violence

Abstract

Quantitative research on ethnic conflict has tended to neglect causal mechanisms. By primarily resorting to macro-level indicators and tests, existing studies fail to give a sufficiently precise account of the actors and their interactions that ultimately lead to conflict. This dissertation shows that a detailed analysis of ethnic settlement patterns can help us get closer to the group processes during conflict. With the help of Geographic Information Systems (GIS), computational modeling and spatial statistics, the analysis presented here provides new insights into the role of group geography as a determinant for conflict. The dissertation is organized around three core questions. The first focuses on ethnic groups as the level of analysis and examines how geographic concentration influences the risk of a group being involved in conflict. Two competing explanations exist: a motivation-driven mechanism which claims that concentrated groups are more likely to fight for their territory, and an opportunity-driven mechanism in which concentrated groups are more prone to collective mobilization because of facilitated social interaction. However, no systematic attempt has been made to test the explanatory impact of these mechanisms. Using detailed GIS data on group settlement regions, my analysis shows that the relationship of concentration and conflict hinges on the increased interaction opportunities in concentrated groups. With the help of a spatial computational model, this dissertation aims to provide a more detailed account of group mobilization in a spatial context. The model simulates the spread of mobilization through group populations, relying on the assumption that direct social interaction is important for an individual’s decision to participate. Consequently, the mobilization influence from one location to the other is proportional to the number of people traveling between these locations. The model computes an indicator for the proportion of a group that can be mobilized given the parameters of the model. Groups with high mobilization values should thus be more susceptible for conflict, a hypothesis which is confirmed in a regression analysis. The second core question of the dissertation increases the spatial resolution and examines how group geography affects the location of violence. By employing high-resolution geographic data on ethnicity and spatially linked reports of violent events from Bosnia, this dissertation sheds new light on the dynamics of civil wars. I show that the application of violence is driven by territorial ambitions of groups. During conflict, groups aim to create and consolidate larger areas of ethnic homogeneity. Violence is likely at a location if more than one group considers this location to be part of its territory. The analysis provides a detailed perspective on different types of ethnic violence by distinguishing between two-sided violence between the groups’ armed factions, and one-sided violence against civilians, both of which seem to be strongly influenced by territorial ambitions. The literature on group settlement patterns and conflict has treated group geography as an exogenous determinant of conflict. However, research on international conflicts tells us that territory is as much a cause of conflict as it is a consequence. The third core question applies this logic to an intrastate setting and analyzes changes in group geography that occur as a result of conflict. Comparing settlement patterns in Bosnia before and after the war, the results show that changes in the territorial configuration occur in a strategic way, such that they reduce the territorial contestation between groups. However, my analysis can only partly explain how the occurrence of violence on the ground relates to these systemic changes. Whereas the occurrence of military confrontations seems to inhibit large changes towards ethnic homogeneity, there is no direct relationship between the targeting of civilians and ethnic unmixing.