| In this dissertation I propose and test a theory of intra-ethnic electoral violence. I argue that intimidation and violence can be used by politicians to control the electoral support of coethnics, specifically, by demobilizing coethnic opposition candidates and by coercing the support of coethnic voters. Two other important theoretical pieces flow from this argument: that violence may be used to mobilize participation (as opposed to depress turnout), and that violence escalates with an increase in size and internal political polarization of the coethnic group.;I evaluate this mechanism by analyzing sub-national patterns of violence in Burundi, a country divided between ethnic groups. I first show that the violence preceding the 2010 elections occurred, in contrast, within these groups, as political parties used violence to signal their strength and their ability to protect supporters, punish political defectors, and coerce political opponents within the same ethnic group. In order to identify the precise conditions that drive violence, I built an original dataset with information on violent acts, socioeconomic factors, public services, and parties' armed capabilities, for each municipality in the country. I show that violence spiraled between the largest parties competing for control of the same ethnic group: was greatest when the majority ethnic group was larger and politically polarized.;Since my argument is about employing violence to boost participation, I also test the impact of violence on voting behavior. I code the perception that voting was influenced by intimidation---through a questionnaire to more than two thousand local election monitors in every municipality---then analyze the sub-national electoral results. My analysis shows that intimidation-driven voting and actual turnout rates were associated with more violence, especially in localities where the incumbent relied on strongmen and therefore was perceived to be a powerful coercive actor.;These results are bolstered by extensive interviews with all political sides and non-governmental organizations. To capture the nuances of political messages, I collected and translated recordings of political-party rallies, TV interviews, and campaign songs. I finally demonstrate that my argument possesses external validity by showing that electoral violence in Zimbabwe and Sri Lanka followed the same pattern as in Burundi. |
