Ethnic fractionalization and Sub-Saharan violence, 1970-1996.

Abstract

This study examines the statistical correlations between metrics of ethnic fractionalization and categories of violence in Sub-Saharan Africa from 1970 to 1995. By examining these correlations both prior to and after controlling for income, the study is able to determine whether or not various types of conflict are linked to patterns of ethnic grouping. The study uses newer, more refined measures to evaluate the correlations between specific categories of violence and specific measures of ethnic fractionalization. Using simple and multivariate linear regressions, the study examines each of the correlations between a total of twenty-two sub-metrics of four categories of violence, per capita income and metrics of ethnic fractionalization on three tiers. This allows the study to gauge the impacts (both separately and in interaction) of dichotomous top-tier cleavages in deeply divided societies, general ethnic fractionalization and nested ethnic sub-grouping. The study finds that the majority of the categories of violence used are not correlated with ethnic fractionalization, neither prior to nor after controlling for income.

(cont.) However, successful coups are highly correlated with the presence and distribution of the dichotomous top-tier cleavages in deeply divided societies, with the frequency of these successful coups increasing linearly as the divide approaches a 50/50 split. The frequency of riots is robustly correlated with general ethnic fractionalization, but that ethnic fractionalization has much less predictive power for this correlation. An implication of these findings for future research is that using these refined definitions allows for a fuller understanding of the behavior being examined: all categories of conflict are not uniformly correlated with all measures of ethnic fractionalization, and vice versa. Hypotheses regarding these correlations must thus enter into detail, both about the claims being made and the groups for which they are made. Additionally, the findings show that income's correlation with conflict and ethnic fractionalization is more complex than has been indicated in much of political science literature, and that further research on the topic is merited.