Population Attitudes and the Spread of Political Violence in Sub-Saharan Africa

Abstract

One of the most powerful predictors of violent political conflict is proximate violence in space and time. This spatiotemporal pattern has been identified between countries as well as within them. What explains this clustering is less clear, and different studies point to different mechanisms. Focusing on sub-Saharan African states, we examine whether population attitudes may contribute to the spread of political violence at subnational scales. In a quasi-experimental research design—using georeferenced survey data of 18,508 respondents for 162 administrative units across 16 countries, paired with precisely georeferenced conflict event data—we find that popular acceptance of (the legitimacy of) the use of physical violence is positively associated with subsequent conflict events. Furthermore, the combined effect of nearby violence and approval of violence is stronger than either condition alone, implying a diffusion effect. While we find some evidence that conflict events affect later public opinion, our final models control for violence that occurred before the survey data were gathered. The fact that we include such violence in our analysis suggests that the reported results cannot be dismissed as merely reflecting a reverse causal relationship.