Trials and Triangulations: Aggregation Problems in Event Data on Political Violence

Abstract

Triangulation constitutes a key approach to analyzing data on political violence, which is sensitive and has many reporting biases. Nevertheless, the triangulation process whereby researchers move aggregate the underlying sources to produce event datasets on political violence remains under-theorized and under-documented. Building on recent calls to improve data transparency by collecting political violence data at the event report level and relying on systematic and replicable aggregation procedures, we discuss both benefits and challenges of this approach. We argue that aggregation procedures are more complicated than often assumed, and examine how both theoretical, empirical, and pragmatic considerations affect the choice of aggregation approach. Drawing on novel event report level data from the Modes and Agents of Election-Related Violence in Côte d’Ivoire and Kenya (MAVERICK) dataset, we further present descriptive evidence and two illustrative regression analyses that examine how sensitive descriptive and correlational inferences can be to aggregation choices.