Estimating Directionality from Event Data

• In Event: Advances in Probabilistic Network Data Analysis

Abstract

Today civil conflict is characterized by multiple warring parties with different motivations, capabilities, and patterns of behavior. Current research on intrastate conflict focuses heavily on questions related to the military behavior of these armed actors and battlefield dynamics---such as in studies on the efficacy of counterinsurgency campaigns, the causes of civilian victimization, or conflict duration. We argue that despite this focus on battle level processes, political science lacks a clear, generalizable measure of offensive and defensive battlefield behavior to capture which actors initiate conflict at the micro-level. Although significant advances have been made in the collection of event data on intrastate conflict, currently available event data does not record information about which actor initiates a clash. Instead, these data only record which two (or more) actors engage in a battle at any given time. We present a novel latent variable framework that parses directionality from event data via a network decomposition approach. We apply this approach to the Nigerian Civil conflict to investigate the relationship between directionality and civilian victimization. We run our model across all conflict based events recorded in Nigeria from 1997 to 2019 and evaluate how civilian victimization (in terms of both intensity and tactics) changes as a function of whether or not the government is on the defensive or offensive.

Authors

• Cassy Dorff, Vanderbilt University

• Shahryar Minhas, Michigan State University