The Network Structure of Conflict Actors and its Effect on Violence against Civilians: The Case of the Conflict in Somalia

​We observe striking differences in the frequency and intensity of civilian targeting. Theories explaining such a variation often do not take into consideration the type and intensity of interactions between the complex array of conflict actors. Although conflicts often consist of many interlinked actors competing for power, resources and support from the local population, current literature tends to oversimplify actors’ fragmentation by focusing on conflict dyads or counts of actors involved. To overcome this problem, I utilise network analysis combined with GIS. Based on the literature on actors’ fragmentation and violence against civilians, I formulate and test two complementary explanations. First, state and non-state actors target civilians to demonstrate their power especially when they are weaker than other conflict actors. Second, violence against civilians is used to control population whose support is competed for with other armed groups. To operationalize the actors’ relative weakness and the intensity of competition they are involved in, I construct network centrality measures combined with geographical factors capturing the dynamic entanglement of conflict actors. I test my propositions on the conflict in Somalia characterized by the complex involvement of various actors. External validity tests are carried out for the conflicts in Colombia and Nigeria.