This thesis examines the relationship between rebel organisation and violence against civilians in sub-Saharan Africa. I argue that rebels use such one-sided violence to enforce popular support when they are unable to secure support otherwise. An important determinant of this strategic use of violence is the rebels' organisational configuration. Organisational factors such as the ideology of a rebel group, the occurrence of leadership divisions, the level of fractionalisation within a rebel group's population base, the existence of external support and the number of competing rebel factions determine whether non-violent strategies to secure support are available or whether rebels can only rely on violent means to enforce support. I test this theoretical model using both quantitative and qualitative methods. First, I conduct a statistical cross-sectional study analysing the relationship between rebel organisation and rebel one-sided violence in sub-Sahara Africa between 1989 and 2007. The analysis reveals that the occurrence of leadership divisions increases the probability of an onset of rebel violence against civilians. I also find that ethno-nationalist rebel groups kill fewer civilians while rebel groups who draw support from highly fractionalised population bases kill more civilians. Second, I use qualitative within-case analyses of the Sudan People's Liberation Movement/Army (SPLM/A) and the Sudan Liberation Movement/Army (SLM/A) to test whether these correlations are actually driven by the causal pathways outlined in my theoretical model. While both case studies provide supporting evidence for this, they also uncover that the causal pathways linking leadership divisions to rebel violence can substantially differ from each other. Moreover, the qualitative analysis reveals that the theoretical model only partly captures the causal pathways between rebel ideology and the rebel groups' population base and the level of one-sided violence.

Violence, Atrocity and the Behaviour of Pro-Government Armed Groups Bronia Naorni Flett ABSTRACT This thesis explores the existence and behaviour of the under-researched and under- theorised pro-government armed group (PGAG). The core research question is: Why do POAOs behave violently? I define a PGAG as a group that is armed, organised and connected to government. Detailed evidence on PGAGs is limited and typically collected on a case-by-case basis by historians, anthropologists and journalists. The pro-government militia database (www.abdn.ac.uk/rnilitias) is the first project to collect evidence on these groups in a systematic manner. The database finds POAOs existing and operating in diverse environments; they are frequently violent and commit human rights abuses. The empirical contribution of this thesis is to collect further detailed evidence on the behaviour of PGAGs in three diverse cases: the notorious militias in Yugoslavia, the under-researched peasant militias in Peru and the well-regarded International Brigades in Spain.