THE SPATIALITY OF CONFLICT IN THE LAKE CHAD REGION A Log-Gaussian Cox Process Approach to Modelling Spatial Patterns of Violence across Conflict Actor Dyads

Most studies treat space and conflict actors as static, overlooking the dynamism both within and between them. This research aims to challenge that perspective by investigating whether actor interactions have an impact on the spatial distribution of conflict events. This thesis thus questions where spatial patterns of conflict differ by actor dyad, and what these variations may reveal about underlying spatial clustering and diffusion mechanisms. As the latter tend to be unobserved, they are often overlooked, which limits our understanding of how violence spreads. To capture these latent spatial effects a Log-Gaussian Cox Process model is employed, which measures conflict diffusion over continuous space. Taking the dyad-level as the unit of analysis allows for a nuanced analysis of the spatial dynamics attributed to different actor constellations. Findings reveal that the same actor can behave differently depending on who they engage with, and this has implications for spatial dynamics. Notably, violence from rebel groups targeting civilians tends to be highly localised; whereas when the same rebel groups engage with state forces conflict is diffuse. These results underscore the importance of actor-specific interactions in shaping the geography of violence.