Students of civil conflict have long identified the role of geographic factors that make rebellion more feasible and thus more likely (Fearon and Laitin, 1999;2003). While the underlying idea remains in the focus of current research, the operationalization of geographic indicators of civil conflict has not gained much momentum in the past decade. Contemporary studies of civil war onset still employ highly aggregated measurements of “mountainous terrain” as explanatory variables. This study offers a much refined analysis both in terms of agency as well as geography. Various terrain conditions, types of land-use, and more realistic estimates for center-periphery distances are used to predict which ethnic groups are likely to be involved in civil war. The results are helpful in terms of narrowing down the set of geography-related causal mechanisms at work that are otherwise meshed into country-level indicators. Since not all types of terrain “roughness” affect the probability of civil war onset alike this high-resolution GIS analysis allows for a better understanding of how geography matters.
