All Peacekeeping is Local: Measuring Subnational Variation in Peacekeeping Effectiveness

Abstract

Understanding whether peacekeepers reduce fatalities at the local level is an important question. We can have increased confidence in peacekeepers’ capabilities by testing whether deaths decrease in the locations where peacekeepers are present. However, commonly used modeling techniques cannot easily test peacekeepers’ local effectiveness. Coefficients from methods such as linear regression, logit, and count models provide average estimates of peacekeepers’ effects on violence. We argue that a solution lies in geographically weighted regression (GWR). GWR can more clearly reveal subnational spatial heterogeneity in peacekeepers’ (in)effectiveness at reducing violence. We conduct an illustrative test of our argument using data on the Democratic Republic of the Congo between 2001 and 2014 and replicate an existing study to show that GWR can also help resolve seemingly contradictory findings of whether peacekeepers are better at reducing violence by government or rebel actors. The article contains important implications for how scholars can more accurately measure peacekeeping effectiveness.