Spatial-temporal modeling of civil war: The example of Bosnia
Nils B. Weidmann (ETH Zurich) and Michael D. Ward (University of
Washington)
Abstract: We describe a spatially and temporally autoregressive discrete regression model, detail and implement an algorithm for estimating the parameters of such a model, following the framework of Geyer and Thompson (1992), as recently updated by Zhu, Zheng, Carroll and Aukema (2008); Zhu, Rasmussen, M ller, Aukema and Ra a (2008); Zheng and Zhu (2008). This model is applied to geo-located data on attributes and conflict events in Bosnia over the period from March 1992 through October 1995. We present an R program library to estimate this class of models. Results show that there is a strong spatial as well as temporal dimension to the outbreak of civil conflicts in Bosnia, dynamics in space and time that standard OLS-type implementations completely miss. Using this approach it is no longer necessary to assume that either the spatial or the temporal dependencies in conflict data are exogenous in order to create predictive and inferential models of civil conflicts. Substantively, we show via inference, simulation, and animations that conflict events in Bosnia did di ffuse spatially as well as temporally.