“…These areas are somewhat smaller than the units of analysis used in some of the previous research using the discrete choice approach (e.g., Bernasco and Nieuwbeerta, ; Clare, Fernandez, and Morgan, ) and indeed are smaller than many previous empirical studies of rioting, which often consider spatial distributions of rioting at a national level (Myers, , ; Olzak and Shanahan, ; Spilerman, , , ). Distributions at finer scales of analysis have shown the complexities inherent in the dynamics of riots and often can seem to raise more questions (e.g., Abudu Stark et al, ; Bohstedt and Williams, ); however, this approach enables us to test the consistency with which targets are chosen during rioting. The advantage of smaller sized units of analysis in the discrete choice approach is that the explanatory variables are more representative of the population and characteristics of each area…”