Behind the frontlines: Identity, competition, and violence in civil wars

Recent research on violence against civilians during wars has emphasized war-related factors (such as territorial control or the characteristics of armed groups) over political ones (such as ideological polarization or political competition). This dissertation distinguishes between direct and indirect violence and presents a theory of the determinants of these two types of violence for conventional civil wars--where armed groups and civilians face a structure of incentives that broadly differs from that in civil wars fought via irregular warfare. In contrast to much of the literature, the dissertation focuses on political factors as well as those set in motion by war. I generate a set of testable implications for conventional wars and suggest that some also occur in other civil war contexts. Specifically, political variables are hypothesized to have a key explanatory role on the perpetration of violence at the local level, and to affect direct and indirect violence in diverging ways.;To test the observable implications of the theory, I draw on archival and historiographic sources to construct a new dataset of victims of lethal violence, pre-war elections results, and geographical and socioeconomic variables in 2,700 municipalities during the Spanish Civil War (1936-1939). 1 show that direct violence against civilians goes up where prewar electoral competition between rival political factions approaches parity; this finding is consistent across armed groups and zones. Using data on bombings by Fascist forces on 1,062 municipalities of Catalonia, I also show that lethal indirect violence is negatively and monotonically associated with levels of prewar support for the group.;Additionally, following the first round of violence, war-related factors are found to gain explanatory relevance at the expense of prewar political variables: on the one hand, there is a clear endogenous trend whereby, at the local level, subsequent levels of direct violence perpetrated by one group are highly correlated with initial levels of direct violence perpetrated by the rival group; on the other hand, the likelihood of bombings is positively affected by executions perpetrated by the rival armed group in a municipality during an earlier period. Overall, the findings indicate that understanding the determinants of violence requires a theory that combines the effects of political cleavages and wartime dynamics.;At a secondary level, this dissertation inductively explores the consequences of civil war victimization on the political identities of individuals. Using qualitative and survey data from Spain, I observe that victimization broadly generates rejection toward the identity of the perpetrating group. Identities endogenous to the war seem to be transmitted across generations. In addition, in the long run, victimization is found to have a negative effect on political participation.;This thesis contributes to the understanding of the relationship between political identities and violence during civil conflict: while identities are explanatory of wartime violence, they are also affected by them. In other words, political identities both exogenous and endogenous to wartime dynamics.