The Dynamics of Violence in Civil War: Evaluating the Impact of Ethnicity on Violence
Stathis Kalyvas, Political Science, New York University
Research Grant, 2000, 2001
My goal in this project was to investigate the assumption that ethnic conflict is associated with higher levels of violence compared to other forms of conflict, such as revolutionary wars. More specifically, I sought to evaluate whether and how ethnic violence in the context of a civil war is really different from nonethnic violence and estimate to what extent ethnic violence is really ethnic; for, ethnic violence is more than violence between individuals of different ethnicities.
The main obstacle in answering this question is methodological: comparing ethnic and nonethnic wars does not allow us to control for the multitude of factors that may be causing whatever differences we observe. In addition, it would be difficult to move from correlations to causal connections.
As a result, I chose to begin with a carefully designed paired comparison at the micro level, as a first step toward a broader investigation to be undertaken in the future. Having completed a microcomporative study of violence in about seventy villages of Southern Greece (the Argolid region) where the civil war took place within a homogeneous ethnic environment, I used the support of the Harry Frank Guggenheim Foundation to conduct a study of a region in Northern Greece that was similar in many respects to the southern one, but was ethnically diverse and polarized (the Almopia region). Fieldwork involving interviews and archival research was conducted in 2000 and 2001.
Insofar as I was able to isolate an independent effect of ethnicity on violence, I found that the two were inversely related: the presence of the ethnic cleavage tended to reduce the levels of violence, by acting as a deterrent under conditions of contested control.
The goal was to find out which of the following three outcomes were supported by the data: ethnic polarization causes (a) higher levels of violence, (b) lower levels of violence, or (c) has no effect. The results support strongly the third hypothesis, while also suggesting some support for the second one. The first hypothesis was rejected. It is important to emphasize here that these findings cannot be generalized to the universe of conflict, but constitute a first step toward more rigorous empirical and theoretical work.
More specifically, I found that violence was considerably less widespread in the ethnically heterogeneous northern region than in the ethnically homogeneous southern one. This cannot be accounted for by patterns of group interaction: ethnic isolation and animosity were real; however, they did not lead to mass violence. In fact, the ethnic cleavage appears to have had little independent effect on the dynamics of violence during the civil war. Factors related to the conduct of warfare (the resources of the rival sides, the level of militarization of the conflict, and the incumbents’ ability to move the population away from mountainous terrain and close to well-defended urban centers) appear to account for the relative absence of lethal violence. Insofar as I was able to isolate an independent effect of ethnicity on violence, I found that the two were inversely related: the presence of the ethnic cleavage tended to reduce the levels of violence, by acting as a deterrent under conditions of contested control.
My research suggests that the observed or perceived relation between ethnic polarization and violence may be the result of a selection bias; as we continue to study ethnic violence, we should take much more seriously into account overlooked factors such as the type of warfare and the resources of the combatants in addition to the nature of the social and political divisions, of which ethnicity is only one.