Published online by Cambridge University Press: 25 May 2012
The research problem driving this paper is the absence of a strong theory that accounts for variation among cases that have similar probabilities of escalating to genocide and similar forms of organized (usually state-led) mass violence against civilians. Much of the existing theory on genocide focuses on explaining under what conditions and by what processes regimes commit large-scale violence against civilians. I argue that a critical missing dimension to studies of genocide, but also more generally to the study of political violence, is a methodological recognition of negative cases and a theoretical recognition of the dynamics of restraint that helps to explain such negative cases. That is, in addition to asking what causes leaders to choose to escalate violence, I argue that scholars should emphasize conditions that prompt moderation, de-escalation, or non-escalation. I propose an alternative framework for how to conceptualize the process of political violence and review the literature to identify key restraint mechanisms at micro, meso, and macro levels of analysis. I further articulate a provisional theory of genocide using this new analytical framework. I illustrate my argument with an empirical analysis of mass violence cases in Sub-Saharan Africa since independence, and with a more in-depth analysis of comparable crises in Rwanda and Côte d'Ivoire, where the trajectories of violence differed significantly. While this paper draws on extensive empirical research, my primary purpose is not to advance a developed new theory or to test particular hypotheses, but rather to outline a research agenda that promises to draw from and contribute to recent work on the comparative politics of violence.