In Chapter 1, I propose two contrasting models of collective killings. The state-policy model traces the source of killings to a state's policy of elimination and expects the policy to be carried out by state organizations and military personnel. My community model, conversely, suggests that collective killings could occur absent a genocidal state policy, and local conditions could turn conflict into atrocities, perpetrated by civilian killers. In Chapter 1, I also present a series of factors based on aspects of state mobilization and state breakdown, which I argued would help promote killings at the local level.

In Chapters 2 through 7, I present extensive evidence supporting the community model and casting doubt on the state-policy model. In 1967 and 1968, there was no central Chinese policy calling for the wholesale destruction of any segment of the population. Neither were the collective killings committed by professional execution squads coordinated by a central command. There are vast variations in the collective killings across provinces, counties, townships, and villages. The state-policy model is generally concerned only with national-level atrocities and conditions, but these patterns of killings call for serious attention to local conditions to explain them. Exploring the implications of these variations is the task I take up in this chapter. I address why some communities had extensive collective killings while similar communities did not – or, to put it differently, which factors account for the cross-unit variations?