The model presented here will clarify two puzzling aspects of the Salvadoran rural insurgency. First, I show how the reasons for action central to my interpretation account for insurgent collective action, even when the primary material benefits sought by these organizations (access to land) were not contingent on participation. Second, I also want to illuminate how radical forms of campesino collective action, which owed their emergence to extraordinary circumstances – the terror of the late '70s and early '80s and the military stalemate of the mid-'80s – could have persisted after those circumstances had passed.

The formal model presented here draws on the rational actor approach in that individuals decide whether to participate or not based on anticipated costs and benefits, which in this case depend on how many others participate. I draw on Thomas Schelling's (1978) representation of collective action as an n-person coordination game among individuals facing a binary choice to participate or not. It differs from most such models, however, in that the likely outcomes of participation are not evaluated in terms of conventional self-regarding and outcome-oriented preferences. Rather, some individuals value defiance, an intrinsic motivation, and/or pleasure in agency, one contingent on both participating oneself and the action's success.

Moreover, Schelling's tipping model leaves unresolved how a group of individuals move from general nonparticipation to general participation. To model that process, I draw on the work of David Lewis (1969) and of Peyton Young (1996, 1998) on conventions.