Replication data for: A Theory of Indiscriminate Violence

This dissertation addresses a simple puzzle: why do governments use indiscriminate violence against civilians? To deter a population from rebelling, a government should make rebellion costlier than the alternatives. Yet indiscriminate violence can make neutrality costlier than rebellion. With the help of mathematical modeling, archival data and micro-comparative evidence from dozens of armed conflicts, I show that indiscriminate violence makes civilians less likely to remain neutral, but not necessarily more likely to support the opponent. There is a threshold level of violence, beyond which it can become safer for civilians to cooperate with the more indiscriminate side. As long as civilians believe that supporting the rebels will be costlier than supporting the government, they will generally support or not actively resist the government -- even if the government is responsible for more civilian deaths overall. The amount of violence needed to meet this threshold depends on the combatants' relative informational endowments. If a combatant can selectively punish her opponents, she can employ a relatively low level of violence. Where she lacks the information for selective punishment, she will use methods more indiscriminate in targeting and more massive in scale. Violence is a substitute for intelligence.