Leaders, States, and Reputations:

Reputational incentives are ubiquitous explanations for war, yet consistent evidence of their effects is elusive for two reasons. First, most work searches for the payment of reputational costs, yet strategic censoring systematically biases observational data against revealing them. Second, the locus of reputation is often ambiguous, yet the choice of leader or state as unit of observation has inferential consequences. Our research design (a) focuses on observable implications of reputational theories in appropriate samples and (b) considers two competing sources of reputational incentives: changes in national leaders and in political institutions. Consistent with our expectations, leadership turnover and regime change are each associated with initially high probabilities that militarized disputes escalate to the use of force before declining over time in the presence of a reasonable expectation of future disputes. Reputations are in evidence, but analysts must look for them in the right place.