Thesis (Ph. D.)--University of Rochester. Dept. of Political Science, 2013.
The dissertation asks the question: do states behave differently when they are being observed? It expands upon the theoretical literature on international conflict by examining the way in which observation by hostile third parties affects crisis bargaining between states. The dissertation consists of three essays that combine theoretical and empirical components to analyze the key research question. Essay one presents a formal theoretical model of state behavior under observation. It shows that states tend to bargain harder when they believe that they are being observed, in order to keep potential future opponents unsure about their strength. This has the effect of increasing the likelihood of war. I elucidate the theoretical mechanisms with an application to the 1962 Sino-Indian War. Essay two uses a new estimator, the local logit, to assess the empirical implications of the previous chapter statistically. It shows that the probability that a conflict escalates is increasing in the likelihood of observation when the attacker's costs for fighting are moderate or high, and that once information is revealed by war, subsequent conflicts become less likely. Finally, essay three applies the framework to mediation both theoretically and empirically. It demonstrates that a mediator can solve the problems uncovered in the first two essays by preventing the dissemination of information. The empirical results suggest that dyads that are more likely to be observed are less likely to experience mediation, but are more likely to resolve the conflict with at least one state making concessions when mediation does occur.