Three essays on international conflict

This thesis is composed of three related essays on international conflict. In the first essay, I examine the link between domestic regime type and two different sources of uncertainty often argued to be prevalent in international conflict: private information about states' preferences over particular outcomes, and suboptimal foreign policy choices by states due to implementation error. I demonstrate that democratic leaders have a greater tendency to make mistakes than non-democratic leaders in international crises. I also find that democracies possess more private information than more authoritarian regimes in a crisis.; The second essay examines heteroskedasticity and non-random sampling as two potentially important sources of bias and incorrect inference for international relations scholars that aim to analyze strategic interaction in their research. The main focus of this essay is the theoretical sources of heteroskedasticity in statistical strategic models of Signorino (2003) and the performance of strategic estimators in dealing with data generated from a form of non-random sampling mechanism called choice-based sampling. With a series of Monte Carlo experiments, the danger of bias and inconsistency in estimates are evaluated. An empirical example is presented for both cases to show the merits of the tools developed.; The third essay focuses on the link between military spending and economic growth, one of the long-lasting empirical debates of political science. I argue that there is a nonlinear indirect relationship between military spending and economic growth, which operates through investment. Very low and very high levels of military spending reduce investment levels, which results in an inverse U-shaped relationship. Previous empirical research for the most part assumed a linear functional form for this relationship, and this essay relaxes this assumption by using a nonparametric approach in estimating a simultaneous equations model on a sample of 125 countries. The results show that military spending affects economic growth only through investment, and this effect is nonlinear. For low values of military spending, an increase in spending has an indirect positive effect on growth. In contrast, the effect is negative for moderate values of military spending.