Essays in comparative political institutions. An adverse selection model of terrorism: Theory and evidence. Constituency service, the incumbency advantage, and divided delegations in multi-member and single-member districts. Informative precedent and

In the first essay I argue that when governments make concessions, only moderate terrorists accept, leaving extremists in control. Nonetheless, the ex ante expected level of terror decreases following concessions because the government's probability of eradicating terror improves as a result of a decrease in active terrorist cells and the collusion of former terrorists. However, should government couterterror fail, the increased radicalization leads to an ex post increase in violence. This suggests that the empirical observation that concessions lead to more violence is the spurious result of selection bias caused by focusing on those cases where terror continued, and thus counterterror failed. The model yields hypotheses regarding terrorist strategies and the likelihood of concessions across regime types, the effect of sources of funding on patterns of violence, the terms of concessions, and incentives for moderate terrorists to recruit extremists. I present five cases: Palestinian terrorists, Basque separatists (ETA), the IRA, Quebec separatists (FLQ), and Zionist terrorists in British Mandate Palestine.; In the second essay I model the comparative effects of multi-member (MMD) and single-member districts (SMD). Voters face both moral hazard and learning problems in assessing legislators' effort and skill. Voters learn less in MMDs be cause there are multiple representatives. As a result, the model predicts: legislators from MMDs do less constituency service, the incumbency advantage is weaker in MMDs than in SMDs, decreasing effort with tenure is present in both systems but attenuated in MMDs, there is less constituency service and a weaker incumbency advantage in unified delegations than in divided delegations, and more intensely partisan voters select unified delegations.; The final essay presents a model of judicial decision-making in which precedent is useful to policy-oriented appellate judges because it improves the accuracy with which they communicate legal rules to trial judges. The model yields hypotheses regarding conditions under which judges will break with precedent, the constraining effect of precedent on judicial decision-making, the voting behavior of Supreme Court Justices, the relationship between a precedent's age and authority, the effect of legal complexity on deference to precedent, the relative stability of rules and standards, and patterns of legal evolution.