9601151 Bennett One key to accumulating knowledge in the social sciences is the regular testing of theories against one another to determine when a new theory has superseded previous ones. However, while political science has numerous theories that claim to explain novel facts, these theories are rarely compared systematically. In this project the researchers perform a broad and inclusive comparative statistical test of the major theories of international conflict. The test is applied to both the initiation of militarized disputes between countries and the escalation of those disputes to war. Such a broad comparative test is useful and necessary for accumulating scientific knowledge about international conflict. The researchers will find that information from multiple levels of analysis is necessary to accurately model conflict behavior. The researchers use a series of relative power tests to show which theories account for more international conflict outcomes than others, suggesting the relative importance of tested theories. ***