Testing Nonnested Models of International Relations: Reevaluating Realism

Unbeknownst to most world politics scholars and political scientists in general, traditional methods of model discrimination such as likelihood ratio tests, F-tests, and artificial nesting fail when applied to nonnested models. That the vast majority of models used throughout international relations research have nonlinear functional forms complicates the problem. The purpose of this paper is to suggest methods of properly discriminating between nonnested models and then demonstrate how these techniques can shed light on substantive debates in international relations. Re-analysis of two well-known articles that compare structural realism to various alternatives suggests that the evidence against realism in both articles is overstated.