Refining known unknowns? Modeling and Measuring Uncertainty

International relations scholars have long emphasized the role of uncertainty in shaping state behavior in international politics, from aggravating conflict to facilitating cooperation. Uncertainty due to private information or about adversary intentions, or fundamental uncertainty about conflict processes generates the bounds of state behavior by influencing how states assess the likelihood they will be successful when engaging with other actors. Information problems are a major determinant of bargaining failures, too. Despite how ubiquitous and theoretically important this concept is in the literature, however, scholars have yet to quantify and include it in empirical analyses. Instead of treating uncertainty as a background condition that cannot be measured, we use a partial observability model of conflict initiation to estimate uncertainty, where values of the unobserved variables are inferred from the relationship of observed variables to outcomes. Toward this end, we leverage structural estimation in two main ways: first, by constructing a statistical model of the bargaining model of war and second, by deriving a measure of dyadic uncertainty over the costs of war. This approach allows us both to extract a measure of uncertainty and to model the heterogeneous causes of war across dyads. By estimating instead of assuming the impact of uncertainty on war—or other outcomes—we encourage scholars to reexamine the effects of existing latent variables on international politics and bargaining behaviors.