Framing an analytic narrative : a tale of three crises

This essay explores a fundamental modeling choice that must be addressed before an explanatory analytic narrative can be developed: whether to rely on a model fashioned around the specific event or political process under investigation using the procedure of "revealed preferences," or to "posit" preferences and situate the narrative in a more general model. It argues that analytic narratives that infer preferences from actual choices not only border on the tautological, but may also fail to take account of underlying preferences that are obscured by strategic behavior. In contrast, explanations derived from a more general model are enhanced when the model's expectations are confirmed empirically. By their very nature, the organizing power inherent in a generic model not only facilitates cross case comparisons, but also isolates the causal role played by critical explanatory variables. The latter advantage of using an 'off-the-shelf" model is demonstrated in an examination of three structurally similar, yet otherwise distinct, acute interstate crises: the Bosnian crisis of 1908-1909, the July crisis of 1914, and the Cuban missile crisis of 1962.