Abstract
Three critical issues involved in quantitative analysis of deterrence are addressed: the criteria for identifying deterrence cases, the problem of selection bias, and understanding the nature of deterrence outcomes. The criteria for establishing the presence of immediate deterrent threats are refined to allow for a more robust selection of cases, and deterrence outcomes are revised to include compromise as a midpoint between success and failure. A new data set of extended-immediate deterrence for the 1895-1985 period is presented and compared to previous data sets. Hypotheses about the effects of relative power and alliances on deterrence outcomes are developed to rectify the problem of selection bias due to unobservable prior beliefs. Alliances are shown to be a more reliable predictor of behavior than previously assumed. The results support theoretical expectations about the consequences of selection bias that more powerful states are less likely to acquiesce. The inclusion of compromise proves significant because only this outcome is strongly associated with the balance of power.
Get full access to this article
View all access options for this article.
