Policy Implications of Statistical Estimates: A General Bayesian Decision-Theoretic Model for Binary Outcomes

How should we evaluate the effect of a policy on likelihood an undesirable event, such as conflict? The significance test has three limitations. First, relying statistical misses fact that uncertainty is continuous scale. Second, focusing standard point estimate overlooks variation in plausible sizes. Third, criterion substantive rarely explained or justified. A new Bayesian decision-theoretic model, “causal binary loss function model,” overcomes these issues. It compares expected under intervention with one no intervention. These losses are computed based particular range sizes policy, probability mass this size range, cost and event intends to address. model more applicable than common models using functions capturing costs terms false positives negatives. I exemplify model’s use through applications provide R package. Supplementary materials for article available online.