Probabilities Over p -Values: A Decision Framework for Evidence-Based Policing

Null-hypothesis significance testing (NHST) continues to dominate policing research, yet binary p-value thresholds offer little guidance for decision-makers navigating operational, fiscal, and political uncertainty. This paper introduces a decision-focused framework designed to help police agencies make practical, risk-informed choices based on the probability that an intervention meets defined effectiveness or cost thresholds. Rather than ask whether a result is statistically significant, the framework focuses on whether the evidence is strong enough to justify proceeding, adjusting, stalling, or scrapping (PASS) a policy or program. It involves three steps: estimating effects that matter to the agency, linking those effects to expected utility, and applying transparent decision thresholds. To demonstrate its application, the paper examines three examples: a synthetic use-of-force training study, a synthetic hot-spots patrol experiment, and a real-world victim-engagement initiative. Across cases, the framework produces actionable guidance that augments conventional statistical output and supports probability-informed decisions that reflect the realities of contemporary police leadership. The paper closes with suggestions for how researchers and analysts can better communicate probabilistic results. By grounding decisions in transparent risk assessments rather than implicit reliance on fixed thresholds, this approach makes statistical evidence more useful, credible, and actionable in advancing evidence-based policing.