Drawing Inferences From Policy Experiments

The article argues that the sctentific credibility and policymaking utility of policy evaluation research is threatened by inappropriate inference strategies. These inference strategies are a consequence of a system of incentives and a methodology that encourage vague, ambiguous, or nonspecification of primary inferences and extended or unwar ranted tangential inferences. The result is policy research designed more to promote the enterprise of policy research than to assist in the useful application of research in policymaking settings. These points are illustrated through an analysis of the inference strategy pursued in the Police Foundation's Kansas City Preventive Patrol Experiment. One solution to the problem of inappropriate inferencing is for policy researchers to adopt the practice ofspecifying an illustrative range of potential primary inferences at the initial stages of analysis.