Evidence and public policy: The example of evaluation research in policing nagin and weisburd policing

Research Summary: In this paper, we argue that both science and the policy process are well served by research with high evidentiary value. We also argue that experimental research is valuable in the policy domain not only because of its high evidentiary content but also because of its transparency. To exemplify this point, we describe how a decade-long research program on hot spots policing overturned the conventional wisdom that police could not affect crime and, in so doing, has profoundly altered police practice. Still, we recognize that for a variety of practical and ethical reasons, policy-relevant research on policing and criminal justice policy more generally cannot be based entirely on experiments. We discuss two key features of randomization-balance in expectation between the treated and the control group on all potential confounders and exogeneity of treatment status-that are the source of the high evidentiary status of randomized experiments. We go on to describe how these same two features can be credibly replicated in quasi-experimental studies that are not subject to the ethical and practical obstacles that may make experiments impractical in some circumstances. Policy Implications: We urge criminologists to take greater advantage of quasi-experimental research opportunities. We also recommend that criminologists go beyond simply identifying such opportunities. They should engage practitioners and policy makers and make clear what is required to allow for strong evidence of program effectiveness.