Police stops to reduce crime: A systematic review and meta‐analysis

Background Police‐initiated pedestrian stops have been one of the most widely used crime prevention tactics in modern policing. Proponents have long considered police stops to be an indispensable component of crime prevention efforts, with many holding them responsible for the significant reductions in violent crime observed across major US cities in recent decades. Critics, however, have taken issue with the overuse of pedestrian stops, linking them to worsening mental and physical health, attitudes toward the police, and elevated delinquent behavior for individuals directly subject to them. To date, there has been no systematic review or meta‐analysis on the effects of these interventions on crime and individual‐level outcomes. Objectives To synthesize the existing evaluation research regarding the impact of police‐initiated pedestrian stops on crime and disorder, mental and physical health, individual attitudes toward the police, self‐reported crime/delinquency, violence in police‐citizen encounters, and police misbehavior. Search Methods We used the Global Policing Database, a repository of all experimental and quasi‐experimental evaluations of policing interventions conducted since 1950, to search for published and unpublished evaluations of pedestrian stop interventions through December of 2019. This overarching search was supplemented by additional searches of academic databases, gray literature sources, and correspondence with subject‐matter experts to capture eligible studies through December 2021. Selection Criteria Eligibility was limited to studies that included a treatment group of people or places experiencing pedestrian stops and a control group of people or places not experiencing pedestrian stops (or experiencing a lower dosage of pedestrian stops). Studies were required to use an experimental or quasi‐experimental design and evaluate the intervention using an outcome of area‐level crime and disorder, mental or physical health, individual or community‐level attitudes toward the police, or self‐reported crime/delinquency. Data Collection and Analysis We adopted standard methodological procedures expected by the Campbell Collaboration. Eligible studies were grouped by conceptually similar outcomes and then analyzed separately using random effects models with restricted maximum likelihood estimation. Treatment effects were represented using relative incident rate ratios, odds ratios, and Hedges' g effect sizes, depending on the unit of analysis and outcome measure. We also conducted sensitivity analyses for several outcome measures using robust variance estimation, with standard errors clustered by each unique study/sample. Risk of bias was assessed using items adapted from the Cochrane randomized and non‐randomized risk of bias tools. Results Our systematic search strategies identified 40 eligible studies corresponding to 58 effect sizes across six outcome groupings, representing 90,904 people and 20,876 places. Police‐initiated pedestrian stop interventions were associated with a statistically significant 13% (95% confidence interval [CI]: −16%, −9%, p