We used administrative data to measure effects. The primary outcomes of interest were documented uses of force and civilian complaints, although we also measure a variety of additional policing activities and judicial outcomes.
Our examination of judicial outcomes was constrained by limitations in the available data. Namely, we did not have access to the full datasets managed by the United States Attorney’s Office (USAO), the Office of the Attorney General (OAG), and the courts. We instead had access to a subset of this data available to MPD, which captures only the initial charges on which an individual was arrested. A consequence is that we are unable to track court outcomes for any changes to those initial charges. For example, if MPD makes an arrest for a felony, and USAO changes those charges to a misdemeanor, then this event is only reflected in our data as a felony not prosecuted. The misdemeanor charge is not captured in our data. As this limitation applies to both control and treatment groups, we can still conduct a preliminary analysis on the evidentiary value of BWCs.
To analyze the data, we compare the average rate of the outcome in the treatment group to the average rate of the outcome in the control group, or a difference-in-means. Results were obtained at the officer level and translated into yearly rates per 1,000 officers.
We also use another approach: a linear regression model that increases precision by controlling for pre-treatment characteristics (e.g., officer age, sex, race, district assignment, use of force prior to the deployment of BWCs). We find the same results using both methods. To simplify discussion, we present only the difference-in-means estimates in the Results section. However, all estimates are reported in the Supplementary Materials, available for download here.
All of our analyses were conducted by two independent statistical teams, to help avoid coding errors and as a check of convergence in results.