Understanding The Results Of Randomized Experiments With Body Worn Cameras: Key considerations for interpreting research findings

James R. "Chip" Coldren, Jr., BWC TTA Director and Managing Director of Justice Programs, CNA, Institute for Public Research

Recent years have seen a number of new research studies addressing the effects of body-worn cameras (BWCs). Several of these studies implemented randomized controlled designs, the strongest designs available to detect the effect of BWCs with high confidence. Under randomized designs, researchers randomly assign an intervention (in this case, BWCs) to a treatment group of officers (those with BWCs) and a control group of officers (those without BWCs). If they implement the randomization correctly, then there is no real difference between the treatment and control groups, except that one received the intervention (BWCs) and the other did not. As a result, we can confidently attribute any observed differences in the outcomes of interest between these two groups to BWCs.
 
Somewhat vexing is the fact that these rigorous studies have produced different findings. For example, a study in Rialto, California, found dramatic reductions in use of force incidents and complaints against officers after the introduction of BWCs. [1] Other studies in Mesa, Arizona [2]; Orlando, Florida [3]; Phoenix, Arizona [4]; and Spokane, Washington [5] produced similar findings regarding the direction of the influence of BWCs—use of force incidents and complaints were reduced—but the reductions were not so dramatic. These studies also found positive outcomes regarding citizen approval of BWCs and positive effects of BWCs on citizen views of police legitimacy. More recently, other studies produced different findings. A study in the United Kingdom found increases in assaults on officers wearing BWCs [6], and a recent study of BWCs in the Metropolitan Washington, DC, Police Department [7] found no differences between treatment and control groups regarding use of force incidents and complaints against police.
What are policy-makers and practitioners to make of these different findings, and why do rigorous scientific research designs produce such different findings? Here we provide some suggestions for how to interpret the recent findings on the effects of BWCs. We focus on three important issues: jurisdictional differences in context, contamination in randomized studies, and compliance with BWC policies.
 
Jurisdictional differences. Rigorous studies of BWCs do not always arrive at the same conclusions about the effects of BWCs because real differences exist among jurisdictions. Simply put, the implementation of BWCs is likely to be different in Rialto, California, and Washington, DC—the communities are different, the histories of police-civilian relationships are different, and the actual implementation of BWCs in each agency might have been different. For example, the implementation of BWCs in Washington, DC occurred after several years of scrutiny and improvements in police operations under a consent decree. This was not the case in Rialto. Thus, it is reasonable to expect more dramatic changes in police-civilian relations after the implementation of BWCs in Rialto than in Washington, where good relations may have already existed. If the baseline frequency of use of force incidents and complaints against police are low at the outset of an experiment with BWCs, large reductions after BWC implementation are less likely.
 
Contamination. One concern with randomized experiments is the contamination issue, which occurs when members of a control group become exposed to the treatment group intervention such that their behavior becomes more like the members of the treatment group. In the case of BWC experiments, if control group (non-camera-wearing) officers are frequently exposed to treatment group (camera-wearing) officers, the control officers will likely react to the presence of the BWCs and act more like they are wearing BWCs themselves. This occurs, for example, when a camera-wearing and a non-camera-wearing officer arrive at the location of the same call for service. In this instance, the non-camera-wearing officer, if he or she notices the camera-wearing officer (or if the camera-wearing officer lets the other officer know a camera is present), may change his or her behavior during the incident. If such contamination happens often, the conditions of the experiment—separation of the treatment and control group—are lost or diminished, and differences between treatment and control groups will be less likely. In the case of a recent randomized BWC study in Las Vegas, Nevada [8], researchers determined that contamination between treatment and control officers occurred less than 20 percent of the time, and they found significant differences between the groups. In the Washington, DC study, in which differences between treatment and control officers were not found, researchers reported up to 70 percent contamination.[7] Thus, contamination is an important factor to consider when reviewing research results.
 
Policy compliance. Another important factor to consider when reviewing BWC research results is policy compliance. BWC policies typically provide guidance to officers on when to activate and when to deactivate BWCs, and whether officers are required to (or advised to) inform civilians that the incident is being recorded. If officers are required to turn their cameras on at certain times and notify civilians of the BWC, and they do not do so on regular basis, then there is a noncompliance issue—officers are not routinely following policy. If officers are noncompliant with BWC policy a majority of the time, the conditions of the experiment are again weakened. Either the camera is not operating, or the civilians do not know the camera is present, or both, in which case the hypothesized “civilizing” effect of the camera is negated. Thus, it is possible to implement an experiment with good randomization between treatment and control groups, and low contamination, but lose the desired effects of the cameras as a result of compliance problems. In our work with BWC technical assistance and through our involvement with BWC research, we have seen wide differences in BWC compliance in different jurisdictions— ranging from less than 40 percent compliance in one, to over 90 percent compliance in another. Recently, the Metropolitan Washington, DC, Police Department reported a compliance rate of approximately 65 percent for BWC activation.[9] This too may have contributed to the finding of no differences between treatment and control groups in that study.
Summary
 
We have illustrated several important features of BWC programs and BWC research projects that readers and research consumers should keep in mind when interpreting BWC research results: jurisdictional differences, contamination issues, and compliance issues. To be sure, these are not simple, clear-cut issues. It is possible to have low contamination or low compliance and still find no differences between treatment and control officers, and it is possible to have high contamination and still find differences between treatment and control officers. Even if these factors are well managed and controlled for in BWC research, it is possible that jurisdictional differences (e.g., different use of force and complaint baselines) will drive the research results.
 
Complexities such as these make clear the need for replication in the sciences. Studies must be done repeatedly under different conditions in different jurisdictions to amass a large body of scientific evidence on which to base decisions and judgements. BWC research has not yet reached this point, though the limited body of evidence in existence points to the conclusion that BWCs are beneficial to the police and their communities. Still, research and analysis must continue, with strong scrutiny of the results, to sort out these confounding issues.
 
REFERENCES
[1] Ariel, Barak, Tony Farrar, and Alex Sutherland, “The Effect of Police Body-Worn Cameras on Use of Force and Citizens’ Complaints Against the Police: A Randomized Controlled Trial,” Journal of Quantitative Criminology 31(2015): 1–27.
 
[2] Mesa Police Department, On-Officer Body Camera System: Program Evaluation and Recommendations (Mesa, AZ: Mesa Police Department, 2013)
 
 [3] Jennings, Wesley G., Mathew D. Lynch, and Lorie Fridell, “Evaluating the impact of police officer body-worn cameras (BWCs) on response-to-resistance and serious external complaints: Evidence from the Orlando Police Department (OPD) experience utilizing a randomized controlled experiment,” Journal of Criminal Justice 43 (2015): 480–486.
 
[4] Hedberg, Eric C., Charles M. Katz, and David E. Choate, “Body-worn cameras and citizen interactions with police officers: Estimating plausible effects given varying compliance levels,” Justice Quarterly 34 (2017): 627-651 
 
[5] White, Michael D., Gaub, Janne E., & Todak, Natalie, “Exploring the potential for body-worn cameras 
to reduce violence in police-citizen encounters” (forthcoming), Policing: A Journal of Policy and Practice, doi:10.1093/police/paw057.
 
 [6] Ariel, Barak, Alex Sutherland, Darren Henstock, Josh Young, Paul Drover, Jayne Sykes, Simon Magicks, and Ryan Henderson, “Wearing Body-Cameras Increases Assaults Against Officers and Do Not Reduce Police-Use of Force: Results from a Global Multisite Experiment,” European Journal of Criminology 13a (2016): 744-755.
 
[7] Yokum, David, Anita Ravishankar, and Alexander Coppock, “Evaluating the Effects of Police Body Worn Cameras: A Randomized Controlled Trial,” The Lab @ DC, Office of the City Administrator, Executive Office of the Mayor, Washington, DC, October 20, 2017.
 
[8] Sousa, William, James R. Coldren Jr., Denise Rodriguez, and Anthony A. Braga, “Research on Body Worn Cameras: Meeting the Challenges of Police Operations, Program Implementation, and Randomized Controlled Trial Designs,” Police Quarterly 0(0): 1–22.
 
[9] Office of Police Complaints, “Annual Report 2017,” Government of the District of Columbia, Police Complaints Board, 18-19.
 

This commentary represents a compilation of the thoughts and suggestions of a number of individuals involved in body-worn camera research and technical assistance, including the following: Michael White, PhD, Arizona State University; John D. Markovic, Bureau of Justice Assistance; Brett Chapman, PhD, National Institute of Justice; Denise Rodriguez, CNA; Craig Uchida, PhD, Justice and Security Strategies; and Anthony Braga, PhD, Northeastern University, among others.

 
Dr. James R. “Chip” Coldren, Jr., is the  Director on BWC TTA and is a Managing Director for Justice Programs at CNA's Instistute for Public Research. He has more than 35 years of experience with research; program and policy evaluation; policy development; advocacy; development, coordination, and delivery of training and technical assistance; and justice system reform. In addition to serving as the Director on BWC TTA, Dr. Coldren is also the National Director for the Bureau of Justice Assistance (BJA) Smart Policing Initiative and BJA Violence Reduction Network. He also serves as Principal Investigator on two National Institute of Justice–funded research projects: a national study of equipment modalities and correctional officer safety, and a randomized experiment with body-worn cameras in the Las Vegas Metropolitan Police Department