Spotlighted Study: “Procedural justice training reduces police use of force and complaints against officers” by George Wood, Tom R. Tyler, and Andrew V. Papachristos. 2020. Proceedings of the National Academy of Sciences.
Program: There is tremendous interest in finding interventions that can improve relationships between community members and law enforcement, and reduce the use of force by police. A new, one-day training developed for police officers in Chicago “emphasized the importance of voice, neutrality, respect, and trustworthiness in policing actions,” consistent with principles of procedural justice. The program’s goals were to increase the level of respect with which officers treat the public, ensure police listen to civilians’ views before making decisions, and, in general, encourage officers to demonstrate an interest in working as members of the community they are sworn to serve and protect.
Research challenge: Officers who voluntarily sign up for procedural justice trainings, or who are assigned by their superiors to undergo such trainings, may be different from other officers. For instance, they may be more or less inclined toward the standard “command and control” approach to policing. For this reason, simply comparing the behavior of officers who have had such trainings with the behavior of officers who haven’t doesn’t tell us much about the effect of the training itself. Similarly, in some cases a specific event (such as a violent incident on the job) might lead an officer to undergo such training, but might also change his attitude toward policing in a way that independently affects his behavior going forward. That means even looking at changes in the same officer’s behavior over time might not demonstrate whether the training itself had any effect. To measure the causal effect of a training program, we need those programs to be implemented in a way that avoids these sources of bias. A previous study randomly assigned officers to receive a procedural justice intervention in Seattle. It found beneficial effects, but the sample was small and might not have been representative; a larger-scale evaluation was needed.
Research strategy: In Chicago, not all police officers could undergo the new training at once, because of capacity constraints (there were over 8,000 officers in the city). Instead, officers were randomly assigned to participate in the program on different days, in groups of about 25 officers per class. Trainings began in January 2012 and continued for several years. In other words, Chicago implemented a staggered rollout of the program, which allowed researchers to measure the causal effect of the training on subsequent behavior. By comparing trends in outcomes for those who received training in a particular month, with trends for those who had not yet undergone the training, researchers were able to quantify the effect of the training itself. Because the timing of when individual officers took the training was not related to anything about their personal characteristics, interest in the program, or recent events, we can be more confident that differences in subsequent behavior are the effect of the training program, and not pre-existing differences across officers.
Results: The training program was highly successful in improving measures of police behavior. Participation reduced complaints against police officers by 10.0%, reduced the frequency of sustained or settled complaints against officers by 15.5%, and reduced the use of force against civilians by 6.4%, in the 24 months following the training. Effects appeared to grow over time, and were largest for those who were assigned to take the training early, which may indicate important spillover effects within the department. (That is, those who took the training later may have already been somewhat “treated” by their previously trained peers.)
What drove the effect?: The effects found here were for the 8-hour training program as a package, and represent the average effects across all officers. It is not yet clear if this program had bigger effects for some officers than others, or whether it would have the same benefits if the program were targeted at such officers.
Cost-benefit: The study did not include a formal cost-benefit analysis. The costs of the program include the instructors’ time (ten police officers who conducted the training in rotating groups of three, across 327 training days), and the time of the police officers (one day each). The benefits include a reduction in payouts for settled complaints (by my calculation, approximately $4.2 million over the 24-month follow-up period). It is not yet clear if the trainings had any effect on crime rates, on injuries to police officers, or on public perceptions of police fairness; such effects should also be considered.
Policy implications: This study demonstrates that a training program based on procedural justice principles can have large benefits in terms of a reduction in complaints against officers and use of force by police. It provides a roadmap for other police departments to follow to implement similar training programs, though the trainings may need to be adjusted for other contexts. Police departments that implement this and other trainings should use a similar staggered rollout design so they can evaluate whether the program has the benefits they seek for their cities.