Study 1
Ariel, Farrar, and Sutherland (2015) conducted a randomized experiment on the effects of police body-worn cameras on use-of-force incidents and citizen complaints against the police in the city of Rialto, in San Bernardino County, California. Covering 28.5 square miles, the Rialto Police Department serves a population of approximately 100,000 residents, and employs 115 sworn police officers and 42 non-sworn staff. The department handles roughly 3,000 property and 500 violent crimes per year; it handled about six to seven homicides per year in the period from 2009 to 2011, which is almost 50 percent higher than the national per capita homicide rate.
Every frontline officer in the Rialto Police Department took part in the experiment (n = 54), although the study authors used shifts as the unit of analysis and randomization. Each 12-hour shift consisted of roughly 10 armed frontline officers patrolling the streets of Rialto, responding to calls for service, and interacting with the public.
Beginning in February 2012, and lasting a year, the study consisted of randomly assigning all shifts on a weekly basis to either an experimental or control condition. In the experimental shifts, officers were required to wear a high-definition, video- recording apparatus fitted to their collar, which would record all of their activities during the shift. Although the device would be visible to citizens interacting with the officers, the police also informed anyone they interacted with that they were being videotaped. A total of 988 shifts were randomly assigned during the 1-year study, resulting in 489 experimental shifts and 499 control shifts, which were patrol shifts “as usual”. The researchers measured the contacts between the police and the public in every shift for all non-casual encounters (calls for service, collecting evidence and statements, formally advising individuals, etc.), which allowed them to compute incident rates per 1,000 police–public encounters.
The devices used in the experimental condition were small enough to fit into the officers’ shirt pockets, and provided high- definition color video and audio, were water resistant, and had a battery life greater than 12 hours. The officers were instructed to use the devices for every police interaction with the public, except for those involving cases of sexual assault of a minor or when dealing with police informants. All the data was automatically uploaded, collated, and inventoried in a web-based, video-management system at the end of each shift, and was available to the researchers.
One of the two outcomes of interest was use-of-force. The Rialto Police Department uses a standardized tracking system that records instances of use-of-force that are beyond basic control and compliance holds, including the use of pepper spray, a baton, a Taser, a canine bite, or a firearm. For the purposes of this study, the researchers operationalized use-of-force as whether or not force was used during a given shift. Use-of-force was counted only as instances of force, and not the degree of force used, the length of that use, or who instigated the use-of-force.
The second outcome of interest was citizen complaints against officers. These data were tracked by the Rialto Police Department through software that records citizens’ complaints of alleged police misconduct or poor performance. The researchers used these data to count the number of complaints of any type filed against officers.
Poisson regression models were used to assess the differences between the experimental and treatment shifts in terms of use-of-force and citizen complaints. The authors acknowledged some limitations of the study, including the Hawthorne and John Henry effects. The Hawthorne effect refers to how individuals change their behavior in response to their awareness of being observed, while the John Henry effect refers to the degree to which comparison subjects change their behavior when they become aware that their performance is being compared with a treatment condition. Given that a frontline police officer may have been involved in both experimental and treatment shifts over the study period, both these effects may be at work in reporting mechanisms. However, the researchers also looked at the overall effect that their experiment had on the Rialto Police Department, by examining the previous years’ reports for both use-of-force and citizen complaints. Subgroup analyses were conducted to examine the effect of officers’ discretion in activating body-worn cameras.