Black, Hispanic, and female police officers in Chicago made fewer stops and arrests than their white male counterparts, according to an analysis of three years of data on daily activities of officers in the Chicago Police Department (CPD) — an agency that has undergone substantial diversification in recent decades.
The differences reported, specific to police responses to non-violent crimes, make the case that diversifying a police force can reduce conflicts between officers and the community without making any trade-offs in public safety. They "provide strong evidence that — at least in some cities — the number of officers who identify with vulnerable groups can matter quite a bit in predicting police behavior," writes Phillip Goff, a professor of psychology at Yale University, in a Perspective on the study.
The findings, published in the February 12 issue of Science, add greater certainty to questions around impacts of diversifying police departments by gender and race, a reform widely proposed to address allegations of abusive and discriminatory policing. A link between these reforms and outcomes — changes in police behaviors — has been difficult to prove.
Around the United States and the world, disparities in police-civilian interactions and high-profile incidents of excessive force continue to raise concern. Diversification of police agencies is often presented as a solution, but consensus is lacking on key matters, including what influence officer demographics may hold on officer behavior.
Scholars often assert such questions go unanswered because of the scarcity of data needed for making causal inference. "As a result," Goff writes, "when scholars identify data that allow for strong inference around race and policing, they are rightly lauded by the field. And, to be sure, [the authors of this Science paper] should be celebrated."
In the past, studies aiming to investigate the impact of diversity on law enforcement relied on data that compared officers within or between departments. Typically, this data could not account for where, when, or whom officers were policing. This made it challenging to ensure officers were only compared against other officers facing the same kinds of civilian behavior.
Other strategies that have led analysts to draw incorrect conclusions because they failed to achieve "apples-to-apples" comparisons between police involved not counting data on shifts where police didn't take any actions. "If officers only show up in the data when they make an arrest, as is the case in many other data sets," said author Jonathan Mummolo, an assistant professor of politics and public affairs at Princeton University, "then we fail to observe all the times officers worked a shift but chose not to make any arrests."
Fewer Stops, Fewer Arrests, Less Force
To overcome this and other limitations, he and colleagues drew on detailed data assembled through three years of open-records requests. "This is a quantum advance for making strong inferences," wrote Goff.
The data they gathered included officer demographics, language skills, daily shift assignments, and career progression. The authors linked these files to time-stamped, geolocated records of the same officers' decisions to stop, arrest, and use force against civilians.
"From the start it was not clear we would obtain all the data we needed for this analysis," wrote Mummolo, "but we knew the data existed. So as long as we had another route we could pursue [to get it,] we did."
As one example of the lengths to which he and colleagues went, Mummolo explained that obtaining information on arrests made by officers in the CPD required an appeal to the Illinois Attorney General's office, which overruled the CPD's initial denial of the authors' FOIA request.
Through their analysis, Mummolo and colleagues report that Black officers make fewer stops and fewer arrests, and use force fewer times than white counterparts, on average. These disparities were not uniform across situations, say the authors, but were driven by a reduced focus on Black civilians by Black officers and a lesser use by Black officers of discretionary stops such as for vaguely defined "suspicious behavior" or for minor violations. (The researchers note they saw little difference in stops and arrests between officer demographics for violent crimes.)
Like their Black colleagues, Hispanic officers conducted fewer stops, made fewer arrests, and used force less than white officers given the same conditions, though the gaps were more modest. The findings also show women make fewer arrests, including of Black civilians, and use force less often than men.
Expanding to Other Cities
The authors say their approach offers a widely applicable template for other scholars to follow when testing whether these findings hold in other places and times. Some cities are particularly well-positioned to use it now, explained Knox in a news briefing on the study at the 2021 AAAS Annual Meeting. He pointed to Philadelphia, Pennsylvania.
"We are doing other work in Philadelphia to understand, and potentially improve, the city's system for filing civilian complaints against police," he said. "Like Chicago, the city has a history of tension between police and minority residents. And like all major departments, the city maintains detailed deployment data. It would be straightforward to replicate our study in Philadelphia or in any other large and diverse city. All that stands in the way is access to the data."
Mummolo emphasized how his team's efforts illuminate the need for data transparency reforms. "When it comes to changing policing, we need to know what works and what doesn't. The data are just sitting there, but very few people can reach it. This issue is too important to be flying blind."
"We hope this research prompts police agencies to readily share similar data so that scholars can carry out the necessary work of testing whether these findings hold [elsewhere]," Mummolo added.
The authors conclude their study acknowledging limitations, including the geographic scope of their study, which focused on just one city.
They also write: "Officers are multidimensional and crafting effective personnel reforms will likely require thinking beyond the coarse demographic categories typically used in diversity initiatives and consideration of how multiple attributes relate police to the civilians they serve."