Read on Twitter
In a new @ScienceMagazine study, we find Black, Hispanic & female officers engage in less enforcement and violence than white & male officers facing common circumstances. @bocar_a @dean_c_knox @romangrivera1 1/ https://science.sciencemag.org/cgi/doi/10.1126/science.abd8694
The central hurdle in evaluating whether officers from different groups treat civilians differently has been accounting for context. Without the ability to compare officers facing common circumstances, any differences in their behavior become ambiguous. 2/
Most police datasets don’t allow these apples-to-apples comparisons. They often aggregate to large units of time & space, or don't account for shifts in which officers were deployed but took no enforcement action. We show any of these can lead to mistaken conclusions. 3/
These data constraints meant much past work had to assume that "white and nonwhite officers are randomly assigned to neighborhoods" (Donohue III & Levitt, 2001). But that’s not true. For example, in Chicago, Black officers tend to work in areas with more reported violent crime.4/
To account for these differences, we merge behavioral data (stops, arrests, & uses of force) w/ microlevel records of where & when Chicago patrol officers were deployed. We then compare the behavior of officers working in the same very narrow slices of time & space. 5/
So what do we find? In short, when facing similar civilians in comparable places and times, officers from marginalized groups engage in much less enforcement activity & violence than their white and male counterparts. 6/
Black officers made far fewer stops & arrests & used force less often than white officers facing common circumstances: differences equal to 29%, 21% & 32% of average white officer behavior respectively. Hispanic & female officers showed reduced activity too, by smaller margins.7/
These reductions are not uniform across encounters—they are concentrated on the treatment of Black civilians. For example, reduced use of force against Black civilians accounts for 83% of the overall force gap between white and Black officers. 8/
We also find Black officers focus less on discretionary enforcement. Reductions are primarily in more minor offenses, like stops for “suspicious behavior”. Arrests for violent crime, on the other hand, look very similar across groups. 9/
Overall, our paper shows that diversity in policing can have a tangible impact. All else equal, deploying a Black, Hispanic, or female officer instead of a white or male officer leads to substantially different on-average treatment of minority civilians. 10/
Some important caveats: 1) This is a study of one city. We need many more analyses in other places and times to know whether these results generalize. We hope our paper provides a straightforward template for doing so. 12/
2) It is difficult to extrapolate these results to learn about the effects of *future* hiring. Recruitment pools can change all the time, as can agency culture and deployment patterns. Evaluating the impact of diversity in policing will require ongoing analysis. 13/
We were only able to conduct this study after years of open records requests to multiple agencies, with some data only released after appeal. This has to change. It should not take several years to find out when, where & how police behave in a single community. 14/
As major changes to policing are debated, it is vital that police agencies become more transparent. Credible answers to many longstanding questions are sitting on police agency hard drives. We need not rely on hunches and intuition when charting the way forward. 15/
Our data and code are available here: https://codeocean.com/capsule/8907164/tree/v1
We encourage readers to probe and extend our analysis.
/end
We encourage readers to probe and extend our analysis.
/end
