Using data science to eliminate racial bias in policing

The Problem

It’s tempting to believe that racially-disparate policing is not about race — that perhaps crime or poverty are to blame. But research indicates that these factors aren’t sufficient to explain the racial disparities in policing outcomes. In the US today, Black people are two to four times more likely to be targets of police force than White people, leading to more injuries, deaths, hospital bills, lost wages, psychological trauma and juvenile crime. The problem doesn’t end there. Police are the most common representatives of the state that communities encounter, and as people trust police less, their trust in democracy erodes. Negative experiences with the police actually increase the likelihood a child will be involved with a crime in the future. Police involvement hamstrings the ability to find employment and even impedes voter turnout.

Big Idea

Racism is most commonly defined as a defect of hearts and minds. And yet, social psychologists have shown that attitudes are poor predictors of behaviors. Behaviors, on the other hand, can be seen and measured — and thus changed. This is what the Center for Policing Equity (CPE) works to do. Nearly every police department already uses COMPSTAT, a data-driven accountability system, to track crime. CPE has partnered with police departments to determine measurable behaviors they can capture with data and work to change – city by city, block by block. It’s a COMPSTAT for Justice. In the 25 cities to which CPE has delivered scientific products, police departments have seen significant change across several outcomes: a 26% drop in use of force incidents, a 13% decrease in injuries to police officers, and 25% fewer arrests — all without an attendant crime surge. Bringing those results to scale can lead to fewer officer injuries, fewer civilian injuries, fewer lawsuits against police departments and dramatic improvements in the lives of Black and Brown communities.

Plan

With support from The Audacious Project, CPE will bring this reform to departments that collectively serve 100 million people a year — approximately one in three Americans — by 2024. As CPE broadens its reach, the organization will refine its analyses, and improve its software to process police data in a matter of minutes rather than months. They will launch demonstration cities to fine-tune their approach, and establish stronger causal evidence to solidify their proven track record of success. Armed with advanced technology, CPE will scale up, building relationships at the state level to save lives, prevent injury, save public dollars and build trust.

Why will this succeed?

CPE is seeing high demand for its services, drafting off an existing partnership with the police chiefs of Connecticut. Until now, invitations from police departments around the country have far exceeded the organization’s capacity. CPE works to establish local partnerships with mayors, city councils and community groups who are vocal about the need for police reform. Collectively, these invested parties work with police departments and use CPE’s findings to take action — solving problems together. For example, CPE helped the Minneapolis Police Department leverage its own data to reduce use of force incidents by 18% over three years. To make sure changes last, CPE establishes frameworks for law enforcement to continue to share data and hold themselves accountable. Basing reform on science lowers the temperature in conversations about race and brings trusted data to the center of a process in which those involved often don’t trust each other.