Can Better Data Fix America’s Policing Crisis?

For decades, the performance of individual police officers and their departments was largely measured with data about crime rates, along with the arrests made by officers and citations handed out. For most police departments that’s still the case. But the widespread push to reform police practices is also spurring a few innovations in the use and collection of data to help drive change.

Examples of the benefits of better data use make a case for reform. The police department in Tucson, Arizona, for example, provides a stellar example of a new approach to the use of police data and performance measurement. It’s still a work in progress, but new data sets are constantly emerging, along with helpful definitions to aid in community understanding. “We’re trying to really be transparent and put out honest information, even if that paints us in a bad light,” says Jacob Cramer, analysis administrator for the Tucson police department.

In one recent case, the Tucson data helped to identify a police officer who was using more force than his peers. Discussions with the officer’s supervisors showed he lacked a solid sense of the tools for de-escalating potentially inflammatory situations and he was sent back to the academy for additional training.

Cramer was hired two and a half years ago to fill an advanced analyst position that was the “brain child” of Tucson police chief Chris Magnus. He has slowly built up a staff of nine team members all with advanced degrees.

A high level of communication with community stakeholders is key to the Tucson effort. In the near future, community surveys will seek input from people in different parts of the city about what will make them feel safer. This fall, multiple meetings with community groups have provided feedback on the city’s dashboards, which have been revised multiple times to make them more user friendly.

In Tucson and other cities that stand out for their use of data, analysts are probing deeply beyond surface data points. When presenting use-of-force data, for example, Tucson looks at when and where force is used, and the details about the kind of force, including the rate a gun or other weapon is displayed, when different restraints are used and how often a taser is fired.

Similarly, in Cambridge, Massachusetts, Police Commissioner Branville Bard has his staff looking not just at the racial breakdown of traffic stops, but developing data sets that examine the duration of a stop to see whether individuals of different races, ethnicities or genders are being detained longer than others. Cambridge is also looking at the results of each stop—whether citations or other police actions are taken—and the reasons given for a stop to see if racial bias or other factors lead to disproportionate interaction for more minor infractions.

The movement to improve police data has an important internal aspect. “There’s a public version and an internal version where commanders can drill down to particular squads and officers and compare them to other officers in their squads. You can see the outliers,” says Cramer.

A growing body of research provides hope for progress in a number of places that haven’t advanced their data-use to the levels now in place in Tucson or Cambridge.  “There’s a growing understanding of the need for a more comprehensive set of data on what police are doing,” says John Hollywood, who runs the Center for Quality Policing at the Rand Corporation.

Hollywood says his center’s work shows the need for a re-examination of what communities want police to do and how to measure what they actually do. Even police agencies with limited data systems can improve with daily reviews of incident reports, using insights to inform next-day operations, he says.

Several large organizations and university programs have gotten involved.  At the National Police Foundation, a project is now underway that focuses on a three-pronged approach that goes beyond simple output measures. Instead, it adopts more sophisticated means of understanding the quality of police, including  public safety risk, community sentiment and internal organizational health—for example, turnover rates and sick leave use.

The technology giant SAS, which has focused over the last decade and a half on helping police departments leverage data analytics to solve crimes more quickly, now has devoted the efforts of 100 individuals to an “officer readiness” project. This effort is aimed at analyzing the often unexamined elements that contribute to officer performance, such as the ability to de-escalate potentially violent situations.  

The Vera Institute of Justice recently published its own take on how to use data differently, suggesting agencies could look at their 911 call data to make changes in how emergency calls are handled. As 911 data shows most calls aren’t about crimes, in many cases people who aren’t police could be dispatched to respond instead.

Those are just a few examples of the research work that’s currently underway. But there are no silver bullets and the path toward developing high-quality police data is not straightforward.

Barriers include union opposition, poor quality data, a lack of staff and fiscal resources for analysis, ingrained culture, and a high level of political partisan tension and disagreement.

The website that houses the much-heralded Police Data Initiative that grew out of President Obama’s Task Force on 21st Century Policing is now described as on “life support,” by James Burch, president of the National Police Foundation, where it was located and funded by the U.S. Department of Justice. At the time, 129 police departments had committed to sharing at least three sets of open data from which others could learn.

The police data initiative website now exists mostly as an archive because the Justice Department shifted all funds out of the effort in 2019. “There are a lot of broken links and updates that are needed,” says Burch. “We can’t make changes or add new agencies.”

Clearly, the research is worth funding. The work being done in Tucson and other pioneering local governments serve as models of how the skillful use of data can help achieve a better understanding of the role of police and contribute to stamping out biased enforcement. “We’re doing this regardless of what it shows,” says Bard. “It forces us to deal with difficult issues and it lets the community know we’re not hiding anything.”