How Can Data Increase Police Accountability?

• June 11, 2020
• Public Safety

In the wake of the most recent murders of Black Americans by police officers, community advocates, policy makers, and the public are asking what can be done to prevent more deaths. Data has an incredibly important role to play in this discussion; it can help reveal prejudicial policing, extreme use-of-force, and racially-motivated patterns of violence. Unfortunately, there is very little standardized data collection on police use-of-force, at either the state or federal level.

According to the United States Federal Bureau of Investigation (FBI), “the opportunity to study use-of-force incidents and discuss their cause is hindered by the lack of enough data to compile nationwide statistics.” In an attempt to rectify this, in 2015 the FBI recommended developing “a new data collection on fatal and nonfatal officer-involved shootings.” Three and a half years later, on January 1, 2019, it launched National Use-of-Force Data Collection nationwide. However, the FBI does not have the ability to enforce participation in this database, and instead is only able to “encourage” state, local, tribal, and federal law enforcement to contribute monthly reports to the collection. As of May 29, 2020, the FBI reported that just over 40 percent of “the nation's sworn law enforcement officers” were even participating.

Despite the hope that this database would allow the FBI to publicly release useful statistics, information on trends, and incident characteristics, the information will not be accurate with this low level of participation from law enforcement entities across the country. The Bureau initially planned to release its first report in March 2019, but now aims to release the inaugural data summer 2020. While this report might provide a snapshot into police use-of-force, it will be difficult to see national patterns and trends.

However, data does have another, preemptive role to play in uncovering and preventing instances of violence and force like the ones tracked in the FBI database. While Early Intervention Systems (EIS) have been used to warn about “problem officers” for about 30 years, there has been increased interest in deploying more data-driven early intervention systems to identify and then support or redirect officers who might be at risk of engaging in harmful and/or deadly conduct.

In 2015, the University of Chicago’s Center for Data Science and Public Policy (DSaPP) worked with the Charlotte-Mecklenburg Police Department (CMPD) to build and test a new, more data-driven EIS system to predict police misconduct. The system worked by pulling together data on officer attributes like length of time in position and individual demographic data; officer activities like number of arrests and dispatches; if the officer has been investigated by internal affairs and, if so, any outcomes; and previous feedback or complaints about the officer.

The DSaPP team found that their EIS system was successful in alerting CMPD to at-risk officers, which allowed CMPD supervisors to stage interventions, provide training or counseling for officers at risk, and rethink how and when they deployed officers that had been dispatched to high-stress situations. However, a 2020 study evaluating the effectiveness of EIS found that many are not significantly capable of flagging potentially problematic officers. The researchers determined that this was due to ineffective “thresholds,” which certain officer behavior must meet in order to be flagged.

So how can the data work described in this article be improved, in order to reveal — and prevent — both systemic and individual violence?

• Every law enforcement agency in the country (including state, local, tribal, and federal) should be mandated to utilize the FBI’s use-of-force database. Currently, the FBI does not have the power to make agencies enter data, but a federal law could be passed to require all law enforcement to participate in the database.
• Also, law enforcement agencies can collaborate with citizen- and journalist-driven data collections, like the Washington Posts’s Fatal Force police shooting database (pictured in header) and the Mapping Police Violence data visualization, for a more accurate count. According to the United States Bureau of Justice Statistics, “the hybrid approach to identifying arrest-related deaths, which combined information from media reviews and agency surveys, resulted in improvements in data completeness and quality.”
• Additionally, the United States Centers for Disease Control and Prevention runs the National Violent Death Reporting System, which is considered one of the best databases for information on fatal shootings. The FBI could use this to supplement parts of the use-of-force database, to focus their efforts on collecting data around non-fatal encounters.
• Despite the difficulty in replicating successful EIS cases, additional research and trials should be run to identify the best way to build and utilize an early intervention warning system in police departments. Further work on EIS includes refining threshold measures; some researchers want to include more complex data (like footage from police body cameras and bystander cell phone footage) in order to improve how officers and incidents are flagged.

The current reality is that Black Americans have a significantly higher risk of being killed by police; research published by the National Academy of Sciences found “that men of color face a nontrivial lifetime risk of being killed” by law enforcement. Efforts to remake police departments will need to embrace and incorporate data to work on finally eliminating racially-biased police violence and murders in the United States.