Modern Tech Suggests Government Data on Police-Related Homicides Flawed

The Federal Bureau of Investigation (FBI) publishes statistics on police-involved homicides, while on duty, of civilians. Its system relies on voluntary self-reporting by the nation’s more than 18,000 local and state police agencies.

Self-reporting works well when there is an incentive to report and an incentive to report accurately. But police departments have no incentive to report this emotionally charged information with possible political ramifications. As a result, some groups have alleged that official FBI numbers undercount the true number of police-related homicides. Thanks to a group of tech entrepreneurs, there is now evidence suggesting government numbers are flawed.

A handful of independent crowd-sourced websites aggregate user-submitted local reports of police-involved homicides. Matthew Green, editor of the blog “The Lowdown: Decoding the News,” part of San Francisco’s NPR affiliate KQED, examined information on one of these crowdsourcing websites, Killed By Police (a Facebook site).

For each of the 156 police-involved homicides listed on Killed By Police for California in 2014, Green checked that each had a link to a verifiable media story, which they did. The FBI reports about 420 police-related homicides nationally each year. So if FBI numbers are accurate, we must believe that California accounts for nearly 40 percent of all police-involved homicides annually. This seems unlikely, and the more plausible explanation is that the FBI undercounts the true number of such deaths. Killed By Police reported 1,103 police-related deaths nationally in 2014.

This exercise says nothing about whether any of the police-related homicides were unjustified, but it does suggest that the FBI’s tracking system is seriously flawed.

The public has a right to know the truth. But because of the incentives, it is likely that modern tech will provide more accurate counts of police-related homicides than government agencies.