Unarmed and Killed by Police: A Closer Look
Calls for a national police use-of-force database have reached the executive branch, with Attorney General Loretta Lynch endorsing the idea publicly last week. As practitioners in the world of law-enforcement technology and data, we at StreetCred Software applaud the idea. The wrongful death of a citizen at the hands of a police officer is among the gravest of crimes.
As FBI director James Comey laments, the best current sources of data are media outlets like the Washington Post and the Guardian. These publications' datasets, while appearing comprehensive, lack contextual data. Casual users may therefore conflate incidents in which the police were called to the relief of all involved, with those in which the police behaved in a manner that was unfair or illegal.
The StreetCred Police Killings in Context Dataset, which catalogs cases in which an unarmed person was killed, is a step toward a more comprehensive and contextual dataset. It is open, non-commercial, and non-partisan. It is reviewed by a volunteer committee from across the political spectrum, comprising experts in law enforcement, use of force, data science, and analysis in the public and private sector, from industries including finance, information technology, insurance, and Internet security.
Both the Post and the Guardian do let users sort, for example, between decedents who were armed or unarmed, or by race, but that's just not enough context to be meaningful. One can't sort, for example, by whether a person was Tased before being shot, indicating that the officer attempted non-deadly force before deadly force. This happened 29 percent of the time in 2015 cases of police killings of unarmed people. To get this information from the Post or Guardian, one would have to read the details of each incident separately.
Other facts that this finer-grained approach reveals: 65 percent of decedents were engaged in either a violent crime (such as domestic violence, robbery, carjacking, or assault) or a property crime (including car theft, vandalism, and burglary). Seven percent were described by 911 callers as dangerous — "crazy," "on drugs," "covered with blood," "yelling," threatening people. Three (2 percent) were wanted fugitives in the act of escape — and one was unarmed when he died but was acting as part of a gang of three who were wanted in a recent homicide and were in the process of a kidnapping a woman. More than one-fifth had injured or killed someone prior to police involvement.
It is especially important to know whether the decedent was specifically named as the person of interest by a 911 caller. Casual users of the Post and Guardian databases may be left with the impression that police are intentionally selecting these decedents. In fact, the data show that about 75 percent of the time, the community is specifically asking the police for protection from the person who ultimately becomes the decedent.
If we want to provide researchers, community leaders, activists, police administrators, sociologists, and journalists the tools with which to frame public debate and the data in which to seek patterns, our methodology must be clear and transparent, and our data must be open. With open data, the community can have confidence that the data are sound and haven't been gamed in the process. We do this in StreetCred PKIC.
Data in context helps everyone examine the facts without adherence to any given narrative. Once we do that, we can look into the horrifying cases in which police abused their authority and took a human life. But until we do that, we're just exchanging anecdotes.
Nick Selby is CEO of StreetCred Software, Inc., a Dallas company that has sponsored the StreetCred Police Killings in Context data project. He also serves as a police detective near Dallas. He is the lead author of "Unarmed Civilians & The Police: Analysis of the StreetCred PKIC Data." The PKIC data is available here. Its methodology is published here.