A new database sheds light on killings by police throughout the country

Journalists trying to investigate police-involved killings often have a difficult time finding good data. The depth and accuracy of information vary from department to department, and there isn’t a comprehensive, centralized database to compare agencies or understand the factors related to police violence, such as officer training. 

Brian Finch is trying to change that.

The researcher at USC Dornsife Center for Economic and Social Research has compiled data from about 17,000 law enforcement agencies over nearly 18 years. It includes whether or not the person killed was armed, whether the autopsy was done by a coroner supervised by the sheriff or working in another county department, and hundreds of other observations for each agency. 

It’s now designed for researchers and data analysts. Finch hopes to release an easily searchable database in the next year.

“Someone could go in and compare LAPD with, say, Houston, Chicago or the NYPD, and do a simple comparison of these rates by the number of arrests, or a simple comparison of the counts, or look at them in terms of the level of violent crime,” he said. “If there's an analyst that's capable of doing that, it’s just something that nobody's really ever seen before.”

Finch’s National Officer-Involved Homicide Database adds to the small but growing body of research tools for journalists trying to shine a light on police violence, nearly two years after the uprisings following the murder of George Floyd in Minneapolis. Even with increased attention on police brutality, it’s still a challenge for reporters, researchers and the public to get information about patterns in their communities.

Finch started down this line of research after reading a 2014 Washington Post article about the lack of data on police shootings. It quoted D. Brian Bughart, a journalist who started a database called Fatal Encounters in 2012. He had enlisted volunteers to scour the news and file public records requests to chronicle these incidents around the country.

“Don’t you find it spooky? This is information, this is the government’s job,” Burghart told The Post. “One of the government’s major jobs is to protect us. How can it protect us if it doesn’t know what the best practices are? If it doesn’t know if one local department is killing people at a higher rate than others? When it can’t make decisions based on real numbers to come up with best practices? That to me is an abdication of responsibilities.”

In October, British medical journal The Lancet published findings that from 1980 to 2018, the U.S. National Vital Statistics System underreported police killings by 55%. 

Finch connected with Burghart, who had compiled incident-level data, to try to make the information more accessible for researchers. Finch secured funding from the National Institutes of Health in 2017 for his dataset. Burghart is now a researcher at USC Dornsife Center for Economic and Social Research.

“It's the first database to ever be released at the level of the law enforcement agencies,” Finch said. “Most everything has been aggregated upward to like counties or states.”

Other resources include Mapping Police Violence, Fatal Force from The Post, and The Guardian’s The Counted.

Researchers have already been busy crunching Finch’s data to spot trends. Preliminary papers, not yet peer-reviewed, have found that officer-involved homicides increase as neighborhoods gentrify and that violent crime rates aren’t necessarily tied to more police-involved homicides. As more affluent people move into new neighborhoods and are unfamiliar with the local norms, the reasoning goes, they make more reports to police, whose increased presence leads to more killings.

Finch said the researchers are also starting to look at certain policies’ relationship to the deaths, such as hiring more women and college-educated officers, training on the proper use of force, eliminating police pursuits, and restricting guns in a given community.

Just because Finch has so much information at his disposal doesn’t mean he’s done.

“We're going to add a lot of new data sources,” he said. “We're going to add stuff about municipal fragmentation and municipal debt, which are factors that may be related to over-policing and then subsequently to the officer-involved homicides. And then we’re just going to keep grinding along and try to find out as much as we can.”

**