University researches officer-involved deaths

In a 2014 Washington Post article, writer Wesley Lowery posed a question: How many police shootings are there in a year? 

The answer was fairly simple: No one knows. The federal government keeps no national database on the number of people killed by police. 

After seeing Lowery’s report on the lack of transparency about police-involved shootings, Brian Finch, director of USC Dornsife’s Center for Economic and Social Research Southern California Population Research Center and a professor of sociology and spatial sciences, set out to answer the same question.

Researchers at the CESR developed the National Officer-Involved Homicide Database, a catalog that documents fatal incidents involving the police that now serves as a tool for researchers to learn more about police department operations. At the time, the team noticed a number of online efforts in place to collect the data, such as Fatal Encounters, a national police homicide database run by D. Brian Burghart, the former editor and publisher of the Reno News & Review. NOIHD builds upon Fatal Encounters by adding more detail, granularity and cross-referencing to the initiative, Burghart said. 

Finch reached out to Burghart, who started Fatal Encounters in 2012, after reading about Burghart in the New York Times. Burghart said the data needed to be “institutionalized,” meaning an institution should manage the data rather than his independent work. 

“The likelihood of finding somebody who is interested in doing this for public service reasons, it’s just not likely,” Burghart said. “However, I think you’ll find that at a university. That’s what the NOIHD data is — the first step toward institutionalizing this data.” 

Though the NOIHD is rather new, researchers began to draw conclusions from the database’s detail and precision, including two papers published in early April at the annual Population Association of America meetings. 

Josh Gagné, a graduate student studying sociology at Stanford University, wrote one of the papers, which estimated a “risk-adjusted propensity to kill civilians” among police departments. He created an estimate for the number of civilian homicides expected from a department based on factors such as demographics or the income of their jurisdiction. Gagné said the NOIHD was particularly useful because he could link several variables together, which was crucial to his research. 

“What the NOIHD allowed me to do is not just have access to this Fatal Encounters data, which was already publicly available, but the NOIHD also links that to a bunch of these other datasets,” Gagné said. “It’s super convenient to go ahead and cross check these things [in NOIHD]. Sometimes, you have missing data that you can impute, but there’s this advantage of having a number of these datasets already conveniently located in one place.” 

For any given police department in the database in any year dating back to 2000, the NOIHD compiles variables like the amount of crimes, violent crimes and homicides, the violent crime rates and the amount of officer-involved homicides. It also records characteristics of the areas officers patrol, such as socioeconomic and racial demographics, information about officers assaulted or killed, whether there’s a use of force policy or a no-pursuit policy and the years of training required to become an officer in that department. 

The NOIHD is the first database with this many variables at the level of law enforcement agencies. The unprecedented detail of the database will let researchers study at a much lower level of aggregation, instead of at the county or state level, which has “never really been done,” Finch said. 

The NOIHD also revealed a potential effect of sheriff-coroner offices. The decision of whether a death is classified as an officer-involved homicide is overseen by a sheriff-coroner in only three states, including California. Usually, a coroner is an independent position, but in these states, the sheriff’s department directly oversees sheriff coroners. Finch said the NOIHD data show “severe undercounts” in jurisdictions with sheriff-coroners compared to those with independent coroners. 

Finch also said several other NOIHD-based papers work through the pipeline. The researchers look to next examine the impact of certain policies on officer-involved homicides, such as local gun restrictions or departmental no-chase policies. 

“There’s over 800 variables in the database,” Finch said. “Researchers that are using it can go in and explore tons of different hypotheses that they’re interested in and analyze them.”