School of Public Health Researchers Find Many Unreported Police Killings

Harvard School of Public Health
Gregory B. Johnston


Significant numbers of deaths from interactions with law enforcement go unrecorded, researchers at the Harvard School of Public Health found in a study released Tuesday.

The study found that 93.1 percent of U.S. police-related deaths in 2015 were recorded by The Counted, a database run by The Guardian, while only 44.9 percent were documented as such by the National Vital Statistics System, which is managed by the federal government.

Justin M. Feldman, the study’s lead author and a doctoral student at the School of Public Health, said their research proved their hypothesis that police-related deaths are more likely to go unreported in low-income counties and for people “killed by non-firearm mechanisms.”

Feldman said that the underreporting of police-related deaths is caused by misclassification, not omission of cases.



Harvard School of Public Health

Harvard School of Public Health

“Everyone who dies goes into the NVSS; the question is whether they are classified as being killed by the police, assigned the diagnostic code of legal intervention, or homicide,” Feldman said.

Feldman added that he sees the relative accuracy of news sources in reporting such deaths as helpful, but a bad sign for the government.

“Local news reports are the best source of data we have right now of police killings… The other side is that it’s not really the role of media to be doing the government’s job,” Feldman said.

To his surprise, Feldman said his team found “no difference” in levels of underreporting in death certificates completed by medical examiners and coroners, whom Nancy Krieger—a Public Health School professor and author of the study—said are often political appointees. And when it comes to race and underreporting, Feldman said, “it gets a little complicated.”

While their study found no direct connection between underreporting and ethnicity, more people of color were killed by the police in regions with the largest underreporting, in states like Texas, Florida, and Oklahoma.

“If you were to analyze the [Centers for Disease Control's] data as is without correcting it, you would undercount the number of black people killed by the police,” Feldman said.

The results of the study raise the question of whether or not this undercounting is intentional. Feldman and Krieger, referring to the myriad of factors affecting the reporting of data, suggested that it would be impossible to pinpoint a cause from their study.

“What we could say from the study,” Krieger said, “because we can only speak to the results, is that the underreporting was highest for people who were black, who were under age eighteen, who lived in low income counties, and who were killed by non-firearm methods.”