America's poorest neighborhoods, and neighborhoods with the greatest number of non-white residents, are the most likely to see deadly police encounters, a new study finds. The only significant exception was that in the country's whitest neighborhoods, black Americans were disproportionately likely to be killed in police interactions. For blacks, the lowest-risk neighborhoods for police-involved deaths were those that were racially mixed—neither having far more people of color than average, nor far more whites.
Knowing the exact neighborhoods in which fatal police interactions occur can help community leaders know where to target programs aimed at reducing those deaths, and know where to look out for other problems that are a consequence of police use of force, researchers say. "In public health, we can think of these killings by police both as a cause of death that we should be concerned about in itself, but also as something that can cause community-wide effects," says Justin Feldman, a public-health researcher at New York University who worked on the study as a part of earning his doctoral degree. Feldman pointed to research showing that police violence can reduce children's school attendance, and increase symptoms of depression among black people in the area.
To figure out which neighborhoods most frequently suffered from police-involved deaths, Feldman and a team of public-health researchers used data gathered by the Guardian. In 2015 and 2016, Guardian reporters tried to collect information about every time a civilian died after interacting with police in the United States, including the deceased's name, race, sex, and age; the location where they interacted with police; and whether they had been armed. Such data isn't reliably available from any federal agency.
Using the locations the Guardian journalists had collected, Feldman's team figured out in which neighborhood each fatal interaction had occurred. Then the researchers used the U.S. Census, which has data on the race and incomes of families living in more than 74,000 neighborhoods throughout the U.S., to calculate which neighborhood characteristics were associated with having more police-involved deaths.
The team's methods make sense and are a good start to studying police-involved deaths as a public-health problem, says Ellicott Matthay, a researcher at the University of California–San Francisco who was not involved in Feldman's study. In November, the American Public Health Association—which also published Feldman's new study—announced it considered law enforcement use of force to be a public-health issue that needs to be reduced. Matthay noted, however, that Feldman's study isn't designed to show what neighborhood characteristics cause deadly police encounters. "It's not saying that living in a low-income neighborhood causes police-involved deaths, or that if you're black and you're in a predominantly white neighborhood, you'll be killed by police," she says. "It's saying, there's a pattern of what's happening and we need to look into it more." The cause might be there, but a different kind of study would be needed to show it.
One problem is the data, which isn't detailed enough to prove cause. "The Guardian is probably the best data that we have available on police deaths," Matthay says. It's just not quite what scientists would typically use in public-health studies. Meanwhile, national efforts to get research-quality data seem to have stalled. A previous study by Feldman and his team has found that the Guardian's database is reasonably accurate and much better than the federal government's National Vital Statistics System, which vastly undercounts civilian deaths resulting from police interactions, including shootings and deaths while in police custody that courts later ruled were homicides. After the Guardian published its work, the government tried to redesign how it counted law enforcement-related deaths. Preliminary results from that redesign, published in December of 2016, are the last data set available on the Bureau of Justice Statistics' "Arrest-Related Deaths" page.