Data on Police Shootings

The high-profile police shootings of 2014 focused attention on the problem of police violence in America. Researchers quickly discovered, though, that the data needed to seriously study the problem did not exist. The Obama administration got the FBI to begin a nationwide effort to track police use of deadly force, and that effort is now up and running and should provide better data going forward. Meanwhile several private groups have started their own efforts, including The Washington Post, which right now probably has the most comprehensive database.

These efforts have revealed that US police kill about 1,100 citizens a year, which is twice what the FBI counted under its old system. With this expanded dataset it is possible to explore many sorts of variation. One thing that jumps out is regional and local variation; the rate of police killings in Arizona is four times what it is in Illinois.

This chart shows variation by race, but counted in a weird way, by lifetime chance of being killed; since the rate of violence in America varies so much from decade to decade I would much prefer a chart with annual death rates. Anyway this shows that a black man in America has about a 1 in 1,000 chance of being killed by the police, 2.5 times the rate for white Americans. (Of course that means a majority of victims of police shootings are white, since there are 5 times as many white Americans as black.)

One thing experts are fighting about right now is the exact mechanism of the disparity. I mean,  racism is surely behind this at some level, but how? There is some evidence that black Americans are several times more likely to be stopped by the police than whites, so it may be that the chance any given encounter will lead to violence might be no more, or even less, for blacks, and the rate of killings is just higher because there are more encounters. But since the data on who the police stop and question is very poor, it's hard to say. Those local disparities also come into play; more blacks may live in the sort of high crime, high violence neighborhoods where the police are on hair-trigger alert. Because while some cops are just brutal and violent, most shootings happen because the officers are scared. Maybe they are more scared than they should be, and quicker to reach for their guns than they should be, but more than 50 cops are shot and killed each year, so the risk to them is real.

The root cause, I believe, is a breakdown in trust between the police and the people they are supposed to be protecting, and since that is a reflection of a general breakdown in how much Americans trust each other, it is going to be a hard problem to solve.

On the other hand there is some evidence, presented here, that the sheer number of guns in the US is a big factor; maybe the police are more scared because the people they deal with are just more likely to be carrying a gun.