Recently I came across a breathless headline in Salon: “Number of fatal shootings by police is expected to reach 1,000 for third year.” That’s an odd way to put it because A) it implies the number has been at a record high the past three years when B) this is only the third of three years the Washington Post has been counting and C) it’s not true.
According to the Washington Post data they cite, there were 991 people shot and killed by police in 2015 and 963 in 2016. The more accurate headline, as see in the actual Washington Post is: “Number of fatal shootings by police is nearly identical to last year.” Well, that’s a bummer if one is trying to hype a crisis. Also “reach” is not synonymous with “approach.” But I’m not hear to quibble about the semantics of a few dozen deaths.
The number of those killed by police seems to be steady at just under 1,000 per year. But given the increase in homicide, it’s actually surprising the number killed by police hasn’t increased. When violence goes down, as it did in the 1990s, police shoot fewer people. When violence goes up, as it did in the late 1980s, police shoot more people. I suspect there are two variables pulling in opposite directions. One involves more violence in general — increasing police-involved shootings — and the other involves less police engagement and fewer interactions with citizens. Less proactive policing means fewer interactions and less that can go wrong (and also more crime).
I looked at the Washington Post dataof those shot and killed by police in 2015 and 2016 and broke it down by states with more and fewer African-Americans. States that are more than 10 percent African American include 21 states plus D.C. (198 million people, 18 percent black, 36 million blacks). There are 29 states less than 10 percent African American (126 million people, 6 percent black, 7 million blacks).
People, all people, are 1.6 times more likely, per capita, to be shot and killed by police in states that are less than 10 percent black compared to states more than 10 percent African American. Blacks are still more likely than whites, per capita to be shot overall. But this ratio (2.6:1) doesn’t change significantly based on how black a state is.
For both whites and blacks, the likelihood of being shot by police is greater in states with fewer blacks. And the difference is rather large. There are seven states less than two percent black. In 2015 and 2016, zero blacks were shot and killed in Maine, New Hampshire, Utah, Vermont, Wyoming, Idaho, and Montana. But if you think cops don’t shoot people in these states, you’re wrong. Compared to the four states with the highest percentage of African-American (Mississippi, Louisiana, Georgia, and Maryland are more than 30 percent black), the overall rate of police-involved killings in states with few blacks is higher. And this is despite a lower rate of overall violence.
The implications of this are many, but for starters, if one wishes to reduce the number of people shot by police, it would make sense to focus on states that have more police-involved shootings in general. New Mexico, Alaska, Oklahoma, Wyoming, and Arizona have rates two and even three times the national average. Even though California ranks only 13 out of 51 (4.2 per million compared to 3.0 nationally), California is significant because it’s so large. In terms of reducing police-involved shootings, these are the low-hanging fruits.
Were the states with fewer blacks able to reduce their rate of lethal force to the level of states with more blacks, there would be an 18 percent nationwide reduction in the number of those killed by police. But this would require a move away from a general focus on policing and race toward a focus on police departments that shoot a lot of people of all races (particularly hispanics, it should be noted, and not examined here). Alas, at the intersection of police, race, and ideology, I’m not holding my breath.
And there’s a caveat: though state-level data is revealing, it’s often too general in terms of policy solutions. Intra-state differences are very large (and not examined here). Policing is local, and some local jurisdictions simply shoot a lot more peoplethan others. (That linked-to post is from two years old, but the basic points are solid).
Here are the raw numbers. As always, click to “embiggen.” Data available on request. Corrections welcomed. Strongly encouraged, even.
Sources: 2016 population, https://www.census.gov/quickfacts/fact/table/US/
Killed by Police, Washington Post, https://github.com/washingtonpost/data-police-shootings
Update July 17: Thanks to I.L., corrected and updated.
Update: 2020 caveat.
Hmmmm. I see an opinion piece under your moniker in the near future. Hopefully in the NYT or WAPO. I would be disappointed if it only appears in Greek American Quarterly.
I wonder how much of this disparity between cops shooting blacks is because of the relatively high number of police shootings in the western states coupled with relatively low numbers of blacks in those states.
I'm going to have to save your table and open it in another online app. Too damn hard to read even when I try to increase the font size.
Are these simple averages of the individual state rates, or are the states weighted by population?
Weighted. Everything is based on population.
Prof, what about the population density? Those high-shooting states seem to be sparsely populated, where a PO might be more likely to be alone and resort to deadly force, as opposed to say NYC where it is much easier to swarm the offender with cops who are then much more confident that they'll survive the encounter without shooting anyone.
And of course more densely populated states tend to have more blacks.
Also as a test of the race hypothesis: have you run the numbers for proportion of Latinos wrt police shootings?
JSM
Population density is probably very important. Looking at it at state level, though, is too crude (though still interesting). Other potentially key variables off the top of my head include percentage of people with guns, officers per capita (for backup/swarming), violent crime rate, and number of tweakers. But I'm probably going to leave the multivariate analysis to somebody else (I'll gladly share my data file).
But I suspect the real variable is officer training, police culture, and threat perception, which none of this data would reveal.
I'm leery of accurate recording of Latinos. And one problem is once you start breaking things down any more, the n's get uselessly low. But I do think there's a lot of there, there.
Anyway, my working hypothesis is the percentage of black people matters independently. I think most white people simply don't care when police kill people. Period. Or, to be more slightly more nuanced: whites are more likely to give police the benefit of doubt in questionable shootings and forgive even egregious (criminal) errors. Police are good, they say; sometimes they mess up; it's the price we pay to have police in a violent society. I call it the Al Sharpton Effect. If there's no pushback when police use lethal force, police have no incentive to use lethal force only as a last resort. I think it's incontrovertible that Sharpton — no matter what one thinks of his methods, moral, and motivations — is partly responsible for helping make the NYPD the least-shooting large police department in the country. Every shooting in NYC gets analyzed and bad shootings get hammered. That makes for better use of force guidelines and a better police department.
I don't think there is any correlation here. Rather, it is the case that western states have particularly high rates of officer involved shootings and blacks have higher populations east of the Mississippi River. Not that you state that there is causation.
Well, I am hypothesizing that there *might* be causation. But even if not, the correlation certainly leans toward something going on other than race.
This is based off of people that I know personally, but perhaps a reason there are more African Americans shot by police in areas where there is actually a greater population of whites has to do with a lack of knowledge or understanding of another culture. Reading that you're probably like "duh", but many people I know who grew up in a predominantly white community are much more racist, or at least prone to laughing at racist jokes, than those who grew up in mixed areas. This has nothing to do with population density or income levels either, just purely off of luck of the draw to who their neighbors were growing up. I know there is always more than one variable in situations like this, but that always comes to mind when I read articles like this.
Peter Moskos,
At one point you state that, you're surprised that police shootings aren't higher, due to an increase in Homicide rates. After you state, "I suspect there are two variables pulling in opposite directions. One involves more violence in general — increasing police-involved shootings — and the other involves less police engagement and fewer interactions with citizens. Less proactive policing means fewer interactions and less that can go wrong (and also more crime)." I was wondering if there is data to back your statement on Less police interaction leading to more crime? Is there a study that was done to confirm this statement? Do you think we should have more or less police interaction as an attempt to reduce police shootings? And if there is more police interaction, is there a better approach for police to avoid shootings?
Do you have any way to slice this data by class? (Rich vs. Poor). I'm proposing a hypothesis that police shootings, if separated by rich neighborhoods vs. poor will reveal a strong correlation. Here is what I propose:
Black per capita is higher only if you are a poor black. If you are rich, it is actually equal to white.
Similarly, white per capita surges higher if it is poor white.
Another way to put it. Your rate of being shot surges based on where you live, not skin color, and it is based on whether that area is rich vs. poor.
Again, it would be nice if the data could prove or disprove it.