How Likely Are You to be Killed by Police?

Trayvon Martin. Michael Brown. Eric Garner. Freddie Gray. Laquan McDonald. Alton Sterling. Philando Castile. Joseph Mann. Keith Scott. The list of black men killed at the hands of police goes on and on. And on and on.

To put it mildly, we have a problem. The first part of that problem is big picture and colorblind: in 2015 and 2016 alone, police in the United States killed over 2,000 people across the country. That’s about seven people per million. If that doesn’t sound like a lot, think of it this way. Every single day, about three people in America die because of our law enforcement officers—those sworn to serve and protect us. Here’s a map I made charting those deaths from the past two years:

Our police forces are far, far more murderous than ones in comparable countries. According to The Guardian, England and Wales combined witnessed 55 fatal police shootings over the past 24 years. The American police killed 55 people within the first 19 days of 2015 alone. Canadian police kills on average 25 people a year, about 2% of our pace. U.S. law enforcement killed more people in one week in 2015 than German police did in two years.

The vast majority (91%) of people killed by U.S. police died via gunshot. The rest died due to a mix of taser, death in custody, and “struck by vehicle.”

The police is inevitably and, by definition, armed. Many of the Americans they killed were not. In 2015 and 2016, less than half (47%) of the people killed by police were carrying a sidearm at the time of their encounter with law enforcement. 24% were not (or at least were not known to be) carrying any kind of weapon — not even a knife or non-lethal armament.

Neither were they running from the cops. Only 30% of those who died at the hands of police in 2015–2016 were trying to flee. Apparently not running away doesn’t get you any goodwill with law enforcement officers (but maybe trying to run away actually improves your chances of survival?).

The second part of our problem is very much not colorblind: killings by police are not random. Who you are and where you live make a big difference. With the recent surge in high-profile police shootings, I wanted to dig deeper into the landscape of police fatalities in this country. How big of an issue is it? (Very big, as mentioned above.) How much does race actually matter? (Spoiler: a lot.) What other factors are significant?

Data on police aggression is incredibly incomplete—an issue in and of itself. Like many other countries, the U.S. government has no comprehensive record on this topic. Thanks to the data science website Kaggle, I discovered a project called “The Counted” that The Guardian has been running to fill in this void. To quote: “The database will combine Guardian reporting with verified crowdsourced information to build a more comprehensive record of such fatalities. The Counted is the most thorough public accounting for deadly use of force in the US, but it will operate as an imperfect work in progress — and will be updated by Guardian reporters and interactive journalists frequently.” For my analysis, I supplemented The Counted with similar data that the Washington Post has been collecting using local news reports, law enforcement websites, social media, and independent databases. To normalize for population sizes, I also tapped into demographic data from the U.S. Census Bureau’s 2015 surveys.

The data spans from January 2015 to December 2016. It’s an incomplete dataset, made even more imperfect because police agencies don’t always report what exactly happened accurately. Nonetheless, it gives us a directionally pretty good picture of what’s going on. And it’s not a pretty picture. I showed you parts of it above. Now let’s take a look at the rest.

A quick note on this data project: As I’ll touch on again later, this analysis probably surfaced more questions than it answered. Police violence is a complicated, multi-faceted subject, and this project is very much just a starting point. It’s meant to be more of a survey course than a graduate-level seminar.

Who you are matters

All are not equal for police fatalities. To start, men are far more likely than women to be killed. For every million men, approximately 13.5 of them died at the hands of law enforcement in 2015 and 2016, compared to 0.7 for every million women. That’s a staggering 20x difference.

Taking into account the relative population size of different age groups, adults under the age of 35 are most likely to be killed. Young folks under 18 and old folks above 65 are least likely to be victims of police aggression, although it still happens. The youngest person to be killed in 2015–2016 was six-year old Jeremy Mardis, who died after two city marshals in Marksville, Louisiana allegedly fired 18 rounds of ammunition into his father’s car in November 2015. The oldest victim was Selma Elling of Indianapolis, killed along with her elderly husband when a police vehicle hit their car in November 2016.

And of course, race. Yes, race matters. A lot. In absolute numbers, more white people are killed by police than any other racial group (compared to blacks, twice as many whites died at the hands of cops in 2015 and 2016). But proportional to their population size, black people are most likely to be victims of police violence. Over the past two years, blacks’ police fatality rate was 2.5x that of whites. The Black Lives Matter movement is alive and well for a reason (because due to police shootings many black people are not).

A group that has received much less limelight is Native Americans, whose police-induced death rate is actually not far behind blacks’. Relative to their population size, Native Americans were killed at 2x the rate as white people in 2015–2016. Often forgotten in the public narrative, it’s perhaps time we started talking more about #NativeLivesMatter as well.

Among victims with known racial identity, Asians / Pacific Islanders have the lowest police fatality rate overall. In 2015 and 2016, blacks were 6x and Native Americans 5x as likely to be killed by law enforcement compared to Asians.

Where you live matters

Not only does police violence discriminate by gender, age, and race, but it also discriminates by location. In 2015 and 2016, someone living in San Francisco was twice as likely as a Chicagoan to die at the hands of law enforcement. In fact, if we looked at the 30 largest cities in the United States and normalized for their relative population sizes, the one with the most violent police (Las Vegas) killed people at more than 21x the rate as the city with the friendliest police (New York City — for real). Also, among the 30 largest cities, all of them except five (NYC, Nashville, Boston, Portland, Philadelphia) had a higher police fatality rate than the national rate of seven deaths per million.

What about states? In absolute numbers, more populous states (such as California, Texas, and Florida) saw more police killings in 2015–2016. If we accounted for relative state populations, though, a different set of “winners” emerge:

The deadliest state in 2015–2016 was North Carolina, where police killed 60 people for every million, which is about 6.5x the national rate. This is followed by Arizona, New Mexico, Washington DC if we count it as a state for the sake of this exercise, and Oklahoma. For some reason the American Northeast seemed to have seen relatively lower police fatality rates. The state with the “friendliest” police was Connecticut, followed by New York, North Dakota, Rhode Island and Maine. In 2015 and 2016, North Carolinians were 24x more likely than Connecticut residents to be killed by law enforcement. (And just to put all this into global perspective again, even as our friendliest state, Connecticut’s law enforcement still killed about 9x as many people per year compared to the entire United Kingdoms police force — and that’s in absolute numbers!)

Does this mean we should all start seriously considering a move to Connecticut? Not so fast. Location and race intersect when it comes to police’s use of force. The Northeast isn’t equally friendly to everyone.

Where you live matters differently depending on who you are

Perhaps not shockingly, which states have friendly versus unfriendly police forces is very much dependent on your race. This table below shows number of people killed by law enforcement per million, cut by race and state. It’s an interactive table, so feel free to play around with it. I’d recommend re-sorting the columns to see how the states rank differently in police fatality rates depending on race (to do so, click on “View larger version” below the table and click the drop-down arrow next to each column header).

There’s a lot going on in this table so I thought I’d point out a few trends that jumped out at me. In 2015–2016:

  • Where you lived mattered the most if you were black. As a black person there was a 66.8% fatality rate difference between Oklahoma (deadliest) and these relatively friendly states: Alaska, Connecticut, Idaho, Montana, North Dakota, New Hampshire, New Mexico, South Dakota, Vermont, Wyoming.
  • Where you lived mattered the least if you were Asian: you were “only” 14.3% more likely to be killed by law enforcement living in Colorado than in most other states.
  • Who you were mattered the most if you lived in Oklahoma. A black person was 66.8% more likely than an Asian person to be killed by police there.
  • Who you were mattered the least if you lived in Tennessee. All races (that The Guardian has data for) were killed by police there at close to the national rate.
  • Latino/as seemed to have had an especially rough time with police in the Mountain Time Zone. New Mexico, Idaho, Wyoming, Colorado, and Arizona all reported above-average Hispanic fatality rates. A Latino/a in New Mexico was almost 4x as likely to be killed compared to a random person living in America.
  • Native Americans were killed by law enforcement at an astronomical rate especially in Alaska, where their likelihood of death at the hands of police was 7.5x the national rate for a random person.
  • The most lethal race-state combination was being a black person in Oklahoma. Your chance of being killed by cops was almost 10x that of a random American.

So to sum up this rather dark data project… U.S. law enforcement fatally hurts people in staggering numbers. But it doesn’t hurt all people equally. Who you are matters. Men, adults under age 35, and black people are most likely to be killed by police. Where you live matters too, and the intersection between where you live and your race matters even more.

As I mentioned earlier, though, this project is very much just the start. It was my attempt to pull back the covers a tiny bit on a very complex issue.

For me, the biggest areas of further exploration are the whys. Why are our law enforcement agencies so much more murderous than counterparts in other countries? Why does Oklahoma’s police have such a violent relationship with black people, when many other southern states (including Mississippi, Tennessee, Georgia, and Arkansas) see relatively few black deaths at the hands of police? Why does New York City have such a low police fatality rate compared to other major cities? Why doesn’t the media talk more about the shocking numbers at which Native Americans are being killed? I have my guesses for these questions, but would want to understand more.

As the wrap-up here, my main recommendation is we absolutely need to get more reliable data on police aggression. I applaud The Guardian and Washington Post for running their own data collection processes on this topic. It’s infinitely better than nothing. But it’s far from perfect. Higher-quality data will give us a better understanding of what’s happening with police killings, why they’re happening, and what we can do to establish friendlier relationships between law enforcement officers and the people they are supposed to protect. To appropriate the wildly misguided #AllLivesMatter hashtag—black lives matter (and #NativeLivesMatter) because all lives should matter. And they certainly don’t today.

People & ops leader | data storyteller & nerd