About . . . . Classes . . . . Books . . . . Vita . . . . Blog. . . . Podcast

by Peter Moskos

April 10, 2015

Killed by Police (1 of 3): New Data!

Two years ago a somewhat shadowy person or group began compiling all media accounts of people killed by police. It's at killedbypolice.net. Best I can tell, he/she/it/they do a pretty good job.

According to the site: "Corporate news reports of people killed by nonmilitary law enforcement officers, whether in the line of duty or not, and regardless of reason or method. Inclusion implies neither wrongdoing nor justification on the part of the person killed or the officer involved. The post merely documents the occurrence of a death."

Not that this list is perfect. Certainly there might be some police-involved homicides that don't make the local paper or TV news broadcast. But there can't be too many. From May 2, 2013 to April 8, 2015, there have been 2,177 documented cases of people dying or being killed by police. (The vast majority are shot... but see died after being tased). And one cop killed his own family. Personally I wouldn't include off-duty and not job related, but that's a minor quibble, statistically.

The compiled data is impressive both in its thoroughness and documented nature. And compared to only other data we have, such as the crappy UCR data on justifiable police-involved homicides, this killed-by-police list is gold. It's the first time -- ever -- we can start looking at who police are shooting.

So I played with the data. I refined it and shined it real nice and turned it into a proper SPSS file (and removed the few who weren't killed in the 50 states plus DC). I have everything from when the list started (May 2, 2013) to April 8, 2015. So it's just under two years of data. N = 2,177

So the first thing we learn, which we knew, is that the UCR is a vast undercount. But now we have some idea about how much: a bit more than 50 percent. [For the more statistically inclined, the missing UCR data, at least with regards to race, does seem to be mostly random (which is good), which means the UCR data might be OK for some analysis.]

So we learn that police in America kill about three people a day. Three police-involved deaths a day may seem like a lot. But is it? America is a big country. It doesn't seem like an epidemic. In a typical day, 38 Americans will be murdered, 90 will die in car crashes, 110 Americans will commit suicide, and 120 will overdose on drugs. Maybe we have to begrudgingly accept three police-involved killings a day as par for a violent nation. But maybe not.

I think we could rather easily cut the number of people killed by police in half. That would save the lives of around 500 people a year. But I'll get to that in the third and final post.

[Part 2]


Anonymous said...

I think your math might be a little off here. KBP says 2200 killed by police over two years, or about 1100 killed per year by police. Cutting those deaths by two thirds would save about 660 lives, not 1500. A minor point, but perhaps one with correcting.

Moskos said...

Thank you! Corrected. (yeah, it's got a bit confusing to deal with a 23-month time frame.)

Anonymous said...

I would also add that according to UCR data, police are assaulted every year about 50,000 times. About 1/5th of the time is with a weapon (10,000 assaults with a weapon). About 6% were assaulted with firearms or knives (3,000 assaults with a deadly weapon).

So considering that police are attacked with deadly weapons about 3,000 times in the US in an average year, 1,100 deaths a year doesn't seems like an excessive amount.


Moskos said...

I think you're right.
But I'm so skeptical of those UCR numbers. How would they even know? I was assaulted as a cop. It never became a stat. I would guess 50,000 is a huge undercount. But sometimes cops can also say they're "assaulted" anytime they're touched. I just don't trust those numbers. But the point is valid.

Anonymous said...

While a suspect pushing an officer might be counted as an "assault" or resisting arrest might be counted as an "assault", I think one general thing we might say about the UCR data and Law Enforcement data in general is that as the severity of an incident goes up, the paper trail gets better.

If we accept that premise and focus solely on the firearm assaults (which I would argue pretty much all constitute and an assault with deadly force) then we're left with 2,500 firearm assaults, in which context the number of people killed by police still seems reasonable to me.

Moskos said...

I agree.

But why are people in California 5(!) times as likely to be killed by cops as people in New York or New Jersey or Michigan (I haven't posted on this yet, but will)? Something else is going on, too.

Anonymous said...

Wow. Great dataset.

How do I get this in an excel spreadsheet?

I like that it can be enhanced to add additional data elements.

After spending an hour, I have some theories about how not to get shot:

1. High speed chase ... bad idea.

2. Domestic problems? Leave the weapons behind.

3. Don't tase me bro ... if only.

4. Largest cities? You can easy get killed, but not by cops. A hell of a lot of killings are in small cities.