Analysis of Washington Post police-shootings data reveals surprising result – nearly 2x more whites than blacks shot by police

Note: While not the usual fare of WUWT, after some discussion with Mr. Eschenbach, I decided that this was an impartial data analysis, and that publishing it would be a public service that may be helpful in these troubled times. However, given the sensitive and inflammatory nature of the subject matter in the context of today’s events, I’ve decided not to allow comments on this article. The data speaks for itself, and the goal here is to present the data analysis minus the rhetoric the topic will generate. I take no position on the topic, but I do feel that better data presentation leads to better understanding of problems, and this data presentation may be helpful to many. There are other forums where I’m certain this data and the results will be discussed. – Anthony Watts (updated, headline was extended at 2PM PST)

When Arrests Go Bad

Guest Essay by Willis Eschenbach

I got to thinking about the issues of race regarding the recent tragic police shootings, both the shootings of police and the shootings by police. The best data is from the Washington Post, which has a detailed site listing all of the people killed by police, which begins in 2015 and goes to the present. I thought I’d analyze their data. I looked at the data for the year 2015 because the full 2016 data is not in yet, and also in order to be able to compare it to other annual datasets.

First, there were 990 fatal police shootings in 2015. How does this compare to other causes of death? Well, I can’t tell you because so few people are killed by police. The number is so small that it is outside the range of the usual mortality lists. I can say that death by police is not in the top fifty causes of death in the US, so it is relatively rare. It is extremely rare for women, because the overwhelming majority of those killed by police were men.  And I would be greatly remiss if I did not highlight that in addition to the 990 civilian deaths, there were 51 police deaths in 2015 …

Regarding the civilians killed by police, more than nine-tenths of them were armed at the time—58% of the people killed had a gun or explosives, 17% wielded a knife or edged weapon, 9% were unarmed, 6% used a vehicle, and tragically, 3% had a toy weapon. The rest used mostly clubs, hatchets, hammers, baseball bats, the usual assortment of your basic stone age deadly weapons that can kill you just as dead now as they could in 1500 BC.


By race and ethnicity, there were 494 whites, 258 blacks, 172 Hispanics, 15 Asians, 14 American Indians or Alaskan natives, 9 “other” races, and 28 deaths with the race not specified. Which leads to the question … is there a racial imbalance? And in particular, are African-American people being killed at an excessive rate?


Now, many folks calculate the death rates of the groups by comparing the numbers killed, to the corresponding numbers of that group in the general population … but you can’t do that. It leads to wildly incorrect conclusions. Here is an example that shows why comparing numbers of police shootings to the corresponding number of individuals in the general population leads to big errors:

Men make up about 50% of the general population, but men comprise 96% of those killed by police. Does this huge number of “excess male deaths” prove that the police are being sexist and that they are biased against men? Does this imbalance in the number of men killed mean that we need a “Male Lives Matter Too” movement?

Of course not. Instead, it simply demonstrates that men both commit and are arrested for far more crimes of violence than women; that men are far more likely than women to both carry and use weapons; that men are far more likely to both threaten and commit serious violence against a police officer than are women; and most importantly, that men are far more likely than women to violently resist arrest.

The one thread in the Washington Post database that is common to the overwhelming majority of police shootings is that almost all of the civilians were killed while resisting arrest. The records of the deaths show that to keep from going to jail, people were fighting with the police; they were shooting at the police; if they didn’t have a weapon handy they were punching the police or trying to drown the police; they were trying to run the officer over with a car or smash the officer’s skull with a baseball bat … but regardless of the endless variations of method, almost everyone killed by the police was in exactly the same situation—no less than 97% of the people fatally shot by police were killed during the course of an arrest that went bad when the people tried to resist.

Now, don’t get me wrong here. Certainly there were people unjustifiably killed during 2015. Even though only about three percent were not resisting arrest, that’s still thirty-four deaths of people who were not resisting the police when they were killed … double-plus ungood. And resisting arrest doesn’t automatically justify a death sentence. I’m sure you all remember the unarmed black man who was shot and killed in February 2015 in a Pennsylvania restaurant, while lying face down on the floor after being tasered. So yes, wrongful and unjustifiable deaths do occur. In addition, as happens most years, there were people killed by accident in 2015, like the bystander to a gun battle between undercover police and violent criminals who was killed by a badly aimed police bullet. And there were a couple of tragic misperceptions of the type where someone was holding a cell phone and the officer thought it was a gun.

But justified or not, 97% of the deaths by police were variations on a simple age-old theme—someone was violently resisting arrest, usually with a weapon of some sort, and the arrest went bad.
Now, I demonstrated above with men versus women that you can’t simply compare killings against the corresponding raw population figures. It gives us very wrong answers. So what should the number of killings be compared to?

Since the common thread in the killings is that the person was resisting arrest, we need to compare how often people of each race get killed by police, with how often people of the same race get arrested by police. But clearly, we’re not interested in arrests for jaywalking and the like. Since 97% of these deaths are occurring in the context of people violently resisting arrest, they are best compared to the corresponding number of arrests for violent crimes.

Here are the results of that comparison for 2015.

  • For every 10,000 white people arrested for a violent crime, 38 white people were killed by police (± 2).
  • For every 10,000 hispanic people arrested for a violent crime, 21 hispanic people were killed by police (± 3).
  • For every 10,000 black people arrested for a violent crime, 21 black people were killed by police (± 2).

Go figure … I was as surprised as you, so I’ve triple checked the numbers, and it’s true—the odds of a given arrest going bad and ending up in a death are much greater for white men than for black or hispanic men.

Please note that this result says nothing about the existence of racist police officers in America. Sadly, while the situation is immensely better than in my youth, we know that there are still far too many racists in the US … including in the police forces.

And it says nothing about racist policing in America. Again the beneficial changes in my lifetime have been huge, but there still remain any number of places where DWB is often a crime, and jurisdictions where people of color are regularly harassed and stopped by police.

Are there unjustified killings among the Washington Post data? You can be certain there are, it is a rare but real issue. Is there more work to be done? Assuredly. But the Post data doesn’t and can’t answer those questions

What the Washington Post data can and does show us is that death at the hands of police is a problem for people of all races and ethnicities. It’s not something happening preferentially to black people, quite the opposite. Thankfully, it doesn’t happen a lot, but whenever it happens, we need to make sure it was justified … and if not, we need to hold the responsible parties to account.

What we don’t need to do is to turn an issue of excessive use of police force into a racial issue.

Now, I’m your basic melanin-deficient guy. My mom said the same thing to me that good moms of every color all over America say to their sons, which can be boiled down to “If you get in trouble with the police, do what they tell you, be respectful, and don’t resist arrest!”.  And like a good son, I followed my mom’s advice the three times that my corpus delecti was deposited in a squad car and hauled off ignominiously to the local cop shop to answer questions, the one time I was arrested and jailed and rather unceremoniously sentenced and tossed into prison for twenty days, and another four times when I was stopped and questioned and searched by police, once unexpectedly at night in an isolated area at gunpoint, scary stuff. Ah, the joys of youth, I was a regular stop-and-frisk magnet as a somewhat scruffy long-haired young man.

Despite my rather colorful past, as a result of listening to my mom I’m not a statistic. I shut my mouth and went along to get along. The numbers are undeniable, the conclusion is obvious. If you don’t want to die at the hands of the police, DON’T RESIST ARREST!

That doesn’t mean that the deaths of the 97% who were resisting arrest are justified, that’s a totally separate question. Undoubtedly, some were justified and some weren’t. And it doesn’t mean there aren’t unjustifiable deaths, they do exist. There just aren’t many of them. All it means is if you don’t want to get killed by the cops, don’t resist arrest.

Look, if you want to avoid shark attacks, it’s simple—stay in the shade of an oak tree. It’s the best shark repellent known. And if you want to avoid dying at the hands of the police, remember what your mom said, and simply and politely do what the police tell you to do. The job of the police is tough enough, they are already on edge, as they should be given the potential danger of every arrest. So smile and go along, stay safe in the shade of the oak tree. It’s not rocket surgery. If you don’t resist, your odds of getting killed are minuscule. Not zero, to be sure, and you also might win the lottery, but if you’re not resisting arrest it’s almost certain that you won’t be killed by the police.

Here’s the final curiosity. According to the actual data above, it appears that the mothers of black and hispanic men are doing a better job of getting that message across to their sons than are the white moms, and that black and hispanic men have paid more attention to their moms’ universal message. After all, arrests of white men end up in their death nearly twice as often as do the arrests of  black and hispanic men … I ascribe it to arrogance. In my experience, white guys of the kind who run afoul of the law are more likely to challenge and mouth off to the cops, and are more apt to believe they’re invincible, bulletproof, and above the law. But hey, that’s just me, your explanation may vary.

Now, the numbers of deaths are low. But still, with over nine hundred citizens killed by police, surely we can do better than that. And the number of unjustified deaths should be zero. So here, in no particular order, is what I’d do:

  • Require that all police be trained in Aikido. Aikido is a martial art which is designed to NOT hurt the other person. Instead, you learn to immobilize someone, disarm them, and prevent their escape without harm to either them or yourself. During the period when I studied Aikido on Maui, the Sensei was a Maui policeman. Many of the Maui cops had spent years studying Aikido at the dojo, and they were very proficient. If they grabbed your sorry okole, or if you tried to attack them, you could count on two things—nobody got hurt, and your okole stayed grabbed.

In Aikido, you never have an opponent—instead, you have a partner. Seeing the person in front of you as your partner instead of your opponent is a very different mindset. It is much more profitable way of approaching violent interactions. Aikido contains no kicks, no punches, nothing designed to harm the other person. Instead, it is all about disarming the other person and ending the situation with nobody getting hurt. “Ai-ki-do” means the path of harmony with energy, it is a non-confrontational martial art.

  • As much effort as we put into training police how to win violent situations, we need to put that same amount of effort into training police how to avoid, defuse, and minimize violent situations. Among other tools, Aikido is very important in this crucial aspect of police work.
  • Increase the involvement of the police with the community, particularly in less formal situations (sports, schools, big brother/big sister programs, martial arts, neighborhood watch meetings, Christmas toy drives, holidays, pancake breakfasts for charity, etc.) The only way to repair and improve the trust between the citizens and the police is for us all to get to know each other. I greatly enjoyed getting to know the police officers I trained with in the Maui dojo, it changed my whole mindset about police.
  • Increase the number of women on the police force, particularly beat cops. The cops on the beat are the backbone of the force, they are the public face of the police, and they are often the ones involved in the high-voltage interactions. We need many more female street cops.
  • Get the majority of the city police out of the cars. A good policewoman walking a beat knows every shopkeeper along the way … and what is more important, every shopkeeper knows the policewoman. A cop driving by in a car knows nobody and nobody knows them … which is a bad condition for any society. Plus walking a beat makes you fit, while sitting in a car makes you fat.
  • Body cameras are no magic bullet, but the truth is good for everyone—it protects police and citizens alike. However, there are many unanswered questions as to exactly how to implement that while protecting the privacy rights of both the police and the citizens.
  • Increase transparency regarding possible police misconduct as far as is consonant with police requirements, officer and citizen rights, and legal restrictions. In particular, it is not enough that justice be done in cases of police misconduct. The public needs to see that justice is done, even if it can not be seen until well after the occurrence when the dust has settled and all the facts are in.

The tragedy to me in all of this is that opposition to the excessive use of force by a small number of police officers, which should be a cause to unite at least the populace and hopefully the police as well, has instead become a divisive racially-based issue. The “Black Lives Matter Too” movement is splitting rather than uniting opposition to excess force, based on the false idea that in terms of killings by police, white lives matter and black lives don’t … but the Washington Post hard facts show otherwise. The actual data on shootings by police clearly demonstrate that nobody is privileged in that regard, no race is singled out or excluded, nobody is exempt.

Deaths in police-civilian interactions, both justified and unjustified, are a concern to everyone. The burden and the pain and the cost of death by police falls on people of all colors—on black, white, brown, yellow, red, and absolutely on blue. All the groups need to unite with each other and work together with good will to solve it. And in the discussion, we need to remember that of all of the colors, the group that suffers the most unjustified deaths by an overwhelmingly wide margin every single year is … blue. Yes, black people do have a legitimate grievance, absolutely so, and we should never minimize it … but regarding police shootings, so do white people and brown people and blue people and all the rest.

Finally, people of all races should welcome the news that death rates per arrest for white people are nearly twice those for black and hispanic people … for a curiously realpolitik reason. Consider: which one is more likely to bring lasting change in the police use of force—news of excess white people being killed by cops, or news of excess black people being killed by cops?

Best regards to all,


1) Be clear that I am NOT saying that a police shooting is justified simply because the person is resisting arrest. However, we are talking about violent encounters. In almost three-quarters of the deaths (74%) the person was actively attacking the officer and a fight was in progress. In another 23% of the deaths the person was actively resisting arrest, and 60% of that group were armed with a deadly weapon.  Heck, more than a third of the unarmed civilians were actively attacking the officer when they were shot. And in all of the deaths, the officer didn’t know at the time whether the person had further concealed weapons. From reading the admittedly extremely brief accounts, I would say that most of the killings seem justified given the circumstances, although a few seem obviously and tragically wrong, some of the others seem very questionable, and a somewhat larger number are not determinable from the reports. Finally, we should work to reduce the numbers of deaths whether they are justified or not.

2) Arrest data is from FBI Table 43, Arrests by Race. I have used the category of arrests for “violent crimes” for the comparison of deaths and arrests. The FBI defines the category as follows: “Violent crimes are offenses of murder and nonnegligent manslaughter, rape, robbery, and aggravated assault.” I have used the 2014 figures because the 2015 figures are not yet available. For the larger populations (white, black, and hispanic) the percentage shares of the arrests don’t vary much from year to year, so any errors will be small. Finally, about 20% of the jurisdictions do not report hispanic vs. non-hispanic arrests, so I have used the percentage values to calculate the probable numbers if all jurisdictions had reported. Again, any error from this will likely be small.

3) If you don’t remember the death of the unarmed black man in the Pennsylvania restaurant who was shot and killed while lying face-down after being tasered, don’t feel bad, blame me …  I fear I have engaged in a bit of constructive deception. I say that the reason you haven’t heard about his death is because contrary to what I stated above, he was actually an unarmed white man in a Pennsylvania restaurant who was shot and killed while lying face-down after being tasered … although as above, your explanation may vary.

4) I’m constitutionally incapable of watching my friends drive a bus off a cliff without saying something. So for my progressive friends who support the public actions of the Black Lives Matter Too movement, let me say that the BLMT movement has devised an ingenious way to make sure that their message gets out to thousands and thousands of people … they’ve taken to commandeering the off-ramps, jamming the streets, and blocking the freeways.

Unfortunately, the thousands and thousands of women and men receiving the BLMT message are mostly just the poor random fools trapped on the freeways in their cars, unable to get to the off-ramp, fuming behind their steering wheels and cursing the @$$&0!#$ who are keeping them from getting home to their husbands, wives, and kids. Anyone who thinks this will increase support for the BLMT movement has not thought it through. When a working single mom ends up cursing black people because she’s unable to make it home to her kids in time, someone desperately needs to reconsider the choice of methods. The BLMT actions are creating new Trump voters at a rate of knots … just saying that unless that’s your objective, you might suggest they re-think their tactics, and you might reconsider your support for their actions …

5) Care must be taken when the numbers are small. Above, I left out the numbers for Asians (28 deaths/10000 arrests, 95% confidence interval [CI] = 15 to 47 deaths/10000) and american indian/alaskan natives (18 deaths/10000 arrests, 95% CI = 10 to 36 deaths/10000) because they are very uncertain. They have wide 95% confidence intervals due to the small numbers of both arrests and deaths in each group. Given those uncertainties, however, the central estimates of both are within the same general range as the other groups, validating the method. As noted above, for the other much larger groups the results are definitely more solid, because the corresponding 95% confidence intervals are small—white, ± 2 deaths; black, ± 2 deaths; and hispanic, ± 3 deaths per 10000. Uncertainties were calculated using the Wilson score interval.

6) The same care must be taken with other small subsets of the 990 deaths. For example, it would appear at first glance that the number of unarmed white people killed (38) is significantly larger than the number of unarmed black people killed (32). However, because of the small amount of data, one standard error for unarmed white deaths gives a value including uncertainty of 32 to 45 unarmed white deaths. Similarly, one standard error regarding unarmed black deaths gives a value including uncertainty of 27 to 38 unarmed black deaths. Since these two uncertainty ranges overlap, there is no statistically significant difference between the unarmed white and unarmed black deaths … and as a result, there’s nothing we can discuss about the unarmed deaths until we have more data, because the numbers are too small to be relied on. As another example, the same is true of the deaths of those 34 individuals who were not resisting arrest—the uncertainties are far too great to draw conclusions from the racial makeup of that small a group.

7) Regarding my use of the term “melanin-deficient” to describe my average white guy skin color, I’m reliably informed that the term “deficient” expresses a negative value judgement reminiscent of “disabled”, implying that the person is lacking in some way. Worse, it embodies a micro-aggression that might decrease the self-esteem and diminish the positive body image of those other unfortunates in my situation. As a result, I’m told that the currently approved term is “melanin-challenged” …

8) After finishing writing up this analysis of the data, I came across a discussion today of the same question in The Economist. To my surprise, although they used totally different and much more complete data, they came to exactly the same conclusion—black people get killed less frequently by police, and white people get killed more frequently by police, than one would expect given the number of interactions they have with police. Doesn’t mean that racism is dead, the study in The Economist agrees with my saying above that there’s still lots of racist policing out there … and it also strongly supports my conclusions about the number of deaths.