Police Killings, Gender and Race
by Mike Kimel
Police Killings, Gender and Race
Police killings of unarmed Black men have garnered a lot of attention lately. Black men are killed by police in numbers disproportionate to their share of the population. I want to put some numbers on that issue later in the post, and those numbers run counter to conventional wisdom. But first I want to warm up on something less controversial, though it is even more disproportionate, namely killings of police by gender. Still, context is in order. So here’s a little table I assembled with data from a number of sources (citations at the bottom of this post):
Figure 1 – Male v. FemaleWhat we see here is that women are disproportionately less likely to be killed by police relative to their share of the population. Women also make a disproportionately small share of police personnel, and female police officers are far less likely to be killed in the line of duty than a male police officer. (I.e., women are 12.2% of the total force, but 4.2% of total police officers killed.) Women are an even lower share of cop killers; only 3.2% of people who kill police officers are women.
To put these deaths in perspective vis a vis the wider population, under 10% of all murders were committed by women, but women make up over 29% of murder victims.
So… women are far less likely to be killers than men are, and even less likely to be cop killers. On the other hand, they are less likely to be police personnel, and among police personnel, they are far less likely to be killed than their male counterparts. Or, to put it a different way, men are are disproportionately likely to be killed by police and to be police. Among police, men are disproportionately likely to be killed. Men are also far more likely to be murderers, murder victims, and cop killers than women are. One piece of interesting information that isn’t in this table because I haven’t found it: what percentage of killings by police are committed by male v. female personnel. One can, of course, hazard a guess.
One last point before we move to the real meat and potatoes of this post: the fact that 96% of the people killed by police are men is not evidence of discrimination against men. After all, about the same percentage of cop killers are men. If 96% of cop killers were men, but men were 50% of those killed by cops, we wouldn’t be praising cops for their even handedness in dealing with men. Why? Well, when dealing with a potentially dangerous civilian, police face a trade-off between risk to themselves and risk to the civilian. (At the extreme, all risk to the civilian is eliminated if that civilian kills the cop, and at the other extreme, all risk to the cop is eliminated if the cop kills the civilian.)
So if 96% of cop killers were men, but men made up 50% of those killed by cops, we’d conclude that a) cops were jeopardizing their own safety by letting men get away with too much threatening behavior, b) cops were way too-hair triggered around women, or c) both. <bold>In other words, on average, if X% of the threat of lethal force against police comes from Group A, we expect to see X% of the use of lethal force by police to fall on Group A.</bold> Remember that relationship – we will be using it again.
Moving along (we’ve got a lot to cover, no time to dawdle), using the same sources I came up with the following two tables. That’s two tables, because the different sources aren’t consistent with their use of Hispanic or Latino people. In some cases, Hispanic and Latino people are folded into Black, White, and other groups, and in some cases they are broken out separately. So the two tables immediately below are as clean (from the sources) as I could make them:
Figure 2 – Racial, First Tables
The first of these two Race-based tables above certainly looks damning. Black people are disproportionately likely to be killed by police. Additionally, White people are more likely to be police than their share of the population would indicate. Conversely, Hispanic people are disproportionately less likely to become police personnel. So far, the story fits the conventional narrative: Racist white cops are excessively prone to killing Black people.
The second table shows that though Black people are disproportionately likely to be killed by cops, they are even more disproportionately represented among those who kill cops, and by quite a bit. One can chicken and egg this a number of different ways, but the wave function collapses when we note that relative to their share of the population, Black people are overly represented among murder victims, and even more so among those who commit murder. I have heard heroic attempts to blame the high murder rate in some areas on excessive policing, but they sound pretty ludicrous.
It is also worth noting that Black police personnel are less likely to be killed than other police, on average, and White police personnel are disproportionately more likely to be killed. Again, data on which police officers kill while on the job is not available, or if it is, I haven’t found it.
While the two tables above are interesting, it might make sense to merge the two. I tried doing it a number of different ways – ideally one could keep Hispanic or Latino as its own separate classification for granularity purposes but I just couldn’t make it work. My best attempt is below. In it, I assigned Hispanic or Latino numbers to Black, White, and Other groups based on their proportion of the general population.
Figure 3 – Final Table, Black & White
I’ve color coded numbers: red means a higher percentage than a group’s share of the population, and blue means a lower a lower percentage than a group’s share of the population. Whites are over-represented among Police personnel, and among Police, Whites are the most likely to be killed even accounting for their share of the police force. They are under-represented on other categories. Black people are a quarter of all people killed by police, and 40% of those who kill police. Additionally, Black people are about 44% of those murdered in the country, and 47% of the murderers.
Now, let’s move to where I point out something in the data that is going to piss people off, and this is it: From what I can tell, the data implies that Police are less, not more likely to kill Black people given the same behavior and/or perceived threat. Here’s why. As noted above, we expect that if police are rational, they will react to perceived threats upon their own lives. If one group contains individuals who are more of a threat to police (and other civilians) than other groups, then that group will be met with lethal force by police more frequently. If a group is less of a threat to police (and other civilians), police will be less likely to use lethal force on that group. In the end, use of lethal force by the police against any given group will approximate the threat posed by that group.
Back to the table. Black people are 13% of the population, 40% of those who kill police, and 26% of those killed by cops. Focus on those last two percentages. 40% of cop killers are black, and 26% of people killed by cops are Black. This means that on average, <ital>the do not react as strictly to threats from Black people as they do to the same threat posed by people who are not Black.</ital> Conversely, White people (77% of the population) are 55% of those who kill police and 65% of those killed by police. Here, on average, police are more likely to meet threatening behavior with lethal force than the threat posed. Similarly, Asians and Native combined (10% of the population) are 3% of those who kill police, and 6% of those who are killed by police. They too are more likely to be met with lethal force by the police than the level threat they impose on police would imply.
In other words, contrary to popular perception, the data seems to show that Black people are less likely to be met with lethal force than non-Black people would be for behavior that genuinely constitutes a threat toward police personnel. At the same time, police are, on average, harsher toward Black people than toward non-Black people because the police are more likely expect a threat from Black people than from non-Black people. (Again, due to the high murder rate in general, and high rate of killing police.) Put plainly: Black people are subject to more low-level police attention (i.e., traffic stops, searches), but less high-level police attention (i.e., actions that reduce violent crime).
These two competing tendencies create winners and losers. The biggest losers are probably non-criminal Black people. Non-criminal Black people pay twice: on the one hand, they suffer the indignity of extra police scrutiny relative to non-Blacks. On the other hand, the most violent members of the Black community are, on average, more likely to be allowed to go about their business unimpeded than the most violent members of the White, Asian, and Native American communities.
How to solve this problem – well, that’s a post for another day, perhaps. But I did want to add one more thing that I see in the data. The current debate going on in society has generated a few recommendations, one of which is that there will be fewer killings of Black men by police if police departments more closely resemble their constituents. In other words, a predominantly Black community should have a predominantly Black police force. Now, the table above indicates that Black people are 13.3% of the total population, but 12.3% of the police force. That means that there may be jurisdictions with large disparities, but overall, nationally, there isn’t a large disparity (and my guess is that its shrinking.)
A slightly larger disparity is the race of the police officers who are killed in the line of duty. While Black people are 12.3% of total police, they are less than 11% of total police officers killed. A positive interpretation is that Black police officers have more understanding of Black civilians and therefore are better at diffusing situations that would otherwise lead to violence. In that case, more Black police officers will reduce total deaths. A negative interpretation is that Black police officers are more likely to escalate to fatal actions than non-Black police officers, getting the drop on potential threats (and perhaps non-threats). A third option is that Black people simply have different jobs than non-Blacks in the police force.
There is a lot the data doesn’t show. There’s too much information that just isn’t being collected. But that’s no excuse for misinterpreting the data we already have available. Ignoring what the data tells us is not only stupid, it is a guarantee that the problem will continue to be framed incorrectly, which in turn means it will not be solved. That gets people killed.
—
data sources
Dep’t of Justice on race & sex of Police Personnel – see tables 4 and 5
Dep’t of Justice on law enforcement personnel killed, by race & gender
Dep’t of Justice on tace & gender information on those who killed law enforcement personnel
Dep’t of Justice on murder info, by offender and victim race & gender
Census, general population demographics
Census, percentage of Hispanics who identify as Black
Washington Post study on people killed by police
One of the issues with the police numeric is the consistency and overall accuracy of it. Some departments report it, some report some, and some do not report it. The FBI numeric is also suspect. We do not have a clear picture.
I suspect they report what is beneficial to them also as any other org. might want to do to improve their stance.
run,
The data the police have the biggest reason to downplay is number of people police kill. As a result, I got that from the Washington Post, which tried to count every such death last year.
On the other hand, there is reason to believe that the FBI figures on the percentage of homicides that are committed by Black men is actually below the real number. The reason is that many of the homicides of Black men go unsolved. Put politely, in some neighborhoods, there is a tendency for witnesses not to talk to the police. (Think the “snitches get stitches” slogan.) Those neighborhoods tend to have high homicide rates, and largely Black populations, suggesting that a large proportion of unsolved homicides are by Black people. The data on homicide offenders only lists the offender when the offender’s race is known.
Additionally, as the DoJ was forced to conclude when investigating the Michael Brown killing, a surprising number of witnesses who do come forward seem to tell stories that completely contradict all physical evidence.
Mike:
I had a similar discussion with the 85 percentile people who propose 75 mph highways. They too rely on reported accidents and how fast the driver was going in citing their safety numbers. Their argument? Higher speeds are safer and it is only the ones going slower which cause the accidents. This is nonsense as speed variation is the issue. Those exceeding the posted speed limit also cause many of the highway deaths as opposed to those going slow. Their study relied on reported accidents. Would you tell a police officer, I was doing 80 mph in a 65 or 70 mph section of the highway? I doubt you would. Faster speeds cause more deaths.
There is enough data and studies out there to refute your hypothesis as once again the DOJ and FBI rely upon reported incidents. “The data underlying the FBI tally “is estimated to cover 46% of officer-involved homicides at best” for the years 2003-2009 and 2011, the BJS report concluded. But the published FBI tallies cover even fewer of the total deaths, closer to 41%, in part because the FBI publishes no data from Florida. A separate tally of “arrest-related deaths”, conducted by BJS itself, was slightly more accurate for the years in question, capturing 49% of law enforcement homicides, at best, the report found.” https://www.theguardian.com/us-news/2015/mar/04/police-killed-people-fbi-data-justifiable-homicides
There is ample reason to suspect the FBI numbers are too low. The numeric does not include anything from Florida for sure and the numbers may be skewed for precisely the same reason I pointed out for speeding. The person reporting the incident is more than likely the same one who did the shooting. They are not going to say; “oops! I got nervous and I pulled the trigger too soon.” They will alter the story or what ever portion they can remember of it. You can not rely upon the data being reported as it is under reported or skewed to protect the shooter.
Using WP data is not accurate either as they do not have access to all of the data and furthermore it relies upon the same reports. There is a major issue here and we do not have the complete picture. Black Americans do have something to fear and furthermore they have little recourse today.
A possible confounding factor which occurs to me: more crimes occur after dark then in daylight. Black people are harder to see than white people in the dark. If both groups are shooting at each other in the dark, more of the white group will be hit (relative to group size).
I am not so much advocating that as a serious factor as advocating that a more detailed look at the cases would be required to form valid conclusions.
We can’t do controlled experiments on these issues. If I could, I would try one in which all people were painted green. The hypothesis given in this post is not unreasonable but here is another which I think historical data supports: most hypotheses which seem reasonable prior to testing turn out to be wrong when tested, e.g., the Michelson-Morely experiment.
Meanwhile, we can only look at the data we have, but it might be interesting to compare USA data with that from other countries – such as England, where the police usually don’t carry guns.
Jim,
Michelson & Morely, not to mention the rest of the physics community, abandoned their previous beliefs when they found those beliefs contradicted the information they were able to collect. Hopefully we as a society can do the same on this subject. As I said, getting it right matters. There are literally lives at stake. You can’t solve a problem if you don’t understand or are unwilling to understand it.
Run,
Bear in mind that arguments imply things about the world.
Your argument restates as: the high death rate in the South side of Chicago is due to over policing.
My interpretation is that the data says there is under policing.
Run,
To be clear, are you stating that the way the FBI collects data understates the percentage (not the number, the percentage) of police killings that are of Black people to a significant degree? Or are you saying that FbBI statistics are overstating the percentage of killings of police that are committed by Black people?
I like good data. It helps me stop wasting $millions when I am rearranging a shop floor. When the data is erroneous how can one determine a direction?
Mike
i used to be a surveyor for a living. many surveyors rely upon statistics to reduce their error. i found it better to rely on very carefully understanding exactly what i was measuring, with what, and where “errors” might come from.
your statistical “analysis” does a pretty good job of completely missing the point.
the argument is not….. or should not be… ‘oo killed ‘oo, but under what circumstances.
a cop sticking his gun in the face of a black man and saying “put your hands up and show me your i.d.” and shooting him when he reaches for his i.d. is not something you can learn much about by gathering statistics.
you might have a case against those who ARE arguing statistics… but i think the problem might be more usefully addressed by applying a little common sense.
Coberly,
1. This is not to minimize the issue of police errors and/or misconduct.
2. With any results, you do a gut check. If the data doesn’t fit your gut, then look for what is wrong. Fortunately, in a simple situation like this one, where all I did was show data (no models were constructed), it’s usually easy to spot the errors in logic. But if you can’t find the error, perhaps your gut is wrong and it’s time to do as Keynes suggested and change your mind.
Mike
best as i understand it at this point you are responding to some people who pointed at the “statistics” of white police killing more blacks than “justified” by the relative numbers of whites and blacks in this country.
your statistics in opposition to their statistics seems to me to be correct.
but i don’t think you will convince them, because their “statistics” is just their poor attempt to make a case and their “gut” feeling… as is mine… is that the police are killing people that in no way “deserve” to be killed. because of the racial issue, and because human beings think racially, they will interepret… or just use… statistics to support their gut feeling.
but it’s the gut feeling that counts. not the statistics.
the police kill too many people. hell, they ARREST too many people. it has always been this way. but we should at least try to look at what is going on case by case and hope that by dealing with them on their individual merits we will gradually create a climate where they are rare.. true “accidents.”
we don’t have this. we have cops routinely excused for killings that should not have happened. racism is undoubtedly part of it. but most of it is just the establishment protecting its protecters.
statistics, like logic in general, is always subject to the error of missing the important facts.
run,
As you note, the FBI seems to capture about 45% of the total incidents. But the tables above don’t look at total numbers, they look at proportions. (I.e., what is the share of the reported total X that is Black, that is White, that is Other?) How biased would the 45% that is captured to be to be so completely unrepresentative of the entire set? Do you trust other federal statistics, such as GDP, which are based on much, much smaller samples of the total? Ditto polls?
As noted upthread, the one data set I used that the FBI has an incentive to bias the proportions in the current political environment, namely the proportion of people of different ethnicities that are shot, I didn’t get from the FBI. The Washington Post may not capture every single death, but they put a fair amount of resources into trying, and they don’t rely just on the police to provide them with information.
coberly,
I included the first table, on men and women, for a reason. See, nobody considers it bias or discrimination that men are arrested and/or killed by cops so much more often than women. It is not inconvenient to our belief structure or our social standing to accept that there are a bigger percentage of men than of women that seem to be willing to have a go at the cops, and who can only be stopped with deadly force. As a society, we aren’t willing to believe that the proportion of Blacks, Whites, Asians, and Native Americans who are willing to have a go at the cops might be different as well, and is roughly correlated with the general likelihood of a person in that community committing a murder. We know that – the likelihood of murder – is different, for whatever reason. After all, you see regular marches against violence in the same neighborhoods, and you don’t see those same marches in a lot of other similarly situated neighborhoods. Reducing this is an important issue, as a lot of people are dying. Every weekend there several new shootings in Chicago. Again, always in the same neighborhoods. I don’t know St. Louis, or Baltimore, or Detroit, but I imagine the same neighborhoods are the source of the shootings there too.
Again, none of this is to say that the cops don’t make mistakes, or that there aren’t bad cops out there. But it is to say that while it is widely accepted that the police have a bias toward shooting Black men, that view is soundly contradicted by the data.
I also disagree with you that there are too many people arrested. The problem in Chicago, the number of shootings in certain neighborhoods, that isn’t coming from too many arrests or trigger happy cops. It’s coming from a segment of the population in that neighborhood. That group of people makes the lives of everyone else in the neighborhood worse, including their friends and family. But in many cases their friends and family will protect them and side with them against the cops. The marches for justice and peace and less killings, and even less killings by cops, will go on until that changes.
Gut feelings may count more than statistics to many, but that doesn’t make the gut feelings true. And the truth is the only thing that can save lives. Gut feelings are making it worse.
well Mike, i expect you to disagree with me that our whole approach to law and law enforcement is wrong.
as for the “soundly contradicted by data”, i will say again about data: you see what you look for.
coberly,
“i will say again about data: you see what you look for” – On this particular topic, my mind was changed by data. True of taxes too – I think I mentioned on a post a long time ago that when I was growing up, St. Ronald the Reagan sounded like he knew what he was talking about when it came to taxes. Eventually I discovered the existence of economic data, and one fine day I tried to fit the Laffer curve and had one hell of a shock. And that led to other shocks. But that’s another story.
Apostasy, the loss of my earlier faith, is a terrible thing
Mike
are you saying you were a Reaganite before you checked the facts?
data is useful when you have made a good analysis. data is misleading when you have not.
i thought you were right about taxes. i think you are missing the point about police killings.
i discovered a fact a few years ago, but i started with no opinion whatsoever.. except that the reported “looming crisis” didn’t sound reasonable considering what i thought i knew about the real world.
it turned out it wasn’t. but i haven’t been able to get anyone interested in that fact at all, except for a few people on Angry Bear, and some of them have decided the fact is not all that important when it comes to their other priorities.
so we are both wandering the world with a lantern looking for an honest man?
“the wave function collapses”
You might have to translate that for people who don’t do QM.
If the question is whether and to what degree there is a racial bias by police in policing activities, then one has to start with the fundamentals.
The first of these fundamentals is to determine why, all other things being equal, the proportion of women police killed is half the rate of males.
For w = women, m = males, k = killed, pop = population, p meaning proportion
0.122 w / 1 = proportion w (pw)
0.04 wk / 1 = proportion wk (pwk)
pwk / pw = 0.04/1 / 0.122/1 = wk/w =0.04/0.122
wk / w = 0.327
popw/1 = 0.508
wk/w / popw/1 = 0.328/0.508
wk/w / popw/1 = wk / (w*popw)
wk / (w*popw) = 0.645
By contrast, for
m / 1 = 0.878
mk / 1 = 0.958
popm = 0.492
then
mk/m = 0.841
Thus
wk/w / mk/m = 0.645 / 0.841
wk*m / mk*w = 0.5197 = 0.52
So wk/w relative to mk/m is 52%
That is, women police killed are 52% of male police killed all other things equal
This is an obvious bias against women police being killed at half the rate of males.
Why this huge bias?
Perhaps women police officers are half as likely to be in policing jobs that involve the use of force or even actual enforcement activities. Or perhaps males are far more likely than women to induce or create conditions for the use of force that that require the use of force. There is no data in this article which even remotely addresses the reasons, yet the reasons for the huge bias is critical to even begin to understand any other parts of the data presented.
Without knowing the answers to the first question then the rest of the report is a purely speculatory use of data. This is not, despite its appearances, an objective analysis of any relevance to the question of whether there is a racially motivated disparity by police. IMO, it is as simplistic as any a high-school senior or college freshman might make.
With the first question unanswered, then come many other unanswerable questions with the data provided.
For example:
– What are the same statistics as a function of minority population densities in policing jurisdictions?
– What proportion of police in each population density jurisdiction are male in actual enforcement activities and of those, what proportion are the same race as the person killed by police?
– And are the proportions the same for white and black police officers in each racial density region?
I think a more precise way to see what is happening between citizens and police is to look at the number of citizens killed and number of officers killed in 2016 as of 7/27/16.
Citizens Killed by cops 611. http://www.theguardian.com/us-news/ng-interactive/2015/jun/01/the-counted-police-killings-us-database
Cops killed by gun fire 31. https://www.odmp.org/search/year
People Killed by police by race & ethnicity per million in 2016 as of 7/26/16.
5.49 Native American
3.78 Black
1.71 Hispanic/Latino
1.51 White
0.56 Asian/Pacific Islander
http://www.theguardian.com/us-news/ng-interactive/2015/jun/01/the-counted-police-killings-us-database
coberly,
I was also a teenager, and back in those days, it was hard to get and analyze quality data, even if you weren’t a teenager. I started having inklings of suspicion something wasn’t right staring at graphs printed occasionally in business publications like the Economist. In 11th grade I had a government teacher who encouraged us to actually look at data.
I’m not sure I’m wandering the wilderness searching for an honest man… at this point, I’ll settle for a person whose incompetence + dishonesty is low enough.
warren,
There are multiple ways you can chicken and egg cops shoot Blacks at high levels and Blacks shoot cops at high levels. But only one fits when you consider the proportionately high murder rate in the Black community. Cops aren’t the cause.
Longtooth,
The data could have allowed me to write many different posts going in multiple directions. I chose one particular focus, the one I think has the most consequence to human life and well-being at this time, and which also happens to be one most people are getting wrong. I’m sure if you chose to write your own post supported by data, Dan would put it up too.
Beene,
I avoided magnitudes for several reasons. 1st, the magnitudes are wrong because the data is incomplete. I assumed it was complete enough that proportions would be relative correct.
Additionally, I wanted to break out “Other” as it was clear that Native American and Asian/Pacific Islander would go in very different directions, but the Wash Post data didn’t break it out that way.
The White Racist cop racist narrative seems to require that they really like Asians a lot. I’m going to guess we can all Occam’s razor this thing a lot better than that.
Also, we expect cops to survive more shooting situations than civilians. For one, they’re trained. Bear in mind also that most of such situations are not “cop randomly walked into a location and shot a guy who was minding his own business.”
Respect.
Was wondering when someone would look at this. And (as far as I can see) you have done it much better and more carefully than I thought it would be done.
My guess is it will be ignored. Hope I’m wrong.
Thank you Nick. Your comment means a lot.
As a last minute follow up on the issue I would recommend taking a look at todays (7-28-16) Paul Craig Roberts.org blog “Police are conducting war against the American people”. This study also contains the Guardian report info.in it…It contains and provides a very critical and in depth must see analysis..
William Ryan,
Thank you. I will look it up.
Here is where the date came from that Ryan posted above.
http://injuryprevention.bmj.com/content/early/2016/06/16/injuryprev-2016-042023.full
In general I think the point is correct, there are too many guns in America and there are some areas that are being destroyed by gangs which has the unfortunate side effect of making the police in those areas trigger happy. Labeling this simply by race though, is I think unfortunate. Poor urban areas as the geographical unit might be more convincing.
I think the point about reporting rates for police shootings made earlier though is probably correct, I would definitely take the statistics with a grain of salt.