Good News from the Manhattan Institute: Police Kill Fewer Black People than Violent Criminals

You heard it here first: Heather Mac Donald’s most recent study from the Manhattan Institute reveals the good (and groundbreaking) news that police kill fewer black people than violent criminals. Mac Donald, as you might recall, was a proponent of the theory that murders were on the rise because of the “Ferguson effect”–that dissatisfaction with the police explained a rise in killings, a claim that wasn’t credible at the time and which has been debunked here. You might be sensing a theme–that statistical analysis isn’t Mac Donald’s strong suit–but this hasn’t deterred her from the important work of continuing to tell people that police violence against black people is no big deal.

Where to start? How about the title: most people are killed by violent criminals because violent criminals are, by definition in the Uniform Crime Reports, those who do violent things.  Police aren’t included in these figures because “The [UCR] program classifies justifiable homicides”–including those by police officers–”separately.” Now you know: most people who are intentionally killed are intentionally killed by murderers. (Note to the HR Department at the Manhattan Institute: if you are looking for another groundbreaking study, I can tell you right now that most people are raped by rapists, not by police, and would be happy to write that up for a fee.)

She states, “It is other black civilians, not the police, who overwhelmingly perpetrate violence in poor, minority communities.”  True.  Most people are not killed by police because most people are not police (only about 618,000 patrol officers in a country of more than 300 million).  Most people are also not killed by dentists, or Duke graduates, or people named Carl, either, because most people aren’t dentists, Duke graduates, or named Carl.  That proves nothing about their moral turpitude or their relative homicidal tendencies. You always bet the field. If, alternatively, there were more homicides from police than non-police (last year that would have meant 13, 472 homicides plus one), then that would mean every year about one in 50 beat officers would kill someone. (These are, in deference to MacDonald, figures based on the killing of all people, not just black people.)

So this latest report is a case of “Duh” (Or, if you prefer, “No duh,” since ”duh” is one of those few words whose antonym is actually a synonym). Is this the discussion MacDonald thinks is newsworthy and worth having? Does she think it’s controversial that police kill fewer people than violent criminals? I sure as heck hope that law enforcement kills less often than lawless citizens.  What next–is she going to tell us that most police are not thieves? (Uh, maybe never mind, since asset forfeiture seizures were greater than the losses from burglary last year.)

I’m not aware of anyone saying that cops are the biggest threat to black people in terms of total numbers of deaths.  Based on general population statistics, cancer, suicide, and accidents are bigger threats.  The issue, of course, is differential risk: that black people die more often than white people from police use of force.  To her credit, Mac Donald acknowledges that but states that black people are killed by police at a rate “lower than their share of violent criminals.”  Let’s set aside some of the endogeneity problems of criminal justice statistics (who we catch and arrest depends on who we look for), as well as what can most charitably be described as a scattershot survey of the data (2009 data from the 75 largest counties, figures from 2014 in New York City).  Does this mean we should expect arrests of robbers and people who assault to end in death for the arrestee?  Does a violent crime mean a violent response?

Maybe it does, when race is involved.  Dig into any of the police implicit bias and shooting studies collected here.  Or watch the police do it.  Here are two examples, one of “police shooting black people“ and one of “police shooting white people“ (which actually turns up mostly videos of police shooting black people). Here’s one where a black man suspected of not wearing his seatbelt is shot. Watch the whole thing. Then watch this video of a white murder suspect not being shot (it’s one of the few videos I could find by searching for police shooting white people).

MacDonald theorizes that this focus on police violence means police will “back off” policing minority neighborhoods, leading to a greater loss of lives.  I think the causal story here is dubious (what’s the evidence that explains either (a) that police aren’t policing as much, (b) that they are doing so because of community hostility, and (c) that more active policing (whatever that is) would reduce the homicide rate) as is the idea that focusing on police violence is the problem, rather than the police violence itself.  I set the bar higher than Mac Donald does.  I want police to be more restrained than the most violent elements of society.  I want police to administer justice fairly.  And I want those who write about public policy to address good arguments, not straw men.

Author: W. David Ball

W. David Ball is an Associate Professor at Santa Clara School of Law. He writes and teaches primarily in the fields of criminal law and criminal procedure, with a special focus on sentencing and corrections. He also serves as the Co-Chair of the Corrections Committee of the American Bar Association.

16 thoughts on “Good News from the Manhattan Institute: Police Kill Fewer Black People than Violent Criminals”

  1. It seems to me there are at least a couple of valid points here you're resistant to engaging.

    First, death by police shooting is a small, small fraction of homicides of blacks. You correctly point out that for much the same reason, death by dentist shooting is a small fraction of homicides of blacks. True!

    But, who is making a big deal of death by dentist shooting? Trying to spark a nation-wide crusade about it? Why, nobody. Nobody is specifically concerned with the number of blacks being shot by dentists, it would be silly to be specifically concerned with this tiny, arbitrary subset of homicides.

    However, people ARE making a big deal of death by police shooting, trying to spark a nation-wide crusade about it. In spite of the fact that only a tiny, tiny fraction of blacks being shot, are shot by police.

    Why? Is the reasoning not the same? If you really want to make a difference in the black homicide rate, shouldn't you be focusing on, say, gang related shootings? Not focusing like a laser on the rare outlier cause of death?

    That's one valid point you're not engaging with.

    The second is that " black people are killed by police at a rate “lower than their share of violent criminals.”" actually does say something significant.

    All things being equal, (The default starting point for this sort of analysis.) you'd expect that people would be killed by police at a rate roughly proportional to the rate they *encounter* police. It's very tough to get shot by somebody you don't meet, wouldn't you say?

    And, again, all things being equal, you'd expect blacks to encounter police at a rate roughly proportional to their share of violent criminals. You rather hope, don't you, that police direct their attention where crime is happening, don't you? Instead of hanging out where the crime isn't happening.

    So, all things being equal, you actually would expect the rate at which blacks are being shot by police to be roughly proportional to the rate at which blacks commit violent crimes.

    And, yet, the rate is actually disproportionately lower than that. This suggests, counter to the impression you might get from, say, BLM complaints, that police are actually LESS inclined to shoot blacks, in any given encounter, than they are whites.

    Now, there were a lot of all things being equal in there, so I wouldn't suggest drawing any strong conclusions from this. But it is at least suggestive of what you might find if you did actual research, instead of listening to BLM.

    All this is pretty important. If you made a big fuss about blacks being shot by dentists, to the point where dentists started actively avoiding treating blacks, you might see an upswing in cavities. Do the same with police, and you might see an upswing in what police are charged with fighting: Crime.

    And you are, and isn't that fairly important?

    1. Hi Brett–

      Thanks for your comment.

      I think the focus on causes of death by police assumes that what we want out of "public safety" is "police not killing us." As I said, I think that sets the bar pretty low. Here's what I want out of police: fair administration of justice, where different outcomes are determined by differences that are relevant to criminal activity. I don't think skin color is relevant to criminal activity, and if certain communities don't trust the police enough to call them and/or cooperate with them (for fear of getting killed), everyone is worse off, so I want to make sure that police responses aren't based on skin color. As I said in my post, if this is a public health concern, we should be spending most of our time and money on cancer research, which is precisely what Biden is talking about. But if we're talking crime, it all comes down to what our goals and objectives are. I want more than just police who don't kill people more than murderers.

      I might add that citizens have much more control over the police than they do each other. There's a chain of command with police, we train them, etc. So if I'm looking for something that I can fix, I'm going to go to an area where I have control. I also think the government always needs to play by the rules no matter what criminals do.

      As to your second point, I'll be happy to discuss that further. Your causal analysis goes something like this: more contact with police leads to more potential for violence, and more contact with violent offenders leads to still more potential for violence. Fair? On that basis, I can see something like Mac Donald's point. The only problem is that she hasn't proven that. It's, at best, a hypothesis. I don't see any evidence that police violence is directed more towards those suspected of violent behavior, nor that there aren't racial differences even within these categories. A good study would compare outcomes per police call–for X thousand calls to respond to a report of violent activity, what percentage of those for whites result in police use of force and what percentage of those result in use of force against blacks? The aggregation she's doing here isn't methodologically sound. If police use of force isn't, in the individual case, being driven by the actual threat posed–but instead by the racial threat that we see in the studies I linked to (and which you should read)–then her hypothesis shows only correlation.

      So, for example, as a result of the Floyd litigation, we have an apples-to-apples comparison of the use of force under similar circumstances in New York City. Analyzing stops, between 2005 and mid-2008, the Center for Constitutional Rights found that 17 percent of Whites, compared to 24 percent of Blacks and Latinos, had physical force used against them during NYPD-initiated encounters. ( So on a per-stop basis, people of color are more likely to be subject to force.

      Her statistics about propensity for violent behavior aren't standardized against the comparisons of police shootings she's using, and she's not accounting for the fact that criminal justice statistics are inherently fuzzy. Do more people get arrested because they're doing more things or do they get arrested because police are arresting them? Yes, I want police to focus on areas of criminal activity, but it all depends on what criminal activity we consider important. Why, for example, weren't more people arrested after Enron?

      As for the idea that I do "actual research, instead of listening to BLM," I'll be happy to take you up on that. I actually did my own research on this. Heather Mac Donald's analysis, however, seems awfully close to that contained in this Trump tweet.

      1. Wait, you seriously want to compare Heather Mac Donald's research, who I've never read an incorrect fact despite her "conservative" take, to Trump and his infamous racist Tweet that stated 81% of whites are killed by blacks. (A flat lie, as 83%+ of whites are killed by whites, and 90%+ of blacks are killed by blacks; a simple Google search for that). She's never come close to emanating propaganda of the sort, and thus, David Ball, I question your fairness. Heck, she's even voted for Obama. <a href="http://(” target=”_blank”>(

        Problem is, you're not looking at the obvious, or showing what the "controls" are in looking at the problem of police violence. Mac Donald does. And so do I. As a liberal, I started making the same points after the “controversial” Ferguson decision not to prosecute the police officer involved. It’s when I first saw the stories on the news didn't match the data I was able to simply gather from FBI, Dept of Justice, and academic studies.

        All below are FBI or Wikipedia stats for Black Americans:

        Homicide offender (and victim) rate of African Americans = 50–55%
        Arrests = 28%.
        Incarcerated = 37%.
        Violent crime arrests = 39%.
        Weapons = 40%.
        43% of cops are killed by black persons
        56% of carjackings are by black persons

        And very important, too, is pointing out that the victims are within communities (49% of male homicides are black, and 35% of female homicides are black): See Department of Justice Report: Black Victims of Violent Crimes. <a href="http://(” target=”_blank”>(

        SO THE KEY QUESTION: If police killings of the black population are by most every measure equal or LESS than their contact points with each other — potentially violent interactions justified by actual crime taking place (case reports, victims stats, arrest rates) — does that mean these killings are not disproportional? And if that’s true, would this debunk the idea that death by the hands of police indicates racial bias in the USA?

        Or super-duper simply: White people and black people, as individual demographic sets, are roughly 40% of those in prison and 40% of criminals and crime victims in many categories. But police in 2015 killed more than 1000 persons, 10% unarmed (90 total in 2015, according to the Washington Post), and about 50% were white and 25% were black.
        Yet what are activists and the media predominantly paying attention to? It’s not killings of unarmed white kids in Utah.

        MORE HERE:

    2. Only a tiny fraction of those killed are killed by Islamic terrorists, so, by your logic, no one should be making any sort of big deal about it.

      Or, maybe, killing someone under the cover of law is something that deserves more scrutiny in each individual case.

      1. Actually, I think you've missed my point–I apologize for not making it clearer.

        I think the terrorism point is exactly Heather Mac Donald's point–if a tiny fraction of people are killed by police/terrorism, then it isn't a problem. That's what I'm criticizing her for.

        Saying merely that police/terrorists kill fewer people than criminals is not a sign that police killings/terrorism isn't a problem. It's not evidence of much of anything.

        It is still a problem, even if cancer, accidental deaths, and suicide kill more folks.

  2. In all too many jurisdictions, instead of being peace officers, police have taken on the role of tax farmers.

  3. "Police kill fewer black people than violent criminals."
    Ambiguous. "Police kill (fewer black people than violent criminals)" or – the intended reading – "Police kill (fewer black people) than violent criminals [kill black people]".

    Heather Mac Donald? How does this name work? In Scots [Scots English), the Mac- or Mc- prefix, meaning "son of", is always attached to the clan identifier. In Gaelic and Irish, it's separated. If she were into the full Gaelic heritage thing, she would be Heather Mac Dhomhnuill.

    1. Another prominent example of an odd-looking Scottish name is the mathematician Saunders Mac Lane (American of Scottish ancestry). The 'Saunders' was his mother's maiden name transmuted into a first name, and the 'Mac Lane' was his personal variation on what was originally 'MacLane'.

  4. David — why is the comparison appropriate? The police work for us, criminals don't. If one of my employees at the hospital screws up, it's my job to intervene, even if a million other people who don't work for me also screwed up that day.

    1. I was being sarcastic! I think the finding is a non-finding, and it's neither "news" nor really all that "good." In fact, that's why I wrote this post–because I think the comparison is inappropriate. I think I need to signal that better…

  5. If the rate of violent crime in a sub-population bore some kind of direct link to the sub-population numbers for death by police, wouldn't one expect the typical death by police to be of a violent criminal in the process of being arrested for a violent crime. Anything else smacks of some kind of (intentional or otherwise) collective punishment.

    Also stating the obvious: homicides are just the most obvious, undeniable marker for how a sub-population gets treated by law enforcement. So each death likely stands in for thousands of other questionable interactions.

  6. How are your feelings about The Ferguson effect now? My home of Chicago has 50% spike in shootings. We had just more than 480 homicides in 2015, but may approach 700 by the end of the year. And police today said they’re going “fetal” on Labor Day, with the Fraternal Order of Police telling members not to volunteer for overtime.

    “The shootings have made this month Chicago’s most violent June in at least six years, DNAinfo reports. About 70 people were killed, with at least 61 of those deaths stemming from shootings. The site reports that Chicago could see 700 homicides by the end of 2016.” <a href="http://(” target=”_blank”>(

    This antagonizing, seemingly pro-police group <a href="http://(” target=”_blank”>( also has some pretty shocking numbers, as well as carry the "police are going fetal" line of thinking. But at least they're reporting real and accurate numbers. Remarkably, police have wounded only 2 persons and killed 4 in the line of duty in first third of the year. Police killings are about average but total wounded is on track to be a record low. Good policing doesn't mean you have to shoot people, but it's a possible indicator of "pulling back" especially in context with huge spikes in crime and shootings in general. That context, which is missed in many of these articles on crime and law enforcement, is key (As of July 27, 2016):
    * Police shot 12 people in the line of duty, killed 5.
    * Year to Date in Chicago (total, Jan 1-July 27)
    – Shot & Killed: 345
    – Shot & Wounded: 2000
    – Total Shot: 2345
    – Total Homicides: 383

    Quick math: That's a 200:1 ratio on shootings, of citizen vs. police.

    One more thing. Are Chicago police more violent given all the news and task force reports critical of them? Last year, the ratio of people killed by police vs general homicide was 60:1. Nationally it's 15:1. Police in Chicago killed 9 persons in a population of 2.7 million. That's exactly the same rate of police nationally, despite vast greater percentage of violence and crime in the city. Yet I rarely if ever see that reported. I find that odd.

Comments are closed.