Grossly misleading claims about black teens being “vastly more likely to be killed by police than whites even after adjusting for crime rates”

12 Oct , 2014  

Sometimes racial discrimination is occurring and should be pointed out, but sometimes false information is being used to unjustifiably inflame concerns. The claims being put forward by Slate and Pro Publica are simply very misleading. From Slate:

The 1,217 deadly police shootings from 2010 to 2012 captured in the federal data show that blacks, age 15 to 19, were killed at a rate of 31.17 per million, while just 1.47 per million white males in that age range died at the hands of police.” . . .

Could higher rates of crime commission by black teens relative to their white peers explain that difference? . . . The data suggest that the answer is no. . . . it looks like black Americans are between two and three times as likely to commit a violent crime as white Americans. But even assuming that black male teenagers are three times as likely as white teenagers to legitimately threaten the life of a police officer doesn’t explain why they’re twenty times more likely to be killed by police.

For the sake of argument, let’s assume that the material in the first quoted paragraph is correct. But it doesn’t make a lot of sense to look at rates of killings by police of young black and white men but then compare it to violent crime for all blacks and whites. If you look at offender data and assume that 92 percent of murders are committed by males, you find that young black males were about 8 times more likely to commit murders than similarly aged white males. However, a lot of murders occur where the offender isn’t identified and that this problem is greater for gang related murders. To try to deal with this, we have made some calculations using victim data. Adjusting the data by the rate that blacks and whites kill members of their own and other races, young black males were 9 times more likely to commit murders than similarly aged white males. (Click on figures to enlarge the calculations made here.)
Screen Shot 2014-10-12 at  Sunday, October 12, 2.25 PM
Screen Shot 2014-10-12 at  Sunday, October 12, 2.28 PM
The Slate piece claims that after adjusting for violent crimes by race, police still kill young blacks at 7 to 10.5 times the rate they kill young whites. Making this one adjustment would still leave a difference, but it would imply that police are killing young blacks at 2.3 times the rate that they kill young whites — about one-fifth to one-third the difference being claimed.

Yet, apparently the reporters at Slate and Pro Publica don’t understand the data that they are using, and even this difference could very well be the result of biases in the data that they use.

— Pro Publica acknowledges that police-involved death data is not complete. But there is a systematic bias in what data is missing. Generally most of this data is collected from urban areas. Obviously most deaths occur there, but these areas also tend to be heavily black. ┬áThe percent of the population that is black in areas reporting justifiable police homicides were 50.3 percent more black (17.8/11.8) than those jurisdictions that didn’t report these homicides.

Screen Shot 2014-10-22 at  Wednesday, October 22, 3.05 AM
— If black murders are more likely to involve gangs and gangs are more likely to get into shootout with police, young black males would be more than 9 times more likely to commit murders than similarly aged relevant white males.

UPDATE: Politicizing Ferguson police shooting?


10 Responses

  1. […] blacks, teenage crime is much more of a problem. Black male teenagers were nine times more likely to commit murders than similarly aged white males, not two to three times as […]

  2. Rob Schwartz says:

    Hi, Thanks for your writing on this, here and in NY Daily News. I don’t get your argument that the unrepresentative sample of police data, overrepresenting urban areas with a higher proportion of blacks, would overestimate shootings of blacks. The 21-times-greater estimate is based on a comparison of proportions: (police killings of blacks/total blacks)/ (police killings of whites/total whites). Right? If we are comparing these proportions, the racial makeup of the population shouldn’t matter, right? So whatever the demographics of the population, 10% or 90% black, “equitable” policing killings should produce a ratio of 1. Hopefully, I’m being clear enough. The only problem I see with the urban bias of the data is that it is a mistake to use it to generalize to the whole country, but this isn’t because of the proportions of the demographic makeup per se. It would be great if you could clarify this point for me.

    I agree, the abuse of this statistic in the media has been very frustrating. Your other argument is just the one I wanted to make, but I did’t know the murder rates for young men. Thanks.

    • TW says:

      What he’s saying is, for an apples-to-apples comparison, the numbers that you’re coming up with have to match.

      The 21x rate is based on 15-19 year old males who live in the selection of police jurisdictions that report data on officer- involved shootings. Those jurisdictions are also heavily biased toward urban areas, and not even proportionally-representative of the population, given that ~50% more blacks and 2x as many Hispanics live in those jurisdictions relative to other jurisdictions across the US.

      So to make a useful comparison against violent crime rates, you’d need to calculate the violent crime ratio for 15-19 year old males who live in the urban areas that report police-involved shooting data for that part of the equation as well – not overall rates for the entire population.

      Otherwise, you’re comparing apples-to-oranges, and the data is worthless for the purpose of drawing useful conclusions.

  3. johnrlott says:

    Dear Rob:

    The problem is that you have the ratio of blacks and white teenage males killed from just one percent of police departments and you are trying to infer what the national ratio is from that one percent of departments. If that one percent of departments isn’t representative of the rest of departments, there is no reason to believe that you can draw national implications from that number. For example, what if the department that you had data for served a virtually all black population? Would you be surprised that blacks were more likely to be killed than whites? Presumably not. That is the general point.

  4. Warren Nickerson says:

    Will someone please inform me of the number of unarmed Black people killed by police who had not committed a crime at the time of the deadly confrontation with police. It is not sufficient to simply roll out the statistical differences in crimes rates between white and black populations, at some point chewing gum or raising your voice may become justifiable for killing unarmed Americans!

    • Henry says:

      I agree that the discussion here doesn’t really address the issue of a racial disparity in the use of force, since the comparison of crime rates and so forth doesn’t tell us anything about whether or not a police shooting was actually justified.
      That statistics and data for this kind of thing are not even kept track of in most of the country should tell us a lot IMO.
      Whether a police killing is “justifiable” basically hinges on whether or not it was *necessary* to protect the officer’s life or the lives of innocents…and the comparison of crime rates tells us nothing about this…

  5. […] which is the study group, are only killed 2.3 times more than for Whites. Because they are also 9 times more likely to commit murder. We also learn from that […]

  6. […] Massey never mentions certain self-evident considerations: e.g., that young black males commit violent crimes at a ratio of 7 to 1 relative to their white counterparts; that most violence against blacks is […]

  7. RS says:

    Warren, The question is whether cops using lethal force are discriminating against blacks. We’d need data on both blacks and whites to make a comparison– your question only asks for data on blacks. ProPublica’s report was very misleading because of its omission of the discrepancies in crime rates between blacks and whites. I can’t even think of what the use of ProPublica’s 21-times statistic is. Taking into account how frequently members of each group are acting in ways that put them at risk of getting shot by police is necessary to come up with a meaningful statistic. Lott provided the most precise estimate to this question that I know of.

    I think that taking into account the discrepancy in murder rates is sufficient for making an inference about whether cops are discriminating against blacks in their use of deadly force. But besides the problems with the police shootings data already mentioned by others, I could see two possible criticisms: (1) The data on murder rates is biased so that the gap between whites and blacks appears wider than it really is. (2) We are using murder rates as a proxy for risky behavior that causes the person to get shot by the police, perhaps murder rates are not a good proxy. Regarding (1), I think the US’s crime data is pretty solid. I’m not an expert but my understanding is that our two data sources (one from crime reports, one from victimization surveys) corroborate each other well. Regarding (2), I see no reason to think murder rates are not highly correlated with the type of behavior that gets one shot by the police.

    Here’s a sociologist’s blog post that links to some studies that might interest you: Her own argument is a muddle though. She seems to be conflating three question: (1) Are African American men more likely to engage in risky behavior generally (like robbery or murder) that increases their chances of getting shot by police? (2), Are African American men more likely to engage in risky behavior in their interactions with police? (3) Are police biased against blacks in their use of deadly force?

Leave a Reply

Your email address will not be published. Required fields are marked *