Authors of study on race and police killings ask for its retraction, citing “continued misuse” in the media

via Tony Webster/Flickr

The authors of a controversial paper on race and police shootings say they are retracting the article, which became a flashpoint in the debate over killings by police, and now amid protests following the murder of George Floyd.

[See an update on this post.]

The 2019 article in the Proceedings of the National Academy of Sciences (PNAS), titled “Officer characteristics and racial disparities in fatal officer-involved shootings,” found “no evidence of anti-Black or anti-Hispanic disparities across shootings, and White officers are not more likely to shoot minority civilians than non-White officers.” It has been cited 14 times, according to Clarivate Analytics’ Web of Science, earning it a “hot paper” designation.

Joseph Cesario, a researcher at Michigan State University, told Retraction Watch that he and David Johnson, of the University of Maryland, College Park and a co-author, have submitted a request for retraction to PNAS. In the request, they write:

We were careless when describing the inferences that could be made from our data. This led to the misuse of our article to support the position that the probability of being shot by police did not differ between Black and White Americans (MacDonald, 2019). To be clear, our work does not speak to this issue and should not be used to support such statements. We accordingly issued a correction to rectify this statement (Johnson & Cesario, 2020).

Although our data and statistical approach were valid to estimate the question we actually tested (the race of civilians fatally shot by police), given continued misuse of the article (e.g., MacDonald, 2020) we felt the right decision was to retract the article rather than publish further corrections. We take full responsibility for not being careful enough with the inferences made in our original article, as this directly led to the misunderstanding of our research.

The MacDonald references are two pieces by Heather MacDonald of the Manhattan Institute, one in the City Journal and the other in the Wall Street Journal

As Cesario and Johnson note, the article earned heavy criticism for its methodology. The article led to an exchange of letters in PNAS in January, and then in April, the journal issued a correction to the paper in which the authors acknowledged flaws in their analysis but stood by their central argument. 

At the time, Jonathan Mummulo and Dean Knox, both of Princeton, called the correction an “opaque half-measure.” Today, Mummolo and Knox told Retraction Watch:

We appreciate the authors’ willingness to take this action. We hope readers and policymakers will now cease relying on this paper when considering the critical issue of racial bias in policing.

May Berenbaum, the editor in chief of PNAS, told Retraction Watch:

I can confirm that the authors contacted us early this morning to request a retraction and we’re working to publish the statement as quickly as possible.

A Wall Street Journal editorial claimed last month that a “Twitter mob” forced the demotion of a Michigan State faculty member, Stephen Hsu, from his post as senior vice president for research and innovation at the school, in part because of his perceived role in having “directed funding to research downplaying racism in bias in police shootings” by funding Cesario’s study.

Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at

51 thoughts on “Authors of study on race and police killings ask for its retraction, citing “continued misuse” in the media”

    1. This study shouldn’t be retracted, even in part, as other professors have already explained. An expert on the admissibility of scientific evidence in court said this study should not be retracted.

      Specifically, that expert, George Mason University law professor David Bernstein, wrote, “It’s absurd to ask that a valid study be retracted b/c you think others are ‘misusing’ it. A study says what it says, and so long as it wasn’t actually flawed it shouldn’t be retracted for political reasons except perhaps under truly extreme circumstances, which this isn’t.”

      Regarding the prior so-called “correction” to this paper, he wrote, “That’s not a correction, that’s narrowing the conclusion drawn from the exact same data, but the extrapolation they are objecting to seems to me a perfectly reasonable one.” See his 2:36 p.m. July 7 tweet on Twitter and responses thereto.

      1. George Mason the notoriously astroturfed university serving as a print shop for the dissemination of “papers” and “studies” “objectively” supporting every facet of conservative ideology in economics and law? That George Mason?

        1. Sam, doesn’t it bother you that nobody – – not one person – – seems to be able to cite ONE example of factual error in this study? Don’t you find that the least bit concerning?

          I mean, what kind of America do you want our kids to grow up in, Sam — one where policy is set by science and by data, or one where policy is based on FEELINGS, and any fact that runs counter to an emotionally-satisfying narrative must be swept under the rug?

          Which one is better for the future, Sam?

        1. It is precisely the study’s facts and accuracy that are getting it retracted, due to those inconvenient true facts being cited in the Wall Street Journal by a conservative. As lawyers and journalists have noted, the study is perfectly accurate, and it was not “misused” by the Wall Street Journal, which accurately quoted it. See, e.g., “Conservative author cites research on police shootings and race. Researchers ask for its retraction in response,” The College Fix, July 8, 2020; “Academics seek to retract scientific study because conservative cited it,” Liberty Unyielding, July 8, 2020.

      2. White men spent over 75% of this country’s history implementing, enforcing and benefiting from anti-Black racist laws and policies. But you claim racism ISN’T part of American culture? Has an unarmed white person ever been murdered for holding up his wallet or cellphone?

    2. The right thing is to specify what was false in the report and until that is done the public can continue to rely on it. Was the information false? It seems that the only thing that changed was the introduction of cancel culture.

  1. The academic willing to stand behind his work even when it doesn’t fit the prevailing narrative is a dying breed.

        1. “This led to the misuse of our article to support the position that the probability of being shot by police did not differ between Black and White Americans (MacDonald, 2019). ” Yes I suppose when you leave out the crucial qualifier, “when controlling for differential rates of offending” then this position would be incorrect, although I’m dubious that someone as bright as Heather MacDonald actually made that mistake. Sounds more like the authors were grasping at straws to create an excuse to ask for retraction.

          1. You’re right to be dubious that Heather MacDonald made that mistake. She did not. The authors of this paper incorrectly (and ironcically) claim that she did, though. MacDonald has published a follow up in the WSJ about this retraction that is well worth the read.

  2. Retraction Watch used to be a place where one could learn about academic malfeasance and other questionable practices. Now it’s a place to track the impact on academia of America’s cultural revolution.

    1. I agree about the impact on academia of America’s cultural revolution – driven by its Red Guards, who are ruling by fear, just as their Chinese counterparts did in the 1960s-1970s.

    1. Are you talking about “masks don’t help” or “there’s no animosity between races” or “it’s just another flu” or “ there’s always animosity between races“ etc.
      Which irrational, anti-science mob are you talking about?

  3. Their article also does support the position that the probability of being the perpetrator of a violent crime does not differ between black and White Americans. Retraction, done for emotional considerations rather than on scientific grounds, will not prevent anyone else from doing the calculations for themselves.

  4. This retraction is being sought for political reasons, even though the study contained valuable information. The study itself, while not perfect (no study ever is), is of better quality than most research on police shootings.

    It is due to the threat of reprisals at odds with academic freedom. It comes after the Vice President for Research at MSU was forced to resign largely because he cited this study, whose conclusions offended political activists. See “And the Truth Shall Get You Fired,” Crime and Consequences, July 2, 2020.

    So the researcher at MSU had good reason to seek this retraction, to avoid facing a similar fate.

    Shootings of black people aren’t any higher than one would expect given the black crime or the black arrest rate, which buttresses research finding that police shootings are not systemically racist. The black crime rate is simply much higher than the general crime rate, as data from the Bureau of Justice Statistics and the FBI’s Uniform Crime Reports shows. Half of all murders are committed by African-Americans, who account for only 13% of the U.S. population.

    Despite all this, most unarmed people shot or killed by the police are white.

    It is quite obvious that most police shootings are not due to racism.

    1. I appreciate the willingness to look at the bigger picture here, though the simple explanation often doesn’t tell the whole story when applied to a complex system.

      The crime are argument assumes that black and white communities are policed at the same rate and are convicted of crimes at the same rate.

      If two groups of people put their hand in a cookie jar at the same rates, but once gets checked in on by the police more often and convicted more often, they will have a crime rate that has been skewed.

      Research has been done on this, finding that even when controlling for crime rate, the disparities exists:

      1. Is the suggestion here that predominantly white communities have the same rate of crime as predominantly black? There is no study on earth that will support that. Police resources are deployed where they are most needed, and your “cookie jar” metaphor implies that there are vast swathes of not just crime but violent crime being ignored en masse in predominantly white communities. That is insane.

        1. I think the question should be do economically equal communities of black and whites have the same crime rate? We know that blacks use marijuana as much as whites, but they’re charged four times as much. Does racism produce more poor black communities? We know that if you take a resume only change the name, Sally, one Marisha, Sally gets called back at a much higher rate than Marisha.
          Is it at all possible that the same person that throws out the Marisha resume also arrest people for marijuana?
          Any studied that says there is no difference, need to explain these facts.

      2. This is not about cookie jars. It’s about violent crime. The idea that the crime rate is being skewed when it comes to murder defies belief. You can’t hide murders—at least not many of them. And high murder rates always go together with high violent crime.

        Of course there are more police in places with high violent crime rates. Or there should be. Because, you know, crime is bad for people and the police are there to stop it.

        And there are no “black communities” or “white communities.” There are communities that are predominantly black or white. But this is always changing, and there are Asians, Hispanics of both races, and others mixed in. They belong to communities also.

        1. You’re conflating violent crime with murder which is not its definition. Battery, assault and other violent crimes (which whites are 40% less likely to be convicted of even when they are arrested) are very easily malleable data. It’s very simple to conclude that police cannot report violent crimes they are not around to see. And since policing is based on instances of crime reported in an area I’m confused how you don’t see the obvious correlation.
          There are no black areas? LOL have you been to the south side of Chicago? You’re more likely to see an alien than a white person. Generations of redlining in areas like Chicago’s south side and MN created condensed poverty specific to people of color. The result being that decent education and access to basic resources near impossible (the only consistent predictor of crime in every race). That would be the discussion to have if anyone actually cared about controlling crime. It makes sense to intentionally create a poverty ridden area and then police the crap out of it and call it fair? Just say you don’t give af and move on.

          1. There are also large victimization surveys that are carried out regularly – where people report the race of their victimizers and the type of crime they were victims of. It largely aligns with the data on arrests and charges. That is to say, it is an independent line of evidence that corroborates the conclusion that blacks commit a disproportionate amount of crime, especially violent crime – they are overrepresented by a large margin.

      3. Thanks Thadryan for that link. Ross et al say, when controlling for crime rate the disparities exist only within the subset of police shootings involving *unarmed* suspects. They state:

        “The formal derivations provided here validate the benchmarking methods developed in Cesario et al. (2019) when applied to the relative counts of police killings of armed suspects. As in their initial analysis, we find reliable evidence that lethal force by police occurs in direct proportion to race-specific rates of violent crime. However, our formal derivations show that the benchmarking methods developed in Cesario et al. (2019) are misleading and statistically biased when applied to the relative counts of police killings of unarmed suspects.”

        Me again… The vast majority of suspects shot by police are not unarmed. Police shoot around 1,000 people per year, [1] fewer than 1/10 of whom were unarmed (e.g., 68 in 2017, 41 in 2018 and 41 in 2019). [2]

        So we’re making large inferences (systemic police racism) from remarkably small numbers, and in fact from the smallest subsets in large datasets. You have to wonder, if police are doing racially motivated shootings, why does statistical evidence for such bias disappear as the number of cases increases? Well, we know that small numbers are more prone to noise. Also, if police exercise anti-Black bias, why would they not exercise it against armed Black suspects? That does not seem sensible and suggests to me the results among the smallest numbers might be a spurious artifact of underpowered data.

        So Ross et al actually bolsters the case Fred (above) and Cesario et al. (2019) make wrt armed suspects. Ross only adds the caveat that anti-Black bias appears to exist only in the subset of police shooting that involve unarmed suspects.


        1. Institutionalized racisms definition is not just police killings. It’s over policing of black neighborhoods , it’s higher interaction rates with police for non-violent drug offenses with equal usage, it’s blacks receiving higher sentences for the same crimes, it’s blacks being intentionally redlined to areas that don’t have access to education and resources (the primary predictor of crime in every race). If you like stats and data there is plenty to back this up. Institutionalized racism isn’t just about cops. But if often starts with them.
          And to your question of why shootings of unarmed people are meniscal on a large scale I would like point to the patchwork of our country. Did you view the data on where the cases of force took place? They are concentrated and repetitive. The overwhelming majority are in areas of high diversity but high segregation….the leader Chicago is a great example. Being that 23 of our states/ territories have less than a 5% black population I’d say it less likely that they are “isolated incidences” as much as they are concentrated incidents where there are actually black people.

          1. ” it’s blacks being intentionally redlined to areas that don’t have access to education and resources (the primary predictor of crime in every race). ”

            Completely false. The biggest predictor of crime in every race is Age (linked to physical capability) . The second biggest predictor is Sex. The third is Culture. If your culture or subculture praises, gives social profits and other advantages even if temporary that increase your status then it is an easy path for teenagers. Status is one most important things for a teen.
            Then we have family and lack of it.

            There are no areas that don’t have access to education except in mountainous lands and sparse populated ones.
            What you might have said is that maybe current education increase crime, because current schools reward bad behavior. School are open fields for bullying and harassment and even cruelty.
            It was already somewhat that way here in Europe in 80’s. But there were other societal brakes.

          2. Nia, you attribute to me a question that I did not ask. Police shootings are w/o question the premiere topic propelling public sentiment that police are racist. And shootings are the topic of the study linked in comments as well as the retracted study.

            Men also receive longer sentences for the same crime than women. I’d like to see how the excess of sentencing for men vs women compares to that of Blacks vs Whites to see if the differences in the corresponding offending of these demographics predicts the differences in sentencing to test the hypothesis that excess sentencing reflects prejudice on the part of judges as to the likelihood a member of a demographic will re-offend. That would still be unacceptable prejudice, but it might be irrespective of race or sex specifically.

            Men get longer sentences for the same crime, but nobody I’ve seen assumes that’s because judges are misandrists. Likewise, I don’t think we should automatically assume a racial disparity in sentencing reflects racism, but it’s certainly a possibility.

          3. Pop Smoke, popular rapper, murdered: “My music is meant for kids growing up the way I did, who carry guns to school because they don’t feel safe”.
            Watch the way he inspires kids:
            Today, three men and two juveniles (policemen?) were arrested for his murder.

        2. Mapping Police Violence actually state that
          “Levels of violent crime in US cities do not determine rates of police violence.”

          The study ( also states that anti-black bias is consistent for non-lethal use of force as well. Caesar et al. claim “if different groups are more or less likely to occupy those situations in which police might use deadly force, then a more appropriate benchmark as a means of testing for bias in officer decision making is the number of citizens within each race who occupy those situations during which police are likely to use deadly force”, one obvious situation being the suspect is armed. However, black people are more likely to be fatally shot despite white people being more likely to be armed.

          And I wouldn’t necessarily say that police shootings is the only way to look at police bias and brutality. (especially since many reports of police shootings aren’t filed.)

      4. “The crime argument” does not assume that black and white populations are policed at an equal rate, in fact it implicitly argues that the higher crime population will have more police encounters. Moreover, if I recall correctly, by looking at crime reporting and victimization surveys, one can see that white people are actually arrested at a higher rate per violent crime compared to black people.

        Unfortunately, the paper you linked to is one that should actually be retracted. Its title misrepresents its findings: they in fact reproduce the findings that police shootings are consistent with crime rate. They then introduce a nonsense method by benchmarking unarmed shootings against the rate of non-criminality.

    2. “Shootings of black people aren’t any higher than one would expect given the black crime or the black arrest rate, which buttresses research finding that police shootings are not systemically racist.” So every Black person shot was committing a crime?

  5. Cultural Revolution marches on.
    Science is only tolerated if says things that the Revolution approves.
    Soon it will manufacture things that Revolution wants approved.

    In Revolutions the first thing that is lost is the Freedom to say No.

  6. So what “mistake” was made, exactly. The mistake of not agreeing with the blacks as eternal victims narrative pushed by BLM and the liberal democrats? Please explain. Thanks.

  7. If someone had told me ten years ago this is what academia would become I would have not believe them. This is a dystopian nightmare. Facts that don’t corroboraye the narrative cease to be facts. I have no words to explain the lunacy…

    1. There ARE no words to explain the lunacy.

      It isn’t just that a factually accurate study was retracted on the grounds that it doesn’t comport with the current “woke” narrative, or that somebody lost his job for funding a study that COULD lead researchers to the “wrong” set of data points in the first place. All of that is bad enough (or should I say, mad enough). But what’s worse is the lack of outrage. You’ll find a few (a rare few) articles as well as the odd jeremiad on readers’ comment threads such as this one. But otherwise – – crickets. Nobody cares. And I just can’t get my head around that. Nobody cares.

    2. White men spent over 75% of this country’s history implementing, enforcing and benefiting from anti-Black racist laws and policies. But you claim racism ISN’T part of American culture? Has an unarmed white person ever been murdered for holding up his wallet or cellphone

  8. An interesting variation of the argumentum ad hominem: Attacking a thesis by attacking the one who presents it by attacking his employer.

  9. The entire tone of this article is indefensible. It is siding with the forces of regression and scientific legerdemain by not noting in clear and unambiguous language that the situation is objectively Orwellian doublespeak on the part of the “now chastened” researchers and indefensible attitudes in academia.

    Very rarely in life are there such easy set ups like “good guys” and “bad guys”, and this is one of them. MacDonald and Hsu are the forces of reason and entire cast of comic book villains with twirling mustaches are arrayed against them. There can be no neutral descriptions of this awful witch hunt and craven bowing.

    Retraction Watch is nearing greatness in its mission. This is a deep disappointment. I understand the fear that comes with calling an orange and orange, but one can’t bend the knee to tyrants forever.

    There has never been a clearer case of bad men doing bad things for bad reasons. It tarnishes Retraction Watch you that you didn’t make that clear. Shame on you.

  10. Someone take their data and their methods and republish their results.

    Now that the study has been retracted (and not because the data/methodology was flawed), there should be no issue in re-publishing their results.

    This seems like pretty low-hanging fruit for anyone willing to take the risk.

  11. @Jedi Master,

    Self report studies actually note somewhat lower overrepresentation, particularly in less serious offenses, than official records do.

    Victimization surveys are optional and there’s no penalty for not participating (although, black and indigenous victims are the most likely to report). So, that could actually lead to crimes going unnoticed. Also, there are limitations, even with the surveys, for example, they don’t usually study suburban or rural areas, cannot verify the claims, and cannot detect cases where the victim is too traumatized to report (I’ll link some studies which delve into fundamental limitations here)

    And “especially violent crime” is false, as excluding all the specific branches that fall under violent crime (murder and non-negligent manslaughter, rape, robbery, aggravated assault, and gang violence) crimes like embezzlement and illegal gambling have a higher proportion of black offenders than the branch of violent crime.

    As a whole, evidence suggests that there’s bias in the justice system, not simply police brutality.

    1. Funny how the non-researcher in your Washington Post link takes the time to question the methodology of the “contrarian studies” but doesn’t do so for those which he thinks fits his narrative and in addition, lists plenty which are not peer-reviewed but are “reviews of data that tend to speak for themselves and don’t require much statistical analysis.”
      Of course, the potentially more credible academic studies even cite potential problems with their analysis such as “A large-scale analysis of racial disparities in police stops across the United States”. It would have been nice, as that journalist claimed…

      “I will note that most (but not all) of these studies do factor in variables that address common claims such as that the criminal justice system discriminates more by class than by race, or that racial discrepancies in sentencing or incarceration can be explained by the fact that black people commit more crimes.”

      …for him to have made a note of which studies (especially those which are “official” or peer-reviewed) do so, as I’m sure he realizes the average reader will only have time to fully read a few of them. I could only scan the conclusions of a few of them and did not see class or crime rate mentioned.

  12. “Although our data and statistical approach were valid to estimate the question we actually tested (the race of civilians fatally shot by police), given continued misuse of the article (e.g., MacDonald, 2020) we felt the right decision was to retract the article rather than publish further corrections. ”

    The info is TAKEN OUT OF CONTEXT.

  13. “Although our data and statistical approach were valid to estimate the question we actually tested (the race of civilians fatally shot by police)”

    This is false, IMO. In one analysis, they have two collinear predictor variables: demographics (what % of the county is black) and “race-specific crime rate” (what % of violent crimes were committed by black people in that county). Being collinear means they are highly correlated and are essentially interchangeable in an analysis. They do the correct thing and only use one of those in the regression because including collinear predictor variables can give you misleading results. But then they do a very WRONG thing and completely forget that those predictors are interchangable. They interpret their results ONLY in terms of “race-specific crime rate” and not demographics. They even go so far as to say, “At a high level, reducing race-specific violent crime should be an effective way to reduce fatal shootings of Black and Hispanic adults.” Their results CANNOT differentiate between the effects of demographics and race-specific crime rates as the choice of which of those collinear predictors to keep in their model should be arbitrary (either would give the same results because they are collinear).

  14. “Ultimately, the number of cases in which someone is fatally shot under clearly wrong conditions is very low. Given the number of police-citizen encounters each day, the likelihood that any one citizen will be shot by the police is extremely low; the likelihood that someone not engaged in criminal activity will be shot by the police is almost zero. ”

    -Joseph Cesario
    Associate Professor of Psychology, Michigan State University

  15. How can you retract an article on the basis of what others chose to use it for?

    After the US dropped the nuclear bombs on Japan, Einstein is reported to have said ” If I knew they were going to do this I would have become a shoe maker”. What he didn’t do was retract his work on the equivalence of mass and energy that provide the theoretical basis for nuclear fission bombs..

Leave a Reply

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

This site uses Akismet to reduce spam. Learn how your comment data is processed.