Patterico's Pontifications

7/7/2020

Authors of Controversial Paper on Police Shootings: We Retract Our Paper So It Won’t Be Misused

Filed under: General — Patterico @ 8:29 am



This is odd, to say the least. From Retraction Watch:

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.

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.

. . . .

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.

Have they discovered further flaws since, that undermine the central thesis of the paper? No. They say they want the paper retracted because they don’t like the way it has been used by people pushing a certain agenda.

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.

I do not have time to dive into this subject further this morning, to see what the data say and what their correction said, to see how Heather MacDonald used the data, and then to place it all in context. I plan to look at those issues in greater depth when I have more time. But the reason offered for the retraction request got my antennae up. Remember: they already corrected the alleged error in their portrayal of the data. They don’t contend the paper remains flawed.

Let’s be clear: researchers are retracting a paper that is racially controversial, not because they believe that the paper is flawed, but because they don’t like the way it has been portrayed by people with an uncomfortable agenda.

That strikes me as similar to the overreactions we have seen with things like The Office editing out a scene making references to blackface, even though the point of the joke was that every sensible person in the show was appalled at blackface.

Society has gone insane, and the insanity looks to be spreading to data itself. If the data is inconvenient to the narrative, we’ll simply suppress the data.

I have a feeling I will saying a lot more about this.

62 Responses to “Authors of Controversial Paper on Police Shootings: We Retract Our Paper So It Won’t Be Misused”

  1. Well, what DO you do when the facts won’t support the narrative or the mission? Get new facts, obviously!

    Kevin M (ab1c11)

  2. Michigan State already fired the professor, Stephen Hsu, who authorized funding for the study. Can’t be funding studies that might undermine the narrative.

    Edoc118 (aa8990)

  3. From your description, it sounds like the study focused on “who does the shooting” rather than “who gets shot.”

    John B Boddie (f44786)

  4. This fits in with the mayor of NY ordering the contact tracers not to ask if people went to the protests. Would not want to gather data that undermines the narrative.

    Bored Lawyer (56c962)

  5. Seems like the PNAS has the ability to reject their retraction request. I assumed it was vetted and peer-reviewed before publication.

    Paul Montagu (e2c658)

  6. “You are entitled to your opinion. But you are not entitled to your own facts.”
    ― Daniel Patrick Moynihan

    It turns out that it doesn’t matter if you are entitled to them. It also turns out that once you’re on your knees it’s hard to stand up straight again.

    frosty (f27e97)

  7. have any of the people loudly bemoaning this bothered to look at the study?

    Time123 (306531)

  8. Time123 (306531) — 7/7/2020 @ 9:58 am

    Have they discovered further flaws since, that undermine the central thesis of the paper? No. They say they want the paper retracted because they don’t like the way it has been used by people pushing a certain agenda.

    Are you saying this isn’t true? That the study has flaws? Or are you about to lay the groundwork for why it made sense to pull an otherwise accurate study because it’s uncomfortable?

    frosty (f27e97)

  9. you can’t tear down the police, if there’s evidence against the proposition, like the joyce foundation,

    narciso (7404b5)

  10. That’s why they retracted it, their facts said that situation A and B, X and Y are not correlated.

    They then clarified and said we were testing for A and B for X and Y, not kumquats and carburetors.

    Now they’re saying Eff it, I’m tired of people taking this and say A is potatoes, b is phone, x is kumquats and y is carburetors.

    It doesn’t paste well. Here’s the original clarification….

    A recent PNAS study, Johnson et al. (1), investigates the role of race in fatal police shootings. Unlike previous studies which focused on victim race alone, the paper features original data about the race of officers who use deadly force and offers a rare accounting of other shooting attributes that contextualize fatal encounters. Johnson et al. (1) discuss possible “discrimination by White officers” (ref. 1, p. 15877), but conclude racial diversity in police agencies brings limited benefits—a claim cited by major news outlets and in US Congressional testimony, inflaming an already contentious policy debate.

    Despite the value of this much-needed research, its approach is mathematically incapable of supporting its central claims. In this letter, we clarify the gap between what Johnson et al.’s study asserts and what it actually estimates, as well as the implications of that difference for policymaking and future scholarship on race and policing.

    Johnson et al.’s study asks “the degree to which Black civilians are more likely to be fatally shot than White civilians” (ref. 1, p. 15877) and prominently asserts “White officers are not more likely to shoot minority civilians than non-White officers” (ref. 1, p. 15877). In the language of probability, Johnson et al.’s study (1) concludes
    Pr(shot∣minority civilian,white officer,X)  −Pr(shot∣minority civilian,minority officer,X)≤0,
    [1]
    where X are encounter attributes.

    Johnson et al.’s (1) analysis cannot recover these shooting rates because all observations in the data involve shootings. Instead, it estimates “whether a person fatally shot was more likely to be Black (or Hispanic) than White” (ref. 1, p. 15880), which does not correspond to the stated assertions. In a preprint response to our concerns, Johnson and Cesario (2) acknowledge the gap between the claim and the quantity estimated. Yet despite this, Johnson et al.’s (1) original paper infers no “evidence of anti-Black or anti-Hispanic disparity…and, if anything, found anti-White disparities” (ref. 1, p. 15880) simply because more fatally shot civilians are White.*

    Johnson et al.’s (1) analysis cannot inform the original claims without accounting for Bayes’ rule:
    Pr(shot∣minority civilian,white officer,X) −Pr(shot∣minority civilian,minority officer,X)  =Pr(min.civ.∣shot,white off.,X)×Pr(shot∣White off.,X)Pr(minority civilian∣White officer,X)    −Pr(min.civ.∣shot,min.off.,X)×Pr(shot∣min.off.,X)Pr(minority civilian∣Minority officer,X).
    [2]
    Johnson et al.’s (1) study examines only part of the numerators in Eq. 2’s right-hand side, terms dealing with Pr(minoritycivilian∣shot,…). Because it does not consider how many minority or White civilians are encountered, Pr(minoritycivilian∣…)—Eq. 2’s denominators— Johnson et al.’s (1) study does not show whether “Black civilians are more likely to be fatally shot than White civilians” (ref. 1, p. 15877); i.e., Pr(shot∣blackcivilian,…)>Pr(shot∣whitecivilian,…). Similarly, the claim that “White officers are not more likely to shoot minority civilians than non-White officers” (ref. 1, p. 15877), i.e., Pr(shot∣minoritycivilian,White officer,X)≤Pr(shot∣minoritycivilian,non−White officer,X), is unsupported. The omission of Pr(shot∣Officer race,X)—the second part of Eq. 2’s numerators—further separates the stated claim and the quantity estimated. As Eq. 2 makes clear, the addition of controls, X, such as the number of crimes committed by each racial group, does not solve these conceptual issues.

    Johnson et al.’s (1) study describes attributes of fatal police shootings. While a contribution, these facts alone cannot inform the relative likelihood of White and non-White officers shooting racial minorities. Readers and policymakers should keep this important limitation in mind when considering this work.

    Colonel Klink (Ret) (305827)

  11. The data did not fit the racist cop narrative. Had to be memory holed.

    Frank (bcb576)

  12. History repeats itself. Different study, same nonsense. Who coulda thunk it?

    Now the damage control begins in earnest. Everyone with a stake in the racial profiling myth, from the state attorney general to the ACLU to defense attorneys who have been getting drug dealers out of jail and back on the streets by charging police racism, is trying to minimize the significance of the findings.

    Heather Mac Donald, eighteen years ago.
    https://www.city-journal.org/html/racial-profiling-myth-debunked-12244.html

    And, note the spineless complicity of the Bush administration.

    beer ‘n pretzels (4d3c08)

  13. I’m sure this will all end once Biden is elected. No wait, it won’t end. it will get worse. Papers like that will never see the light of day.

    Democrats are pushing censorship. Most Americans can’t see that yet. But they will soon enough.

    Hoi Polloi (7cefeb)

  14. History repeats itself. Different study, same nonsense. Who coulda thunk it?

    Again, their study didn’t test for this, wasn’t for this, this is exactly why they’re retracting it, because people keep conflating two things that look similar but are in fact not the same thing.

    You see, one thing isn’t another thing, even if you want to believe it is. The people who made the one thing are telling you, stop saying its this other thing because it’s not, because WE MADE THE THING!!

    Colonel Klink (Ret) (305827)

  15. Frosty, I didn’t say the part you quoted. I haven’t had a chance to look into this yet and come up with an opinion.

    Time123 (7cca75)

  16. Colonel Klink (Ret) (305827) — 7/7/2020 @ 10:58 am

    Again, their study didn’t test for this, wasn’t for this, this is exactly why they’re retracting it, because people keep conflating two things that look similar but are in fact not the same thing.

    This isn’t why you retract an otherwise valid study.

    The people who made the one thing are telling you, stop saying its this other thing because it’s not

    This is only a problem because the mob does exactly what you do. The mob doesn’t don’t like a group, these people have a study the group is using, rather than deal with the actual issue it’s much easier to just attack the people who made the study, and get them to pull it. Problem solved and now the mob can say the bad group used a discredited story.

    frosty (f27e97)

  17. Frosty, How much of the study have you read?

    Time123 (7cca75)

  18. What Klink describes as “the original clarification” is actually the original criticism, namely that what Cesario and Johnson claimed could be inferred from their data cannot, in fact be inferred.

    And the criticism, which is based on an elementary foundation of probability called Bayes’ Theorem, is correct.

    In simple terms, the article claimed to show that the probability of a minority being shot by a minority officer is higher than the probability of a minority being shot by a white officer.

    But what the data they used actually showed is that the probability of an officer who shot a minority being a minority is higher than the probability that an officer who shot a minority was white.

    While those sound similar, they are quite different quantities mathematically. What they claimed they showed and what they really showed can be related to each other using Bayes Theorem – but to do so you need other information that they didn’t have (and which would be very difficult to collect, as a practical matter).

    Another pretty easy way to understand the problem is that they were trying to determine the probabilities of minorities getting shot by officers of different races, but their data only included shootings. To show what they claimed they would also need records of all encounters between minorities and police (broken down by the officer’s race) where there *wasn’t* a shooting.

    The original paper should never have been accepted in the form it appeared.

    Dave (1bb933)

  19. Somewhere Mark Regnerus is sadly shaking his head. Garbage work reaching pre-approved progressive conclusions is published regularly, with only the lightest and most cursory effort at peer review. Only occasionally are they challenged, Yet those unicorn studies that go against established leftist academic dogma get pored through with every fact and figure subject to intense challenge, until the authors are either excommunicated from the field (Regnerus) or forced to confess their crimes (Cesario and Johnson).

    JVW (ee64e4)

  20. Time123 (7cca75) — 7/7/2020 @ 11:28 am

    Not enough to debate it. Which means I have to rely on signaling. The PNAS is a peer-reviewed journal and that has some weight. The authors are also fairly consistent in their replies to criticism and don’t themselves seem to have an agenda. The criticism also seems valid and technical enough that it would require some understanding of the study and of the criticism to evaluate it.

    But the decision to retract the research doesn’t seem to be based on technical grounds. At least not as far as I can tell. The decision to pull it seems to be political and that is a very big red flag. Add to that the fact that there will be a lot of people with no concern for the effect this will have on future research, in fact, that effect will be seen as a good result, and I’m at least skeptical enough to play devil’s advocate.

    If the argument is that this research needs to be retracted for no scientific reason but instead based on the political use to which it’s being put and you’re in favor of that can we review that Copernicus and the Church issue again?

    frosty (f27e97)

  21. Here is an example of what I meant above when I said that garbage work reaching pre-approved progressive messages is accepted after only cursory peer review. I forgot the link in my original comment. This study stood as gospel among the social sciences crew and the progressive elite for four years, until one lone graduate student decided he wanted to build upon the author’s work but discovered their error when he was trying to understand their methodology. There’s no excuse for this study being originally published; any competent reviewer should have seen the same error. Yet it fit the narrative, so now we have it.

    How many similarly-flawed studies do you think exist in fields such as education, which have caused policy-makers to make laws and regulations based upon the mistakes?

    JVW (ee64e4)

  22. Oh, is that how social SCIENCE works?

    We studied Y and found X.
    Someone used our study in way we don’t like
    Therefore, our study is no longer valid.

    Uh, no. your study and your conclusion remain valid. It doesn’t matter that you have “retracted” your study because of your left-wing politics. Too bad, you don’t want to defend your study against your Stalinist co-workers, but too bad.

    rcocean (fcc23e)

  23. unlike the garbage surgisphere, which was fed to the lancet, and was the pretext for the who suspending hcq trials,

    narciso (7404b5)

  24. Not enough to debate it. Which means I have to rely on signaling. The PNAS is a peer-reviewed journal and that has some weight. The authors are also fairly consistent in their replies to criticism and don’t themselves seem to have an agenda. The criticism also seems valid and technical enough that it would require some understanding of the study and of the criticism to evaluate it.

    means that you aren’t actually basing any of that on reality, as you would know if you read their reasons for retracting, the questions from PNAS, and the clarification from them. It’s in plain English.

    But the decision to retract the research doesn’t seem to be based on technical grounds.

    It’s specifically because people are ignoring the technical and conflating an unrelated thing and using their research both incorrectly, and disingenuously, to infer that X is Z when they at no point was Z on the table.

    You don’t have to play devils advocate, they told you the reason.

    Colonel Klink (Ret) (305827)

  25. Uh, no. your study and your conclusion remain valid.

    It was never valid.

    It was objectively, mathematically wrong, like saying the probability of a senator being a republican is equal to the probability of a republican being a senator.

    Dave (1bb933)

  26. Just like the bogus boston fed study that fed the subprime bubble. Or warrens debunked bankruptcy research.

    Narciso (7404b5)

  27. You don’t have to play devils advocate, they told you the reason.

    No they didn’t. They did this because the conclusions that should be drawn from this research does not fit the Democrat/BLM narrative that police are racist monsters.

    Colonel Haiku (2601c0)

  28. Without knowing what the data set looks like, I am wondering if the authors could revise the study and draw new conclusions. At the very least they could tweak their original conclusions as they appear wont to do, yet the fact that they are instead retracting the study suggests to me just what Patterico has postulated, that the authors don’t have the heart to face down an academic mob who is enraged that their data runs counter to the preferred narrative of racist white cops hunting innocent black men and women on the streets.

    JVW (ee64e4)

  29. Or warrens debunked bankruptcy research.

    That’s a great example because it is still quoted as gospel in left-wing circles.

    JVW (ee64e4)

  30. Without knowing what the data set looks like, I am wondering if the authors could revise the study and draw new conclusions. At the very least they could tweak their original conclusions as they appear wont to do, yet the fact that they are instead retracting the study suggests to me just what Patterico has postulated, that the authors don’t have the heart to face down an academic mob who is enraged that their data runs counter to the preferred narrative of racist white cops hunting innocent black men and women on the streets.

    But that’s not what the authors are complaining about, they’re specifically not even concerned with the “academic mob” that’s just a fantasy. They’re retracting it because it’s being used at best incorrectly, by people trying to say that police violence against blacks isn’t a thing.

    This is exactly the problem, they’re trying to avert, blaming the “academic mob” for a thing they aren’t doing, while some caucasian enthusiasts use their research to promote something their research doesn’t say.

    Colonel Klink (Ret) (305827)

  31. that was the building blocks behind the tarp fund, the stimulus and obamacare, as well as the consumer bureau, alas her lead critic passed away some years ago

    narciso (7404b5)

  32. Colonel Klink (Ret) (305827) — 7/7/2020 @ 11:59 am

    means that you aren’t actually basing any of that on reality

    I keep asking myself if you are unable or unwilling to make a good faith argument. Now, I’m asking myself why I keep asking myself that.

    It’s specifically because people are ignoring the technical and conflating an unrelated thing and using their research both incorrectly, and disingenuously, to infer that X is Z when they at no point was Z on the table.

    You don’t have to play devils advocate, they told you the reason.

    You aren’t even listening to yourself. Right here, in this comment, you claim they are retracting it for political reasons. You yourself say that the technical issues are being ignored, i.e. the retraction doesn’t have anything to do with the technical details of the study.

    You’re just making the usual leftist case that of course it should be pulled because it’s being used to undermine the narrative. Maybe you should let Dave run with this one since he’s actually making a better argument on the merits.

    frosty (f27e97)

  33. 28 Sorry JVW, but this uninformed conspiracy mongering.

    The letter Klink linked to explains exactly, in math, why it’s impossible to answer the question the study claimed to without data they don’t have and can’t get.

    I also tried my best to explain it above.

    Dave (1bb933)

  34. You aren’t even listening to yourself. Right here, in this comment, you claim they are retracting it for political reasons. You yourself say that the technical issues are being ignored, i.e. the retraction doesn’t have anything to do with the technical details of the study.

    Duh, as I’ve said in every post, they’re not retracting it for technical reasons, they’re retracting it because idiots…ahem…keep trying to use it to justify something that it doesn’t say. So they don’t want people (you) claiming it does, but you don’t even want to read it, you’d rather FEELZ about it not facts (got one in for you). Basically they are saying “Keep my name out of your dirty mouth” but you keep mispronouncing their name, after being told that’s not what it says, over and over.

    Colonel Klink (Ret) (305827)

  35. But that’s not what the authors are complaining about, they’re specifically not even concerned with the “academic mob” that’s just a fantasy.

    As if these authors would actually come out and say they’re concerned about being ostracized and/or cancelled by the “academic mob”. Wake up and smell teh cat food, Clyde.

    Colonel Haiku (2601c0)

  36. Democrats don’t embarrass easy.

    Colonel Haiku (2601c0)

  37. Dave (1bb933) — 7/7/2020 @ 12:33 pm

    From JVW;

    At the very least they could tweak their original conclusions as they appear wont to do

    This is my understanding of the original round of updates. They published the study that said X based on Y. It was pointed out that X couldn’t be based on Y. The authors agreed and then said Z could be based on Y.

    The response to that, which is the current debate, was that there are still people saying Y -> X so the entire thing, even with the Y -> Z updates had to go.

    frosty (f27e97)

  38. Colonel Klink (Ret) (305827) — 7/7/2020 @ 12:45 pm

    Why do you keep making personal insults and lying about what people are saying? Do you even realize you are lying or do you read comments and the words on the screen get transcribed as you try to understand them? It’s almost as if your cognitive biases are so extreme that any position in any way different from yours has to be the worst version of whatever it is.

    frosty (f27e97)

  39. @20, that’s fair. What bugs me are the liars and frauds strongly proclaiming this or that without any real understanding of what the study says.

    Time123 (7cca75)

  40. just like ferguson’s bogus model, shut down most of the planet’s economy, bostic (who ended up the head of the boston fed) fueled the subprime crisis, michael mann was the pretext for all the green pelf, same with joyce meaning ayers, pushing gun registration,

    narciso (7404b5)

  41. Time123 (7cca75) — 7/7/2020 @ 12:59 pm

    That’s fair too. I’d prefer a world we don’t seem to be able to live in, i.e. one where peer review meant something and science wasn’t political, really one where everything wasn’t political. But we’ve got what we’ve got.

    frosty (f27e97)

  42. Their premise was to research whether black cops were more or less likely to shoot black people more/less than white cops. That will never answer the question “is there racism in policing”. They seem related, and maybe have some shared data, but are not the same thing, were never intended to even ask the question.

    They keep having their research quoted as answering the second question that they didn’t ask, instead of the first question that they did. I’m sure it’s tiring for them, it’s tiring for me.

    Colonel Klink (Ret) (305827)

  43. When you have the leaders of the black lives movement who are now talking about, ‘If we don’t get our demands we’re going to burn it down,’ other black people who are talking about working with other whites and other races being viewed as sellouts or called Uncle Toms, you start to understand that you are now being controlled… someone wants to control the narrative, and I view it as a very, very dangerous self-righteousness.

    — Actor Terry Crews

    Colonel Haiku (2601c0)

  44. I’ll try again.

    Conditional probability is expressed mathematically like this:

    P(A | B)

    Which reads: “The probability of A given B.” In other words, in what fraction of the cases where B occurs does A also occur?

    If A means “it rained” and B means “it’s July”, then P(A | B) is the probability of a rainy day in July, or in other words the fraction of days in July that it rains.

    On the other hand, P(B | A) is “the probability of B given A.” For the definitions of A and B from the previous example, that means probability that it’s July if it rained today – the fraction of all rainy days that occur in July.

    Conditional probabilities are related to unconditional ones in an obvious way:

    P(AB) = P(A | B) P(B)

    This says that the probability of A AND B is the (conditional) probability of A given B multiplied by the probability of B.

    But we can also write P(AB) in terms of the second conditional probability, P(B | A):

    P(AB) = P(B | A) P(A)

    Since both ways of calculating P(AB) must agree:

    P(A | B) P(B) = P(B | A) P(A)

    If we know any three of the probabilities in this equation, we can compute the fourth. In particular:

    P(A | B) = P(B | A) P(A) / P(B)

    which is Bayes Theorem.

    In the retracted study, A meant “fatally shot by a police officer”, and B meant either “civilian is minority and officer is minority” or “civilian is minority and officer is non-minority”.

    They claimed to be comparing P(A | B) where the officer is minority with P(A | B) where the officer is not minority.

    But the data they actually had was P(B | A) , which is the probability that a minority shot by the police was shot by a minority/non-minority officer.

    Bayes Theorem would allow them to compute what they wanted from what they had, but only if they also know the other two quantities on the right-hand side: P(A) and P(B). Which they don’t.

    P(A) is the fraction of minority encounters with police (of any race) that result in the minority being fatally shot.

    P(B) is the fraction of minority encounters with police of the race in question (regardless of whether the minority is fatally shot).

    Since P(A) is the same for officers of both races, it’s not strictly necessary to do a relative comparison (it would cancel out if you took the ratio of P(A | B) for minority cops and non-minority cops).

    But if you don’t know P(B) for both categories of police officers, you’re just out of luck.

    In effect it means you have to control for how often minorities encounter officers of the two race categories. The data does not exist so all you could do is make a very uncertain guess.

    Dave (1bb933)

  45. Their premise was to research whether black cops were more or less likely to shoot black people more/less than white cops.

    For what purpose?

    To find out if certain races require more training?

    If three times as many minorities where shot by white cops, would the study have then looked as to why and have concluded something other than “racism.”

    BuDuh (588820)

  46. Dave (1bb933) — 7/7/2020 @ 1:38 pm

    In effect it means you have to control for how often minorities encounter officers of the two race categories. The data does not exist so all you could do is make a very uncertain guess.

    Is this the complaint against the original version of the story or the revised version?

    frosty (f27e97)

  47. For what purpose?

    To find out if certain races require more training?

    Who knows, I didn’t read their grant proposal. But it makes no difference why they wanted to do something, it matters what the something is, or specifically in this case, isn’t

    Colonel Klink (ret) (305827)

  48. Speaking hypothetically then, if their grant proposal was to look for racial bias in policing, are you saying that it is irrelevant?

    BuDuh (588820)

  49. BuDuh (588820) — 7/7/2020 @ 1:43 pm

    As I understand it, the purpose is related to

    [The research] suggests that increasing diversity among officers by itself is unlikely to reduce racial disparity in police shootings.

    Said differently, we’ve all heard by now the argument that having fewer white cops will reduce the number of black people being shot. The claim is that this study suggests that white cops aren’t more likely to shot non-white people, i.e. more diversity won’t matter.

    Said differently again, this is one of those studies being used to argue that black people, specifically black men, are more likely to be shot because they are more likely to be violent criminals.

    Said differently differently, this is one of those studies that is being used to say white cops aren’t any more or less racists than non-white cops.

    I’m not saying the study says any of those things.

    frosty (f27e97)

  50. I had a post disappear.

    Thank you Frosty.

    BuDuh (17c20a)

  51. Is this the complaint against the original version of the story or the revised version?

    The revision was acknowledging the problem I described, but they still left in wording suggesting that their result had relevance to the question of whether minority or non-minority officers are more likely to use lethal force.

    The two guys who pointed out the flaw in the first place strongly criticized the correction as inadequate:

    While the authors now acknowledge they described a statistical quantity in their original study which they never estimated, they still maintain the result of the test they performed “is consistent with our framing of the results in the abstract and main text. … To be clear, this issue does not invalidate the findings with regards to Pr(race|shot, X) discussed in the report.”

    These statements are false. The original article claimed that, “Racial disparities” in the quantity they estimated “are a necessary but not sufficient, requirement for the existence of racial biases… .” In their significance statement, they also claimed that their results suggest “that increasing diversity among officers by itself is unlikely to reduce racial disparity in police shootings.” In other words, the article claims to shed light on whether white and nonwhite officers exhibit differential levels of racial bias in the use of lethal force, with the aim of informing the promise of personnel reforms in police agencies. Since the study’s publication, the authors have also amplified these strong claims in interviews with several major media outlets.

    But when properly understood, the test that was conducted in the original article sheds no light on racial bias or the efficacy of diversity initiatives in policing, and a meaningful correction would acknowledge this. Because every observation in the study’s data involved the use of lethal force, the study cannot possibly reveal whether white and nonwhite officers are differentially likely to shoot minority civilians. And as we show formally in our published comment, what the study can show—the number of racial minorities killed by white and nonwhite officers—is simply not sufficient to support claims about differential officer behavior without knowing how many times officers encountered racial minorities to begin with.

    [emphasis added]

    The objection to the correction is, again, that the corrected study purports to answer a question it cannot.

    Dave (1bb933)

  52. Maybe we should be less concerned with questions of race and instead concern ourselves about how we can trace police officers from all walks of life to be members of the communities they are sworn to serve.

    Gryph (08c844)

  53. the objective is what we see here:

    ttps://hotair.com/archives/ed-morrissey/2020/07/07/minneapolis-neighborhood-checked-privileged-eschewed-police-hows-working/

    narciso (7404b5)

  54. Dave (1bb933) — 7/7/2020 @ 2:41 pm

    Because every observation in the study’s data involved the use of lethal force, the study cannot possibly reveal whether white and nonwhite officers are differentially likely to shoot minority civilians.

    This part makes sense. There are several statements made by the authors and this one

    We find 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.

    is unsupported. However, this

    As the proportion of Black or Hispanic officers in a FOIS increases, a person shot is more likely to be Black or Hispanic than White, a disparity explained by county demographics;

    isn’t the same statement and doesn’t seemed to be undermined by the objection. Same for

    race-specific county-level violent crime strongly predicts the race of the civilian shot

    And I think the first one does start to undermine the argument that more diversity will improve the situation.

    frosty (f27e97)

  55. “What exactly does being told “I’m going to do something you won’t like” give one license to do?”

    Isn’t that irrelevant. The questions that are relevant is whether calling the police was a reasonable response and whether that was reasonable while still approaching and arguing with CCooper. Much larger man, confrontational, trying to lure the dog away with a treat, in a park with few people around, at night, not backing down, making a cryptic threat….I think many women would be uncomfortable….now how that fear manifests….fight, flight, or paralysis….depends on that individual. You can still be pretty scared yet assume an aggressive posture…because that’s what your instinct (hippocampus and amygdala) tells you is the best option.

    AJ_Liberty (ec7f74)

  56. For the record, I had a year of stats and research design in undergrad and a second year of stats and research design at the graduate level and did quantitative research at the graduate level as part of my program.

    The study itself.

    So, my commentary as I read the thing.

    This is a very poorly written study. I’m not even talking about the data yet, the writing is just bad. It is also severely opaque. Who proofed this? This is why English majors exist, to create coherently written pieces. Why didn’t you people hire one, there are enough out there.

    The data sets they are using are… incomplete. Not great. really kind of a bad idea.

    Do they state a null hypothesis? No, no they do not.

    When shooting at a suspect, officers miss more than 60% of the time. Why do we give them guns?

    From their less than ideal data set, they are only using shootings where the officer killed a person. If an officer is shooting at a suspect, they are supposed to be aiming to kill that person. Why didn’t they use all shootings?

    We are now going to use opaque predictive methodology that will not be clear to anyone who doesn’t know something about stats. AND we are going to write about it in a manner that is almost impossible to understand. AND also requires another set of data analysis for which the methodology is… interesting. This will not go wrong at all.

    Good, they finally controlled for something. (higher percentage of minority officers in minority communities, controlling for that ended up changing some racial disparities in their truly unfortunate data set)

    I have doubts about their predictive methodology. It relies on several assumptions that they contradict in earlier sections and nears the realm of “it’s magic, just trust us”. IDEK.

    Summary:

    They ask two questions. 1. Did the race of the officer have an effect on shootings (based on their incomplete data set.) and 2. Did the race of the shot person have an effect on the shootings (based on their incomplete data set and um, interesting, methodology.)

    The answers that should have been taken from the data as they present it (they do say this in their conclusion, but it isn’t what is described in the abstract at the beginning of the article, which is the thing everyone reads. Sigh.) is 1. that officers are equally likely to fatally shoot people regardless of the race of the officer and 2. That officers are most likely to shoot someone who “looks like” the demographic most likely to commit violent crime in their area.

    My analysis:

    I don’t think their data is sufficiently complete and I think their selection of which data to use is also not ideal. The way they asked their original questions and the way the phrased their results doesn’t match. The methodology for this study is pretty esoteric (and possibly not terrible valid), so is the way they reported their results. This is not a study that should have escaped from academic circles.

    Should it be withdrawn because the general population doesn’t understand what it really say and or is trying to do? Probably not. Honestly there are a lot of social science studies out there that don’t really say what a regular person might think they say, it’s just that the general public doesn’t necessarily latch onto some of them. I suspect that the legal field has similar problems where the legal meaning of something and the general meaning of that same phrase don’t mean the same thing or don’t mean what they appear to mean on the surface.

    Should it be withdrawn because it isn’t very good? Possibly. It’s very poorly written and their methodology seems less than ideal.

    I don’t think it’s a great study.

    Nic (896fdf)

  57. Nic (896fdf) — 7/7/2020 @ 3:28 pm

    This is not a study that should have escaped from academic circles.

    Should it be withdrawn because the general population doesn’t understand what it really say and or is trying to do? Probably not.

    Both of these are true but I’m not sure either would help. A very long time ago, I sat through, what I thought, was a decent talk on using machine learning to solve a variety of physics problems. This seemed like fairly benign stuff, it wasn’t a graduate-level talk, it was just part of a basic program for potential candidates so it was undergrad level content. It was going as well as things like this go until the guy mentioned using genetic algorithms to solve problems involving discontinuous search spaces. A guy in the audience interrupts with questions during the talk about the ethics of using this sort of methodology which got really weird fast.

    frosty (f27e97)

  58. When shooting at a suspect, officers miss more than 60% of the time. Why do we give them guns?

    That’s actually a pretty good percentage.

    If an officer is shooting at a suspect, they are supposed to be aiming to kill that person. Why didn’t they use all shootings?

    That data set, where an officer wounds a suspect or misses altogether, is probably kept in a separate case as those three situations (killed, wounded, missed) are treated differently administratively.

    Hoi Polloi (dc4124)

  59. I wondered how long it would be before it was purged.

    Narciso (7404b5)

  60. This is not new.

    These facts have been suppressed or ignored for years.

    Just this morning I watched Don Lemon on CNN talk over and accuse actor Terry Crews PF ‘stepping in it’ for having the nerve to mention BLM intentions and their ignoring the black-on-black crime victims who have been mowed down by the bushel in NYC, Chicago, Detroit recently.

    To say these shootings and deaths have nothing to do with BLM is as ignorant and racist a statement as one could make.
    _

    harkin (5af287)

  61. @58 The data set probably is probably kept in a separate group, but the researchers were basically creating their own data-set, because the official data set may have been up to 50% off, so they should have been able to include that data.

    Nic (896fdf)

  62. OT, but pray it’s true…

    The media and many politicians inside the Democratic Party continue to shriek over the recent jump in the number of recorded coronavirus cases, seeking to keep the economy closed at all costs — and we mean that literally. Don’t fall for the argument. The data show that, in fact, our pandemic nightmare might well be coming to an end.

    https://issuesinsights.com/2020/07/07/is-the-covid-19-epidemic-coming-to-an-end-at-last/

    Colonel Haiku (2601c0)


Powered by WordPress.

Page loaded in: 0.3849 secs.