The AP’s Biased and Incompetent Racial Attitudes Poll

“There those whites go again, increasing racial tensions!”

All over the internet, the results of the AP’s just released “Racial Attitudes Poll” are being headlined as “proof” that racism is alive and well in America, and that racial bias has increased in the last four years. Either the poll is being released now to attempt to make Americans feel guilty about not wanting to vote for Barack Obama, or it is setting up the excuse for Obama’s defeat, should it occur, that only racism can account for such a successful, brilliant, eloquent leader being defeated. I apologize for the cynicism. With all the talk about “firewalls,” however, it seems self-evident that white guilt, which has been the target of pro-Obama racial politics from the beginning, is one of the most obvious, odious, and desperate.

The poll, in my analysis, is garbage, and unethical garbage as well. It is an accumulation of confirmation bias, locked in with horrible methodology.

To measure “explicit” racial biases, for example, the poll asked its subjects  how well  certain words, such as “friendly,” ”hardworking,” ”violent” and “lazy,” described “most” blacks, whites and Hispanics. That question is a trap. To answer it at all (some did refuse) is to automatically confess irrational bias. As Marissa Tomei charmingly says to the questioning D. A. in “My Cousin Vinnie,” “It’s a bullshit question!” “Most” whites, blacks and Hispanics aren’t any of these things, within my experience, as I haven’t met or gotten to know “most” whites, blacks and Hispanics. But I do know that even if I thought “most” Hispanics, say, spoke Spanish, I would never presume that any one of them did, because that would be foolish.  And if, as I did for two years, I happened to live in a mostly black area in D.C.’s Capital Hill where almost all the young black men I encountered looked at me and spoke to me as if I had ten minutes to live, and was asked, “Are most blacks unfriendly?”, I’d refuse to answer that question as well. As it happened, the vast majority of blacks I encountered every day were outright hostile, but I would never project that experience onto any individual. I could easily see, however, how the AP would find an answer based on my experience at the time “explicitly racist.”

To their measure implicit racism, respondents get questions like this:  “Irish, Italians, Jewish, and other minorities overcame prejudice and worked their way up. Blacks should do the same without special favors.”  This is a matter of debate and opinion, and for researchers to call a positive response proof of racism shows their bias (and political orientation), not the respondents. Similarly, the AP asks questions like this: How much of the racial tension that exists in the United States today do you think whites are responsible for creating?   “Whites”?  I don’t know: I’d say Chris Matthews, Rachel Maddow and other race-baiting MSNBC hosts create a lot of it—are they “whites”? Because I refuse to take responsibility for the slander of those hacks, and I’m white. Are white supremacy groups “whites”?  Are criminals who happen to be black, “blacks”? What are you, if you categorize people who engage in certain conduct by their race, rather than by what they do and say? I think you are a racist. Does the poll’s question mean whites acting as a race? If so, the AP is guilty of overt racism for suggesting such a thing is even possible.

How would any thoughtful person answer this question: “Some people say that black leaders have been trying to push too fast. Others feel that they haven’t pushed fast enough. What do you think?”  I think it’s a vague and useless question, unless I know what the questioner thinks they are “pushing” for. Reparations? Any pushing is too much. (I must be racist.) Hunting down George Zimmerman? The black leaders who advocated that were thugs. (I guess I’m racist.) No voter ID laws? I think they are wrong. (And the AP presumably thinks I’m racist.) Affirmative action? Well, I think they should stop pushing that entirely, because, you see, I’m racist. The AP knows.

What “black leaders?” Al Sharpton, or Allen West? Oprah, or Cornell West? Maxine Waters, or Colin Powell?  Condi Rice, or Sheila Jackson Lee? Are they all “pushing” the same thing? If so, I have no idea what that is, so I can hardly judge its collective speed.

The entire survey is a faux scientific method of validating what liberals and progressives do routinely: assert that if you don’t agree with their preferred race-related policies, you are, by definition, a racist. For news organizations to use  its dubious  conclusions based on slanted questions to justify a headline like the Boston Globe’s unforgivable “Poll Finds Majority Hold Racist Views” only shows where the real bias is located in this country.

You can read the whole survey, if you have a sufficiently strong stomach, here.

_________________________________

Pointer: James Taranto

Facts: AP

Source: Boston Globe

Graphic: eb9

 

 

27 thoughts on “The AP’s Biased and Incompetent Racial Attitudes Poll

  1. ”Irish, Italians, Jewish, and other minorities overcame prejudice and worked their way up. Blacks should do the same without special favors.”

    This assumes that irish, Italians, Jewish, and other minorities actually overcame prejudice.

    Did they?

  2. Whites are not the only ones who will be voting against Obama in this election. Just because Obama is black doesn’t mean that he represents the views of all blacks in the country. Quite a few blacks will not be voting for Obama. They say the reason is because he did not accomplish what he said he was going to do during the past four years. I won’t deny that racism still exists in the country today and some of those racists are the people who will be voting for Obama just because he is black.

  3. Michael, thanks for the laugh. It’s a classic movie that unfortunately probably couldn’t be made today.

    Jack, if we assume for a moment that attitudes on race are important to quantify then what would be a more appropriate way to uncover those attitudes?

    • Oh, I don’t know—an unbiased poll that wasn’t tainted by biased and ideological definitions of “racism”? Questions that didn’t use value loaded vague terms? Unbiased researchers? How about observing conduct that cannot be explained by non-racist attitudes? Is someone with anti-black biases who recognizes that they are emotional rather than rational and thus does act on them still a racist? Lincoln had racist attitudes—so what?

      There are some things surveys can’t determine with sufficient accuracy for the inflammatory nature of the exercise to be justifiable. This is one.

    • That’s the truth, Eric. The term “nigger” is all over the screenplay. Now, as a guy with a clot of Teutonic blood in him, I likewise protest that movie’s depiction of German men as military martinets and the women as sex-crazed Marlene Deitrich clones. In fact, who DIDN’T Mel Brooks insult in that film? Did well, though. It must have been Slim Pickens.

  4. I don’t see any reason why this attempt to explore whether or not there’s still racism has ANY direct connection to the current Presidential campaign. Obama’s election has brought some conversation about racism forward, but the AP’s work isn’t about him.

    • Really? Then why did the researchers apply their dubious findings to the potential popular vote, hypothesizing that racial bias would cost Obama 2-3% points? Why did they put out such a study a week before the election? SOME conversation about racism? Where have you been the last four years? Have you listened to Chris Matthews, Sheila Jackson Lee, Morgan Freeman, Nancy Pelosi, Charles Blow, Andrew Sullivan, or Obama himself? Do you ever check out Twitter? The Daily Kos? The Huffington Post? The Daily Beast?

      I bet you have a bridge in Brooklyn to sell me too.

      • Oh, I don’t think it takes much logic to conjecture that a certain number of voters dislike Obama and won’t vote for him because of his skin color. As for your laundry list, no, I don’t watch TV news, use social media or read “chic” internet opinion sites. I raise guide dogs and work in an Urgent Care helping sick and injured people. Within my values system, it’s a better use of time than your version of being “in the know”. But you are free to use your time as you see fit.

        • What an obnoxious response. Translation: I think it is just as valid to just shoot from the hip without actually knowing what your talking about. You said that there was “some” race-baiting without, by your own admission, paying enough attention to know how much there really was–which was a lot. Then you denigrate the values of someone who does pay attention, ridiculing the process of being becoming informed. Your version of being “in the know” is to fake it. Some values.

  5. I agree with you that the questions you quote are unpersuasive.

    However, the method they used to measure implicit racism was not what you described in your post. They measured implicit racism this way:

    Respondents look at a series of Chinese ideographs and sort them into those that are more pleasant and those that are less pleasant. Preceding each ideograph is a very fast flash of a photograph of the face of an African-American or White person, which respondents are told to ignore. When an ideograph is preceded by a face, people’s affective reactions to the face spill over onto their assessments of the ideograph. People who have favorable feelings toward the face are more likely to label the ideograph as more pleasant, and people who have unfavorable feelings toward the face are more likely to label the ideograph as less pleasant. A summary score subtracting pleasantness judgments of ideographs following African-American faces from pleasantness judgments of ideographs following White faces yields a measure of anti-African-American affect that has manifested reliability and validity as a measure of racism in studies of convenience samples of participants.

    That strikes me as a lot more interesting and valid. using this method, they found that about 56% of their sample exhibited implicit anti-black bias. Measured this way, a majority of both Dems and Repubs exhibit racial bias. (Broken down by political self-identification: 49% of Independents, 55% of Democrats, 64% of Republicans.)

    • It kind of depends on the photographs, now doesn’t it? If the picture of the “white person” was a toddler laughing while playing with a puppy and the picture of the “black person” was Mike Tyson with part of Evander Hollyfield’s ear in his mouth, I bet you would find some “anti-black bias”. All tests of this nature are subjective and can be easily manipulated to give the desired result.

      “The only possible conclusion the social science can draw is: some do, some don’t”. -Ernest Rutherford

      • If the picture of the “white person” was a toddler laughing while playing with a puppy and the picture of the “black person” was Mike Tyson with part of Evander Hollyfield’s ear in his mouth, I bet you would find some “anti-black bias”.

        You’re suggesting that the researchers aren’t merely mistaken, or overlooking some relevant factor, but actually engaged in a conspiracy to produce fake results by cheating.

        That’s not plausible, especially since it would have to be a wide-ranging conspiracy involving not just the researchers behind this one study, but also by the dozens of other researchers who have ever conducted this sort of study.

        • (1) I was giving an explicitly ridiculous example so that my point was not missed.
          (2) The research of this type I have seen is much worse than that produced in medical research. I remember a Science article that found that over half of the medical studies published in JAMA and the New England Journal of Medicine has such fundamental flaws that the results were meaningless.
          (3) You forget that community that produces such studies is not made up of people who want a result that shows that the United States is becoming less biased. This results in a bias to the design and interpretation of the results, that are always speculative in this type of research and especially with sample sizes this small.

          Below are some articles to demonstrate my points.

          It is difficult to find a free version, but here is one where the abstract nicely summarizes the findings of a study of medical clinical studies.
          http://humrep.oxfordjournals.org/content/18/5/1000.short

          An nice article that shows that, yes, you can have such a ‘conspiracy’ as you so casually discount.
          http://mescal.imag.fr/membres/arnaud.legrand/teaching/2011/EP_lies.pdf

          This article nicely outlines the many ways the researcher can bias the results of such a study.
          http://www.epidemiology.ch/history/PDF%20bg/Sackett%20DL%201979%20bias%20in%20analytic%20research.pdf

    • The study was advertised as a poll in all news reports. What you described was an experiment, not a poll. With a poll, we can review the questions. With an experiment, I have to see the pictures, watch the tapes, observe the behavior of the researchers for tainting conduct, and most all, see the results replicated in a different setting by different researchers. Even then, it doesn’t “prove” racism or racist attitudes. It proves that on this occasion, some individuals reacted more negatively to an Asian stereotype picture after viewing a black stereotype picture. Does that tell us anything at all about how such individuals will actually treat blacks, Hispanics and Asians? Everyone has biases that they should be able to recognize and adjust for–biases in favor of pretty women, handsome men, tall men, shapely women, healthy people, people who speak our language, share our preferences—biases against fat people, old people, not-so-bright people, people who use bad grammar. The individual who acknowledges these and refuses to allow himself or herself to let these biases dictate unfair, disrespectful or unjust treatment is just as ethical and virtuous, and worthy of admiration and respect, as some theoretical being without any biases at all. You can’t mistreat a picture.

      I think it’s the epitome of junk science, yet the media swallowed it whole.

      • 1) It’s not true that “all” news reports described it badly. The reason I knew about the implicit bias testing (and thus knew to google it) is that the news report I read about this study was more accurate than the news sources you relied on.

        In general, when a news report makes researchers seem like idiots who haven’t considered obvious factors, you should at least consider the possibility that the research is good but the reporting is bad.

        2) The results of this study are not surprising. It’s not like they found that racism is a huge and overwhelming factor; on the contrary, they found that direct racism based on a candidate’s race can explain only a small percentage of the vote, and the vast majority of voting decisions are explained by other factors..

        If you find that an implausible result, then I have to ask: Do you deny that there is such a thing as anti-Black racism? If not, then why do you get so angry because a study found racism?

        3) There have been a whole bunch of studies of implicit bias, looking not only at racism but at homophobia, sexism, and other areas. The scientific validity of implicit association tests isn’t ironclad, but it’s pretty well established. According to an article in Research in Organizational Behavior, “participants’ implicit associations do predict socially and organizationally significant behaviors, including employment, medical, and voting decisions made by working adults.”

        4) I’m not married to this study’s results regarding how people vote (which I find dubious). Racism, like any belief system that people are motivated to keep secret, is inherently difficult to study, and results will always have potential areas of ambiguity. But I don’t think that means that it’s wrong to study racism, or that all results are nothing but bias.

        Your angry and unfounded smears of the people who ran this study as biased and incompetent, suggests that you’re not just calmly and objectively assessing the evidence. Jon Krosnick is not an incompetent, nor do you have any foundation for accusing him of dishonesty; he’s one of the country’s leading experts on survey design and measuring attitudes, and he knows a hundred times as much about survey design issues than you or I.

        That doesn’t mean he’s right, of course, but it does suggest that your knee-jerk expressions of contempt is based in your political bias, not because you know a thing about the research.

        • 1) Or that the research was devised to give predictable press coverage the opportunity to make an illegitimate point. The study was for public consumption, and overstated dubious findings.
          2. That, however, wasn’t the conclusion the researchers stated, because the emphasized the bias angle, and the take away was that the nation is more racially biased than four years ago. Based on that data?
          3. “According to an article in Research in Organizational Behavior, “participants’ implicit associations do predict socially and organizationally significant behaviors, including employment, medical, and voting decisions made by working adults.” All gut level supposition. They can’t track those decisions, like voting, so they can’t confirm the thesis. I tend t agree, but that’s not the same as reliable data.
          4. Of course we should study racism. I said this study is junk, and it is junk. It would be junk if it proved there was no racism at all, too. Were those clear questions with terms fairly described? You are arguing for the validity of a study because you like one component of it. My view: those questions are so slanted and poorly written that they forfeit trust.

          My reactions were not knee-jerk, and you should not impugn my process or motivations. I looked to the study objectively—I use and review social science studies regularly. And I took the poll. I’m a reasonably smart guy, and when I read a question about what qualities “most” Hispanics or blacks have, I think: Why would I answer a question that proves bias even by answering it? What a stupid, dishonest, biased and incompetent question! Are these researchers incompetents, or do they have an agenda? That’s no knee-jerk reaction. I don’t necessarily disagree with the OPINION of the researchers at all. Mostly, I think their gut instincts are correct. But their lazy, incoherent, biased study doesn’t “prove” a thing, and it shouldn’t have been released.

  6. I looked to the study objectively—I use and review social science studies regularly.

    Jack, it’s obvious from our discussion that you weren’t even aware of a major component of this study until I pointed it out to you. Yet I’m supposed to accept that your opinion is based on an objective assessment of a study that you obviously didn’t read or fully understand?

    Regarding the questions, they do seem dodgy to me, but that in and of itself isn’t enough for declaring this junk science. Those questions have been used in prior studies, after all. I’m dubious about them because they look (on their face) as if they might be really asking about political ideology rather than race. However, if it turns out that study after study shows that those questions correspond in a significant fashion with anti-Black attitudes after controlling for political ideology, then logically I’d have to concede that the questions seem to be measuring something relevant after all.

    I don’t know if that’s the case, because I haven’t done the relevant background reading on these questions. But you haven’t done it, either.

    I think that my assessment – which I’d describe as “this element of the study seems solid, this element seems dubious but I’d need to know more” – is more reasonable than dismissing the entire study, including the portions that you don’t even have a shadow of a rational argument against, as junk science.

    • Wrong. First of all, I was very aware of the face/character association component. Again, the study was repeatedly referred to as a poll, and I used the poll portion of it, which I could fully examine unlike the photo test. I also wasn’t satisfied that the descriptions of the face/race test gave me enough detail to comment on, absent the actual photos, without knowing how much time elapsed, and actually seeing the process. I know from various suspect ID methods how easily the conduct, words and attitudes of the individual presenting the photos can influence the results. Second, if a major component of a study, which the poll certainly was, shows sloppiness, poor design and bias, that is plenty of data to conclude that the findings, plausible or not, should be ignored, not plastered in headlines.

      Finally, this is as junky as the study:

      “Regarding the questions, they do seem dodgy to me, but that in and of itself isn’t enough for declaring this junk science. Those questions have been used in prior studies, after all. I’m dubious about them because they look (on their face) as if they might be really asking about political ideology rather than race. However, if it turns out that study after study shows that those questions correspond in a significant fashion with anti-Black attitudes after controlling for political ideology, then logically I’d have to concede that the questions seem to be measuring something relevant after all.”

      Yes, I recognize some of the questions from a similarly slanted study produced to show that tea party supporters are racists. So if “dodgy” questions that assign racial attitudes to legitimate policy positions repeatedly are used to claim racial bias when in fact the researchers are the ones who are biased, that repetition validates the use of the questions? The fact that such questions are used. it seems obvious to note, prove the bias of the researchers. How can you assume that they correct the data to control for political ideology?

      We have been bombarded by such studies, showing that Republicans are cruel and dumb, conservatives are rigid and stupid, and that all of them are racist, since the scholarly community decided that it could use distort its weapons into political weapons, almost always in the service of the Left. This began around the time that Bush was elected, and it dove-tailed nicely with new found popularity of pop-psychology studies championed by the likes of Gladwell and Arielly. What it did was convince me that social science had forfeited its trustworthiness. This study just reinforces that conclusion.

      OK, you want the study to be true, so you can be sleep soundly convinced that those who disagree with you dream secretly of lynchings and and mass deportations. But the whole idea behind the study is based on a logically false premise, and the kind of premise that researchers, oddly, never seem to use to show phony links between liberal ideology of liberal sacred cows–theirs—and unattractive qualities and characteristics. When a study is contrived to show a dubious link that offends the liberal establishment, say, Jensen’s race IQ treatise, it is not reported as happy fact, and the academic community aggressively dissects it into shreds. A study like this, using vague wording, value judgments and dubious assumptions to conclude that white racism will determine the results of the presidential relationship, we get accepting headlines and rationalizations like, “those questions must be good ones, because they’ve proved racial attitudes before.

      • 1) I don’t think you can blame me for inferring that you didn’t actually know what the “implicit” measure used was, when your post unambiguously gets that fact completely wrong.

        2) “OK, you want the study to be true, so you can be sleep soundly convinced that those who disagree with you dream secretly of lynchings and and mass deportations.”

        That’s not at all what I believe, and nothing I’ve ever said justifies accusing me of thinking that Republicans secretly dream of lynchings.

        (On mass deportations, I’d argue that many Republicans have explicitly called for removing large masses of undocumented immigrants from the USA, and furthermore that Obama’s policies for most of the last three years could fairly be described as mass deportation of undocumented immigrants.)

        I do think most Republicans are racist, but I also think most Democrats are racist. That doesn’t mean that I think most of us want lynchings.

        3) “How can you assume that they correct the data to control for political ideology?”

        I didn’t assume it, but I also couldn’t rule it out, since it would be an obvious step for any competent researcher to take, and – unlike you – I’m not assuming that the researchers are either incompetent or dishonest. Having just done what I should have done in the first place – looked it up — I can offer you this, from a paper co-written by Krosnick:

        Recent research suggests that associations of symbolic racism with posited attitudinal consequences of it are not reduced notably by controlling for conservatism and an array of other potential confounds (e.g., Rabinowitz, Sears, Sidanius, and Krosnick in press). Nonetheless, any analyses done with measures of symbolic racism should be done controlling for political conservatism in order to minimize the likelihood of spurious associations being misdiagnosed as effects of racism.

        • “I do think most Republicans are racist, but I also think most Democrats are racist. That doesn’t mean that I think most of us want lynchings.”

          I assumed, as a cartoonist, that you were instinctively sensitive to the use of hyperbole. I was not seriously suggesting that you think all, most or any rightists dream of lynchings.

          • Okay, but now I’m just puzzled. You weren’t actually suggesting that I think most Republicans dream of lynchings, that was hyperbolic for… for what, exactly?

            I think my cartoons, hyperbolic as they are, actually clarify my opinion (at least when the cartoon is a good one) for any readers. Your hyperbole has just spread mud in the discussion, and now I have no idea what the heck it is that you’re objecting to in my views.

            The truth is, I think most conservatives are biased against Blacks, either in the sense of holding to racial stereotypes, or in the sense of being overly quick to sacrifice the well-being of Black Americans in pursuit of other goals, or both. I don’t see this as an all-consuming hatred that absolutely dictates all decisions, but rather as one bias mixed in with a myriad of other beliefs and goals.

            Is that an objectionable view?

            (I also think that most white Democrats are biased against Blacks, but in practice the party acts a little better both because a larger proportion of important Democrats are Black, and because Black voters are a crucial part of the Democratic party’s voting block.)

  7. Thanks for pointing this one out. I read the Washington Post article last week and just had to click through to see the questions. I had the same take as you, garbage.

    It’s garbage because the premise of the poll is to ask the respondents to explicitly stereotype people. Who knew, when you ask people to stereotype others they default into us vs. them mode, it’s classic priming.

  8. As a scientist who is a liberal (Rachel Maddow is an insightful commentator, in my view), putting aside the authors comments on liberals, I entirely support his premise that this survey on race (I read a the questions and wrote to gfk asking them to defend the scientific validity of their survey design) is weak, and maybe worthless, science. Science is the foundation upon which a prosperous society thrives, and this survey has been flagged by the author as suspicious and agenda-driven, as I see it. I’d welcome the authors attempt to demonstrate the validity of the survey’s objectivity and credibility. We’ll see if gfk provides same. Well done, Mr. Taranto

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

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