Main Error Mesage Here
More detailed message would go here to provide context for the user and how to proceed
Main Error Mesage Here
More detailed message would go here to provide context for the user and how to proceed

City Journal

Close Nav

The “Science” Behind Implicit Bias


The “Science” Behind Implicit Bias

October 18, 2017
The Social Order

Heather Mac Donald joins City Journal associate editor Seth Barron to discuss the dubious scientific and statistical bases of the trendy academic theory known as “implicit bias.” The implicit association test (IAT), first introduced in 1998, uses a computerized response-time test to measure an individual’s bias, particularly regarding race. 

Despite scientific challenges to the test’s validity, the implicit-bias idea has taken firm root in popular culture and in the media. Police forces and corporate HR departments are spending millions every year reeducating employees on how to recognize their presumptive hidden prejudices.

Heather discusses the problems with implicit bias, the impact that the concept is having on academia and in the corporate world, and the real reasons for racial disparities in educational achievement and income levels.

Heather Mac Donald is the Thomas W. Smith Fellow at the Manhattan Institute, a contributing editor of City Journal, and author of the New York Times bestseller The War on CopsHer article in the Autumn 2017 issue of City Journal is entitled, “Are We All Unconscious Racists?

Audio Transcript

Brian Anderson: Over the last two decades, the idea of implicit bias has made its way from the academic world into the mainstream of public discourse.  Implicit bias purports to answer the question, “Why do racial disparities persist when explicit racism has greatly diminished over the last half-century?”  The answer, according to proponents, lies deep within the human brain.  The academic creators of the implicit association test, first introduced in 1998, claimed that they could scientifically measure the level of an individual’s unconscious bias, especially against blacks.  From university hiring committees to human resources departments, the idea took off.  An entire consulting industry has spawned, advising organizations on how to combat this unconscious prejudice.  Every year government agencies and private firms pay millions of dollars to send workers to implicit bias training.  In recent years, however, social scientists have begun to dispute the scientific reliability and validity of the test.  And the original authors themselves have even walked back some of their earlier claims.  With the statistical basis behind the implicit bias theory beginning to crumble, can we expect its influence on popular culture to disappear?  Probably not without a battle.  In the latest issue of City Journal, longtime contributing editor Heather Mac Donald takes a deep dive into the literature surrounding the idea of implicit bias in an essay entitled “Are We All Unconscious Racists?”  Coming up on the podcast, associate editor of City Journal, Seth Barron, talks with Heather about her article and more.  Their conversation begins after this.

Hello, I am City Journal editor Brian Anderson.  Thanks for joining us for the 10 Blocks Podcast, featuring urban policy and cultural commentary with City Journal editors, contributors, and special guests.

Seth Barron: So, Heather, what is the principle behind the theory of implicit bias or unconscious racism?  How does that work?

Heather Mac Donald: Well, the principle is that even if we believe ourselves to be colorblind and firm opponents of any kind of racial discrimination, our minds are impregnated with so many messages of bigotry that we can’t control our thought processes and we can’t control our actions.  So, even if we think we are unbiased, we are, in fact, going to be making biased decisions against blacks.

Seth Barron: So, what’s the science behind, this though?  Because from what I understand, implicit bias theory is supposed to be grounded in science and have a lot of evidence supporting it.  What are the metrics or tests to determine the presence of implicit bias?

Heather Mac Donald: Well, there is a test that has gotten a heck of a lot of public attention thanks to Malcom Gladwell’s book, Blink.  The media routinely takes this test and penitently it cops to its own alleged racism.  It is called the implicit association test.  And it is modeled on something that is actually sort of common in social psychology and cognitive psychology to try and measure response times to concepts and it is a technique that traditionally has been used to test the closeness of concepts in memory.  But the developers of this specific implicit association test took this methodology into vastly radical political realms.  But I’ll describe it.  You – on a computer screen the test-taker is shown a series of black and white faces.  And you are supposed to categorize them into the white and black categories with two computer keys.  And then you are shown a series of words that are positive and negative, like happy, or death, and you are supposed to categorize them into good words and negative words with the same two keys that you have used to categorize the faces.  Then those sorting tasks are intermingled, and at one stage of the implicit association test you are supposed to sort a white face with the same key that you are going to sort a positive word and a black face with the same computer key that you sort a negative word, and then that’s reversed and you sort the black face with the key for a positive word, and a white face with a key for a negative word.

Seth Barron: Hmm.

Heather Mac Donald: It turns out that the majority of test-takers are faster by milliseconds in sorting white faces into the positive category than they are in sorting black faces into the positive category.  The developers of this implicit association test announced to the world as soon as it came out that it showed the ubiquity of prejudice and that the prejudice that it allegedly under – that it revealed had an active effect in the world and would result in behavior that was discriminatory.

Seth Barron: I see.  So, well, two things.  What is the connection between this implicit bias, these unconscious reflections we have that operate below the level of awareness and actual discrimination?  I mean, how does this manifest itself?

Heather Mac Donald: There is no connection.  That is the problem.  Since this test was announced, there has been a handful of courageous social psychologists that have pushed back against the methodology, and they have found that there’s not a single aspect of the test that is not vulnerable to rigorous methodological critique.  A: Any individual test-taker’s scores on the implicit bias test can vary wildly from one taking to the next, so the test fails what is known in the social psychology literature as the measure of reliability.  It is not reliable from one test.  But it also fails what is known in the literature as validity in that it does not predict what it purports to predict.  It turns out when they try to measure whether your score on the implicit association test relates to discriminatory behavior, A: What counts as discriminatory behavior is completely artificial and trivial.  It is how, whether you make eye contact or the placement of your chair in a mock interview in a college psych lab, or whether you decide to donate in a hypothetical charity experiment whether you donate to children in Columbia slums, versus – Columbian slums – versus South African slums.  That’s the extent of what they call discriminatory behavior.  In other words, this is not about a black candidate walking into an accounting firm and getting turned down because he is black.  It’s these artificial constructs.  But even if we accept that those artificial lab constructs count as discriminatory behavior we should care about, it turns out there is no relationship between your score on the implicit association test and these artificial discriminatory behaviors.  So, the IAT and its social and political significance is falling apart as we speak, and yet it continues to have enormous effect on the corporate world, on the policing world, on the foundation world, and on the educational world.

Seth Barron: What are some of those effects?  I mean, how is this theory being applied in, you know, HR departments, search committees?  I mean, what do they do with it?

Heather Mac Donald: Well, some corporations are starting to screen employees using the IAT to say well, we don’t want any biased employees here.  Hiring committees inevitably, and promotion committees in business inevitably have to go through their IAT self-examination and, you know, go to classes in implicit bias.  There is a huge consulting scam built around this thing.  Recently up to 200 CEOs signed a pledge that was spearheaded by PricewaterhouseCoopers to send all of their employees to implicit bias training.  They will all take the IAT.  This is enormous cost, enormous waste of resources.  The CEOs in this pledge have also pledged to encourage their employees to have more conversations about race on the job, which is the last thing anybody wants to talk about given how fraught it is and given the reality, which is contrary to discrimination, it is the reality of racial preferences.  And, so, race is understandably a charged topic.  But these businesses can’t afford to send their employees to these classes instead of doing work.  Some of the, for instance, Procter & Gamble is on the steering committee of this CEO pledge.  They’re currently facing a corporate takeover bid and have had lagging profits.  The idea that their employees are best spent – their time best spent in implicit bias training is ludicrous.

Seth Barron: Do the proponents of implicit bias training and the theory, I mean, since this all takes place below the level of awareness, what can the individual do, according to the theory, to fix themselves?

Heather Mac Donald: Well, that’s a superb question.  And even one of the developers of the test, Anthony Greenwald, was quoted earlier this year saying that implicit bias training is snake oil.  That hasn’t stopped the consultants from pulling in millions of dollars a year from gullible corporations.  What, at best, what they argue is that you should have procedures in place that will block the working of implicit bias.  You know, then they’ll say well, if you are aware of it that helps but, ironically, what some of the implicit bias proponents call for is hiring procedures that mask the race of the applicant to the extent possible.  I say bring it on.

Seth Barron: Uh-huh.

Heather Mac Donald: You know, they seem to be unaware that such a procedure would demolish any institution’s ability to engage in racial preferences, which is a practice that is ubiquitous in every elite institution, as well as university, in the country.

Seth Barron: Yeah.  In your article there was an interesting part where you talk about interviewing HR professionals and faculty search people and asking them well, how many qualified minority applicants were not hired because of implicit bias.  What kind of response did you get from them?

Heather Mac Donald: Complete evasion.  One would hope that a corporation that is worried about the bottom line would actually generate an evidentiary basis for some massive employee training and would know that there is – they can point to instances of qualified minority or female candidates who are not getting hired or promoted because of implicit bias ditto universities.  But I asked one of the HR managers at PricewaterhouseCoopers who is spearheading this ridiculous corporate CEO pledge for diversity, well, please give me some examples of, you know, are you aware of instances at PricewaterhouseCoopers or elsewhere where people have not gotten the job because of implicit bias.  She completely ducked it.  Now, she may be worried about a lawsuit, you know, if she says…

Seth Barron: Uh-huh.

Heather Mac Donald: …but the fact is, the reality is, it doesn’t happen.  I asked Anthony Greenwald, one of the two cocreators of the implicit association test, can you give me any example anywhere, not just at University of Washington, where he teaches, but any college, give me one example of a female or black, competitively qualified faculty candidate who didn’t get a job or promoted because of implicit bias?  He refused.  He changed the topic.

Seth Barron: Really?

Heather Mac Donald: So, there is no – there’s not examples of this happening.  In fact, the reality is just the opposite.  Anybody who has observed any faculty search knows it is one desperate effort to find competitively qualified, black or female candidates who have not been snapped up already by better endowed institutions.

Seth Barron: I see.  So, well, if it can’t be tackled at the level of the individual and, you know, from what you just said, it doesn’t sound as though it is actually impacting individual hiring decisions.  Do they, do the implicit bias advocates, do they have a broader social agenda?  What is the goal, if you will, of this theory?  What are they looking for?

Heather Mac Donald: Well, there’s a big movement in the legal academy piggybacking on implicit bias to remove the idea of voluntary intent from the law.  It already – that kind of already exists with disparate impact theory, so that you’ve got a body of caselaw and statutes that say to make a case of discrimination in hiring you don’t actually need to show that the employer intentionally discriminated.  You just need to show that any given hiring practice such as requiring say, a high school degree, or a certain standard on a verbal test, that that has a disparate impact on blacks.  But they want to go further than that and really remove completely any idea of intentionality on the theory that human beings are so vulnerable to this implicit bias that any conscious cognition is just irrelevant.  And this would mean that every single personnel decision could be challenged on the grounds that it was motivated by implicit bias and the only way to guarantee fairness in the workplace would be through a set of even more explicit racial quotas than already exist.

Seth Barron: It sounds like – I mean, I am reminded of that recent case at Google where the engineer was fired for having written a, basically a little essay questioning whether or not women were really being discriminated against in hiring at Google, or maybe there aren’t just that many women qualified to be software engineers at that level.  Where does this go if you are really, I mean, in your essay, in your article you said that only 1%, I believe, of Ph.D.s in one of the hard sciences in the last year were black, so how would you, I mean, if there’s really very few candidates, how would racial quotas even work?

Heather Mac Donald: Well, it’s very scary.  I mean, in the STEM fields the game is more about female preferences than race preferences because, as you say, Seth, there are so few blacks and Hispanics in the pipeline.  It’s – we are talking about 1% in computer science and computer engineering.  Nevertheless, the pressure would mean, to the extent it applies in the race field as well as gender, accepting very mediocre hires.  And, to a certain extent, this is simply about ideology for its own sake.  The other developer of the implicit association test who is a social psychologist at Harvard was involved in an email exchange with a journalist from New York Magazine earlier this year and she just exploded at him and it was clear that she sees herself on a social justice crusade and views anybody who questions the scientific methodology of the IAT, she said that they should take this up with their priest or their psychiatrist.  You know, there are some people who believe we should go back to a world where there’s coloreds only fountains.

Seth Barron: Uh-huh.

Heather Mac Donald: This is absurd.

Seth Barron: Right.

Heather Mac Donald: This is absurd, but she views any criticism as itself a symptom of bias.

Seth Barron: Ironically, the whole notion of implicit bias sort of sounds like, I mean, it sounds very theological, in a way.  This idea that there is this devil in you making you do things that you can’t even, you know, become aware of.  But let’s shift gears, because in the area of law enforcement, you know, I know that implicit bias has been discussed there, too, and surely it must play a role in the racial disparity in shootings of people by police, no?  Does that – is this not a – is not bias a major factor there?

Heather Mac Donald: Well, federal law enforcement agents were all sent off to implicit bias training by President Obama at enormous cost.  Local police departments are sending off their departments to implicit bias training at enormous cost.  The New York Police Department next year is going to put every officer and every recruit through a full-day seminar on implicit bias, all, again, without any evidentiary basis.  There is a psychologist at Cornell named Josh Correll who has been studying police shooting decisions for years.  They call it Shoot, Don’t Shoot Decisions.  And, he, clearly, if you read his writing, he’s not particularly pro-cop, but against his obvious will he has found that for police officers, they are not making errors.  They take a little bit longer to observe a situation involving an unarmed black that is a counter-stereotypical threat, because most of the shootings of police officers are disproportionately done by blacks.  In fact, a police officer is eighteen-and-a-half times more likely to be killed by a black male than an unarmed black male is to be killed by a police officer.  That’s because black males have made up 42% of all cop killers over the last decade even though they are only 6% of the population.  So, police officers are taking a little longer to assess threats and has shown that their neurological fear symptom is a little higher, but that is not resulting in what is known as bad shoots, that is, shooting unarmed civilians, black or white.

Seth Barron: Uh-huh.

Heather Mac Donald: And they are not, not shooting armed whites because they can’t imagine a white person being armed.  So, and Correll’s findings were reinforced and strengthened by another study that came out last year by a University of Washington researcher, Lois James, that found that officers were really, really hesitating when confronting black targets, even armed targets, because they are so second-guessing themselves due to the black lives matter narrative.

Seth Barron: So, in fact, according to you, bias is not really playing into shootings of people by police, that it is, more or less, what you would expect?

Heather Mac Donald: It is predicted by crime rates.

Seth Barron: So, where are we headed with all of this science?  I mean, there was a recent crisis in the replication crisis in the social sciences.  Is implicit bias going to have a kind of meltdown, or is the institutional support and momentum too great?  What is going to happen?

Heather Mac Donald: Well, it’s a very interesting situation because two researchers who have been working with the implicit bias original team of Banaji and Greenwald, one is the student of Banaji, who is a social psychologist at Harvard, he was the one responsible for putting the implicit association test on the web.  Anybody can take the IAT.  They can go right now to the website known as Project Implicit and they’ll get their IAT scores, take the test.  But these two associates, Brian Nosek and Calvin Lai, were participants in a recent meta-analysis that was the most damning yet about the IAT, that found no connection between IAT scores and behavior, so they, to their enormous credit, is going where the data is taking them.  Greenwald, one of the originators, dismissed their work.  So, within the crabbed area of purely statistical analysis, it is hard to know who is going to win.  The political realm is different.  There is just, the IAT is part of our country’s massive commitment to racial victimology.  There are institution after institution, whether it’s universities, or foundations, or the media, which is wedded to the idea that it is racism, not behavioral differences, that explain socioeconomic disparities.  So, it is going to take something really cataclysmic to pry from their white knuckles the implicit association test and the implicit bias idea.

Seth Barron: So, when we look at things like the achievement gap between the races in terms of test scores, college education, life outcomes, you are saying bias really doesn’t play into that.  Well, what are the other factors?

Heather Mac Donald: Behavior, and that’s the biggest taboo on college campuses and in the public discourse today.  You are not allowed to speak about the behavioral disparities that are what is really generating socioeconomic differences among the races.  They are things like blacks 73% national rate of out-of-wedlock childrearing, which gets probably as high as 85% in inner city areas, the academic skills gap is enormous on the math SAT, the difference between black and white scores is nearly one standard deviation, if you look at the tales of the math SAT distribution, Asians, and to a lesser extent, whites, are clustered at the top end of the math distribution.  You find whites and Asians, almost no blacks, at the bottom end of the distribution of math SAT scores, you find blacks and no, virtually no Asians and some whites.  Those skills are going to affect the availability of blacks for the hiring pipeline in anything that has to do with analytic skills or quantitative reasoning, and the same is true in verbal skills.  The verbal SAT shows the same huge score gap as the math SAT, the LSAT for law school admission shows an even larger skill gap of over one standard deviation between black and white scores, black law students end up overwhelmingly in the bottom quintile of their law school classes, even though law school exams are graded blind, so this cannot be chalked up to phantom police bias, and that affects bar passage rates.  Blacks, about – over a fifth of all blacks, law student’s graduates never pass the Bar exam after five attempts compared to maybe 5% of white, 2% of white students.  So, unless we address the inputs, whether it is educational skills, behavior, crime rates, we are going to continue seeing racial disparities in our world.  But we are prohibited from talking about those behavioral disparities.

Seth Barron: Heather, it has been a pleasure talking to you about this.  Very enlightening and interesting, so thank you.

Heather Mac Donald: My pleasure, Seth.  Thank you for having me on.

Brian Anderson: You can subscribe to this and other Manhattan Institute podcasts in the iTunes store.  The audio edition and transcript is available on our website,  This is City Journal editor Brian Anderson.  Thanks again for listening to the 10 Blocks Podcast.

Read More


Send a question or comment using the form below. This message may be routed through support staff.