‘Implicit Bias’ Is Real, But Racism Doesn’t Cause It

Black Americans broadly report they need to work harder and be better qualified than white Americans merely to receive equal consideration for jobs and other opportunities. Whites generally say they hold no racial animus toward blacks. These reports seem to contradict each other, but they aren’t. The fact that both can be true at the same time points both to barriers to, and possibilities for, racial progress in America.

The irony is that the racial inequalities in America today can sustain themselves without racial bias among whites. This is the implication of a large swath of the literature on “implicit bias.” The inapt phrase itself—implicit bias—masks the implication. The phrase misfocuses attention on what much of its own literature suggests is the wrong variable, and, hence, on the wrong remedy.

Lest I be misunderstood, let me state the obvious: racial economic inequality today certainly reflects the legacy of racism in America. The puzzle, however, is this: Why hasn’t racial inequality decreased along with the decrease in racial animus?

Many on both the left and the right deny there is a puzzle. On the left, white protestations that they do not harbor racial animus are discounted as reflections only of social desirability bias—that most whites will not admit what they in fact believe about blacks. On the right, legal prohibition of racial discrimination, and widespread affirmative action policies, result in discounting the continuing reports from black Americans of needing to work harder and be better qualified than whites to be considered for a given job or opportunity.

The seeming contradiction in beliefs stokes racial resentment in both groups: among whites because they’re accused of racism even when they harbor no racial animus, and among blacks because whites discount what they experience every day.

To be sure, while bias certainly is a sufficient condition for the patterns of racial inequality we see today, bias is not a necessary condition for the patterns of racial inequality we see today. Yet, ironically, continuing racial bias is not a necessary condition for blacks accurately to understand they generally need to better qualified than whites to compete successfully for a given job or opportunity.

How can that be?

Before turning to the explanation we might note the two important policy implications that follow from recognizing both experiences can be true at the same time. First, even if or when racial prejudice has truly disappeared, that fact alone would not necessarily eliminate racial economic disparities. That’s the challenge. The good news, however, is this: attention to individual merit would minimize the effects of “implicit bias” for individual job candidates. That’s the direct effect. But there would also be an indirect, more systemic effect by establishing for black Americans the same rewards for achievement that white Americans already enjoy.

Now let’s turn to the explanation.

There are a cluster of theoretical accounts that get grouped under the title of “implicit bias.” What I might call the “information-cost” theory of racial disparities, or perhaps the “search-cost” theory sits somewhat awkwardly under the title. Not least because it can account for the continuation of those disparities even without racial bias, whether implicit or not.

There are at least two components to this account. The first is that, in interacting with people, all individuals draw on what they know of the characteristics of groups to which those people belong. There is a time component to this, however. We draw on what we believe of group characteristics particularly when they do not know the other person very well or are just getting to know them.

And, of course, people have all sorts of identifiable features—not just race—that associate us with different with different groups. When others “see” us, particularly for the first time, they often will use what they know about common or “average” characteristics of these groups to form their initial, preliminary beliefs about us. In essence, when we first me other people, they are largely a blank canvas to us. So we use group averages to fill in the vast blank areas on the canvas as we (implicitly) consider how we will interact with them initially.

The critical two words here are “group averages.” A person looks at me, sees, say, my gray hair. From that, before they know me more particularly—before they have time to know me more particularly—they make certain initial assessments about what I am like based on what they know about the population of “gray-haired males.” These assessments are contingent, to be sure, but they are assessments nonetheless.

Mind you, all of these assessments are simply based on averages or perceived commonalities among gray-haired males.

To be sure, people see more than my gray hair—there are lots of “intersectionalities.” That I wear a wedding band. That I’m overweight. That I look haggard, or have laugh wrinkles around my eyes, and so forth. How I’m dressed. And, certainly also, that I’m white.

If the person gets to know me better, then many of the suppositions made based on population averages will get updated based on the information they learn about me as an individual. But, critically, most of the people whom I meet during the day will never know me better than from our brief interaction.

This process is almost always an “implicit” psychological process. But while it is typically an implicit process, it is not necessarily a “biased” process.

What the other people believe regarding the commonalities or averages of the different populations to which I belong can be entirely correct in their assessment of those averages. This provides a serviceable standard from which to define “bias”: A person is “biased” when that person’s belief about the average characteristic of a population diverges from the true average of that population.

But even when a person holds true beliefs about a population—when he or she holds “unbiased” beliefs about a population—individuals within those populations deviate from those averages all the time, sometimes significantly.

This deserves emphasis. People can have entirely accurate beliefs about averages or commonalities in populations to which others belong. But averages are just that, averages. It takes additional time and attention to update one’s assessment of a person’s individual characteristics and capabilities.

Let’s now bring the discussion home to race-based economic inequality.

Consider the large set of experimental results in which substantively identical résumés are submitted for consideration for a job. What differs is the use of “racially identifiable” names on the résumés. Some of the résumés have names commonly identified as “African American names,” others have names commonly identified as “white names.”

The consistent outcome is that callbacks for résumés with names identified with African Americans are significantly lower than callbacks for résumés with names identified with whites.

To be sure, this outcome is entirely consistent with the existence of racial bias. Yet the outcome is not necessarily the result of racial animus.

Consider a hiring manager who has 200 résumés to review before lunch. The amount of time the manager can spend on any one résumé at the preliminary review stage is maybe fifteen seconds. As a general matter, the manager is looking for candidates with good reading, writing, and arithmetic skills. The hiring manager also knows because of access to worse educational opportunities on average, African Americans on average have lower reading, writing, and arithmetic skills than white Americans. In the press to sort the résumés within the limited time allotted the manager makes quick judgments on the set of finalists.  

Because of time-pressured reliance on population averages, the manager selects a disproportionately high number of whites for the callback file and a disproportionately low number of African Americans. Not because of prejudice or bias, but as a result of population averages and time constraints.

Several notable implications follow from this. First, note there is a misalignment between the individual incentives the manager faces and the company’s goals. The company wants the hiring manager to hire the most qualified applicants for the jobs it has. But because of the time pressure the hiring manager faces to review the initial round of applications quickly, he or she often draws on population averages to fill in the blanks on applicants. These “blanks” may not in fact literally be blanks on the résumés. But the hiring manager doesn’t see the information in skimming over the résumés.

As a result of this context, however, the hiring manager actually excludes qualified African Americans from the callback list because of the informational shortcuts the hiring manager has to use to get the job done.

Mitigating the misaligned incentives facing the hiring manager—often simply a matter of allocating a little more time for a more in-depth review of the initial set of applications—means the company would now interview and hire more African American applicants because the review process now better identifies the most qualified applicants. Note that this problem of racial inequality is in fact mitigated by sharper focus on the individual merits of the applicants.

Now flip the focus to the black experience. The theory states that to get the attention of hiring managers African American applicants on average need to be better qualified than the corresponding white applicant. This is what African Americans report as their experience. To get the same attention, African Americans need to have better qualifications.

This, however, is only the direct effect. Consider the feedback loop created by the current experience. As a systematic matter, black Americans on average receive lower payoffs for the same level of education and experience as white Americans. Well, just do the math: Less profit, less investment. The feedback loop sustains race-based economic inequality. Simply hiring qualified African American applicants at the same rate as similarly-qualified whites are hired would be a big step to equalizing incentives for achievement across the races in America, and thereby equalizing achievement.

“Implicit bias” is an unfortunate title for this line of research. It focuses attention on something—the existence of “bias”—that is not necessary to account for the existence of racial economic disparities. As a result, it can point toward ostensible remedies that don’t in fact help to remedy the problem.

The irony is that the problem is actually more mundane than the continuing existence of racial animus, and the fixes aren’t really all that complicated. A good swath of the literature suggests that greater attention to individual qualifications—that is, providing enough time and the platforms actually to notice individual qualifications—would be a big step toward remedying the problem. And that’s something both whites and blacks, both liberals and conservatives, can get behind.

Reader Discussion

Law & Liberty welcomes civil and lively discussion of its articles. Abusive comments will not be tolerated. We reserve the right to delete comments - or ban users - without notification or explanation.

on April 11, 2019 at 13:25:54 pm

With the increasing use of Artificial Intelligence to review applicants I wonder, aloud, if maybe the school, applicant's name, and address should be hidden in the initial review of applicants. Could this enable a lowering of the artificial screening and focus more on the accomplishments of the applicant?

Look at the difficulty of Facebook and Google in developing systems to screen applicants. By using data from successful employees they screen applicants and got the same results. While this be no surprise the couldn't fix the problem. Maybe they are focusing too much on the wrong information.

Of course it will be surprise when the applicant arrives for an interview and it turns out to be someone who does not appear in the reviewer's mental profile but that could just begin to expand their mental database.

What do you think?

read full comment
Image of Arthur
on April 11, 2019 at 13:30:30 pm

"Whites generally say they hold no racial animus toward blacks."

Yes, as a result of PC and the innate act of most everyone to be kind to strangers or to those who are constantly ululating their misfortunate station in life.

But that's why the compassion is only cosmetic--that's why it's called implicit.

That make it appear as if progress is being made in race relations but the progress we see is due to the past decades of legal enforcements of overt behavior.

I would enjoy reading what the Professor has to say about what he sees as the meaning of race progress--progress to what? When all a "People of color" ? Is that it?

No, it is much more complicated than that

read full comment
Image of Martin Kessler
Martin Kessler
on April 11, 2019 at 14:00:41 pm

You'll just be displacing the bias charge onto some other stage(s) of the process--the interview, perhaps, and/or the "disproportionately" low number of black applications containing the "right" list of "accomplishments."

But this article seems anachronistic to me. In the world I inhabit, black applicants and hires are such a huge priority, eclipsing all others, that applications with indicia, overt or tacit, of black ethnicity are privileged. The problem is that the applicants must still possess certain minimum qualifications and too few black applicants do.

read full comment
Image of QET
on April 11, 2019 at 15:23:40 pm

“People of color”

I have always hated that phrase - or at least I believe it characterizes the WRONG people.

It may be that caucasians are ACTUALLY people of "color."
Do not caucasians have a wide variety of skin tones, do they not have a broader range of eye color, hair color, lip color?

I must be missing something here.

Perhaps, this *appropriation* in reverse?

What kind of *wizardry* is this?

read full comment
Image of gargamel rules smurfs
gargamel rules smurfs
on April 11, 2019 at 15:33:15 pm


How right you are or put another way and to paraphrase Ralph Cramden, "How [un]sweet it is" as seen below:


The efforts made in pursuit of this ever elusive and ever more "inclusive" goal is quite astounding. It would make one believe that the doctrine of "disparate impact" had not been proven to be a folly.

read full comment
Image of gabe
on April 12, 2019 at 12:30:47 pm

Your have two candidates, one white, one black. If things go south and employment must be terminated, which one can lay down the racist card??? As a manager, I have personally experienced this...

read full comment
Image of JN
on April 12, 2019 at 14:30:51 pm

And just why do accept that as a given? "racial economic inequality today certainly reflects the legacy of racism in America" Perhaps the dramatic increase in single mothers had an effect? perhaps high school drop rate (though the differential has decreases significantly since the early 2000s) has an effect. Perhaps the issue is college majors. Black college students are over-represented in service-oriented fields: humanities, education and social work . One of the majors common among African-Americans with a bachelor’s degree is early childhood education and the median earnings is only $38,000 annually compared to $65,000 for computer science . Look a little deeper than accepting consensus "wisdom."

read full comment
Image of -------
on April 13, 2019 at 09:16:06 am

Great article. Most of what the author outlined does ring true.

Nevertheless, I will say, over a 40 yr career of managing mostly professionals, I always went out of my way to identity qualified black candidates and to prefer them in hiring. I felt it was a social obligation to do my small part to help heal the racial divide and to redress historical wrongs. Among my management colleagues, I was not alone in thinking this. And, outreach was not limited to hiring decisions. I, along with many colleagues, focused time on these employees to coach them and to provide opportunities to succeed.

This experience is what makes the accusations of racism and, frankly, the perceived attitudes of President Obama’s during his term in office, so frustrating. We try to do our part for the benefit of the country but whatever we do is never enough. To researchers and the indoctrinated young, who typically have never done a thing to help anyone, let alone a black, we’re still racists. It’s seemingly in our DNA.

read full comment
Image of RJones
on April 13, 2019 at 09:40:42 am

James Rogers himself seems biased. Without citation, he claims " As a systematic matter, black Americans on average receive lower payoffs for the same level of education and experience as white Americans. " But statistics strongly suggest otherwise, that once you actually reach a relevant education level (college), holding all else equal (performance, grades, family structure, criminal background), black Americans out-earn white Americans.
The bias James is applying is that he is lumping that "average" together. The FBI Uniform Crime Statistics tell us that the "average" male black American is five times more likely to have murdered somebody, for example, a fact he glosses over and which clearly impacts earning potential. The U.S. Census tells us that black children are more than twice as likely to be living with a single mother than white children (54% vs 22%), which correlates more with success than ethnicity does... and which Mr. Rogers glosses over and ignores.
Certainly you could claim that these "inform" the implicit biases, but the trouble is, to do that, you first have to refuse to hold those variables constant in your analysis. Because, with those all equal (i.e. a college-educated black male from an intact nuclear family with college educated parents and no crimes in the family, compared to an identical-but-white male), the gap is reversed.

read full comment
Image of Mack
on April 13, 2019 at 14:01:20 pm

If we dwell on the example of a hiring manager screening 200 resumes, we can look much deeper than trying to invent a innocent-sounding-but-still-racist explanation, and find a cause which is actually implicit and doesn't reflexively try to play a race card.

People are most confident doing what they know, so a hiring manager will be most confident about candidates with resume items the reviewer can relate to. If the manager has had a traditional career after graduating from a mainstream college, they will be most comfortable with applicants who have followed a similarly standard path. An applicant with non-traditional work experience after earning and unusual degree or attending a less-known school will be at a disadvantage.

If the manager is cutting 200 resumes to a shortlist of 10-20, and will interview five, the first screening is going to be brutally biased towards producing solid candidates in a specific mold. The middle-class applicant who went to a name-brand college is going to have a strong advantage out of the gate in almost every situation, and that group is historically much whiter than the nation as a whole.

read full comment
Image of James B.
James B.
on April 14, 2019 at 05:12:44 am

In 2015 the average income of an African-American household was 3/5 that of that of a white one. This 3/5 ratio has been the case in America since the original 3/5 compromise of April 1783. This should mean nothing as far as individual situations are concerned, but a 15 second scan of a resume doesn't involve much individual consideration. Group reputation counts for way too much, and the reputation of African-Americans as employees is not as good as for whites. Too bad. How sad. Get over it.

read full comment
Image of Loran Tritter
Loran Tritter

Law & Liberty welcomes civil and lively discussion of its articles. Abusive comments will not be tolerated. We reserve the right to delete comments - or ban users - without notification or explanation.