Posted on December 9, 2020

The Bias Fallacy

Heather Mac Donald, City Journal, Autumn 2020

The United States is being torn apart by an idea: that racism defines America. The death of George Floyd under the knee of a Minneapolis police officer in late May 2020 catapulted this claim into national prominence; riots and the desecration of national symbols followed. Now, activists and their media allies are marshaling a more sweeping set of facts to prove the dominance of white supremacy: the absence of a proportional representation of blacks in a range of organizations. That insufficient diversity results from racial bias, claim the activists, and every few days, the press serves up another exposé of this industry or that company’s too-white workforce to drive home the point.


These current diversity exposés are distinguished from their predecessors not just by their frequency. Virtually all mainstream institutions now agree that a nonracially proportionate workforce proves racism, including their own. Organizations preemptively accuse themselves of discrimination without waiting for the inevitable indictment. The publisher of the Science family of journals denounced his own publications for “the discrimination, subjugation, and silencing of minority colleagues and voices.” The American Mathematical Society pledged to “address systemic inequities that exist in our mathematics community.” The dean of the Harvard Business School apologized that “we have not fought racism as effectively as we could have and have not served our Black community members better.”

The remedy proposed for this alleged systemic bias is an even greater emphasis on race as a job qualification. Increasing the role of skin color in employment decisions, however, will compromise the caliber of American institutions, if bias does not, in fact, explain workforce disparities. Moreover, the insistence on America’s racism undermines the legitimacy of our polity. It is worth examining in some detail, therefore, the charge of systemic bias and the counterevidence against it—especially since that counterevidence is excluded from public discourse.


Given management’s prostration before every charge of bias, it is no surprise that diversity demands are increasing in stridency and scale. The Los Angeles Times must hire 18 black journalists, regardless of need or revenues, and issue a public apology for racism. Public radio station WNYC must hire two black reporters and two black producers in 100 days, to make amends for hiring a white woman as editor-in-chief. {snip}

Companies’ self-initiated diversity goals are no less sweeping. Half of Facebook’s workforce will come from “underrepresented communities” (i.e., black and Hispanic) by 2023, according to Chief Operating Officer Sheryl Sandberg—a remarkably ambitious pledge in light of Facebook’s current 3.8 percent black workforce.


But the expectation of proportional representation in every profession is groundless, thanks to the academic skills gap. The unequal distribution of skills, not bias, explains the lack of racial proportionality in employment.

The median black eighth-grader does not possess even basic math skills. “Basic” skills, as defined by the National Assessment of Education Progress exam, means partial mastery of grade-related knowledge. Fifty-three percent of black eighth-graders scored “below basic” on math in 2017. Only 11 percent of black eighth-graders were proficient in math, and 2 percent were advanced. By contrast, 20 percent of white eighth-graders were below basic in 2017, 31 percent were proficient, and 13 percent were advanced. Only 12 percent of Asian eighth-graders were below basic, 32 percent were proficient, and 32 percent were advanced.

The picture was not much better in reading. Forty percent of black eighth-graders were below basic in reading in 2017, 17 percent were proficient readers, and 1 percent were advanced readers. Sixteen percent of white eighth-graders were below basic in reading, 39 percent of white eighth-graders were proficient readers, and 6 percent were advanced readers. Thirteen percent of Asian eighth-graders were below basic, 45 percent were proficient, and 12 percent were advanced readers.

Black students never catch up to their white and Asian peers. There aren’t many white-collar professions where possessing partial mastery of basic reading and math will qualify one for employment. The SAT measures a more selective group of students than the NAEP, but even within that smaller pool of college-intending high school students, the gaps remain wide. On the math SAT, the average score of blacks in 2015 was 428 (on an 800-point scale); for whites, it was 534, and for Asians it was 598—a difference of nearly a standard deviation between blacks and whites, and well over a standard deviation between blacks and Asians. The tails of the distribution were even more imbalanced, according to the Brookings Institution. Blacks made up 2 percent of all test takers with a math SAT between 750 and 800. Sixty percent of those high scorers were Asian, and 33 percent were white. Blacks were 35 percent of all test takers with scores between 300 and 350. Whites were 21 percent of such low scorers, and Asians 6 percent.

In 2005, the Journal of Blacks in Education estimated that there were only 244 black students in the U.S. with a math SAT above 750. Brookings used an estimation procedure that maximized the number of high-scoring black students and came up with, at most, 1,000 blacks nationwide with scores of 750 and above. Whether the number is 250 or 1,000, it means that the STEM fields, medical research, and the ever-more mathematical world of finance cannot all have a 13 percent black participation rate, at least if meritocratic standards remain in place.

The SAT gap is replicated in graduate-level standardized tests. Between 2014 and 2017, the average score on the quantitative section of the Graduate Record Exams (GRE) was 150.05 out of 170. The Asian average was 154.1; the white average, 151; and the black average, 144. MIT’s entering engineering class in fall 2017 had an average GRE quantitative score of 167; students in the University of California, Berkeley, civil and environmental engineering program averaged 160, as did graduate students in USC’s engineering program. Even if the curve for blacks on the quantitative GRE is normally distributed in a bell curve, unlike for the math SATs, there will still be fewer blacks with higher-end scores than whites and Asians, given that the average black quantitative score is so much lower.

The organizers of the various STEM antiracism protests, such as #ShutDownSTEM, #ShutDownAcademia, and #BlackInTheIvory, argue that bias drives the lack of black representation in quantitative STEM fields. {snip}

But there are simply not enough black STEM Ph.D.s to go around. In 2017, blacks made up 1.2 percent of all doctorates awarded in physics to U.S. citizens and permanent residents, according to the annual Survey of Earned Doctorates from the National Center for Science and Engineering Statistics. Blacks earned 0.9 percent of all mathematics and statistics doctorates, 1 percent of all doctorates in computer science, 2 percent of all doctorates in chemistry, and 1.7 percent of all doctorates awarded in engineering disciplines. There were no black Ph.D. graduates in medical physics, atmospheric physics, chemical and physical oceanography, plasma/high-temperature physics, logic, number theory, robotics, or structural engineering. How academic STEM departments and Silicon Valley tech firms are going to fulfill their diversity pledges in light of that dearth of supply is a mystery. Yet in July 2020, MIT’s president blamed his own institution for not making headway on “racial equity and inclusion,” despite years of quota-izing effort. Virtually every other college leader has issued the same self-indictment.

The Law School Admission Test is similarly skewed. The gap between white and black scores on the LSAT in 2013 was a 1.06 standard deviation. In 2004, only 29 blacks, representing 0.3 percent of all black LSAT takers, scored 170 or above on the LSAT, according to the Journal of Blacks in Higher Education. The average entrance score was 170 for the top-ranked law schools. There were 1,900 whites who scored at least a 170, representing 3.1 percent of all white test takers. Of black test takers, 1 percent—or 108 blacks nationwide—scored at least 165 in 2004, 165 being the average for the top ten law schools. Over 10 percent of white test takers—or 6,689 whites nationwide—scored at least 165. That gap has only grown, and it affects law school outcomes. Of black law school graduates, 22 percent never pass the bar exam after five tries, compared with 3 percent of white test takers, according to a study by the Law School Admissions Council. The skills measured by the LSAT and the bar exam—verbal reasoning, command of English, logical thinking—are required for a range of professions, whether corporate management and consulting, banking, or journalism. As for the law itself, if corporate law firms do not have a proportional number of black partners, it is not for lack of trying.

The mean black score on the business school admissions test (GMAT) in 2017 was 453, on an 800-point scale. The average for whites was 565, for Asians 586, and the overall average was 549. Only 15 percent of black test takers scored 600 or higher in 2017, compared with 45 percent of whites and Asians. Sixty-six percent of black GMAT takers scored less than 500. In 2018, students entering the top ten business schools, as ranked by The Economist, had average GMATs ranging from 690 to 732. The Journal of Blacks in Education estimated in 2006 that only 1 percent to 2 percent of black GMAT takers scored above 700, a proportion that has not likely changed. A former Harvard Business School professor told the Wall Street Journal in June that he had quit to protest the school’s systemic “antiblack practices.” An expectation of high-level quantitative skills was likely among those antiblack practices. Far from discriminating against blacks, the Harvard Business School admits at least twice as many black applicants as their average GMAT scores would predict. Nevertheless, Dean Nitin Nohria claims that the school’s efforts at racial justice have been “painfully insufficient.” California has now mandated minority representation on the boards of corporations headquartered in the state, regardless of whether the quota hires possess business expertise; other states will follow suit.

Medical schools have instituted large racial preferences because blacks’ Medical College Admission Test (MCAT) scores would otherwise make a “critical mass” of black students unattainable. As with the math SAT curve, black MCAT scores cluster at the bottom-left tail of the distribution. With each higher tranche of scores, the black percentage drops to low single digits. Between 2013 and 2015, about 53 percent of black MCAT takers scored 23 or lower, corresponding to 496 and below on the new MCAT scale. A 23 (or 496) is the 34th percentile and would be disqualifying for whites and Asians. The Princeton Review recommends a score at least in the 80th percentile for medical school applicants.

Blacks in the next, still very low, tranche of MCAT scores and with low GPAs are admitted to medical school at nearly ten times the rate of similarly situated Asians and at seven times the rate of similarly situated whites. Over 56 percent of blacks with low MCATs and low GPAs are admitted to medical school, even though such scores predict failure on the U.S. Medical Licensing Examination Step 1. Less than 6 percent of Asians and 8 percent of whites are admitted under similar circumstances. Race plays so large a role in medical school admissions that a free online calculator of acceptance likelihood uses only three variables: GPA, MCAT score, and race.


The myth of bias, whether in medicine, technology, or finance, can be maintained only by ignoring the skills gap. There simply are not enough competitively qualified black candidates to go around. {snip}

As long as data on the skills and behavior gap remain available, it is possible to challenge the myth of bias, at least in theory. So those facts must themselves be canceled, as well as anyone who publicizes them. That is the ultimate motivation for the movement to end the use of standardized tests in admissions. All minimally selective institutions of higher education practice race-norming in the evaluation of test results already, holding blacks and Hispanics to a lower standard than whites and Asians. The tests do not impede racial preferences; colleges simply ignore the results for the sake of diversity.

The reason to eliminate standardized assessments is rather to put the College Board and the Educational Testing Service out of business entirely—and with them, any possibility of an objective measure of intellectual skills. The College Board was dealt a body blow this May when the University of California announced that it was ending the use of the SAT and the ACT in admissions for state residents. The SAT was a “racist test,” said the Vice Chair of the UC Regents, Cecilia Estolano. More than 1,000 colleges have already made the SAT optional or eliminated it entirely, but none has the financial clout of the University of California, the largest consumer of College Board products.

Graduate programs have been dropping the GREs because of their disparate impact. Half of all physics and astronomy departments, and nearly as many molecular biology and neuroscience programs, no longer require them. Cornell’s department of biomedical engineering, known for its diversity advocacy, and the University of Michigan’s biomedical sciences program have eliminated them. At the University of California, San Diego, the computer science and engineering, bioengineering, and nanoengineering programs are eliminating or waiving the GREs. UC San Diego’s graduate division dean recommended in July 2020 that the rest of the school’s graduate programs follow suit. Harvard and Caltech’s physics departments suspended the general GRE and the physics subject test in 2020, citing coronavirus concerns; they will likely not reinstate them.

The MCATs are in the crosshairs because of their disparate impact, even though they predict performance on the medical licensing exam. Medical school grades are also a racist practice. This spring, Columbia University medical students demanded that their professors stop grading them. Protest organizers warned that any medical student who refrained from signing the anti-grading petition would be publicized as someone content to “sit in their own privilege” at the “expense of their black and brown peers.”

At least one medical residency program is ignoring grades already. The University of Pennsylvania postgraduate surgery program disregards medical students’ grades, their scores on the medical licensing exam, and their membership or nonmembership in the medical school honors society. Instead, the surgery program focuses on “leadership, teamwork, altruism, and research activity” in admitting graduates to the surgery residency. Miraculously, those criteria favor blacks.


America’s elites have apparently decided that if, after five decades of massive financial outlays and obsessive attention from policymakers, the academic skills gap has not closed, it is time to break up the objective yardsticks that measure it. The same will happen regarding crime data. At present, data that show the vastly disproportionate rate of black criminal offending are still available from a few local police departments. Expect to find such facts harder to obtain from government agencies until they are disappeared entirely.