Restrictions on Asian American admissions into top STEM programs—ironically, imposed in the name of DEI—are receding. Thanks to the Supreme Court’s 2023 decision banning race-based admissions and the Trump administration’s commitment to enforcement, several universities have abandoned former 20-25% caps. Their freshmen admissions are trending 40%-50% Asian American, broadly in line with data on top performance in AP math. This is a welcome step toward reducing universities’ obsession with race and renewing attention to merit. However, since racial preferences for Blacks and Hispanics have not been rolled back, the brunt of adjustment has been borne so far by Whites. White shares that averaged 40% to 50% in 2014 shrank to 20%-30% in 2025—approaching total Black/Hispanic shares, despite an average White edge over the latter groups that exceeds the average Asian edge over Whites.
My investigation was triggered by a stupendously ridiculous headline: “Asian enrollment at Johns Hopkins is skyrocketing. No one can say why”. No siree. None of their spokesmen can, not without confessing to prior discrimination, which would shock as much as gambling in Casablanca. That got me wondering about trends in leading US universities. Most of them maintain a “Common Data Set” with statistical breakdowns for freshmen and undergrads as a whole, which LLMs (I used Grok) can easily explore. Examining a dozen or so—here is Harvard’s—revealed some striking features. I will elaborate on them here, with two caveats. First, more thorough researchers will doubtless amend my conclusions, hopefully with a crisper definition of “leading US universities” than I can offer. Second, the data on ethnicity/race is inherently flawed.
The data is flawed because no one is sure what it measures. The categorizations trace back to crude views of skin color, mix different races, get gamed by self-reporting, and are blurred by intermarriage. From a genetic perspective, most Americans qualify as “two or more-races”, but only 3% of adults and 5% of schoolchildren are recorded as such. The “ethnicity” qualification lumps together a host of different cultures, particularly in Asia. A fairer description is geographic-origin of ancestors, but that must be qualified too or we would all be African. However, the standard shorthand description is “race”, with White, Black, Hispanic and Asian as the four main groupings. That is the data I will rely on.
Back to the main story. DEI claims about persistent white supremacist education in the US have long been undercut by evidence that some minority groups significantly outperform Whites academically. The first group was Jews, but DEI advocates wriggled out of that by redefining Jews as White. Asians provide a tougher challenge, since they are officially classified as non-White and historically faced far more legal discrimination in the US than did Hispanics. Asians comprised 5% of US school students in the 2010s but 40-45% of top scorers on AP Calculus BC. Blacks and Hispanics, eight times as numerous as Asians, made up less than 9% of top scorers. Success in increasing the shares of Blacks and Hispanics taking AP Calculus did not significantly change this; in 2020 the College Board stopped reporting those breakdowns to reduce embarrassment. Here are more details on comparative math performance.
Differentials on verbal/reading tests are slightly milder but move in the same direction. To be clear, no average differential justifies ignoring the heterogeneity within these four broad groupings or the even greater variance across individuals. We should treasure excellence wherever we can find it. But the statistics refute beyond all reasonable doubt the DEI myth that every kind of talent or passion is equally distributed across every large pseudo-racial or gender grouping.
Anyone who watches football or basketball recognizes that top players are disproportionately Black and that Asian superstars are rare. Insisting on ethnic quotas for wide receivers or three-point shooters would stoke more racial derision and do nothing to improve team performance. Everyone realizes that. Why then should anyone begrudge differentials in academic performance that turn the other way?
Asian Shares Rising
Nevertheless, most universities came to apply a DEI rule of thumb that Asian undergrads should not significantly outnumber Black and Hispanic undergrads. In 2013, when Students for Fair Admissions (SFFA) first sued Harvard and the University of North Carolina (UNC), Asian shares of undergrads were 21% at Harvard and 9% at UNC, while Black and Hispanic shares were 19% at Harvard and 15% at UNC. The Asian share at Harvard had been relatively stable for 20 years despite mounting evidence of Asian outperformance.
Harvard and UNC consistently denied discrimination, invoking vague “holistic” criteria that coincidentally turned against Asians. In another remarkable coincidence, their Asian admissions started rising after the suit was filed. By 2022 when the Supreme Court heard the lawsuit, Asian American shares of freshmen were 28% at Harvard and 25% at UNC. While the Asian share at UNC has not changed much since the Supreme Court ruling, it has since surged to 41% of freshman at Harvard.
In still more coincidence, Asian Americans comprised 18% of Johns Hopkins undergrads in 2014, 27% in 2022 and 49% of freshmen now. To repeat, “Asian enrollment at Johns Hopkins is skyrocketing. No one can say why”. Asians now comprise 40%-50% of freshmen at MIT, Carnegie Mellon, and Rice. Coincidentally, that broadly matches their shares at more merit-based Cal Tech. Asian shares of American undergrads are lower at other top universities but usually exceed 25%.
White Shares Falling
On tests of STEM competence, Whites average roughly 0.5 standard deviations lower than Asians but roughly 0.8 standard deviations higher than Hispanics and 1.0 standard deviations higher than Blacks. Since two randomly chosen individuals tend to differ by 1.4 standard deviations, this should warn against judging individuals by group averages. However, the differentials imply significant group disparities at extremes. For example, in football and basketball, where the differentials are roughly reversed, Blacks are substantially more overrepresented on college teams than high school teams and even more overrepresented on pro teams. Conversely, Asian American get substantially more underrepresented on better teams.
Hence, if judged on merit alone, removing the ceilings on Asian admissions should have come disproportionately at the expense of Black and Hispanic admissions. Yet most top universities (Rice and MIT were exceptions) admitted significantly higher shares of Blacks and Hispanics in 2025 than in 2014. White admissions bore the brunt of adjustment. White shares that averaged 40% to 50% in 2014 shrank to 20%-30% in 2025.
From a DEI perspective, Whites deserved to bear the brunt since they still outnumber Blacks and Hispanics combined, despite near-parity in college-age White versus Black/Hispanic populations. However, as our sports examples showed, that’s not a relevant metric for gauging top performers. In a previous essay, I combined statistics on relative math competence and population sizes to estimate that Blacks and Hispanics comprise 8% of the top 1% in US math competence and 4% of the top 0.1%. When we take into account the extra heterogeneity within Asian and White groups—northeast Asians considerably outperform other Asians, while Ashkenazi Jews outperform both other Whites and likely northeast Asians too—the estimated Black and Hispanic shares decline further.
Granted, math competence might be too narrow a criterion to warrant judgment on fair shares. Let’s redo calculations assuming only that a margin of 0.5 standard deviations in average performance warrants overrepresentation by a factor OVR. That is, for two groups that differ by that 0.5 margin, their fair-share ratio in universities should equal OVR times their ratio in the generally population. What does the evidence suggest for the Asian-to-White OVR?
If White shares of undergrads at leading universities had stayed at 2014 levels, they would roughly match Asian shares now. In that case, OVR would roughly match the 7-to-1 ratio of White to Asian college-age populations. Since the average White margin over Blacks and Hispanics exceeds 0.5, the corresponding OVR should exceed 7. Near parity in populations then implies a fair Black and Hispanic share of at most 6%–less than a third of current levels.
The estimates get even starker if we assume that the 40%+ reduction in White shares since 2013 is justified. Then the OVR is 11, implying a fair Black and Hispanic share of less than 3%. Civil rights lawyers take note: There is no logical way to defend both 40%+ Asian shares and 20%+ Black and Hispanic shares on grounds of merit.
DEI Double Standards
How about grounds of merit for Asian shares and some other grounds for Black and Hispanic shares? Those “other grounds” are inevitably shaky, since the juxtaposition of two different standards begs the question of why. Asians historically faced more vicious discrimination in the US than Hispanics. Blacks excelled in sports once Jim Crow restrictions were lifted. Top universities pride themselves on their commitment to fair treatment. Why do they need blatant double standards for different minority groups?
One answer is “social justice for Blacks and Hispanics.” While the sentiment sounds noble and often is, it is not something elite universities can deliver. They educate a privileged few for careers that will likely ensconce them in privileged communities with privileged rewards. Hopefully graduates prove worthy of their privileges through their know-how, competence, and leadership. But it is racist to confine them to serving their ethnic community or to rule out people from other backgrounds serving their ethnic community better.
A better answer is “cultivating talented Black and Hispanics,” both for their own sake and for inspiration to others. That aim is best served—especially given the historical context--by churning out graduates just as qualified as White and Asian graduates. The numbers matter less than the demonstration of excellence that transcends and refutes kneejerk categorization. Lower standards for Black and Hispanic applicants work against that. Most entrants cluster near the bottom of highly competitive pools. Some work hard to catch up but the lag is stressful and tends to leave many gaps; they would likely benefit more from a year of intense preparation somewhere else. Others are so disconcerted that they lower their own aims and transfer to softer majors.
The only legitimate defense of lower admission standards is offset for objective constraints on preparation. But these should be documented for each individual with clear explanation, and the degree of offset adjusted in light of subsequent statistical results. Is there any empirical evidence that Black and Hispanic entrants strongly outperform what their application test scores predict? If so, it would be glorious news highlighted in every DEI presentation. Instead we hear nothing, apart from quieter studies that reaffirm the predictive power of test scores and find no evidence of prejudice.
The DEI-inspired double standards have severe consequences:
• They stir grievance, obsession with pseudo-racial identity, and tension across groupings—the very opposite of what universities ostensibly intend.
• They tend to lower standards throughout. Princeton, Stanford, Columbia, Cornell, Duke, Vanderbilt and the University of California now offer pre-calculus courses for credit, while Harvard and Yale fold pre-calculus review into one of their calculus courses. It’s a waste of top-notch faculty to use them as high school teachers.
• They aggravate Bayesian bias, whereby ignorant or undiscerning observers impute group averages to exceptional individuals. To say this baldly, they make race a more useful signal of graduates’ likely quality, when we would prefer it to be useless. This makes it harder for genuinely qualified Black and Hispanic scholars to get the respect they deserve.
• They set up underprepared students for failure. Discouraged, they drop out at disproportionately high rates, when they might have flourished given more ramp-up time and less immediate pressure.
Still, I am cautiously optimistic about the shift in DEI. It is a big step toward restoring high-end merit, which is more important to R&D and the economy than what happens in lower ranks—especially with AI squeezing the latter. It encourages qualified-but-denied White students to head South for college, raising the standards and improving the human capital of what had been a lagging region. And it makes the anti-White biases more blatant, exposing them to further litigation and ridicule.
Important essay, indicating that admissions are still doing DEI. "Civil rights lawyers take note"!
This is indeed an important essay because it demonstrates the continuing deviation from merit designed to protect and promote particular demographic groups. Ironically, the "release" of Asians from the suppression of their enrollments while maintaining separate and unequal standards for Blacks and Hispanics has come not just at the expense of Whites generally...but specifically of Jews whose numbers in elite colleges are plummeting. Since Jews were reclassified as White in the eyes of the woke bean counters they are now starting to suffer the consequences of anti-White racism. We see complaints from the Jewish community that this is unfairly reducing the number of Jews of campus back to the levels when elite colleges had defacto quotas on Jewish students like they just dropped for Asians. Here's where the problem lies however. Jewish students still remain the most over-represented demographic group in elite colleges...so whining about how it is "unfair" for Jewish students to suffer from anti-White racism and referring to it as a form of anti-semitism (which is admittedly rampant in these colleges) just doesn't pass the smell test. Trying to make affirmative action/DEI style arguments to defend Jewish over-representation on the basis of "merit" just because it has now been allowed for Asian students isn't going to fly with the non-Jewish white majority in the US that always seems to be on the losing end of these anti-merit interventions. It is not only stirring contempt for the idea that merit plays any role in academic institutional decisions...but it is destroying sympathy for all minority groups from the justly aggrieved white population. The white majority can honestly question why "fairness" only applies to members of "protected" groups but not all citizens as the laws were intended. Asians and Jews, as the over-represented groups, can't have it both ways and will pay the price in public opinion if the double standard of "merit" AND "protected" status is used simultaneously for their benefit. What is the solution? End all protected status practices and impose strict merit in all such decision regardless of the impact on enrollment and "representation".