The data released as part of the Harvard litigation reveal the scope of the problem with respect to academic qualifications and ability of URM.
Harvard is a household name that attracts a hugely disproportionate share of the country’s ablest students. Despite this fact, there simply are not very many academically elite URM in the applicant pool. .
The data show that in the top decile of academic performance in the applicant pool (excluding legacies, development and athletes), in any given year there are approximately 1000 Asian applicants, 700 white, 50 Hispanic and 15 black.
No doubt adding in the legacy, development and athlete groups would add some to those numbers, but mostly in the white group (since they disproportionately benefit from those preferences). Unfortunately, I have not found an easy way to get at those numbers. Maybe at the trial phase more information will leak out?
That’s just the way it is. (See Table 5.1 in the plaintiff’s expert report for the clearest presentation of these data.)
I have a question for the thread. What is at stake with a verdict for the plaintiff? If Harvard appeals and a verdict for the plaintiff is eventually had at the Supreme Court, what does that mean for elite admissions, or would one expect a narrow ruling (One that affects only Harvard)?
If you are hoping for a resolution of this case at the Supreme Court I have some really bad news for you: Harvard College’s legacy admits this year include the daughter of Chief Justice John Roberts, a Harvard College alumnus. I saw him on the Visitas day and expect to see him again next month on freshman move in day. (Unless Josie Roberts gets Z-listed.)
If VMI (Virginia Military Institute) case where Clarence Thomas had to recuse himself because of his son’s attendance at VMI at the time is any guide, Roberts will likely have to recuse himself in the next four years from cases involving Harvard College. And you will have a 4-4 decision assuming Kavanaugh is confirmed and votes with the conservatives. Oh BTW, John Roberts has a son who is going to be a senior in two years. If Harvard admits him too SFFA and Blum can forget about a resolution from SCOTUS for the next seven years.
Indeed, these are academic superstars in the country. Sadly, most of them will be lopped off at some point in the process, even members of the world champion US Math Olympic team this year. I know two gold medalists including one with perfect score (only two in the world this year) were turned down.
^ Very useful information about the IMO gold winners turned down at Harvard. I had always heard vague rumors that it was possible but could never confirm it. I guess they’ll just have to content themselves with MIT
Not sure if the top decile are truly the academic superstars though. It’s just a mechanical application of scores and GPA and given the rampant inflation over the past two decades, I’d be suspect.
There are a very select few in the pool who are labeled as “1” in academics by the adcoms at Harvard. IIRC there are only about 100 of these kids total ( all races and statuses combined) in the entire applicant pool in any given year. Most of these are admitted, and I would have to imagine IMO gold gets you entree into this group. Sad that these two were not admitted, it takes tremendous natural ability and dedication to win a gold at the IMO.
Just as an aside, kudos to Team USA this year again. It should come as no surprise that there is tremendous natural mathematical talent in the USA. Our lousy education system has obscured this for a while and made identification of talent very much a hit or miss sort of thing. We should take nothing away from the accomplishments of the Team USA kids themselves, but the rise of AoPS and other online avenues for education deserves special recognition. Up until recently, talented kids had to listen to generally weak elementary and high school teachers telling them to slow down and make sure there were no “gaps” in their knowledge.
@SatchelSF Harvard’s approach is not without cost. Out of six IMO members this year, four are seniors, all applied to Harvard. Only two were offered admission. Unfortunately, the two who really wanted to go to Harvard were turned down. And now all four are going to MIT, which means that Harvard will be overshadowed by MIT in Putnam for years to come.
“The principal conflicts of interest that I would encounter arise from my service as solicitor general,” Kagan wrote on her Senate questionnaire, adding that the only others she was aware of would involve Harvard litigation."
Then it’s 4-3 against Harvard, and using race would be considered discriminatory in college admissions and colleges would have to stop using race.
If it’s 4-4 then it goes back to the lower court ruling and if Harvard appealed that, would mean they lost that ruling and would have to stop using race in admissions. Not sure what that would mean for other colleges though. If Harvard won that lower court case then they don’t have to change the way they evaluate applicants.
Being overshadowed by MIT in the Putnam has basically been the case for at least the last five years, for Harvard and every other school. They’ll survive no problem of course.
Getting back to the topic, there should be no issue with Harvard choosing its class as it see fit - if it works for them, and it has, more power to them - but they shouldn’t be able to discriminate on the basis of race under the guise of virtue. Private employers and landlords, for instance, cannot hide behind “holistic” dissembling. If SCOTUS takes this up, regardless of ruling, it should have implications for all private schools, as it would represent perhaps the first high profile examination of race preference in the private college world.
As a general legal matter, there is potential for extending the admittedly muddled jurisprudence into an entirely new area, thereby confusing more of the players. Perhaps the best we can hope for is that race preference dies away in the face of potential reputational and legal implications, though I’m not holding my breath…
Private employment and rental housing decisions are typically made holistically and opaquely. Of course, this means that lawsuits regarding illegal discrimination in such may be difficult to prove, and individual acts of illegal discrimination in such may not be obvious to either the applicant or any other observer.
Note the deciles rank by Academic Index, not “academic performance.” Academic Index is a function of academic stats, with 2/3 of AI score based on SAT/ACT test scores and 1/3 based on GPA. Harvard’s current 75th percentile ACT score is 35. To have stats in the top 10% of applicants, I’d expect that you need a 35+ ACT, combined with perfect GPA and perfect SAT II.
The academic 1-6 ranking is a more broad measure of academic performance, including additional factors, such as rigor of classes. Academic Index was notably correlated with academic rating, yet still explained only a minority of the variance in academic rating, suggesting that Harvard is looking at much more than just AI decile and other stats when evaluating academic performance.
Sure, but even using Harvard’s 1-6 Academic Rating, URM applicants are notably weak.
To be in the running at all, an applicant ordinarily needs a rating of 2 or higher (+ and - are available). Table B.5.3 shows that for the unhooked (other than race) pool, 60% of Asian candidates were scored by the Harvard adcoms higher than 3, 46% of whites, 18% of Hispanics and 9% of blacks. So that’s Harvard’s own proprietary if you will evaluation of the academic aptitude of its candidate pool.
Just eyeballing without a calculator, that implies in any given year approximately 5,000 academically “viable” white candidates, 4,000 Asians, 500 Hispanics and 225 blacks. Not a whole lot to work with there as regards URM.
I wouldn’t call the academic rating evaluation of “aptitude,” just as I wouldn’t focus on test scores in predicting academic rating. The rating includes both previous academic achievement and future potential. It considers things like LORs, as well as submitted academic work. The small portion of applicants who get the maximum score of 1 often have academic work reviewed by faculty, and getting a 1 is associated with a bigger boost in chance of admission than any hook evaluated in the lawsuit besides being a recruited athlete.
That said, yes URM applicants are underrepresented among those receiving high ratings in the academic category. Nevertheless, among admitted students, URM’s academic ratings have relatively small differences from white students in academic rating, as well as relatively small differences in academic index and other scores. I previously posted the stats below. All except SAT are for the unhooked baseline group.
White Admits
SAT CR: +0.72 SDs (0.43)
SAT Math: +0.55 SDs (0.52)
HS GPA: +0.50 SDs (0.52)
Mean SAT Score in 2017: ~745 per section
Mean 5.90 AP classes with average AP score of 4.73
Black Admits
SAT CR: +0.41 SDs (0.56)
SAT Math: +0.11 SDs (0.68)
HS GPA: +0.33 SDs (0.73)
Mean SAT Score in 2017: ~720 per section
Mean 5.08 AP classes with average AP score of 4.50
Ending up with zero IMO gold medalists is a notable admissions failure.
Suppose we replace the applications of the “nation’s four best math students” with the “nation’s four best football players” instead. Anyone doubt that all four would have been accepted, assuming they met minimum academic standards?
The bigger problem is that this is will be well publicized among math students at this level, and could reduce future applications to Harvard among these students. Remember that in the application process, students choose twice–where to apply and where to attend. Colleges only choose once–who to accept.
One of my kid’s fellow math classmates (at a relatively elite private high school) turned down Harvard this cycle to go to CalTech. She didn’t think the opportunities were as compelling at Harvard for a girl looking to do astrophysics and math.
While the differences between black and white admits at Harvard might seem small, as @Data10 listed above, it’s useful to look at them in context.
The z- score and sd information provided in the appendix of the Arcidiacono report imply that on the academic index calculated by Harvard, 87% of unhooked white admits exceed the average unhooked (other than race) black admit. 97% of Asian admits exceed the black average.
Viewed from a different perspective, because of the higher sd of the black admits, 20% of the black admits are at or above the average white admit, while 13% are at or above the average Asian admit.
Notably, these figures are after all holistic criteria have been applied (they are descriptive of the actual holistically admitted class). They are consistent with Harvard’s own internal OIR analysis that removal of race preference would reduce black admits by approximately 80% and are also roughly consistent with Bok and Bowen’s observation in Shape of the River a generation ago that without race preference, black representation would fall between 80 and 90%.
In my opinion those sorts of differences are too large. They are readily observable by students and faculty and lead to all sorts of undesirable consequences, not the least of which are cementing stereotypes and weakening standards.
This is the crux of the matter. It is not surprising that there are differences in admissions statistics of black applicants. That is the entire point of considering ethnicity in admissions. The question is whether this kind of affirmative action is beneficial to the recipiants and to society at large. We can all make make the arguments pro and con, mostly based on our own observed anecdotal experience. I would be interested in seeing some good research on this question.
The Black admit AI z-score was +0.33 (0.52). The White admit AI z-score was +0.76 (0.38). Across the full sample “unhooked” (only RD), it was +0.69 (0.46). This suggests that in a normal distribution, 24% of “unhooked” Black admits would have scored above the “unhooked” admit average, and 57% of White admits scored above average. As noted in the previous post, the specific score differences (tes scores make up bulk of AI) in the latest year were ~720 average for Black and ~745 for White. In the most recent math SAT, the difference between 720 and 745 was 1 question (2 wrong = 750, 3 wrong = 720). How important this score difference is, which can be just a single careless error per subject, is debatable.
What is clear is that ~97% of Black entering students are graduating. Hardly anyone is failing out. Surveys also suggest Black students’ average cumulative GPA is in the A- average range, not far from White students. Yes, self reporting may be imprecise and inflate, but again it suggests that Black students as a whole are academically capable, with a good GPA. Similarly Black graduates as a whole appear to have success after graduation and report starting salaries not far from White graduates.
@Data10
Your data are powerful! Yes, math SAT test score of 720 vs 745 might seem big, when in reality it is only the difference of answering one more question wrong.
Many kids have enough intelligence to succeed in Harvard, you really don’t need to be a genius or even be (the over-used adjective) “gifted” to make it in those tippy top schools. To most people, the magic (“to succeed” in life) is not intelligence itself, but to have the opportunity to apply the sufficient intelligence, of which a Harvard/other selective schools would provide amply.
Except, you can’t use a normal distribution for admitted students in a particular subject area. Too large a percentage of the admitted students have perfect scores.
