“Asian – 20% Harvard, 5.4% US students”
With such an over-representation of Asians, how can there be discrimination? This seems to go against how discrimination is determined. The comparison to the NBA doesn’t seem to hold water. Don’t seem to be too many Asians attempting to become NBA players.
The Inside Higher Ed article on the earlier page suggests that roughly 1/4 of US applicants to Harvard were Asian. Asian students were tremendously over-represented compared to the overall US student population among both applicants and matriculating students. The lawsuit claims that nearly 1/2 of admitted students would be Asian, if only Harvard’s 1-6 academic rating was considered in admissions decision without holistic factors, legacy, athletes, demographics, etc.
Over/under representation of one group in a broad pool (such as the general population or even the population of students) when dealing with a more selective subgroup (the highest achieving students) does not indicate discrimination or the lack thereof. We have to see if among equally qualified students (academics and certain subjective “holistic” standards) race has become a factor greater than what is permissible under Fisher II (assuming Fisher II itself is not overturned). Based on pure objective academic criteria, and even EC ratings, the scales at this glance are tipped against Asian applicants based on the data coming out of the litigation. A factor that would favor Harvard would be that Asians disproportionately indicate and/or have certain areas of interests or qualifications so that within those sub-buckets which themselves have limited spaces or perhaps even higher median standards, the Asian rejects skew the overall Asian admissions rate. If on the other hand, admissions rates for Asians across a broad swath of interests are suspiciously similar so that it indicates race being an almost overriding factor, indeed a quota, H could be in trouble even under Fisher II.
https://www.wsj.com/articles/harvard-is-too-discriminating-1529363694 While the author clearly feels Harvard’s admissions process is impermissibly discriminatory, it is a good legal breakdown of the requirements of Fisher II and how Harvard stacks up in his view.
I bet that if NBA announces tomorrow that there would be a cap of 60% for African American players in the league or for each team I think there would be a rush of interest in becoming an NBA player from all other racial groups, including Asians if they were guaranteed a percentage too. NBA could even claim it served their institutional priority—driving up TV rating in all racial groups and around the world.
But such an initiative would garner hardly any public support including from Asians. And most importantly NBA would be doing something illegal, subject to many lawsuits…
I know someone (female white college student) who just interviewed for a summer internship with a big bank who made it to the last phase, but did not get the offer. It was a diversity event and females are considered diversity, as are Asians (male and female).
Anyway, she knows of Asian girls that received the offer. She also spoke with a recruiter to told her that Asians are considered more of a URM for banking and have an advantage in the banking internship selection process over Whites.
I thought that was interesting, given all of this debate.
Asians have always been thought to have a bit of an edge in law school admission, and perhaps surprisingly have very slightly lower test scores and GPA than whites after they enroll in law school (although to a slight degree the Asian stats reflect inclusion of lower performing Asian groups from SE Asia and the Pacific Islands). This slight underperformance on testing and GPA has been evident for decades.
What we are discovering, of course, is the reality that diversocrats refuse to accept: different racial groups have reliably different strengths and weaknesses, and interests which are likely shaped by innate inclinations. The reason that one group or another is preferred at one time or another in one endeavor or another is mostly just a desire by the elites to virtue signal.
@BKSquared A good article, I agree its going to be a tall order for Harvard to win the case (nothing against Harvard College personally, in fact the opposite as I send my kid there too). Since its behind a pay wall I think the following passage is devastating to Harvard’s case.
@jzducol , the passage that particularly struck me:
“Nor is Harvard’s use of race narrowly tailored to achieve any particular measure of diversity. The idea of “critical mass” of minority enrollment played a central role in Fisher, and in a pair of landmark 2003 cases involving the University of Michigan. But Harvard officials’ depositions show that, as the plaintiffs put it, “Harvard concedes that it has no interest in achieving critical mass and has never given the concept serious thought.” Citing redacted testimony from the dean of admissions, the plaintiffs conclude that “Harvard is adamant that racial preferences are indispensable to its mission—and always will be.” In other words, race isn’t just a “plus factor,” which would be acceptable under the Michigan precedents, but often the dominant consideration—which again the Supreme Court has held to violate the law.”
The plaintiffs of course are being selective in the facts they present to fit their legal theory, but if the totality of the facts is that otherwise acceptable subjective holistic criteria are being manipulated to achieve overall racial quotas within a desired range, I think Harvard will have some issues.
“…Harvard’s own documents show that Asians have higher extracurricular and alumni-interview scores than any other racial group, and scores from teachers and guidance counselors nearly identical to whites (and higher than African-Americans and Hispanics)…”
Semantics may matter here --and inference and value may be missing --it may not though. My point is that the word “scores” may be simply quantitative. Asians tend to have huge numbers of extracurriculars. We see it here on CC in both the prep school and college discussions. Sheer quantity doesn’t mean ii makes the candidate special or desirable or have a potent influence on admissions. Interview ability can be prepped as well. Lower socio economic status applicants may fare more poorly on interviews and have lower access to things like robotics, model UN and squash to name but a few oft mentioned ECs.
The lawsuit does show some suspicious numbers that may indicate a bias against Asian applicants, but it is also important to consider the actual numbers, rather taking cherry picked “higher”/“lower” claims without considering statistical significance. For example, on a scale of 1 to 6, Asian applicants averaged 0.04 above White applicants on the EC rating catgory. No SD is listed, but I’d be surprised if this 0.04 difference is statistically significant. Among interviews, Asian applicants only averaged higher ratings than white students on the 2nd alumni interview/rating, not the 1st. I’m not sure when a 2nd alumini interview is given at Harvard, but I expect this is a special sample group that is not representative of the full applicant pool.
The more significant differences are Asian applicants averaged 0.23 to 0.32 above White applicants in all academic and test score rating categories. And Asian students averaged 0.13 below White students in the “personal” rating category. It’s not clear from the lawsuit whether the 0.13 difference is statistically significant , as well as whether the 0.13 difference is deserved based on LORs, essays, and the other criteria on which the “personal” category was judged. The prosecution says Harvard did not provide a large enough sample of this material to evaluate. What seems more suspicious are the claims about larger differences between races in specific academic deciles and the testimony from Harvard administrators, including not remembering anything about their internal review on racial influences in admissions decisions.
People seem to be missing the forest for the trees here. Peter Arcidiacono, the expert witness in the case whose prior analysis of race and GPA progression has been cited approvingly many times on CC, shows that Harvard is setting a minimum floor on the number of its African American admits. From his rebuttal report to defendants’ expert:
It’s pretty simple. In the case of African Americans, Arcidiacono estimates that at least 2/3rds of Harvard’s admits are unqualified based on its criteria - which, importantly, include all holistic criteria, not just academic ones. These figures are based on Harvard’s own expert witness. (See discussion on p. 47 of the report linked above.)
If Harvard is willing to admit unqualified applicants in order to meet a minimum racial quota for one race, is it really so difficult to imagine that it is denying qualified applicants admission in order to meet a maximum racial quota for another?
(Just to anticipate the rebuttal that many or most applicants are “qualified” for admissions, that is nonsense. If Harvard believed that, it could simply set a minimum qualification, perhaps internally, and then simply lottery all those above the threshold. Clearly it believes it is picking a particular class deliberately, and in a competitive situation, those denied admission are properly considered unqualified, by definition.)
Race should not be used in admissions, period. If colleges truly wanted to create a “diverse” environment, giving “extra points” based on socioeconomic status and life experience would bring diversity. Case in point: My upper-middle class kids have gone to private school from elementary on with many of the same kids, all from affluent backgrounds. One kid has a 1/2 Cuban grandmother, not much in the picture. No holiday celebrations or traditions ever included any Cuban reference. In fact, the child never identified as “Hispanic” until college admissions. He and the other affluent URM kids, who didn’t do nearly as well in school or on the SAT/ACT as many white and Asian kids are practically the only kids in the class matriculating to top schools. How will they enrich a school environment? On the other hand, a kid who had to work throughout high school to help put food on the table, or one who immigrated from another country, or one who cared for a dying parent, is bound to bring a different perspective to a discussion. So, in my perfect world, college admissions would look beyond skin color, especially in the affluent, to bring diversity to campus.
Re, the Asian issue, rating subjective personal qualities is a way to keep Harvard from becoming the UCs. Yes, it’s used to discriminate, same as it was more openly used in the past to keep down the number of Jews.
[/quote]
“Qualified” can mean different things. Black students at Harvard have a ~97% graduation rate, with a self-reported average GPA in the A- range (average GPA is more than 3.5). As a whole Black admits appear to quite capable of succeeding at Harvard, as well as after graduating. Whether this suggests being “qualified” or not is debatable. I’d agree that the report suggests that being Black is a major hook in admissions decisions.
I couldn’t agree more.
Just a few observations, but in the interests of time I will omit page cites and just post impressions.
First, there is convincing evidence of bias in the personal ratings, derived from correlation with observable metrics. These personal ratings are unobservables, subjectively assigned by the admissions office. It appears that the admissions office is artificially boosting black applicants’ ratings and penalizing Asians.
Second, SAT scores of black *admits/i are lower than white and Asian applicants and far below white (~1480) and Asian (~1540) admits. We often hear that the Espenshade and Bok (1999) data are irrelevant for today, but they would appear to be persistent.
Third, there is a boost for low-income across all races but black students. Black students receive NO boost from disadvantaged SES. I have to continue reading and thinking, but my interpretation is that there are basically NO disadvantaged black students in the pool that are even remotely admissible. Far from being granted a boost for facing disadvantages, high SES black students are simply being given a large preference for skin color. Affirmative action at the elites is simply virtue signalling, and a way to avoid facing the tough work of boosting black achievement at the elementary and secondary school levels.
Fourth, and last for now, contrary to what we often hear on CC, most applicants to Harvard are basically instant rejects, utterly uncompetitive regardless of race. It depends on how you slice the data, but approximately 80% of applicants would appear to be instant rejects. Intriguingly, there are only a few hundred at most who receive the highest academic rating (these no doubt are the USAMO winners, Intel Finalists, extraordinary TASP humanities people, etc.). This is consistent with the view that only about the top 10% of the class at most at any Ivy is truly exceptional on an academic basis.
Ok, one last observation. Legacy is very powerful for white applicants - roughly 20% of the admits - and so is athletics (~15%) and development and other special cases at ~14%. Yes, it would appear that the Kushners and Kennedys of this world are quite a lot more common than might be imagined.
In addition to the rebuttal report cited above, some of these data can be seen in Arcidiacono’s initial report here: http://samv91khoyt2i553a2t1s05i-wpengine.netdna-ssl.com/wp-content/uploads/2018/06/Doc-415-1-Arcidiacono-Expert-Report.pdf
All in all, no surprise to anyone who has attended any of HYP.
I’ll keep reading.
HYPSM… type holistic colleges favoring URMs in admissions decisions is well accepted, with ample evidence. Claiming such colleges have affirmative action administration policies in place for grading of individual courses is something else. I’ve never seen evidence of this or even someone making such a claim before. If I was guessing without evidence, I’d expect that some individual professors (or TAs grading papers/exams) have racial biases and/or expectations. And in more subjective classes, these racial biases/expectations can influence grades. I’d guess the net effect would be slightly negative for Black students, although there would likely be effects in both directions.
It’s far from an ideal data source, but it’s also consistent with other metrics, such as the more officially reported graduation rate. If the Black average GPA was abnormally high due to sample size issues and/or percent feeling marginalized, then I’d expect see the Black GPA being abnormally high compared to Hispanic students. Instead Black students fell slightly lower than Hispanic students.
This is probably esoteric for most people reading this, but I know you will understand it. Take a look at the overprediction by UGPA (as measured by mean residuals) in Table 7 (p. 14) here:
https://www.lsac.org/docs/default-source/research-(lsac-resources)/tr-17-01.pdf
This is not anomalous, see same presentation for different years in Table 7 (p.17) here:
https://www.lsac.org/docs/default-source/research-(lsac-resources)/tr-12-02.pdf
If you compare residuals of UGPA versus residuals of test score alone, you will see that blacks are a special case of dramatic overprediction. Either they took course work in college that wholly unprepared them or they were assigned grades higher than their true performance. Perhaps you can come up with an alternative explanation that fits the data.
Also, take a look at the summary statistics of GPA by race for some truly shocking distances (in SD terms) and note the narrowness of the white distribution of GPA. We essentially see two entirely different schools in each school, segregated by ability. (See Tables 5 in both studies.)
These are very large data sets (N=~80,000) and they are well done. They regress GPA against UGPA (so, college grading) and standardized tests, but they appear to do it correctly. (Transforming GPA by classroom and school into z-scores effectively to remove systematic effects of different grading systems and idiosyncratic effects of different curves).
These studies are also particularly useful because they examine GPA of a standardized first year curriculum as explained by two variables: (1) UGPA (which will vary by choice of major, difficulty of school, affirmative grading by race, etc.) and (2) a single standardized test. Unlike in the much more sloppy HSGPA/SAT-FGPA studies, these more careful studies show that the standardized test is a much better predictor of GPA, particularly for blacks and Hispanics. In fact, it explains roughly 4 times the variance for black students, for example, than UGPA alone. (See Tables 6 in both studies). See what you think - to me they look convincing.
I only skimmed through the first paper, but it looks like they are predicting first year law school GPA based on a combination of undergrad GPA and LSAT score. Undergrad GPA alone overpredicts Black students’ first year law school GPA more than other races, where as LSAT + UGPA does not to the same extent. As you’ve previously pointed out in this thread, Black students tend to do poorer than other races on the LSAT, and law schools tend to have affirmative action policies to compensate. I have no doubt that at a particular UGPA, Black students average lower LSAT scores than other races. And it seems first year law school GPA depends on additional factors not captured well by UGPA alone that are better reflected in these lower LSAT scores. There are many possible reasons why first year law school GPA might depend on additional factors besides UGPA other than just “affirmative action grading”, and those additional factors are not expected to occur at the same rate for law school students of all races.
There are many differences between predicting law school GPA from undergrad GPA and LSAT, and predicting undergrad GPA from HS GPA, SAT, HS course rgior, LORs, essays, and the rest of the criteria in a holistic evaluation. What best predicts academic success at Harvard undergrad no doubt has significant differences from what best predicts first year GPA at a composite average of various law schools. This is reflected in the Duke study you linked, which found both being femaile and HS curriculum to be statistically significant predictors for switching out of techy majors with all levels of controls, yet SAT score was not a statistically significant predictor with full controls.
Thanks for looking at those studies, @Data10. Sure, I agree that there are a few alternative explanations for the massive overprediction by UGPA for black students, and we’ll never know for sure whether fingers are on the grading scales.
What do you think of the mean FYA for black students? Can it really be that - in terms of the white SD of FYA - that black students are ten SDs away? (~3 SDs in terms of black GPA.) I’d love to hear your input, because to me that basically is two schools, with virtually zero white students ever having the same grades as the average black student. I mean, at 10SDs we are talking longer than the age of the universe before we would expect to see a white student with the same grades as the average black student, no?
Also, the data are interesting for Asians. Though a few lower performing Asian groups members are doubtless thrown in there (Native Hawaiian, Pacific Islander, SE Asians), the lower scores and grades for Asians as a whole in law school might surprise some people.
About the Duke study, isn’t one of the controls SAT score itself? In any event, let’s not get lost in statistical mumbo jumbo. The Duke paper establishes that the black admits are the weakest on all metrics (particularly so on measures of academic achievement, including scores). Upon entering Duke, they predictably achieve the lowest GPA (at the 25%) of any group. And they remain at the bottom of the GPA heap each year thereafter. All groups show rising GPA through the years, with the gap between black GPA and other groups narrowing. Arcidiacono shows that this narrowing is fully explained by black students switching into easier majors. Period.
It’s sad that those black students who bailed out at Duke might well have stuck with sciences had they been at an easier school. As Duke knows the data about switching, one would think ethically Duke should either apply tougher standards to its prospective black science admits, or at least provide some disclosure. After all, once someone drops out of a science major, one will never work in the field, and many of these bright black kids no doubt were dreaming of working in science for years and may well have succeeded had they been matched to a more appropriate school.
Using that logic, Asian students are ~6 SDs below White students, so you’d also have to wait many years before a White student ever did as poorly as the average Asian student. The more likely explanation is that the individual student SD is on the 50/10 scale indicated in the quote below, and the 0.78 SD is per overall law school average scores not per individual student scores. This makes the mean of Black students, as well as all other races, less than 1 SD from the overall mean.
“FYA values were transformed to a scale having a mean of 50 and a standard deviation of 10. Results presented in this report are on the transformed 50/10 scale”
Well, thanks for your explanation, @Data10. I’ll admit that I am a bit perplexed at that aspect of the presentation, as well. But the idea that black students’ mean GPA is within 1SD of whites is contradicted by the actual grade data examined by Professor Sander in numerous studies, which showed 50% of black students falling in the bottom 10% of the overall class, with the mean GPA of blacks at approximately the 6% of whites if I recall correctly. That’s quite a lot more than 1SD wouldn’t you say?