"Race" in College Applications FAQ & Discussion 12

Applying to College
2221

In the senior survey, Harvard students are asked about the difficulty of their concentrations. Each year, the majority say their concentrations are either “somewhat difficult” or “very difficult.” In the most recent one, Harvard seniors reported spending an average of 33 hours per week on academics. That’s a significant amount of time and double the national average, but also not a ridiculously high workload and probably lower than certain other schools.

I attended Stanford, which has a similar grade distribution to Harvard. I also took quite a few classes at SUNYA and RPI, prior to attending. My experience was that equivalent Stanford classes where generally more challenging than SUNYA and RPI classes, with a higher workload, although there were numerous specific class exceptions. Instead of being really easy, the primary reason why there were so many A grades was because a large portion of the student body was composed of excellent students You see the same trend in overall GPA reports. As a general rules, as selectivity of the student body increases, so does the percentage of A grades and overall average GPA. Few highly selective colleges appear to use ‘A’ to mean top <30% of a particular HYPSM… class and instead use it to mean students who are doing very well on the assignments and exams. I believe most professors at highly selective colleges would be okay with allowing the vast majority of students in the a particular class to receive A’s, if the vast majority were doing well. Along the same lines, while there were far more C’s in SUNYA classes and to a lesser extent RPI, I didn’t feel this was because of making it easier to get C’s. Instead a larger portion of students were not doing well on assignments/exams at SUNYA than at Stanford, so a larger portion of the class received lower grades.

2222

http://faculty.ucr.edu/~mmarks/Papers/babcock2010falling.pdf suggests that the national average was around 27 hours per week in 2004 (down from 40 in 1961), so 33 hours per week may not be double the national average.

However, a nominal full time course load (15-16 credit-hours or equivalent) is supposed to represent 45-48 hours per week of work. But it is certainly possible that technology has reduced the amount of time to do the same amount of work. For example, decades ago, looking for a reference may require going to the library, looking in the card catalog for candidate books, and then looking in the books to find whether they have the needed reference. But now, a quick web search may yield the needed information, or where the needed information can be found, so that you can go directly to that book in the library to read it (or order your own copy on-line if you want). Also, stronger students like the ones who can get admitted to Harvard may need to spend less time than the average student for a given volume and difficulty of academic work.

Of course, not all Harvard students choose the harder courses that are available.

2223

It looks like those results are from 2004. The 2018 NSSE survey of more than 100 colleges at http://nsse.indiana.edu/2018_institutional_report/pdf/Means/Mean%20-%20Sex.pdf found freshman and seniors both spent an average of ~15 hours per week “Preparing for class (studying, reading, writing,doing homework or lab work, analyzing data,rehearsing, and other academic activities)”. The exact number of hours depends on definitions. I expect most highly selective colleges will not have huge dfferences from Harvard, even ones with a reputation for being especially rigorous and/or having a high concentration of engineering majors. For example, in the MIT survey at http://web.mit.edu/ir/surveys/pdf/2011_ESS_Overall.pdf , MIT seniors reported a median of 11-15 hours attending class + a median of 21-25 time spent outside of class, suggesting a combined median near 36 hours. That’s higher than the 33 hours reported in the Harvard survey, but not tremendously higher.

2224

This looks like out-of-class time only, since a full time student probably spends about 15 hours per week in class (or more if taking courses with labs, art studio, or music performance, or less if taking a light full time schedule of 12 credits that typically comes with 12 hours per week of class time). So the total academic time per week would be 27-30 hours.

2225

If we’re discussing how much time Harvard students spend studying, we’re getting far afield from the primary question of whether Harvard uses racial/ethnicity quotas/buckets in undergrad admissions.

2226

Recently in the news Yale student was reported as “suspicious” by her fellow student just because she was black. Now Smith College had similar situation. These incidents made me wonder, how such elite schools admission process really like? Even though they admit minority, are all students equally treated?

2227

Whether the college is “elite” in admissions does not determine the chance of encountering racism from other students or others in the area.

2228

People keep mentioning SAT scores as if they are the sole component of the academic index or Harvard’s proprietary Academic Rating. They’re not. The index and rating also include GPA (and derived class rank) and SAT2 subject test scores. Presumably AP scores and rigor factor into the 1-6 rating as well (there is a world of difference between Physics C and Human Geography, e.g.).

Also, the 745 versus 720 figures thrown about are nowhere explicitly listed in the litigation material, but rather are being eyeballed from a graph. Notably, that graph reflects all hooked candidates as well, including those legacies, development and athletes who present with relatively lower scores ( not all do, of course). As these preferences disproportionately favor white students, we would expect the apparent differences between black and white to narrow.

IMO the Asian scores and Academic Ratings are more representative of the unhooked pool than the broader figures (because Asians tend to receive some of the least preferences in athletics, development, and legacy). On SAT scores, it’s worth noting that Asian ** rejects ** are * higher * than black ** admits ** in every year examined.

Also, many people on here talk about blacks being from disadvantaged backgrounds, but if you look at the information you’ll see that approximately 30% of black admits are tagged as disadvantaged by Harvard while 25% of Asian admits are. Fairly close imo.

Moreover, although only about 15% of white admits are disadvantaged, approximately 35% of white admits are legacy and/or development iirc, and another 15% are athletes often in very expensive sports. I’d guess the unhooked white group is comparable to Asian and black as far as disadvantaged goes.

The GPA discussions are not serious. Anyone who’s been to HYP knows the score here. There’s a world of difference between majoring in physics or math versus sociology or political science. Also, most courses (and all upper level ones) are not blind graded, and do not have objective standards anyway. To believe that there is not strong institutional pressure to narrow apparent differences and to compress the grade scale is naive.

2229

Table B.3.1R in the Arcidiacono rebuttal report sheds some light on the academic differences between unhooked (other than race) black admits versus unhooked white and Asian admits. (However, the data excludes unhooked early decision applicants, and so don’t tell the whole story.)

On the academic index, white admit z-score +0.76 (0.38 sd) versus black admit z-score score of +0.33 implies a greater than 1 standard deviation difference, as measures by white standard deviation. In other words, approximately 87% of unhooked white admits will present an index higher than the average black admit.

For Asians, +0.91 z-score (0.32) implies almost a two standard deviation difference, with approximately 97% of Asian admits presenting higher indices than the average black admit.

It’s ironic that after literally countless posts on cc that SAT scores are not that important, now many posters seem to be focusing on smaller differences in SAT scores and ignoring the academic index that takes account of all standardized testing plus high school GPA.

The table is on page 144 of the rebuttal report as filed.

While stats mumbo jumbo like model outputs and regression coefficients are interesting, I find the raw data to be much more compelling in thinking about the extent of preferences. What they tell me fwiw is that Harvard is looking for a reason to admit black candidates who are at all within the reason of reason on academic measures (and judging from the assigned academic ratings only about 7% of black applicants are), while holding Asian applicants - and presumably unhooked white applicants - to perfection. Hispanic applicants appear to be about half way between the scrutiny accorded to unhooked whites and Asians, and blacks.

2230

Nobody has stated that. However, 2/3 of the academic index is composed of SAT/ACT scores, and the majority of Harvard students mention having a 4.0+ UW GPA in the freshman survey; so AI of Harvard admits is primarily a measure of test scores.

In contrast, academic rating is not primarily a measure of AI or of scores. It includes far more than just stats + rigor. AI is correlated with academic rating, but AI only explains a small minority of variance in academic rating. For example, ~1-2% of applicants receive an academic rating of 1, which is associated with being one of the best academic scholars in the class and often involves having academic work reviewed by faculty. These few top academic “1” applicants were found in all 10 academic deciles. It is possible to get the maximum academic rating with the minimum AI decile.

The lawsuit provides this in comparison with baseline and expanded sample. You reference these Z-scores in your following post, so I’m sure you saw how little the difference in average scores is between the baseline sample without legacies, development, althletes… and the expanded sample with these groups… Specific numbers are below:

White “unhooked” baseline: Verbal = +0.72, Math = +0.56
White full sample including hooks: Verbal = +0.69, Math = +0.52

The average Z-score delta between black and white students was 0.335 in the expanded sample and 0.365 in the baseline sample, or a 9% increase. If we assume the score delta in the graph of the full class also increases by 9%, then the estimated scores would change as follows for the “unhooked” baseline sample:

Class of 2017: “Expanded” Full Sample: White = 745, Black = 720
Class of 2017: “Unhooked” Baseline Sample:: White = 747, Black = 720

Extrapolated Current Class of 2022: “Expanded” Full Sample: White = 750, Black = 735
Extrapolated Current Class of 2022: “Unhooked” Baseline Sample: White = 751, Black = 735

I’ve never heard anyone suggest that HYP professors are pressured to increase black grades. However, to suggest that the bias is all in a positive direction and no professors have any kind of bias against minorities, as has been confirmed in numerous studies of the education system going back all the way back to elementary school, is naive.

2231

@Data10 just as a note, the AI index for all Ivies is 2/3 SAT and/or SAT 2 scores and @SatchelSF the AI no longer includes class rank.

2232

It does not imply this level of difference. A huge portion of the group is hitting 4.0 + 800 on at least one of the tests, which reduces the SD lower than would occur in a normal distribution. If you try to use that lower SD, then you underestimate the portion with lower scores. A similar issue occurs with White scores, although to a lesser extent. As has been discussed numerous times, if you instead try to predict the number of Black students who score above Asians, you get 13%. The Black SD is not compressed to the same extent by hitting 800 (although it is compressed by hitting 4.0), so the normal distribution approximation is not as innaccurate as it would be for Asian admits.

The HS GPA difference between races was much smaller than the score differences. As has been stated, the majority of Harvard students in the freshman survey report having a 4.0+ UW GPA. The reason much of the discussion has been on SAT scores, rather than just standard deviations of AI is because it’s more meaningful to put in to context. Just looking at SDs of AI alone is not meaningful unless you also consider how meaningful that SD difference is. For example, suppose the AI difference was 1 SD, and that 1 SD corresponds to a 10 point score difference on the SAT for students that have a GPA near 4.0 UW (a hypothetical example, not actual numbers). I think most persons would not consider a 10 point difference in SAT score extremely important, even if is 1 SD.

2233

I would imagine that elite school admission had some kind of a racist student candidate profile system, unsure if the system is reliable

2234

Harvard doesn’t have ED, @SatchelSF

“(However, the data excludes unhooked early decision applicants, and so don’t tell the whole story.)”

2235

Harvard does have an early decision program for most of the years examined in the report, it’s just called restrictive early action. Sure, there is no requirement that the accepted student be bound by the decision, but the decision is received early nonetheless. Arcidiacono excluded these early admits from his baseline dataset because if the large differences observed in admit rates.

2236

Ed and scea are not the same thing. One requires acceptance of the admissions offer and one doesn’t.

2237

Of course, as I mentioned above. No need to distract with irrelevancies. The real point is the disparities in academic indexes by race, which disparities are enormous.

2238

@SatchelSF Correct me if I’m wrong. You seem to be bothered by the fact that some 13% of Harvard’s class is black, and of those, some portion is academically less qualified than most of the other students. It doesn’t bother you that some portion of legacies and development admits are also less qualified. Harvard has 1000 arguments to explain why both kinds of less-qualified students are institutional priorities – (they’re future leaders!) – but I’m curious to know why you’re so hawkish on the URMs but forgiving of the legacies. Surely whatever harm is being done by juicing URM admissions is at least equaled by the iniquitous perpetuation (expansion?) of social inequality that comes with favoring a Jared Kushner over, I don’t know, an IMO champion.

2239

That’s a very fair question, @LadyMeowMeow, so let me offer a brief explanation. (As an aside, it is not that I am not bothered by all preferences - I’d rather see a more academically focused admissions system - but I recognize that private schools have the right to shape their class as they see fit, subject to law.)

I am bothered by these huge preferences on the basis of skin color first and foremost because the rest of society is paying a huge price in trying to eradicate discrimination on the basis of race, both in the private sphere as well as the public one. Corporations and landlords could never get away with the disparate impact that, for instance, elite admissions at Harvard gets away with. The elite eggheads at Harvard should have to live within the system of law that Harvard had such a large role in creating. The sort of racial gerrymandering that Harvard is undertaking here - with admit rates of blacks in the middle Academic deciles of 10x or more equivalent decile white and Asian admits (see Table 5.2 in the Arcidiacono report) should be ipso facto illegal.

In addition, I believe that large preferences are stigmatizing. As I point out often - and not just for virtue signaling but because I know it to be true - there are many black and Hispanic admits who would have been admitted without race preferences. Nevertheless, those students will be painted with the same brush. People who insist that the problem lies only with those people who stereotype simply are being disingenuous in my opinion.

Moreover, I believe that race preference is special in that it immediately marks the recipients because of an attribute that they cannot change. In a somewhat different context, Thomas Sowell wrote that if you put a second quintile black student in a first quintile school, he will see racism in every nook and cranny of the school and curriculum; put him in a second quintile school and everything will be fine. This accords with my own personal experiences. I don’t see quite the same problems with legacies and development kids, not even athletes. When is the last time you saw a protest by those groups on campus?

Last, I believe that the presence of wide disparities in academic ability, especially when combined with the natural tendency of people to identify with their race, inexorably leads to lower standards and less ability for the grading system to make distinctions. I’d argue that the desire to narrow apparent differences by race has in part led to inflated high school gpa and watered down standardized tests. This phenomenon makes the college application process harder for all students looking at “top” colleges, as the typical metrics of achievement have been so watered down as to make an “arms race” in different metrics almost inevitable. There is dramatic deadweight loss in tolerating a system that forces our best and brightest to jump through meaningless hoops in order to impress adcoms. I can’t tell you how many kids I know who waste $6k of their parents’ money pretending to build houses in Costa Rica or teachng English in Colombia, lol…

I could go on, but those are some of my reasons. I am mostly interested in this from a societal standpoint. For myself and my own kids - allow me my own virtue signaling here - I am not so concerned as they will be Ivy legacies and also (hopefully) race preference URM (Hispanic).

2240

Race is useful to indicate in the admission process insofar as it allows officers to contextualize students. It’s just a way to add more data for a fuller picture. But certainly there are those cases where it can hurt or help.