"Race" in College Applications FAQ & Discussion 12

Applying to College
3961

Well, @tpike12, it’s difficult to estimate what percent of unhooked whites are Jewish, just from the overall numbers of Jewish students reputed to be at Ivies. Also, there are many Jewish students no doubt who are legacy and development (and probably a few recruited athletes and faculty affiliation), so it is not fair to think of all Jewish students as belonging to the unhooked whites group. Even the overall numbers are disputed, and not dispositive because they are typically based on survey data, and of course some Jewish people identify as “ethnically” Jewish but might not consider themselves belonging to the religion. I just have no insight into the numbers, other than to say, as is obvious, that there is a very large overrepresentation of Jewish students at the Ivies based upon their population share of the 18-22 year old cohort generally. Perhaps on the order of 5x to 8x population share.

But those figures I provided above in Post 3958 above are admit rates, not shares of the respective groups. So, the way to read them is, for example, 15.3% of unhooked whites who apply having stats that place them in the top 10% of stats are admitted to Harvard, while 57.1% of white legacies/development who apply with those stats are admitted. Or, 55.2% of nonhooked black applicants with stats placing them in the top 20% of stats are admitted (so this group will also include all those who fall into the top 10%), while 85.7% of legacy and development black applicants with those stats are accepted.

Clearly, the notion that all groups and categories of students face miniscule odds at Harvard is “horsepucky,” but everyone reasonable already knew this. Just looking at the numbers, one can see the very large advantages enjoyed by race preference groups, and by the special status preference groups.

I’ll go back and try to figure out what percent of white admits only are actually “open,” in Harvard’s construction. Just going by memory, I think legacy and development together account for fully 40+% of the admitted white group, and I think white recruited athletes constitute another 10%, so fully more than 50% of the “white spots” are already “allocated” before the competition even opens up. The other race groups do not enjoy the same legacy, development and athletic preference, and so are a little more “open.”

I hope that data like these make it clear that what Harvard is doing is implementing a rough quota system, with carveouts within each race group for certain preference candidates. Different admission criteria by race. I’ve finally built a spreadsheet with all this information, so if anyone wants to know something specific, like “what percent of Hispanic admits are in the bottom 40% of stats?” or “how many Asian students on average a year have top 10% stats,” just ask. Considering how difficult it has been for Harvard to get its message out there regarding whom it is really seeking, it’s only fair that we try to help disseminate the data.

3962

There is little doubt there were buckets for various applicants. So far, data on tippy top holistic process have only come from the Harvard case. Though its a treasure trove of data it probably captures less than 20% of the information on an admission file. When my DD showed me her Harvard admission file last fall—admits could request the file under fed privacy act—I was stunned how much information the evaluation folder contained beyond the summary page. The seven page dossier has two readers evaluations, alum interviewer’s report, a faculty reviewer’s comment and some AOs’ email communications on the case. It talked about the things from region’s applicants landscape that year to my kid’s interview attire. Though the summary page is all about the data it doesn’t quite reveal the thinking behind Harvard decisions.
By my conservative estimate there was at least 2.0 person hrs spent on my DD’s application at Harvard admissions, which doesn’t even include alum’s one hour interview and another hour on her two page interview report. I doubt many schools are in Harvard’s position resource wise to conduct such non-data driven holistic process.
To make the matter more complicated even the summary page which has all the data, doesn’t quite present the data the way people would understand them. Here is the blank summary sheet, the middle portion of the ratings is where the litigation drew most of its data from.

https://www.wsj.com/articles/breakdown-of-the-harvard-admissions-process-1540287000
The importance of the data seems to decrease from left to right. So the academic rating is far more important than academic index. My DD’s academic rating was a 2/2-; but her academic index was a solid 1, as her test scores and GPAs were perfect and she had national academic awards too. I don’t know how those deciles get translated into academic ratings but they are certainly not a direct translation and are probably de-emphasized in the process.

3963

Deciles in the litigation appear to be just a formula, based upon GPA + SAT/ACT + SAT2 + AP Scores. There is presumably an adjustment for rigor, but this is not clear. National academic awards would only flow through to the Academic Rating.

I can’t remember exactly where I saw it, but the correlations between deciles and Ratings were pretty high, but not perfect of course, with effectively 5 deciles (the top half) mapping to only three available Ratings (2-, 2 and 2+). A rating of 2/2- frankly sounds low for a perfect stats applicant; perhaps race plays a role here (not all Academic 2s are equal?).

No doubt that Harvard spends a great deal of time on candidates after the first screen, but of course many applications must simply be throwaways.

3964

My daughter was a “throwaway” application for Harvard. Great stats (Valedictorian, near perfect SAT and subject tests, tons of awards for service, 4 year varsity athlete, blah blah blah). I know she was a throwaway because a good friend of mine used to interview for Harvard. I mentioned to her that she was never contacted for an interview and she contacted a friend of hers that still interviews for Harvard. She looked her up and said she didn’t make the cut for interviews.

We never thought for a minute that my daughter was getting into Harvard and just laughed it off. For the record, she is a white girl from NJ (non-Jewish but I think it wouldn’t have mattered if she was). She is happily thriving at her non-Ivy college right now, not feeling slighted at all.

3965

I doubt that your daughter’s application was a throwaway, @collegemomjam. Valedictorian plus near perfect SAT would put her somewhere in the top 20% I estimate of Harvard’s applicant pool, at least by academic stats. Harvard cannot afford to look at those kids as throwaways, because it needs to ensure enough top academic kids are in the admit pool.

About 1800 unhooked white students apply per year with those sorts of stats, or higher. On average, 230 are admitted, for an admit rate ~13%.

About 130 white legacy and development applicants have those sorts of stats per year. On average, 74 are admitted, for an admit rate ~57%.

There is not enough granularity in the data to estimate the male/female split at any level of academic stats. Similarly, I cannot disaggregate the legacy and development candidates, although it can be inferred that approximately 20-25% of development candidates are also legacy.

Looking through the academic stats of the legacy and development candidates, it is very clear that they are a little weaker than the unhooked applicants, which makes sense because the knowledge of the preference encourages application from weaker candidates (relative of course to the universe of kids generally who would consider an application to Harvard).

3966

“She looked her up and said she didn’t make the cut for interviews.”

How do you not make the cut for the interview?

3967

I believe Arcidiacono is simply grouping the Academic Index by decile. Academic Index is a calculation the Ivy League athletic conference schools use to set a stat minimum for athletes compared to the overall class (x athletes can have an AI y SDs below overall class). Some may also use it for non-athletic admission purposes. It’s my understanding AI is composed of GPA, SAT I, and SAT II, as listed below. During a portion of the lawsuit analysis period, rank could be substituted for GPA. It’s my understanding, the Ivy League used to have a straight sum of these 3 sections, making 2/3 of AI composed of scores, then they switched to a 50% GPA / 50% score weighting.

80 Points for GPA
98%+ = 80, 4.0+ unweighted = 80, 4.30+ weighted = 80
95% = 77, 3.75 unweighted = 77, 4.0 weighted = 77
90% = 69, 3.3 unweighted = 69, 3.5 weighted = 69

80 Points for SAT/ACT – Superscored SAT / 20
1600 SAT = 80, 36 ACT = 80
1500 SAT = 75, ACT = concordance table equivalent

80 Points for best 2 SAT II Subject Tests – SAT / 20
1600 SAT = 80
1500 SAT = 75

Admitted Class of 2019
Mean Converted GPA = 77.1 (3.75 UW)
Average SAT = 2241 (74.7 points)
Average ACT = 33.1
Average SAT II = 760 per test
Average Academic Index = 228

The academic rating is on a 1-5 scale, rather than a 240 scale like academic index. The academic rating includes more than stats. The lawsuit states, “The academic rating summarizes the applicant’s academic achievement and potential based on grades, testing results, letters of recommendation, academic prizes, and any submitted academic work.”

Perfect stats alone is not enough to get the maximum 1 academic rating. The reader guidelines for giving a 1 state, “(in most cases combined with unusual creativity and possible evidence of original scholarship, often substantiated by our faculty or other academic mentors.) Possible national or international level recognition in academic competitions.” However, there is a decent correlation between academic index and academic rating. The percentage of applicants receiving a 2- or better academic rating among different AI stat deciles is below. The percentages were reasonably similar, regardless of race. Black students were slightly less likely to get the high academic rating for the same AI, perhaps because of being slightly less likely to have the other criteria.

Top AI Decile – 97% receive 2- academic rating or better
2nd AI Decile – 94% receive 2- academic rating or better
3rd AI Decile – 84% receive 2- academic rating or better
4th AI Decile – 70% receive 2- academic rating or better
5th AI Decile – 51% receive 2- academic rating or better
6th AI Decile – 26% receive 2- academic rating or better
7th AI Decile – 8% receive 2- academic rating or better
Overall – 42% of unhooked applicants and 82% of unhooked admits receive a 2- academic rating or better

However, academic rating is far from the end all to admissions decisions. Instead academic rating only explained 9% of variance in admission decisions – roughly the same influence as the EC rating or athletic rating, and substantially less influence than the personal qualities rating.

3968

On the construction of “academic index” as used in the Arcidiacono report, see footnote 29 (pp.22-23):

This would imply that, at least as used in the report’s analysis, the academic index is based on 1/3 SAT/ACT, 1/3 SAT2 subject tests, and 1/3 GPA/class rank.

Frankly, the whole relationship of academic index to the Academic Rating could have been elucidated a bit more clearly. It would seem to be fertile ground to correlate academic index to Academic Rating by race, to test if there are any systematic differences. As the Academic Rating is a subjective determination, it would likely reflect any bias in the readers.

3969

@Canuckguy I don’t think that employers will do anything. U of Chicago and many other schools are now test optional and I don’t think that employers will care if students complete degree requirements without lowering those degree requirements at those institutions. I don’t mind test optional schools in the educational landscape, but I always disagree with the GPA vs. Standardized Test Score comparison over which one is better. All of the data I have previously seen says that putting both together (GPA and test scores) is better than either is separately at predicting college success. If GPA alone was better than the GPA/test scores correlation, then I would be all for getting rid of standardized tests everywhere, but that is not the case. High schools today have more grade inflation and provide very different educational experiences so AP exam scores, ACT, SAT, and SAT subject tests are an important way to distinguish students.

3970

@Data10 and @SatchelSF Thanks for all of the data. For my son, it basically means that my son would probably have a 50/50 shot at Harvard if he decided to apply based on stats alone (assuming he took the subject tests and did as well on them as he did on the ACT). It would be tough for me to watch my son deal with the perception of his “worthiness” due to AA, but I would do it if he chose to apply to any T-20 schools.

One thing that has been different between my daughter and son’s admissions journey is that my son has received much more “junk e-mail” with many more elite schools corresponding (he has received 103 e-mails from universities in the last 7 days including 6 Ivys and many other top institutions). His test scores are slightly higher, but she was a slightly better high school student. I am not sure if this is just because of his higher standardized test scores, more recruitment of students today versus 2 years ago or because he is a part of a smaller AA subgroup (African American male), that is even more underrepresented in elite admissions.

3971

Harvard’s admissions dean did say the following in 2009 (though this may have been before widespread gaming of the SAT/ACT writing sections by test prep companies):

https://thechoice.blogs.nytimes.com/author/william-r-fitzsimmons/

Looking at each of these, we see:

AP/IB tests: measure achievement in high* level material in a standardized way
SAT subject tests: measure achievement in medium* to high* level material in a standardized way
HS grades: measure achievement in low* to high* level material in a less standardized way, but also including factors not easily measurable in a 1-3 hour test (e.g. achievement in larger projects or papers).
SAT/ACT: measure “aptitude” or something like that, based on low* to medium* level material, in a standardized way.

*for high school.

Of course, when Harvard has an applicant pool loaded with high achievers in all of the above (plus a few super-achievers thrown in), it has the luxury of considering other aspects that it wants as it is building its next class of future management consultants, Wall Street investment bankers, etc… Predicting academic performance may not be an admission priority at Harvard when just about every admitted student is likely to do well.

A more typical moderately selective college may place a higher priority on predicting college achievement to raise its retention and graduation rates. However, it also may be looking at USNWR rankings, which use SAT/ACT scores for most of the direct student selectivity measures.

3972

The admit rates listed above are not for “stats alone.” They are for applicants who have stellar ECc/awards, LORs, essays and similar, at the level that is typical for highest stat decile applicants at Harvard. If the applicant instead had highest decile “stats alone” and nothing special in the rest of the application, I’d expect it’s extremely likely the applicant would be rejected, regardless of race. The relates to how highly selective holistic colleges get a reputation for making random admission decisions or being reaches for everyone. The admission decisions are reasonably predictable if you can see the full application and reader ratings (both Plantiff’s and Havard’s experts lawsuit models could explain majority of variance in decisions). However, looking at academic rating alone could only explain 9% of variance in decisions. Also note that the analysis sample period started in 2010, when admit rates were higher than they are today.

3973

Employers as a whole generally focus far more on college major than college name. For example, in the survey at https://chronicle-assets.s3.amazonaws.com/5/items/biz/pdf/Employers%20Survey.pdf , hundreds of employers in a variety of industries are asked to rate the relative importance of different criteria in evaluating the resumes of new grads for hiring decisions. The overall relative importance they found is below (linear, higher is better).

Internships – 23
Employment During College – 21
College Major – 13
Volunteer Experience – 12
ECs – 10
Relevance of Coursework – 8
College GPA – 8
College Reputation – 5

There are some issue with their methodology, particularly in the way they separate academics from ECs. However, the point remains that in every surveyed industry, employers appear to give little weight to college reputation. When asked about specific types of colleges, the pattern was similar. Poorly known colleges were ranked poorly, but there was little difference between “elite” colleges and other well known colleges, such as state flagships.

Nationally Known – 3.77 (57% prefer, negative factor for 1%)
Elite Colleges – 3.69 (56% prefer, negative factor for 3%)
Local College – 3.57 (47% prefer, negative factor for 2%)
Unknown College – 2.67 (6% prefer, negative factor for 34%)

In contrast, 97% of employers said college major is important to some degree. The preferred majors varied dramatically and generally fit with the type of industry. Surveyed tech employers most preferred majors were CS and EE. Business preferred degrees in accounting/finance or business. Health care favored degrees in nursing or other health related professions. This fits with my personal experience in tech. Most jobs I considered as a new grad stated that they required a degree in a relevant major, generally electrical engineering. All interviews tested me on electrical engineering related knowledge, at a level that students in other majors would most likely not pass.

You may be talking about elite banking and consulting, which generally composes a significant portion of Harvard’s class, but is a small portion of the overall job market and probably negligible portion of the above survey. These industries do often focus far more on college reputation, which I expect partially relates to their popularity among Harvard grads. Students who are focused on a highly prestigious college name are more likely to favor a highly prestigious employer name, that primarily hires from highly prestigious colleges and has various other hiring barriers/screens that contribute to an aura of prestige and selectivity. I expect the hiring is not just about choosing the candidate who is most likely to be successful on the job and would continue to favor graduates from Harvard, regardless of how Harvard changes admission criteria. Some of these employers already already have stat filters, in addition to college name filters, such as expecting college GPA to be close to 4.0 (median GPA at Harvard is likely ~3.8) and/or asking for high test scores. College major is generally less critical, which I understand is one of the reasons that they were able to admit such a notable portion of Harvard’s class. For example, the article at https://www.vox.com/2014/5/15/5720596/how-wall-street-recruits-so-many-insecure-ivy-league-grads mentions:

“But in the early 80s Goldman Sachs and others figured out they could broaden their net and get lots of really smart people if they made it a temporary position rather than a permanent one. So they created the two-and-out program. The idea is you’re there for two years and then you move onto something else. That let them attract not just hardcore econ majors but people majoring in other subjects who had a passing interest in finance and didn’t know what else to do.”

3974

@ucbalumnus Thanks for that blog link. Seeing AP/IB exams at the top of the list makes sense, but it seems like quite a few schools have started to move away from mandatory SAT Subject tests recently so I would not have guessed they were a close 2nd.

@Data10 Thanks for clarifying for me what is incorporated in that decile look. I consider my son’s EC’s to be subpar in comparison to those he would be going up against at most schools as all of my son’s EC’s revolve around giving back/volunteering outside of school grounds. Hopefully, doing MITES (if he gets in) or the Carnegie Mellon SAMS Summer programs will help, but I think most AdComs would have trouble seeing his strong STEM slant besides looking at his classes.

3975

The main issue that colleges find with SAT subject tests is likely that they are a non-default testing option or requirement, so few high school students take them. Requiring them would screen out potentially desirable applicants who may not have had them on their radar (e.g. if attending high schools where hardly anyone applies to colleges that require them). That is likely why Harvard changed them from “required” to “recommended” recently, basically hinting that students in academically high end high schools really should take them, but they will accept applications from students in high schools where they are not on the radar and hence do not know about them until it is too late.

With AP/IB scores, the issue is often availability, since many of them are taken in 12th grade, so the scores are not available for college admission decisions.

So the SAT and ACT, with their default or incumbent advantages (knowing that these tests need to be taken for college admissions is widespread, although there is variation in depth of knowledge as it relates to test-taking strategy*), remain the primary standardized test for college admissions in the US.

*For example, some with less sophisticated knowledge may just do one-and-done in fall of 12th grade, giving no opportunity for a retry if the first try is not enough for the target colleges, and not having scores before 12th grade makes building an application list more difficult in terms of assessing reach/match/likely/safety.

3976

The SAT subject tests aren’t some hidden secret. They’re clearly listed in most requirements pages for colleges that require them. My experience has been that everyone who is somewhat competitive and cares about college knows about SAT subject tests. Fulfilling the SAT subject test requirements itself isn’t hard. To get to 2 SAT subject tests, do Math 2 (anyone who is competitive can easily pass this without studying) and pick your best subject in school.

3977

Harvard and others presumably do not want to auto-reject applicants who come from high schools where SAT subject tests are basically unknown because hardly anyone ever applies to a college that wants them (perhaps because the most selective college that students normally apply to is the local non-flagship public that is not that selective), and overworked counselors are too busy with other problems like just getting students not to drop out.

3978

@ucbalumnus I have always hated the excuse of subject tests being unknown. I took 2 subject tests in 1993 (Math and US History) and it was not fun scraping up my own money to pay for those tests. I did not have internet access (not sure my school even had it available for students) and I did not know anything about the subject tests, but I saw through my own research at a public library that I might need them based on one of the schools I wanted to apply to. Today, with the advent of fee waivers and Internet access just a smartphone away, it feels like a convenient excuse. If the tests show a strong correlation towards college achievement, use them for all students, and if not, get rid of them. Making the tests seem required for some and optional for others is just not a good look (just my personal opinion).

The SAT subject tests seem to be destined to the fate of the dinosaurs. Only 17 schools require or recommend the SAT Subject Tests today and a couple of those (McGill and Duke) will take the ACT without needing a SAT Subject Test. Avoiding taking a SAT Subject test has become easier to do, and it is another way that elite college admissions policies have recently tilted to reach institutional objectives.

3979

It’s 2019. The requirements for universities from their official websites are 1 click away. Most guides on applying to college also mention the subject tests. If someone doesn’t know about SAT subject tests, chances are that they didn’t bother to do any research on applying to college. Filtering out people who didn’t have the initiative to do this research would be a feature not a bug.

3980

Every time an objective metric like a subject test requirement is eliminated, wealthy but weak students breathe another sigh of relief. None of this is about helping underprivileged kids. None.