A new (and larger) Chetty study on elite college admissions is released today

Parents Forum
871

Actually, there was a confusing moment during the podcast where Brenzel talked about his prior experience as head of the Yale Alumni Association immediately before his appointment as Dean of Admissions (around 38:00). A lot of alumni certainly hoped he had a “personal dog” in the fight.

Which is interesting because that was precisely one of the reasons Wesleyan’s president gave for eliminating the legacy boost at his university:

Legacy status played a “negligible role” in admissions, Michael S. Roth, Wesleyan’s president, said in an interview. But, he added, the practice was becoming a distraction and “a sign of unfairness to the outside world.” Wesleyan University Ends Legacy Admissions - The New York Times (nytimes.com)

4 Likes
872

The bump that legacies got in this study is astounding. The fact that they got into colleges in this rarified sample where they are not legacies at a rate 32% higher than the field also upends what has been so confidently assumed by so many about their qualifications.

5 Likes
873

I’ve always assumed that most legacies are what’s called “Average Excellent” here on CC. In other words, the sort of nicely accomplished student who has a negligible chance of admission to an Ivy+ if unhooked.

1 Like
874

So Figure A.11(c) suggests if you look at legacy admits, their model indicated about 1/3rd of them would still be admitted to that same college, or indeed another Ivy+. If you look at legacy applicants, it is about 1/8th (note about 5/8ths of legacy applicants were not admitted, so that leaves 3/8ths admitted, a third of which is 1/8th).

I’m not sure what everyone means by “average excellent,” but generally speaking that is consistent with other things we learned in the Harvard lawsuit. Basically, if you had an overall Harvard 2 and were unhooked, you had like a 65% chance of admission. If you had Harvard 2s for academics and activities/athletics (I think a fair interpretation of “average excellent”), you would likely still need a Personal 2 to get an overall 2.

About 21% of applicants got a Personal 2. Crudely, .21 * .65 is 13.7%, so close enough to 1/8th for these purposes.

Meaning this is consistent with unhooked “average excellent” applicants having around the same admit rate as legacy applicants had at other Ivy+ colleges, or in the model where legacy wasn’t a hook.

Or to put it the other way, if unhooked “average excellent” applicants had materially less than a 1/8th admission chance to an Ivy+, then legacies would appear to be skewing a bit higher than “average excellent”. But that is really a question of how you define “average excellent”.

875

Screenshot 2023-08-04 at 14.52.42
Screenshot 2023-08-04 at 14.52.421070×1078 85.8 KB

I’m more fascinated about this graph.

SAT score of Ivy-plus matriculants was much more correlated with career success/elite graduate schools/prestigious firms than GPA of Ivy-plus matriculants.

I also strongly suspect the same would apply to leadership in society more generally.

So are Ivy League schools taking the wrong approach by making the SAT/ACT test optional when it’s clear it’s a much better predictor of success (in the way those schools tend to define success) than GPA?

1 Like
876

@Lindagaf coined this term back when her child was going through the college admissions process. Here is the original thread: Truthful advice about getting into top colleges, for your "average" excellent student.

1 Like
877

So as I understand it, most US admissions officers would agree they need to do a LOT of work to normalize GPAs, not least among the most competitive applicants. That is why they need full transcripts, school reports, and counselor reports, part of why they try to assign reviewers familiar with various high schools, and so on. All that is required because our curriculum and evaluation standards as reflected in HS GPAs is so inconsistent.

Given that, I am not sure it is a surprise HS GPA appears to have little or no information in it as such. What I think you would want is whatever sort of academic score a college gave after factoring in all that other stuff, minus whatever role test scores and other non-grade factors played–which sounds like a measure not easily accessible.

878

Yes, I was familiar with that post (not the whole thread) from 2016.

But I think it was in 2018 that there was a public release of the discovery information and expert reports in the Harvard lawsuit. From my perspective, that allowed us to get a much better idea of what actually explained Harvard admissions, and likely holistic admissions at many similar colleges. Among other things, I think it was a surprise to many what a large role the Personal factor was playing in determining unhooked admissions, and conversely how difficult it was to make up for a merely “generally positive” Personal score with some sort of academic or activities “hook”.

And the study we are looking at in this thread was, of course, done by some of the Harvard lawsuit experts.

So, sitting here in 2023, trying to figure out how “average excellent” maps onto the analysis in this study–I am not sure what that means today.

Specifically, as I was suggesting before, there are one of two notable options given the post-2018 way of understanding holistic admissions at colleges like Harvard:

(A) “Average excellent” means unhooked applicants who have very good academics and very good activities/athletic who have a good chance of admissions if, but only if, they also get a very strong Personal/Fit score; or

(B) “Average excellent” means unhooked applicants who do not have a good chance of admissions even if they do get a very strong Personal/Fit score, presumably because they fell short of very good either academically, or in terms of activities/athletics, or both.

And my point is if (A) is the definition, then the legacy admit rates at other Ivy+ and in the model without legacy preferences are consistent with legacy applicants typically being “average excellent”.

If instead (B) is the definition, then the legacy admit rates in the above senses are too high, and legacy applicants must therefore skew better than “average excellent”.

881

Then the argument becomes… why get rid of standardized tests?

Surely this is the most effective measure of normalizing GPAs?

Why get rid of the main measure that helps standardize things?

It’s the non-academic scoring that’s causing the disparity in acceptance rate among equally qualified applicants of different income levels, not the test scores.

If schools are arguing that test scores privilege the rich, it’s clear from this study that it’s actually non-academic ratings that privilege the rich the most.

1 Like
882

Perhaps because some of those elite employers screen college graduate applicants by SAT scores?

1 Like
883

That’s because:

A. Family income is strongly correlated to admission, and much more importantly, to the ability to afford prestigious business and law programs. Income is more highly correlated with test scores than with GPA.

B. Around 40% of the students who matriculate to Ivy+ colleges attended “prestigious” private high schools. The grading policies and practices of these high schools vary a lot, meaning that the correlation that Chetty et al. is attempting between HSGPA and anything else really is not insightful.

C. “Predicted top 1%” is not a useful metric, since the prediction is a function of graduate school attended and the company that they work for. So it is nothing more than a modified version of the two graphs below it, not an additional independent measure of “success”.

D. The definition of “prestigious” employers is based on the number of Ivy+ graduates who are employed there. So this is not an indication that students whose SAT scores were higher had a higher chance of success, just that they had a higher chances of working at the firms for which Ivy+ schools are feeders.

Finally, I am sorry, but “working at a wealthy financial firm/wealthy law firm which employs a lot of Ivy+ graduates” is a VERY VERY VERY narrow and limited definition of “success”.

In short - all that these graphs demonstrate, and all that they are meant to demonstrate, is that family wealth not only helps kids be accepted to Ivy+ colleges, but also continues to help them later in life. This is especially true for students who choose careers in law and business.

6 Likes
884

And elite graduate schools heavily admit by test score. People who are terrific at the SAT tend to do well at the LSAT, GRE, GMAT, etc.

3 Likes
885

Finally, I am sorry, but “working at a wealthy financial firm/wealthy law firm which employs a lot of Ivy+ graduates” is a VERY VERY VERY narrow and limited definition of “success”.

I mean surely looking at career outcomes of elite schools is what elite schools would define as success. That seems like the most accurate definition of what success is (from the perspective of the elite schools themselves).

Put simply, elite schools disproportionately send their graduates to financial firms, law firms and tech firms. Elite schools know they do and they don’t particularly attempt to change the student body such that different career outcomes exist.

That would seem to suggest that they’re pretty happy with the careers that those kids go into.

But I take your point.

So let’s take leadership for example which would be a metric that isn’t to do with career income necessarily.

Harvard says they produce leaders and that’s their aim - does anyone have data showing leadership outcomes by high school GPA vs SAT/ACT score?

Is income also correlated with leadership?

all that they are meant to demonstrate, is that family wealth not only helps kids be accepted to Ivy+ colleges, but also continues to help them later in life. This is especially true for students who choose careers in law and business.

So what about other metrics then? What about in terms of leadership of society?

I wonder whether there’s a link between wealth and leadership outcomes?

But in any case, the question still remains, why remove test scores at all?

Anecdotally, the leadership at both the American companies I’ve worked at (Investment Banking → Asset Management) with went to elite schools but they definitely don’t reflect the demographics of elite schools.

Which is why I’d be curious to see leadership outcomes among Ivy League schools and whether there’s a correlation between income and leadership/other ‘success’ metrics.

Even the leadership of universities isn’t representative of elite school student bodies in terms of gender and other demographics.

886

Never heard of elite employers asking college grads for their SAT scores. Certainly not seen it in cases I am personally familiar with.

Could it be that standardized test scores simply measure something that better correlates with success at elite job interviews years later than high-school GPAs that lack any common yardstick?

887

Goldman Sachs, Bain, McKinsey: Job Candidates SAT Scores suggests that the practice was common, at least until relatively recently.

1 Like
888

“Jonathan Wai, an intelligence expert and researcher in Duke University’s Talent Identification Program (TIP), says the SATs are considered to be a measure of “general intelligence and general ability.””

Anathema! No wonder Duke TIP got shut down.

889

So the colleges can attract more applications from the students who fit their institutional priorities…URMs, limited income, first gen, other disadvantaged students. This is simple. Don’t overthink it.

5 Likes
890

I think most AOs would disagree as to the SAT and ACT, or any other general aptitude test. Even APs are pretty weak because they are not particularly sophisticated and individual high schools can and do argue their advanced courses are more rigorous than AP courses. Things like the IB HL system or UK A-Level system are really what you need to have easily comparable HS grades, and we don’t have those systems in the United States.

Other posters have pointed out that is not really what the study suggests, and that often people who think that have failed to realize they are looking at something where test scores were part of the control.

Personally, I think it is a fool’s errand to try to disentangle which common US college admissions factors favor high socioeconomic status applicants “the most”. They all do, and they all work together. And high socioeconomic status families get great advice on all that, and typically take a similarly holistic strategy to maximizing their kids’ chances, including careful evaluation of relative marginal benefits of different ways of spending their money to help their kids in the context of their kids’ overall admissions situation.

Given this, even if you could, say, figure out how much standardized testing was helping high SES families in the past, that wouldn’t tell you what would happen if you changed the system to make standardized testing even more valuable at the margins. Because if that happened, then high SES families would devote even more resources to maximizing their kids’ standardized test scores. If the standardized testing companies tried to maintain the same score distributions, this would likely force down the scores of many applicants from non-high-SES families. The marginal value of what the high SES families were doing in terms of standardized testing would then go up.

In other words, I think it is hard to just tweak our system and expect it not to be a form of whack-a-mole where even if you drove down the high-SES advantage in one area, it would not just reappear in another.

If you were actually serious about this, you’d really have to completely undo the whole system of privilege that starts from before birth and continues up to when kids submit an application. There would have to be no more local control of schools, no more private K-12s, universal early childhood and summer programs, and so on.

If you instead just fiddle with the end parameters, I am confident that system of privilege will easily adjust.

4 Likes
891

These are not aptitude tests, nor IQ tests. They are merely tests that test what a student has learned…and we know that because K-12 education is sorely lacking in the US, students don’t learn the same things, and some students don’t have the same depth and breadth of learning as compared to others. That doesn’t mean these students don’t have tons of academic potential. AOs (and many others) know that test scores are not necessarily correlated to one’s academic potential, and there is plenty of data that support that.

Multiple colleges (DePaul, Ithaca, Bates) have published data showing students admitted without test scores (to their schools) have similar GPAs and graduation rates as compared to those admitted with test scores. I know these aren’t Ivy/Ivy Plus schools, but these schools will have that data sooner rather than later, and I hope they publish it.

892

Actually GRE scores at least became optional during the pandemic the same time as SAT/ACT were and many graduate programs, including elite ones, have either continued to make them optional or not considered. The same complaints about bias leveled against the SAT/ACT have also been levelled against the GRE and many programs are weighing them less or not using them at all with a lense towards DEI.

3 Likes