- EphBlog - http://ephblog.com -

Harvard Admissions Trial, 4

See The Wall Street Journal for background on the Harvard admissions trial which starts this week. Best commentary is from Razib Khan at Gene Expression. Should we provide daily coverage, connecting news from the courtroom to EphBlog’s coverage of admissions issues at Williams over the last 15 years? In the meantime, let’s spend this week reviewing some of aspects of the debate. Day 4.

Yesterday, I wrote (slight edit):

If admitted students in a category, like legacies, have similar academic qualifications than other students in the class then, almost by definition, they did not received a large advantage in admissions.

This seems obvious to me. Now, a nihilist might disagree, might claim that the “holistic” admissions that Harvard and Williams practice make it impossible to compare categories. The galaxy brain of Admissions Director Sulgi Kim ’06 is a multi-splendored emerald of infinite complexity. Mere mortals can no more discern its inner workings than dogs can interpret the movements of the heavenly spheres. I think this position is nuts — and both sides of the Harvard case disagree with it, hence all the regression modelling and expert testimony — but it is logically consistent.

abl, one of EphBlog’s best writers, disagrees with me about legacies:

That’s not true. If literally every legacy applicant was an AR1 and AR2, and legacies were guaranteed admission, you would simultaneously see (1) that legacies have, on average, higher academic credentials than the average student (because the average student at Williams is not an AR1 / AR2); (2) that legacies received a large boost in admissions (because a 100% acceptance rate would represent a very significant advantage for AR1 and AR2 applicants).

[A]ny proportionate (or disproportionately high) number of legacy high performers in the class does not imply that legacy applicants are given no meaningful advantage in the process.

abl continues:

My point is that, without inside knowledge of the process, or without very good knowledge of the applicant pool and relative admit rates, it’s difficult to look at the class of admitted students and draw these sorts of inferences about the advantage of applying as a legacy student.

This is dangerously close to the Sulgi Kim Inestimable Galaxy Brain view of admissions. I have (almost) no “inside knowledge” of the process and not “very good knowledge” of the applicant pool. Almost all my information comes from public statements by college officials. Are you claiming that I can’t know — without inside knowledge — that Williams provides significant advantages to recruited athletes and African-Americans? And, if you agree that I can make those judgments, just what is stopping me from making a judgment, using the same tools and analysis, about the lack of advantages given to legacy applicants?

abl is certainly correct (and he uses a similar example later in his excellent comment) that weird stuff might be happening behind the scenes. There are 55 legacies in the class of 2021. Perhaps 35 are AR1 geniuses, applicants who were also accepted at HYPS but turned them down to come to Williams. The other 20 are AR4 chuckleheads, who never would have gotten in if it were not for their legacy status. This scenario would create a group (55 legacies) with similar academic credentials to the class as a whole, but it would still be the case that 20 of them received a “meaningful” admissions advantage (and 35 did not).

I agree that this is possible, but it also strikes me as highly unlikely. The world is a continuous place.

Facebooktwitter
Comments Disabled (Open | Close)

Comments Disabled To "Harvard Admissions Trial, 4"

#1 Comment By Fendertweed On October 18, 2018 @ 10:37 am

What does the repeat ad hominem snark have to do with the point – whatever it is? … “galaxy brain” … wtf?

#2 Comment By David Dudley Field ’25 On October 18, 2018 @ 11:21 am

> repeat ad hominem snark

That is not ad hominem. EphBlog loves Sulgi and is glad that she runs admissions.

I am making fun of the attitude, not necessarily expressed by abl, but not uncommon in these discussions, that “holistic” admissions is so complex and difficult that outsiders can’t really understand what is going on.

#3 Comment By Whitney Wilson ’90 On October 18, 2018 @ 1:50 pm

I am making fun of the attitude, not necessarily expressed by abl, but not uncommon in these discussions, that “holistic” admissions is so complex and difficult that outsiders can’t really understand what is going on.

Its not so much that the concept or even the application of holistic admissions is so tough, its that outsiders in general don’t have all of the information that the admissions committee does, so discerning their thinking can be problematic.

#4 Comment By frank uible On October 18, 2018 @ 6:20 pm

Holistic admissions? As Harry Caray would have said, Holy Cow!

#5 Comment By David Dudley Field ’25 On October 19, 2018 @ 5:24 am

> discerning their thinking can be problematic.

Which us why you have EphBlog!

#6 Comment By abl On October 19, 2018 @ 9:37 pm

I have (almost) no “inside knowledge” of the process and not “very good knowledge” of the applicant pool. Almost all my information comes from public statements by college officials. Are you claiming that I can’t know — without inside knowledge — that Williams provides significant advantages to recruited athletes and African-Americans?

No — in part because discerning a large advantage is obviously going to be easier than discerning a small advantage, which will be easier than discerning a lack of a small advantage–and in part because there is a lot of public information about the significant advantage that being a recruited athlete or African American confers in the admissions process.

And, if you agree that I can make those judgments, just what is stopping me from making a judgment, using the same tools and analysis, about the lack of advantages given to legacy applicants?

See above. It is basically always going to be the case that there are certain inferences that cannot be drawn from a given limited data set. You have no epistemic entitlement to being able to draw legitimate inferences about small and complicated effects (or the lack thereof) from a partial and limited data set. That doesn’t leave us with the “Sulgi Kim Inestimable Galaxy Brain view of admissions”–even if admissions worked as a perfectly horizontally consistent mechanical process not in any way beholden to the individual whims of adcoms (it doesn’t), you would still lack the information you need here to draw the sort of inferences that you want to draw.

Finally, my point wasn’t “that weird stuff might be happening behind the scenes,” but that stuff that is not particularly weird and that you and I agree on is almost certainly happening behind the scenes. You don’t have to invent a strange hypothetical, as you do, to end up with a relatively academically strong class of legacies who also receive a meaningful admissions boost. If, as you and I both believe, legacies represent a stronger-than-average group of applicants; and if, as you and I both believe, AR1 and AR2 legacies are more likely than average to matriculate to Williams than the average AR1 and AR2 applicant, then you would expect to have a class of legacies with stronger-than-average academic credentials irrespective of any legacy admissions advantage or lack thereof.

Here’s another way to think about it, using entirely invented but, I think, reasonable numbers. It’s easy to imagine that the average legacy applicant to Williams scores 30 points higher on average on the SAT than the average Williams applicant. It’s, moreover, easy to imagine that the higher yield rate of legacies adds another effective 20 points to the average legacy matriculant’s SAT scores (relative to non-legacies). In a world in which there was no legacy advantage in admissions, you would expect the average legacy student at Williams to have SAT scores that are 50 points higher than the average non-legacy student at Williams. Now, imagine, that there is a legacy academic advantage in the admissions process in which legacy “counts” for approximately 30 points in the SAT (note: this is roughly the SAT difference between the Williams class and the Princeton class, I think). Because every legacy applicant will have +30 tacked on to his or her SAT score, this will result in a class of legacies who underperform (-30) on the SATs relative to what would otherwise be expected. The combined results of these factors–(a) the particularly strong stats of legacy admits (+50); and (b) a small but meaningful SAT bump for legacy applicants (-30)–is a class of legacies who have received a meaningful advantage and yet have higher-than-average academic stats (+20 SATS when compared to the class).

Obviously, this is a drastic oversimplification of the admissions process (although the complications generally work against your point), but I hope it helps illustrate why you can’t conclude that there is no meaningful legacy admissions boost on the basis of the fact that legacy matriculants score about as well as non-legacy matriculants.