Weeding out the undesirables

This New Yorker piece by Malcolm Gladwell shows once again why, despite all the hype surrounding him, Gladwell really is one of the most perceptive social observers of our time. Gladwell is reviewing The Chosen by Jerome Karabel, which relates the history of the undergraduate admissions process at Harvard, Princeton, and Yale.

Have you ever wondered why that process places so much emphasis on sports, extracurriculars, personality, “leadership,” “character,” and suchlike, as opposed to the more obvious intellectual criteria? The answer, it turns out, is that in the early 1920’s, Harvard and Co. had to find some way to limit the number of Jewish admits:

By 1922, [Jews] made up more than a fifth of Harvard’s freshman class. The administration and alumni were up in arms. Jews were thought to be sickly and grasping, grade-grubbing and insular. They displaced the sons of wealthy Wasp alumni, which did not bode well for fund-raising. A. Lawrence Lowell, Harvard’s president in the nineteen-twenties, stated flatly that too many Jews would destroy the school … Finally, Lowell — and his counterparts at Yale and Princeton — realized that if a definition of merit based on academic prowess was leading to the wrong kind of student, the solution was to change the definition of merit. Karabel argues that it was at this moment that the history and nature of the Ivy League took a significant turn.

Gladwell writes that from that point forward,

The admissions office at Harvard became much more interested in the details of an applicant’s personal life. Lowell told his admissions officers to elicit information about the “character” of candidates from “persons who know the applicants well,” and so the letter of reference became mandatory. Harvard started asking applicants to provide a photograph. Candidates had to write personal essays, demonstrating their aptitude for leadership, and list their extracurricular activities … The personal interview became a key component of admissions in order, Karabel writes, “to ensure that ‘undesirables’ were identified and to assess important but subtle indicators of background and breeding such as speech, dress, deportment and physical appearance.”

The byzantine admissions process that Harvard and the other name-brand schools set up pretty much remains in place to this day. (I still remember the “Potpourri” section of the Princeton application, which asked applicants to list their favorite movies, music, etc. It reminded me of the bridgekeeper from Monty Python and the Holy Grail: “What is your name? What is your quest? What is your favorite color?”)

Granted, the Ivy League admissions process no longer serves its original purpose, possibly because most American Jews have become so assimilated themselves. But today, the enormous preferences given to legacies, athletes, and other students of “character” work very similarly to limit the number of Asians. To which I can only say: huzzah! For as Gladwell explains: “If Harvard had too many Asians, it wouldn’t be Harvard, just as Harvard wouldn’t be Harvard with too many Jews or pansies or parlor pinks or shy types or short people with big ears.”

18 Responses to “Weeding out the undesirables”

1. Pyracantha Says:

Hi there…are you at Perimeter Institute?
I found your blog thru Michael Nielsen. I am a middle-aged artist studying physics. My own blog is ELECTRON BLUE at my Website http://www.pyracantha.com
I look forward to reading lots of interesting posts on math and physics from “Shtetl-Optimized.”

2. Osias Says:

I wonder why there are so many people complaining about the “vestibular”, here on Brazil… 🙂

3. Miss HT Psych Says:

Another possibility: it may serve as a way to weed out those of lower classes from applying. Two guys I know were competing for the biggest scholarship Queen’s University had to offer. One was a son of two doctors, the other had to work since public school to help support his family. Guess which one got the 4 years of full support from the university? The son of the two doctors. Why? He had more extracurricular activities. Queen’s didn’t care that one of them needed that money to go to school and the other didn’t. They didn’t seem to notice that one boy could do all the extra-curricular activities because he’d never had to work a day in his life. I think it’s sad that those things can be a major determinant of your ability to get into a world-class university

4. Andy Drucker Says:

Is it obvious that private colleges should select uniformly for academic/cognitive ability rather than for a more complex bundle of virtues and attributes?

Is the latter stance necessarily invalid because it has at least historically been used as a cover for racial discrimination? ‘Intelligence’ as a selection concept has at least as murky a history; for an amusing example see
http://historymatters.gmu.edu/d/5293

What if a plausible set of indicators for a defensible list of virtues happens to show some race or class bias? E.g. community service seems to speak well of a person, yet participation has opportunity costs and is a privilege. (Well, so is SAT prep…)

Perhaps because I’m coming from CS, my tendency is to think that tuning the weights assigned to these indicators in the admissions process is a hard optimization problem among multiple social values, and formal guarantees of *real* equal-opportunity are both chimeric and undesirable. That doesn’t mean admission policies are above criticism (and I accept the Queen’s U critique, on an anecdotal basis); but it’s not enough to identify a negative statistical tendency in the process, e.g. class selection; one has to argue that this outweighs the positive effects of selection for merit or ‘virtue’, or to argue that these can be selected for in a way that cuts less across race/class lines..

Take community service. One might try to deflate high school service somewhat as a signal of ‘virtue’, defined say as likelihood of ‘helping the community’ after graduation. If it’s rewarded in the college admissions process, then plenty of nonvirtuous students, especially well-off ones with time, will play the game with no intention of making it a way of life.

On the other hand, truly virtuous students will also tend to participate, both because it’s what they do and because it’s the most natural and rewarding way to build their resumé. So service as a selection instrument has partial validity, albeit with a built-in class bias. As a side benefit, service helps its targets, and might raise the virtue level of its participants.

So… what? There must be empirical studies on service that could help. But there remains a question related to the previous post: if extenuating circumstances like having to work serious hours give one no opportunity to produce extracurricular ‘virtue signals’, should admissions assume you are of ‘average virtue’, or give you more benefit of the doubt as a fairness practice, or is excellence in the face of hardship itself a virtue signal?

Sorry this was so long.. looking forward to your posts, Scott.

5. Anonymous Says:

Your expression “Huzzah!” has the definition:

Main Entry: huz·zah
Variant(s): or huz·za /(“)h&-‘zä/
Function: noun
Etymology: origin unknown
: an expression or shout of acclaim — often used interjectionally to express joy or approbation

— why are you happy about the number of Asians being limited?

6. Greg Kuperberg Says:

I have to infer that your “huzzah” was sarcastic. Otherwise I don’t know what you really mean.

That aside, politics has gone full circle. At Berkeley, Asian undergraduates significantly outnumber non-Hispanic Caucasians. To the extent that Berkeley wants to boost under-represented minorities, there are now too few Caucasians left to be supplanted. (Or rather, the Caucasian applicants would have to be hit directly in order to spare the Asian applicants.) So what methods does Berkeley (and UC in general) use, given that it can’t use overt quotas? Character points, life stories, and outreach.

Personally I do not see anything wrong with admissions offices giving bonus points to the downtrodden. But I think that it would be better if they directly targeted downtroddenness as measured by family income.

7. Miss HT Psych Says:

Andy: I would think that excellence in the face of hardship would be a virtue in itself. Or that, at least, hours spent doing work unrelated to future endevours should be counted as an equal weight to time spent in extra-curricular activities.

It’s funny that you should have provided a link that mentioned the Army Alpha intelligence tests. I’ll give you one guess as to what other famous aptitude test was based on the Army Alpha test. It was (wait for it!)… the SAT’s! An equally interesting point is that the Army Alpha tests (and Army Beta, for that matter) were notoriously culturally biased against anyone who wasn’t a white American male with the privilage of having some education. Results from these tests spawned the American eugenics movement and laws limiting the amount of immigrants from many countries (mostly eastern europe). So, it begs the question if the SAT’s should even be used in admission processes as they are not an objective measure of achievement or intelligence.

8. Anonymous Says:

I recall reading that Feynman had trouble getting into Princeton because he was a Jew.

9. andy Says:

thanks miss ht, a few thoughts:

i) the SAT’s origins don’t invalidate it as it has evolved over time, just as the multidimensional selection concept for admissions has evolved beyond its anti-Semitic role.

ii) The SAT doesn’t market itself as an objective measure of intelligence. It has retreated from its claim to measure ‘aptitude’ and even ‘achievement’ (‘SAT’ now stands for.. nothing at all!). Its chief statistical claim is to predict 1st-year college grades, and it seems not do this very well, while introducing racial and class biases, see http://hypertextbook.com/eworld/sat.shtml

But this kind of prediction, if done better, seems a valid enough admissions tool, and I mainly want to ask, even supposing we had more effective predictors (call one SAT’), what do we do if there are ineliminable biases?

iii) Members of the affected groups need to know that they are facing statistical discrimination by the SAT’: it underpredicts their college performance, and they would seem within their rights to demand some corrective. If SAT’ just systematically underestimated everyone in an identifiable group G by 1 GPA unit, the corrective would be simple, fair to G, and wouldn’t compromise expected performance of the incoming class.

But say SAT’ was largely uninformative about G–just squashed in-group differences. Then (supposing we have no other comparable tests) meritocracy and fairness to G can come into conflict.

iii) This setup is aimed at identifying a type of structural dilemma which we might have to face, not making a strong case that it’s in effect now. But my (ad hoc) account of community service in the last post suggests that service records tend to behave like SAT’ with respect to measuring civic virtue: not just biased against, but also underinformative about, disadvantaged students.
(And I agree, excellence thru hardship is a virtue in itself, tho possibly of a different kind, and I find it tough to adjudicate between the two so am looking for a viewpoint that doesn’t predecide the issue for admissions offices.)

10. Anonymous Says:

I haven’t read “The Chosen” so maybe this is explained there, but why did universities feel the need to come up with arcane “admissions criteria” rather than just outright limit the number of accepted Jews per class? Were they trying to pretend to someone that their admissions criteria were unbiased? To whom?

11. Scott Says:

Greg (and anonymous): Yes, my “huzzah” was sarcastic. I apologize. People often seem to have trouble telling when I’m being sarcastic. 🙂

12. Scott Says:

pyracantha: I’m actually at the University of Waterloo, but I like hanging out at Perimeter, and apparently they’re even going to give me an office there.

13. Scott Says:

anonymous: As for why Harvard didn’t just institute a quota, Gladwell says this in his review:

“Lowell’s first idea — a quota limiting Jews to fifteen per cent of the student body — was roundly criticized. Lowell tried restricting the number of scholarships given to Jewish students, and made an effort to bring in students from public schools in the West, where there were fewer Jews. Neither strategy worked.”

14. Miss HT Psych Says:

Andy: You bring up many good points…

1) No, the SAT’s origins don’t invalidate it. However it continues to show a racial and gender gap in test scores. Although, I do recognize that the racial differences likely have more to do with access to resources (education, \$), environment, etc. With this in mind then, is it fair to ask students to subject themselves to these tests as admissions requirements if unchangeable factors work against them (gender, parents income, neighbourhood, school board, etc.)?

2) I did not realize that the Educational Testing Service had dropped the “Scholastic Aptitude/Assessment Test” name. I suppose I wrote the “old” SAT when I was thinking about going to the States for university. The “new” structure is interesting… very much like the GRE’s I had to write for admission to grad school. However, the GRE’s are not a fabulous predictor of performance either. The subject tests aren’t bad, but the general tests tend not to be great predictors.

3) You’re quite correct… there isn’t any better estimator of scholastic performance in university. The only option I would suggest is give people a battery of tests to complete, with each assessing different aspects of performance. Although, it would appear that the ETS has been trying to move in this direction with the introduction of the SAT II (subject-specific tests). It’s always difficult to apply some corrective to culturally-biased tests, because within group variance is usually quite high (in some cases, the within groups variance is higher than the between groups variances, which calls group means into question). To this problem, I have no solution… and from my research in sex differences I haven’t found any literature which has suggested one.

4) The question of extra-curricular participation is tied into this entire subject. Those that have the time to participate, usually go to better schools, have access to more resources, and also tend to perform better on achievement tests like the SAT’s (culture-bias, expecially in the US, could be viewed as class-bias and nothing more). If I was in an admissions office, I would find it very difficult to find an objective measure to distinguish between classes of students. All applicants may have some reason that makes them equally valuable to a school. In my opinion, these kinds of decisions will remain enitrely subjective. Again, I wish I had a solution…

15. Sasha G Says:

A while back the Economist did a Lexington column on extolling SAT. It seems like the exam was requested by a later Harvard president, James Bryant Conant, to introduce some measure of objectivity into admissions, and perhaps to reverse Lowell’s “innovations”.
I’m not saying that SAT is all cake.

16. Anonymous Says:

Note that the new yorker article actually justifies Harvard et al in their selection criteria, and says that it seems that one is shooting for the most successful graduates (as opposed to the academically strongest students) then such “subjective” criteria are the way to go.

17. Anonymous Says:

“Its chief statistical claim is to predict 1st-year college grades, and it seems not do this very well, while introducing racial and class biases”

Yes, the SATs have a racial bias–they overpredict black 1st year college grades.

18. Shtetl-Optimized » Blog Archive » Admissions unhooked Says:

[…] I harped about this issue in one of my very first posts, almost a year ago. I don’t know what else to say. If idealism won’t goad us Americans (yes, I’m still an American) into overhauling our crooked, anti-intellectual admissions system, then maybe it will help to see just how absurd that system looks to the rest of the world. […]