Thoughts on high-stakes college admissions

Maybe they're bad

The thesis

1. I wouldn’t suggest literally dismantling Harvard. (Caution is advised before destroying your most successful institutions.) My real thesis is more like:

   College admissions are (1) highly competitive and (2) consequential. Maybe those alone are bad? Maybe we should think about them instead of exclusively arguing about admissions criteria?

   When I tried to write that down in an even-handed way, I kept getting tangled up in counterarguments, which is why I resorted in the last post to having the evil twin write a primal scream. Thankfully, most of you interpreted as close to the real thesis anyway ♡.

2. I only half-believe that high-stakes college admissions are bad. But I think it’s weird that this view is so absent from public debate.

3. Any critique of college admissions should acknowledge one central fact: Most of the reason Ivy grads do so well is not because they learn so much in college, nor because they make powerful friends in college, nor because Ivy degrees have magical gold star signaling power. Mostly it’s that Ivy admits are smart and rich and would do well no matter what.

   If you took the accepted classes at Harvard and William & Mary and swapped the schools they attended, the original Harvard accepts would still do better. (I use William & Mary not to pick on it but because it is an excellent state school.)

4. But college admissions do matter. Here’s a graph that compares the percentage of Ivy+ grads in different jobs to the causal impact of going to an Ivy+ school on the odds of getting those jobs, as estimated by Chetty et al. The fourth blue dot shows that 8.1% of people in the top 1% of incomes went to Ivy+ schools, and that going to an Ivy+ school increases your odds of reaching the top 1% by a factor of 1.44 (relative to a still-very-selective state school). The orange dots are a totally unreliable linear extrapolation for things where there isn’t enough data to directly estimate causality.

   

   If college admissions double your chance of becoming a senator, then that’s worth worrying about, no?

5. It’s not just the Ivies. Arizona State admits more people than all the Ivy schools combined. If you think high-stakes college admissions are bad, then probably you should worry at least as much about who gets into Michigan (rejects 80%) and UT Austin (rejects 71%). We don’t really know how much it matters if you go to Michigan instead of Lake Superior State because everyone studies the Ivies because everyone is obsessed with the bloody Ivies. But common sense says it probably matters a lot.

Counterarguments to that thesis

1. Several people suggested that it’s pointless to try to stop college admissions from awarding gold stars because elites will always find some signaling mechanism.

   I’m sure that’s true. But if elite signaling is bad, then isn’t “elite signaling Whac-A-Mole” a game worth playing? It’s probably impossible to eradicate littering, but we still try to reduce it.

   Or—more relevant—consider the Anglosphere norm to not include pictures on resumes. Does this eliminate discrimination based on appearance? No. But surely it helps. Why is elite signaling so uniquely intractable?

2. Another counter (e.g. G2F4E6E7E8) was that high-stakes college admissions are good because if we didn’t have this explicit signaling mechanism, then people would rely on implicit signaling mechanisms like class or parental connections or your accent or manners. College admissions might be bad but it’s not as bad as that.

   My thesis implicitly assumes that markets work—that the “default” is that people are evaluated based on their performance and colleges are screwing that up by coming in and awarding gold stars.

   But maybe that’s wrong! Maybe society is so broken and un-meritocratic that the only way to even approximate justice is for powerful institutions to confer status and exclusivity to the right people.

   Not sure I agree, but it’s a sobering thought, and I was enlightened to see it make explicit.

3. The polar opposite view is that I’m right that the current system is flawed and:

   This is a solved problem, the solution is called markets. Education is a private good, resources are scarce, markets will take it from there. (Mr T.)
   

   Maybe our problem is thinking that education is a “public good” that benefits everyone like a clean environment or national defense. In reality, it’s a “private good” that benefits only the people who get it like oil or pony rides. When private goods are scarce, we usually let people bid for them so the people who want them the most will get them.

   I guess this view would imply eliminating subsidies and tax-exemption for colleges and having them charge market-rate prices. (If you auctioned off the last slot at Harvard, what would it go for?)

4. A third counter is that the current admissions for elite universities are good because they mix the meritorious with the privileged. Maybe Harvard’s causal impact comes from (somewhat less rich) eggheads hanging out with (somewhat less brilliant) elites. The eggheads get connections and the elites get to look like they deserve their privilege. If you got rid of Harvard, you’d hurt the eggheads without much impacting anyone else and overall meritocracy would go down.

   I feel like this is somewhat contradicted by the data which suggests that largely the eggheads and the elites are the same people. And I also think it’s hard to square this with the fact that elites do care about where they go to college, suggesting they would be impacted. But still, maybe there’s some truth in it.

   But if this argument is right, doesn’t that mean we should keep legacy admissions? Or even expand them and reserve some fraction of slots for the richest students? Should we do this at Michigan, too? What’s the right percentage?

5. A final counterargument is that Harvard is good because in allows the smartest students study with each other. The idea is that teachers are less important than who you go to college with. (It’s all those late-night Category Theory parties and Hegel debates that help you reach the top 1%.) If society benefits from having the strongest students together, then we have no choice but to take our best guess about everyone’s potential when they are 18 so we can assemble the best cohort.

   I think this is a pretty strong objection. But if it’s right, why just do it 18? Shouldn’t we also have strong elitism for students in high-school and middle-school, too? Some people are certainly willing to bite that bullet, but given that this isn’t how things work now, I suspect most wouldn’t.

6. It’s hard not to think that there’s some status-quo bias in defending the current regime. I strongly agree that we should be cautious and avoid abrupt sweeping changes. (For all policy changes in all domains, please? OK good talk.) But we could experiment without strangling the golden goose.

Subtleties

1. Our current culture says meritocracy is, by definition, good. But why?

   Say there’s no college so everyone enters the workforce at 18 and rises in the world based solely on their performance. Who wins? I guess people who (a) work hard and (b) were lucky enough to have had stable childhoods, good educations before 18, or just be born smart. Besides work ethic, it’s hard to argue any of that means they “deserve” it.

   The utilitarian argument for meritocracy isn’t “justice” but the idea that rewarding high performers means the most effective people are placed so they contribute the most to society. Or at least that trying to “fix” the injustice would do more harm than good. I mostly buy that, but it’s surprisingly tenuous, and I’m surprised meritocracy doesn’t see more direct attack.

2. In pre-college education, the world is trending towards “fluid gating”. For example, Singapore currently divides high-school students into three streams. But starting in 2024, this will be reformed so each subject has three different bands, and students can take a mixture and move between tracks. This seems like a step away from “sorting students by quality” and towards “providing experiences to maximally benefit the student”.

3. Everything is a matter of degree. Canadians are happy if they get into U. Toronto. Australians are happy if they get into U. Melbourne. New Zealanders are happy if they get into U. Auckland. But this doesn’t seem to be the kind of obsession that Americans have with Harvard or Indians have with IIT or Britons have with Cambrford or Chinese have with Tsinghua. (True? Tell me about other countries.) If we want elitism in college, what’s the right amount?

4. Even if college admissions have a strong causal impact, why blame colleges? Why not blame the people who use the signals (all of society?) rather than the ones creating them?

   From a utilitarian view, blame is irrelevant. The only question is what interventions exist. Surely there are other places this could be addressed. But if you want to attack the influence of college admissions, where is the obvious centralized point of weakness?

Asides

1. Yes, I know that country clubs can be non-profits. In some sense, anything can be non-profit—just don’t earn any profits! But they aren’t tax-exempt in the way that Harvard is. Your non-profit country club pays property taxes and isn’t allowed to operate a $50 billion dollar hedge fund tax-free because that would be unrelated business income.

2. I argued that high-stakes college admissions are bad by analogy to handing out Executive Platinum driver’s licenses with higher speed limits. But maybe we should have graduated driver’s licenses (sugarpile).

3. EDU-GATTACA (h/t PolymorphicWetware)

Mechanisms

1. So if you agree that it’s bad that colleges serve as gating functions—if—then what to do? This is an important question. Arguably it’s the only question and anything else is building a castle in the sky.

   I don’t propose to know! I wish people would think about this more. But there are a few possible mechanisms.

2. The simplest would be to simply make Harvard bigger. This would both (a) give the gold star to more people and (b) deflate its value. Both good!

   (Incidentally, why hasn’t Harvard already tried much harder to become bigger? The obvious explanations all seem kind of damning.)

3. Another mechanism would be to expand transfers. Part of my objection to high-stakes college admissions is that this amounts to handing out a gold star at 18 even though people mostly study the same stuff at different colleges, so if you had to do gold starts, there’s no structural reason it couldn’t wait until 22. But if people could transfer between different colleges, then at least there it would be possible for someone to distinguish themselves early on and then “upgrade”. But guess how many transfer students Harvard accepts every year? Zero! Just kidding, it’s 12 to 17.

4. Another mechanism is: Allow employers to use standardized tests in hiring (Bostonian, Sebastian Jensen). The theory here is that the reason college admissions are so important is that it’s illegal for employers to use IQ tests in hiring, but college admissions (used to?) use standardized tests that amount to IQ tests, so employers use them as a proxy for IQ.

   I’m skeptical. For one thing, as far as I can tell it isn’t illegal to use IQ tests now, not exactly. Certainly, there’s no law to that effect. But there is a Supreme Court ruling (Griggs v. Duke Power Co.) that says employers can’t have any job requirements that (a) do not directly pertain to the applicants’ ability to do the job, and (b) would have a disparate impact on protected classes. It’s unclear exactly what that prohibits, because it’s super vague and so you only find out the rules after someone sues you and you lose in court. So I think a lot of employers stay away from “IQ tests” for safety.

   But there are already various tests (e.g. leetcode) that are surely IQ-loaded. And if employers really wanted to hire on the basis of IQ, wouldn’t they use more of those and push the limits of this ruling much harder? Are IQ tests heavily used in other countries?

   I think the fact that IQ-like-tests aren’t more popular now shows that employers don’t think they contain that much information. Say you make laminated cardboard products and you want a new laminated cardboard project manager. Across the population, will someone’s performance be correlated with how they score on an IQ-test? Probably. But there are lots of other factors like interpersonal skills and actually knowing stuff about laminated cardboard. Once you know how well people have done managing cardboard projects at other companies, how much signal is left in the IQ test? My guess is not much.

5. My favorite mechanism is sortition. If Harvard really wants to be in the business of educating rather than handing out gold stars, they could do this:

   • Establish some absolute threshold for “has the ability to succeed at Harvard”.

   • Take everyone who applies, and classify them as above or below that threshold.

   • Among the people above the threshold, send out acceptances completely at random.

   • Mail everyone else a fancy framed “You Were Admitted to Harvard but lost the lottery” certificate.

   This would diminish the value of the gold stars with no need to build any new dorms! You can debate exactly how high that threshold should be, but currently 98% of Harvard admits graduate. How far can you go down the list before that drops to 95%? I’d think pretty far.