Why you should attend the University of Luck
Much hullabaloo out of America. Apparently elite universities can no long engage in "Affirmative Action". How can they now admit a balanced and fair selection of the population0?
My suggestion is, as always, sortition.
Let me explain.
Most top flight universities around the world have the same problem. They have space for 100 students on a specific course. 15,000 apply. How do they select the best-of-the-best-of-the-best?
My answer is - they don't.
They should ignore extra-curricular activities (that tend to bias against poorer students with less free time). Ignore admission essays (which aren't read, can be creepily voyeuristic, and are probably copy-and-pasted). Ignore relevant experience (which, again, can be bought). Ignore whether someone's parents went to the school. Ignore skin colour, religion, sex, gender, credit-score, disability, and everything else1.
Instead, find a somewhat objective method of filtering the applicants. For example, academic score2.
Take the names of everyone who meets or exceeds the minimum standard. Put the names in a big hat. Pull out 100 names. Those are the students you admit. Done.
It's cheap, fast, simple, and - above all - ruthlessly fair.
Everyone who is eligible has an equal chance of getting in. There can be no accusations of favouritism, discrimination, or bias. You can't easily buy your way in, seduce a professor, or concoct a heartbreaking story to sway someone's opinion.
Might the university miss out on some stellar minds it really wanted to attract? Probably. Will the university get some undiscovered gems they would have otherwise overlooked? Probably. If the people applying to the university are broadly representative of society, will the intake be similarly representative? Probably.
I am convinced that sortition (and similar stochastic methods) are fairer way of apportioning limited resources.
Now, before you go storming my barricades, I'd like to point out some of the flaws in this scheme.
Firstly, it's probably impossible to set a completely unbiased academic test. Even if it were, rich students have always been able to buy academic tutors to help them cram and study. It also means that a talented student who has, for example, gone to a crappy school or suffered a bereavement may not meet the strict criteria.
Secondly, if a university puts all of its recruitment efforts into specific schools - they're the ones more likely to apply. For example, if they never bother to send recruiters to schools for blind students, will those students be less inclined to apply?
Thirdly, what happens if one year the admissions happen to be 100% white students? Or zero disabled students make the cut? Or no men get in? That is statistically likely to happen. And our monkey-brains just aren't wired to cope with that level of "unfairness". We want randomness which looks random, not that is random.
Fourthly, how does this make students feel? If you know that you only got in because of a dice roll, what does that do to your mental health? I notice no one seems to worry about the psychology of students who only got in on a legacy basis. Or a sports scholarship. But I think sortition might cause both imposter syndrome in those that get in, and hostility in those that didn't.
Perhaps the answer is a mix of traditional and sortition? Give 50% of places to those who can write a good essay, wow an interviewer, and generally play the game. Give the remaining 50% to those students who are merely lucky.
Of course, unless such admissions data are kept secret, there will always be an uncomfortable urge to find out who "deserved" their place and who were merely lucky. Which, in a sense, is no different to how opponents of Affirmative Action feel today.
I am quite serious about this. It seems that the two options facing universities - and anyone else oversubscribed with qualified candidates - are either to fully embrace your current discriminatory practices, or just pick people at random.
I'll go one step further. If US universities are no longer allowed to positively discriminate on any characteristic, then it strikes me that a lottery of qualified applicants is the only solution they can adopt.
I don't think the world is ready for this level of cold-blooded rationality. We want processes which feel fair and which appear reward deserving students. Sortition asks us to confront the very real possibility that our traditions and processes have been monumentally unfair.
-
Assuming, of course, that is what they wish to do. ↩︎
-
Yes, you can already see that this could be problematic. But is there a line between a reasonable adjustment and positive discrimination? ↩︎
-
This isn't perfect, and certainly isn't neutral. It might bias against people who are gifted academically but suck at exams. Or show bias towards those who are assessed favourably by some other condition. ↩︎
ethan said on mastodon.xyz:
@Edent I’ll try to find the anecdote (I might have the details wrong) but a bit ago (10 years?) a US university got in trouble when someone saw them sort the middle three quintiles (top 20% got in, bottom 20% excluded) using the staircase method - dropped the stack of applications on a staircase and chose two steps at random for further review The others were to be rejected. ai think maybe good allocation of resources but really bad feeling for rejected students in this publicized case.
Jill the Pill said on nutmeg.social:
@Edent this isn’t that far off from (some?) Canadian universities’ admissions systems: no recs, no extracurriculars, no essay, just make the grade and score cutoff until they run out of spaces.
Shih Ching Fu said on bayes.club:
@Edent This is how the Statistical Society of Australia (SSA) distributes it's four PhD/Masters Top-up #Scholarships each year (https://statsoc.org.au/top_ups). The application process is not very onerous and there's some stratification by gender. I think these were introduced when @aidybarnett was SSA President. Full disclosure, I am a happy recipient of one of these scholarships. #StatSocAu #RandomSampling randomSampling scholarships statsocau HDR top ups
Jeff Kaufman said on mastodon.mit.edu:
@Edent another benefit of admitting (some) students randomly is that we can determine causality: how much does being admitted to Harvard actually help you, vs Harvard picking the students who were going to do well regardless?
michiel@social.tchncs.de said on social.tchncs.de:
@Edent https://meridian.allenpress.com/jgme/article/13/5/612/472072/Rationales-for-a-Lottery-Among-the-Qualified-to
^ Dutch Medical schools have been doing this since 1972 (with a brief gap between 2017 and 2023, it seems) Rationales for a Lottery Among the Qualified to Select Medical Trainees: Decades of Dutch Experience
Stefan says:
"Secondly, if a university puts all of its recruitment efforts into specific schools - they're the ones more likely to apply. "
Just apply second level sortition - randomly allocate places to schools, and then select individual students (by whatever method) from those schools.
Not an idea original to me - I came across it years ago, where the argument was that it reverses some of the incentives to game secondary school choices.
More comments on Mastodon.