Thread by @rothstein_jesse

I’m hesitant to disagree with @mattyglesias after he says such nice things about my work. But I think he gets the headline wrong. That’s especially strange because he gets the key points right. A thread. (1/17)

https://twitter.com/mattyglesias/status/1395685527467266055

Right #1: The best thing we could do to promote equity within higher education is to better fund the less selective schools, especially community colleges, where most students from lower-income and Black/Brown families go. (2/17)

Right #2: Selective colleges should grow, a lot. It would be better if highly selective colleges set a bar and admitted everyone above it, or randomly selected above the bar until big enough to take all of them.

https://twitter.com/mattyglesias/status/1395686407000125446?s=20

(3/17)

If those were the takeaway, it would be a great piece. But @mattyglesias ties this to an argument that “the anti-SAT push is misguided” that really doesn’t hold up. (4/17)

His argument: (a) SATs do measure college preparedness, (b) in our current system, we could improve our “teen sorting” by making sure all low-inc students take the SAT, (c) bad arguments are made against the SAT. All are correct. But they don’t support the conclusion. (5/17)

On (a): SATs predict college grades & everything that predicts preparedness correlates w SES. But this doesn’t show much. Magnitudes matter. SAT is *more* correlated with SES than other measures. Unlike grades, SAT especially good at capturing the SES part of preparedness. (6/17)

Aside: Predicting college grades should not be the be-all and end-all of admissions.

https://twitter.com/matt_barnum/status/1394723847182733316?s=20

(7/17)

Consider an example: I develop a new test where I interview each student and give them a score. I label it a perfect “meritocratic sorting tool for academic skill.” It is also a black box – I don’t tell you how I construct scores. (8/17)

This score very strongly predicts college grades. It is correlated with SES, just like other measures. It is only modestly coachable. And in an admissions system based on it, it would be very much equity-promoting to make sure all students have the opportunity to take it. (9/17)

Now I open the black box. It turns out that scores on this test are based entirely on how well the student knows the rules of lacrosse (except in Maryland), when to tack when sailing upwind, and the difference between béchamel and hollandaise. (10/17)

This obv. not meritocratic. SAT isn’t this bad, but it is of this type (tho the black box stays closed). It is more predictive between high- and low-SES students than within SES; low-inc students w high SATs don’t do especially well in college (but those w high GPAs do). (11/17)

Upshot: All meritocratic measures will yield an unrepresentative class. But SAT is worse than others. If sorting is only goal, just consider income directly. If we care about equity, we should prefer measures that don’t derive their predictive power by laundering SES. (12/17)

Also growing evidence (tho still suggestive) that going to a better college just doesn’t matter much for high-inc students’ prospects, and matters more for disadvantaged students. That would imply that leaning toward equity is efficient as well. (13/17)

Matt suggests “automatically admit the Top X% of SAT-scorers.” That would yield a class of rich students from elite pvt schools, and v few from low-inc schools. Very different from Top X% by class rank as in Texas or CA. (See @zbleemer CA paper: zacharybleemer.com/wp-content/upl…) (14/17)

@dynarski & others emphasize a few low-inc students who have/could get high SATs, and would be left out if SAT is dropped. But there are more low-inc students with high GPAs who can’t get high SATs and are excluded now. And they would do better in college if admitted. (15/17)

Dropping SAT won’t fix equity on its own. (Fund community colleges!) There’s no tweak to highly selective college admissions that makes it a force for equity overall. But it could be less bad than it is; dropping SAT would help at the margin. (16/17)

Some references:
My testimony to UC: eml.berkeley.edu/~jrothst/other…
Background paper: edpolicyinca.org/newsroom/uc-re…
Technical paper: dx.doi.org/10.1016/j.jeco…
Texas 10% paper: eml.berkeley.edu/~jrothst/worki…
@zbleemer paper on CA % policy: zacharybleemer.com/wp-content/upl…
(17/17)

Tagging various coauthors, though none are responsible for the above: @mkurlaender @econsarahreber @Econ_Sandy @JeffDenning