Eric Levitz in New York:
This doesn’t mean that it would be good to base admissions on the SAT alone. The test’s biggest problem is that it doesn’t predict success in college very well. But as an equity matter, the SAT is one of the least-bad metrics we’ve got.
So I think it’s really important that we get this straight, as this sort of thing has somehow gone uncorrected in the media for decades: this is just wrong. It’s an understandable mistake given the dissemination of information on this topic but it’s just wrong. The SAT’s seemingly-low correlations with college GPA are a product of systematic range restriction. When adjusting for that range restriction, we find that the SAT-college GPA correlation is robust. (Same with the ACT.) This is particularly true given the modest correlations found in all kinds of human research, which is inherently noisy. What’s intensely frustrating about this is that we’ve known about this issue for forever and yet nobody updates their understanding of the test. I wrote this piece about this issue in 2017! Range restriction is a problem we know all about, and fixing it is something we do very well. The SAT is a strong predictor of college success, period. Even in grad school, where grades are notoriously inflated, entrance exams are strong predictors of success. I find the willful ignorance about this stuff so frustrating.