On Essays and SATs: Some Students are Just More Prepared Than Others

Whoops! It turns out that, far from being a more “holistic” measure that rewards the well-rounded, college admissions essays correlate more strongly with family income than SATs do. (This is in part a function of the fact that, found to be .25 in the largest sample I am aware of, the correlation between SAT scores and family income is vastly overstated by liberals.) The inevitable result will simply to be to kick the can further; “we need even more holistic criteria than before!” But it might be more useful to have a hard conversation.

My piece on why holistic criteria actually hurts marginalized students has been lost to the inevitable linkrot of time (edit: here!), but the point’s not particularly hard to grasp. The book Measuring Success amounts to a book-length refutation of all of the progressive myths about the SAT, and anyway, there is absolutely no reason to believe that holistic admissions aren’t subject to capture by the moneyed. Since test prep doesn’t work, family money can’t fix low SAT or ACT scores, whereas money can absolutely get a privileged teen into a summer program building houses for low-income families in the Andes (intersectional as well as holistic!), fencing lessons (decent at rare sport > great at common sport, unless it’s football or basketball), and a writing workshop (obsessive about one interest rather than casual about several, in today’s admissions game).

And since holistic criteria are so vague, you can’t do a simple correlation to see the impact of SES, like you can with the SAT and ACT. We know there is some relationship between test scores and family income precisely because those tests are quantitatively transparent; we can never have that with the black box of holistic admissions. Cynics might suggest that affluent white liberals are such strong proponents of holistic admissions precisely because there is less transparency, as less transparency in the system inevitably benefits the moneyed. There is no system of reward for academic achievement that the wealthy can’t game, but the assumption that holistic criteria favors diverse candidates while hard numbers favor well-off students remains totally unproven. It’s just liberal theology.

But the bigger issue is this: at some point you have to consider the possibility that all of these things are correlating together because they are measuring some stable construct of “academic ability.” The correlation between socioeconomic status (SES) and college essays only invalidates those essays if the SES differences are causing differences in perceived essay quality. If instead higher SES students actually write better essays, then the instrument is doing its job. (If you think “better essays” is too vague a target, well, that’s why people have tried to develop more objective criteria… like the SATs.) Could essays be subject to elite capture? Sure; I imagine a remarkable percentage of college essays are written by adults, paid or not. But we’re also aware that writing skills, like all academic abilities, correlate with other academic abilities. Despite the common notion of “math people” or “language people,” measures of verbal and quantitative reasoning ability tend to correlate more strongly than many assume. Smart kids tend to be smart in all areas, just like when you went to K-12 yourself. As long as college admissions are at least partially justified as selecting for the most prepared, they’ll favor high SES kids.

Why would richer students be better at writing essays than the norm, aside from various types of illegitimate manipulation? All kinds of things related to their learning environment, the type a dozen breathless dissertations are written about every year. You can say that those beneficial environmental factors are themselves the product of family wealth, and I wouldn’t disagree. But if we forbid considering factors that correlate with SES, we’ll eliminate literally every selection criteria. If the idealized point of college admissions is to select for predictive validity - the ability of these measures to predict which students will flourish in college - then measures that correlate with socioeconomic status will necessarily be prized in admissions, as SES is a significant direct predictor of performance in college. And even if you somehow come up with some way to avoid this universal association, you’d inevitably end up rewarding another unearned and arbitrarily distributed factor: the individual student’s intrinsic academic ability. (Read all about it in my book.)

I’ve said this many times: our school system is asked to do two flatly contradictory things at once, promoting equality and sorting students into a hierarchy of performance on academic tasks. These are totally contrary goals, and 21st century education is so fucked up because no one will look at this essential tension and take it seriously. We now speak as if promoting socioeconomic equality and superficial diversity are the only job of colleges, but this conception is very new and has resulted in awkwardly grafting a progressive ideology onto institutions that are inherently, inevitably, and existentially vehicles for creating inequality. The very act of saying “Student A did better than Student B” is identifying inequality, inequality in performance. A degree makes its holder unalike some other job applicants and it’s that inequality that is sold on the labor market. Just like you can’t actually diversify the British royal family, you can’t make colleges tools of equality; creating hierarchy is their very function.

Some kids are better at school than others. Colleges sort the good from the bad and pursue the former. How can we ever make this a progressive project? Well, we can’t.

Update: A fundamental point I somehow forgot to include: the book I cited above, Measuring Success, includes a large study looking at the student body of schools that go “test optional” - that is, that previously required the SAT/ACT and no longer do - and found that going test optional made their student bodies no more diverse.