What would it take to fix our wasteful and unjust system of university admissions?, Steven Pinker asks:
Let’s daydream for a moment. If only we had some way to divine the suitability of a student for an elite education, without ethnic bias, undeserved advantages to the wealthy, or pointless gaming of the system. If only we had some way to match jobs with candidates that was not distorted by the halo of prestige. A sample of behavior that could be gathered quickly and cheaply, assessed objectively, and double-checked for its ability to predict the qualities we value….
We do have this magic measuring stick, of course: it’s called standardized testing. I suspect that a major reason we slid into this madness and can’t seem to figure out how to get out of it is that the American intelligentsia has lost the ability to think straight about objective tests. After all, if the Ivies admitted the highest scoring kids at one end, and companies hired the highest scoring graduates across all universities at the other (with tests that tap knowledge and skill as well as aptitude), many of the perversities of the current system would vanish overnight. Other industrialized countries, lacking our squeamishness about testing, pick their elite students this way, as do our firms in high technology. And as Adrian Wooldridge pointed out in these pages two decades ago, test-based selection used to be the enlightened policy among liberals and progressives, since it can level a hereditary caste system by favoring the Jenny Cavilleris (poor and smart) over the Oliver Barretts (rich and stupid).
If, for various reasons, a university didn’t want a freshman class composed solely of scary-smart kids, there are simple ways to shake up the mixture. Unz suggests that Ivies fill a certain fraction of the incoming class with the highest-scoring applicants, and select the remainder from among the qualified applicant pool by lottery. One can imagine various numerical tweaks, including ones that pull up the number of minorities or legacies to the extent that those goals can be publicly justified. Grades or class rank could also be folded into the calculation. Details aside, it’s hard to see how a simple, transparent, and objective formula would be worse than the eye-of-newt-wing-of-bat mysticism that jerks teenagers and their moms around and conceals unknown mischief.
So why aren’t creative alternatives like this even on the table? A major reason is that popular writers like Stephen Jay Gould and Malcolm Gladwell, pushing a leftist or heart-above-head egalitarianism, have poisoned their readers against aptitude testing. They have insisted that the tests don’t predict anything, or that they do but only up to a limited point on the scale, or that they do but only because affluent parents can goose their children’s scores by buying them test-prep courses.
But all of these hypotheses have been empirically refuted. We have already seen that test scores, as far up the upper tail as you can go, predict a vast range of intellectual, practical, and artistic accomplishments. They’re not perfect, but intuitive judgments based on interviews and other subjective impressions have been shown to be far worse. Test preparation courses, notwithstanding their hard-sell ads, increase scores by a trifling seventh of a standard deviation (with most of the gains in the math component). As for Deresiewicz’s pronouncement that “SAT is supposed to measure aptitude, but what it actually measures is parental income, which it tracks quite closely,” this is bad social science. SAT correlates with parental income (more relevantly, socioeconomic status or SES), but that doesn’t mean it measures it; the correlation could simply mean that smarter parents have smarter kids who get higher SAT scores, and that smarter parents have more intellectually demanding and thus higher-paying jobs. Fortunately, SAT doesn’t track SES all that closely (only about 0.25 on a scale from -1 to 1), and this opens the statistical door to see what it really does measure. The answer is: aptitude. Paul Sackett and his collaborators have shown that SAT scores predict future university grades, holding all else constant, whereas parental SES does not. Matt McGue has shown, moreover, that adolescents’ test scores track the SES only of their biological parents, not (for adopted kids) of their adoptive parents, suggesting that the tracking reflects shared genes, not economic privilege.
Regardless of the role that you think aptitude testing should play in the admissions process, any discussion of meritocracy that pretends that aptitude does not exist or cannot be measured is not playing with a full deck.