MIT brings back a test that, despite its reputation, helps low-income students in an inequitable society.
Critics of standardized tests have had plenty of reasons to celebrate lately. More than three-quarters of colleges are not requiring the SAT or the ACT for admission this fall, an all-time high, and more than 400 Ph.D. programs have dropped the GRE, up from a mere handful a few years ago. MIT’s announcement on Monday that it is reinstating a testing requirement for fall 2023 admissions was a major departure from these recent trends. Just as striking, amid the widespread perception of standardized testing as an engine of inequality, was MIT’s rationale: “Not having SATs/ACT scores to consider,” MIT’s dean of admissions, Stu Schmill, wrote, “tends to raise socioeconomic barriers to demonstrating readiness for our education.” Dropping the SAT, it turns out, actually hurts low-income students, rather than helping them.
MIT’s conclusion is counterintuitive because students from richer families, on average, score higher on the SAT and other standardized tests than students from poorer ones. The correlation between family background and SAT performance is from about .25 to .40—that is, meaningful but far from perfect. Still, it’s strong enough that some researchers dismiss standardized tests as nothing more than a proxy for asking, “Are you rich?” (The ACT measures roughly the same skills as the more widely used SAT, and the arguments for and against both tests are similar.)
But the income-related disparities we see in SAT scores are not evidence of an unfair test. They are evidence of an unfair society. The test measures differences in academic preparedness, including the ability to write a clear sentence, to understand a complex passage, and to solve a mathematical problem. The SAT doesn’t create inequalities in these academic skills. It revealsthem. Throwing the measurement away doesn’t remedy underlying injustices in children’s academic opportunities, any more than throwing a thermometer away changes the weather.