Posted on June 21, 2010

New Evidence That SAT Hurts Blacks

Jay Matthews, Washington Post, June 17, 2010

Roy Freedle is 76 now, with a research psychologist’s innate patience. He knows that decades often pass before valid ideas take root. When the notion is as radical as his, that the SAT is racially biased, an even longer wait might be expected. But after 23 years the research he has done on the surprising reaction of black students to hard words versus easy words seems to be gaining new respectability.


Now, in the latest issue of the Harvard Educational Review, the two scholars who took on that project have published a paper saying Freedle was right about a flaw in the SAT, even in its current form. They say “the SAT, a high-stakes test with significant consequences for the educational opportunities available to young people in the United States, favors one ethnic group over another.”

“The confirmation of unfair test results throws into question the validity of the test and, consequently, all decisions based on its results,” said Maria Veronica Santelices, now at the Pontificia Universidad Catolica de Chile in Santiago, and Mark Wilson of UC Berkeley. “All admissions decisions based exclusively or predominantly on SAT performance–and therefore access to higher education institutions and subsequent job placement and professional success–appear to be biased against the African American minority group and could be exposed to legal challenge.”


{snip} Back in 2003, when I wrote a long article for the Atlantic Monthly on Freedle’s work, I relied heavily on him and other experts to explain what they were talking about. Much of it had to do with a method of test analysis called differential item functioning, or DIF (rhymes with cliff). Psychometricians like Freedle and his colleagues at ETS, which was then managing the SAT, looked at how different ethnicities that were matched at different scoring levels (those who had scored 360 on the SAT verbal test, then those who had scored 380, and so on) did on each item.

At each level of ability, but particularly in the lower-scoring groups, white students on average did better than blacks on the easier items, whereas blacks on average did better than whites on the harder ones. (Whites, however, as a group did better overall.)


Hard questions, those that produced more wrong answers, tended to have longer, less common words. Easy questions tended to have shorter, more common words. Freedle thought this was key to the relative success African American students had with the harder ones. Simpler words tended to have more meanings, and in some cases different meanings in white middle class neighborhoods than they had in underprivileged minority neighborhoods, he concluded. This, he said, could help explain why African American students did worse on questions with common words than on questions that depended on harder, but less ambiguous words they studied at school.

On average, he said, black students were performing only slightly above matched-ability whites on hard questions. But averages did not submit applications to colleges. Individual students did. Some of those individuals, he discovered, would have gotten a boost of a hundred points or more on the SAT if the score was weighted toward the hard items. He proposed that the College Board offer a supplement to SAT scores, called the Revised-SAT, or R-SAT, which would be calculated based only on the hard items. This, he said, would “greatly increase the number of high-scoring minority individuals.”

In their paper, Santelices and Wilson rule out Freedle’s suggestion that the bias he found in the test might affect all kinds of multiple-choice questions, or minorities other than blacks. But they did find it in sentence completion and reading comprehension sections of the SAT.