A Second Study Raises Questions About Potential Bias in SAT
Is the SAT biased against African-American students? That's the question that Washington Post reporter Jay Mathews raises in a riveting blog post published last night. He cites a study in the latest issue of the Harvard Educational Review by Maria Veronica Santelices at the Pontificia Universidad Catolica de Chile in Santiago and Mark Wilson of UC Berkeley.
The study replicates and updates some work published in 2003 by Roy Freedle, a retired research psychologist from the Educational Testing Service about whom Mathews once wrote. In both studies, researchers looked at how white and black students who were matched by ability levels did on different sections of newer SAT tests. They found that white students outperformed black students overall, and on easier, more common words on verbal sections of the tests, while black students tended to outperform white students on the harder words. (Those would be the ones that drew fewer correct answers, in other words.)
According to Mathews, Freedle's theory is that the easier words tend to have more than one meaning. And some of those meanings may be different in white, middle-class communities than they are in disadvantaged African-American communities. If the test had included less ambiguous words, Freedle figured, the overall scores for black students might have been higher.
Unlike Freedle's study, the new research does not, however, show the same pattern for other minorities.
Predictably, the College Board, which administers the SAT, is not happy, calling the Santelices-Wilson paper "fundamentally flawed." A more-formal critical analysis of the results is forthcoming.
Lots of studies have tackled this question before and few, besides Freedle, have found much evidence of bias. It will be interesting to see if this new study renews the debate on this conversation.