How Fair Is Your Test?

So, which news stories about the College Board’s new report on the addition of a writing section to the Scholastic Aptitude Test (SAT) do you believe?

Here’s the Pittsburgh Post-Gazette:

“The results echo preliminary findings, released in April, that the new exam is just about as good as high school grades – and in some cases better – at predicting college freshman grades.”

And here’s New York Newsday:

“A new report on the SAT’s controversial writing section finds that it provides colleges with little help in predicting how well applicants will do during their first year on campus, beyond help provided by other data such as high school grades.”

It probably depends on which source you prefer to believe: the College Board itself, which evaluated data concerning about 150,000 students from 110 four-year colleges who have taken the SAT since the writing section was added in March 2005, or its chief critic, the National Center for Fair and Open Testing, or FairTest, a Cambridge, Massachusetts-based advocacy organization that has never met a standardized test it likes. Most of the news stories included quotations from FairTest spokesman Robert Schaeffer pooh-poohing the College Board’s conclusions. Many of the news stories also pointed to critics’ concerns that females, blacks and Hispanics continue to lag behind white males in SAT scores, raising the specter of “cultural insensitivity,” as the Washington Examiner put it.

True, the College Board report admitted that the new writing section, which adds 45 minutes to the SAT’s three-hour length and has driven its cost up to $45 this year from $28.50 in 2003, has boosted only slightly the SAT’s predictive value as to college freshmen’s academic success. The College Board also reminds colleges not to rely on SAT scores alone in making admissions decisions, but to look at a combination of SAT scores and high school grades. Still, the board insists that the “SAT continues to be an excellent predictor of how students will perform in their first year of college” and is a better predictor than high-school grades for all minority groups: blacks, Hispanics, American Indians, and Asians. And while the writing section may add only a little predictive value to the test, it is still “the most predictive of the three SAT sections” (the other two are the traditional verbal and quantitative sections),” not only overall but for all subgroups. At the very least, the College Board maintains, the SAT is a useful hedge against the grade inflation rampant at many high schools.


These findings are unlikely to stanch the hemorrhaging of colleges and universities from the ranks of those that require SAT scores from their bachelor’s-degree applicants. On June 19 Marlboro College, a small liberal-arts institution in Vermont, announced that it would no longer use standardized tests in making admissions decisions unless applicants submitted them voluntarily. That adds Marlboro to the more than 750 four-year colleges that no longer use the SAT and other tests; at least 40 of those have dropped the SAT during the last two years alone.

Abandoning standardized tests can make it easier for colleges to justify racial and gender-based admissions decisions regarding applicants whose SAT scores would predict a lower likelihood of academic success than those of other applicants. Sadly, what no one—except perhaps spokesmen for the College Board—wants to face is that the reason females, blacks, and Hispanics generally perform less well on the SAT than white males has little to do with the test itself and everything to do with second-rate schooling, dysfunctional cultural values, and, in the case of females, what seem to be hard-wired biological differences that affect quantitative reasoning—which is why boys as a group have consistently outperform girls on the SAT’s quantitative session despite many individual exceptions..

Charlotte Allen

Charlotte Allen blogs for the Los Angeles Times and writes frequently about cultural trends for the Weekly Standard.

Leave a Reply

Your email address will not be published.