Study of Gender Bias in S.A.T. Prompts Debate
New research has added fuel to the long-running debate over whether the Scholastic Assessment Test, the nation's best-known college-entrance examination, is biased against women.
Critics for years have raised questions about possible bias in the test, which is used predominantly by colleges on the East and West coasts and in parts of the South to predict applicants' academic potential. Once known as the Scholastic Aptitude Test, the exam has been revamped and rechristened as the S.A.T.-1.
The Educational Testing Service, which constructs the test, and the College Board, which administers it, have argued that the apparent gender differences in the scores were small.
Now, in a study discussed here last month, two researchers from the University of California at Berkeley take the test developers' claims to task. David K. Leonard and Jiming Jiang followed 10,000 freshmen admitted to the university between 1986 and 1988. Among men and women who had the same S.A.T. scores upon entering the university, the women tended to have higher grade-point averages by the time they left. The courses they took were no easier than those in which their male classmates enrolled.
Mr. Leonard, who presented his findings at the April 18-22 meeting of the American Educational Research Association, laid the blame for the grade discrepancies on the S.A.T., which the university uses as the basis for more than half of its admissions decisions.
But Nancy Burton, a researcher for the Princeton, N.J.-based E.T.S., defended the test.
"Certainly we don't think men and women should do equally well on the test," she said. "We could do that with a flip of a coin."
The Math Factor
Ms. Burton presented findings from her own study, which looked at whether new mathematics items on the test favored males.
The changes made in the S.A.T. last year were designed to make the test reflect changing classroom practice, which calls for more emphasis on critical thinking and reasoning.
In the math portion of the test, students are being asked to "grid in" responses for 10 questions, rather than choose among several multiple-choice alternatives. The math section is 15 minutes longer than before, and students are now allowed to use calculators.
The new test was administered for the first time last spring. Looking at those results, Ms. Burton said that, while women appeared to have more difficulty solving problems that required critical-thinking skills, the new items led to only slight differences in favor of men on overall mathematics scores. On the verbal portion of the test, the gap between men's and women's scores was reduced slightly.
"I believe the results are tentative and, for math, they are susceptible to change due to improved instruction," Ms. Burton said. "I would not be at all surprised if you would see some closing of the gaps over time on this test."
But Mr. Leonard, the Berkeley researcher, said he was not convinced those changes have "fixed the problem."