An abundance of poverty, windy moments in Washington, New York's smart boys, and other matters. FURTHER MYSTERIES OF THE SAT
By DANIEL SELIGMAN REPORTER ASSOCIATE Patty de Llosa

(FORTUNE Magazine) – Around our own house these days, the biggest question about the Scholastic Aptitude Test is how well your servant scored. (As breathlessly described in this space two fortnights ago, he went and took the test for some reason.) Alas, the scores are not yet in. That brings us to the second-biggest question, which arose out of this crazy lawsuit brought by ten young ladies here in New York State and their ''next friends,'' who mainly turn out to be their parents. The question is whether the SATs discriminate against femmes. Backed by numerous big guns of the civil rights establishment, the plaintiffs say you bet they do. If the tests really did discriminate, they would be costing some next friends a pretty penny. New York State annually awards $40 million worth of merit scholarships to superior students, and at the moment the awards are based solely on SAT scores. This means that the boys, who score higher on the tests -- 60 points higher on average -- get more of the scholarships. The most valuable are the Empire State scholarships (worth $2,000 a year for up to five years), of which boys win around 70% when SATs are the only measuring rod. In one recent year, when the state legislature briefly mandated equal weight for grade point averages (GPA), that figure fell to 62%. Girls definitely do better than boys on GPAs. In New York the mean grade point average of girls who take the SATs has recently been 3.04, vs. 2.95 for the boys. Girls are believed to get better grades because they drink less beer, raise less hell, and are more likely to turn in their assignments on time. What the New York civil rights folks are now demanding is a permanent scholarship standard giving more weight to GPA, less to beer and bias. The fact that boys score higher does not itself prove that the SATs are biased against girls. It might mean only that smart boys outnumber smart girls in the relatively elite group that takes the test. Matter of fact, that is what it means (see below). To prove the test was biased against girls, the plaintiffs would have to prove it had a tendency to underpredict their future grades. To be sure, there are some studies showing women's grades to be underpredicted. But the studies typically do not adjust for the fact, virtually uncontested, that boys take tougher courses in college than do girls. When you control for this fact, the underprediction basically disappears. At Harvard, where the management has looked at hundreds of correlations relating SAT scores to student grades, the conclusion is that test scores predict equally well for men and women. Candidates for the top New York scholarships are competing in a somewhat similar world of elites, and we assume the same judgment applies to them. But about the relative smartness of boys and girls competing for those prizes: A highly subversive thought about this matter, previously elaborated in this space (May 11, 1987), is that the male sex figures to be significantly overrepresented in any group of superior intellects. The reasons for this male predominance are in some dispute, but the underlying facts seem fairly clear. While men and women have the same average IQs, the data on men show more variation. In any representative sample of test takers, more men than women will be ''outliers,'' far from the group average. The standard deviation -- a statistical measure indicating the extent to which members of a statistical series deviate from the mean of the series -- is noticeably larger for male than female IQs. The most common estimates of SD imply that there will be 35% to 40% more men than women at IQs above 140 or so. (To be sure, males predominate even more sharply at very low IQ levels.) One of the more conservative estimates is that of Arthur Jensen of the University of California at Berkeley. Jensen has estimated 20% more men than women above IQ 140 (and 60% more men below IQ 70). For some reason, judges and politicians dealing with bias never get around to these data. Not even in Albany.