The Scientist did an excellent job reporting FairTest's recent study demonstrating that the lion's share of National Academy for Science, Space, and Technology (NASST) scholarships went to boys because of reliance on American College Testing (ACT) exam scores to choose winners (F. Hoke, Jan. 10, 1994, page 1). However, the reaction statements attributed to spokesmen for ACT contain such significant errors of fact and logic that further clarification is necessary.
For example, the statement by ACT's Kelley Hayden that the test has a predictive validity of 95 percent for both males and females is not supported by a single piece of data. The test-maker's own research shows that ACT scores have a correlation of about 0.4 with first-year college grades, the only outcome the test claims to predict. That means the ACT explains less than 20 percent of the variance in grades. Moreover, ACT admits that young women receive higher college grades, despite their lower test scores.
Equally wrong is ACT vice president Thomas Saterfiel's logic in claiming that the 3:1 male/female scholarship ratio is justified because the gender breakdown of college graduates in science, engineering, and math is similar. In fact, his argument reverses the causal chain and ignores the possibility that the low percentage of female science students is partly a result of test-based admissions and financial aid decisions.
FairTest does agree with Hayden that the ACT was never intended to be used as the sole criterion for scholarships and that "test scores should not be used alone to make high-stakes decisions." Why, then, did ACT allow its scores to be used in the NASST program, and why does ACT continue to allow test scores to be the sole factor in determining freshman eligibility under the National Collegiate Athletic Association's Proposition 48?
It is true that Congress, not ACT, mandates use of test scores. But ACT apparently complied without hesitation. The entire incident demonstrates that relying on unaccountable testing companies to police the use of their products is insufficient.
Cambridge, Mass. 02139