AFTER FOUR YEARS of work and often heated debate, a panel of the National Academy of Sciences has finally produced its report on the uses and abuses of standardized ability tests. This is one of the prickliest subjects around.
The study found that the tests are capable of predicting future performance in both the classroom and the work place. The degree of accuracy depends on many things. But if a test is properly developed, if it is shown to correlate with the particular job or course of study for which it is being used and if the limits of these tests are understood (they do not measure creativity, motivation, experience) then ability tests are, in the study's words, "important predictive tools for our society." This holds true for both achievement tests and so-called aptitude tests, which largely measure the same thing--what a person knows at the time he or she takes the test, not inherent ability to learn.
The report's most important conclusion is that ability tests are equally valid for whites and blacks. The tests do not, as has frequently been claimed, "underpredict" minority group performance. Studies of college and graduate school entrance exams and of a variety of employment tests show that test scores either predict performance equally well for whites and blacks or somewhat over-predict for blacks.
The evidence also shows, however, that blacks score less well than whites on these tests. In general, the spread in average scores is what statisticians call one standard deviation. This is a very substantial difference. It means, for example, that a cutoff score that included the top 50 percent of the whites on a certain test would include only 16 percent of the blacks. A cutoff that included the top fifth would include only 4 percent of blacks.
There are people who resist and quarrel with such findings because, in our view, they misread the implications of them. In fact, there is something condescending, even racist in its way, about those who protest that there is an anti-black message to be found in the conclusion that the tests don't lie. Their assumption seems to be that what the test discrepancies reveal is an innate racial difference-- as distinct from a difference in background and opportunity. But the fact is that these tests are not measuring inherent ability. Nor do the differences in average scores say anything about individual potential. That more minorities than whites score poorly on ability tests means neither, as is so often claimed, that the tests are discriminatory nor that there is something wrong with the people taking the tests. It means that people differ, very often for reasons embedded in national and personal history.
Those who would throw out standardized tests because of poor minority group scores, and those (like the federal government at the present time) who try to rig tests so that they produce equal passing rates are doing no one--especially their theoretical beneficiaries--a favor. On the contrary, they are undermining the credible argument for a stronger effort to reconcile these disparities--and thereby doing those minorities and the society as a whole great harm.