Testing: The dilemma of group differences.

Abstract

A revival of test bashing has followed the growing public disillusionment with racial preferences as a means for abrogating the adverse impact of tests on certain groups in educational and occupational selection. This, added to the racial differences in g and the slight and ephemeral effects of Head Start and other more intensive interventions aimed at decreasing racial group differences in scholastic achievement, is the dilemma of group differences. In the face of the apparent failure of equal educational opportunity to make all groups in American society equal in scholastic performance or in the test scores used in selection for higher education, jobs, and the Armed Forces, psychological tests are again being blamed and scorned. This could be averted in the public's perception by emphasizing tests' face validity in test construction without lessening the latent traits that account for the tests' validity. The new wave of opposition against the use of mental tests in educational and personnel selection really has nothing to do with psychometric science. It is a public, political, and popular1 media phenomenon entirely driven by the fact that identifiable groups in our society have different distributions of test scores, all of them more or less normally distributed but with different means and different standard deviations, and with the means of some groups differing by more than one standard deviation. This well known fact, in and of itself, would be trivial were it not for the fact that the best professionally constructed tests are valid correlates and predictors of a person's actual performance in school, in college, in Armed Forces training programs, and in many jobs. Therefore tests are useful in selection, the need for which is inescapable whenever the number of applicants exceeds the number of openings or the costs of failure are prohibitive (e.g., pilots). Professionally constructed, standardized, objective tests have a better proven track record for valid meritocratic selection than any other single means that has been tried, including letters of recommendation, personal interviews, biographical information, essays, and portfolios. Because high school grades in English, history, math, science, and a foreign language may also reflect a student's level of academic motivation and effective study habits—variables that are not as well reflected by test scores alone—a combination of grade point average (GPA) and test scores is the best predictor of academic performance in college that psychometricians have come up with. If a typical selective college decided to admit a random sample of youths from the general population and maintained its usual grading standards, the validity