ABSTRACT
Members of several demographic groups receive systematically different scores on many of the tests and other assessments used to make high-stakes decisions, such as admission to college or graduate school or selection into a job or an organization. Research on the impact of such tests on the opportunities of members of lower-scoring groups has focused for several reasons largely on standardized tests of cognitive ability. First, these tests are widely recognized as valid predictors of performance in an extraordinarily wide range of settings (Gottfredson, 1986, 1988; Hartigan & Wigdor, 1989; Jensen, 1980; Neisser et al., 1996; Ree & Earles, 1991, 1992; Ree, Earles, & Teachout, 1994; Schmidt & Hunter, 1981, 1999) and are often among the best-available predictors when considered in terms of the trade-off between their cost and their predictive validity. As a result, there is often a strong argument for using these tests as an important part of making decisions about applicants. On the other hand, the use of these tests will result in substantial adverse impact against members of lower-scoring groups (Gottfredson, 1986; Herrnstein & Murray, 1994; Jensen, 1980; Scarr, 1981). For example, because of the relatively large differences in the mean scores obtained by white, Hispanic, and black examinees on standardized tests of cognitive ability, the use of these tests in contexts for which there are a large number of applicants for a small number of positions (e.g., medical school) will virtually eliminate black and Hispanic applicants from consideration. The continued use of these tests in academic admissions and personnel selection is almost certain to contribute to the racial and ethnic segregation of many jobs and institutions.
