Predicting Black Student Grades with Nonintellectual Measures

Abstract

In previous research, questions have been raised about the validity of standard academic predictors for black students. It has been found in many cases that correlations of standardized tests are at least as high for Blacks as for Whites (Hills, 1964; McKelpin, 1965; Munday, 1965; Boney, 1966; Cleary, 1966; Hills and Gladney, 1966; Harris and Reitzel, 1967; Pfeifer and Sedlacek, 1971). However, in several instances, standardized tests have been found to be less predictive for Blacks (Green and Farquhar, 1965; Cleary, 1966). Moreover, evidence suggests that high school grades may not be valid predictors of college grades for all Blacks (Thomas and Stanley, 1969; Pfeifer and Sedlacek, 1971). The Cultural Study Center at the University of Maryland has begun a three-phase program of research on predictors and criteria of black student success. The first phase involved an examination of the predictors currently employed by the University (Pfeifer and Sedlacek, 1971). Results indicated that a regression equation weighting high school grades and SAT worked about as well for black students as it did for Whites (cross validated R = .62), but beta co-efficients varied considerably by race-sex groupings. For example, high school grades received relatively little weight for black males. Bowers (1970) and Temp (1971) also found separate regression equations most appropriate for racial subgroups. Pfeifer and Sedlacek (1971) also raised the methodological point that often samples of blacks employed in studies are not representative of all Blacks capable of college work. Self selection of institutions by students and selective loss of students were cited as serious problems limiting generalization of results.