Prediction of Graduate Major from Cognitive and Self-Report Test Scores Obtained during the High School Years

Abstract

Using information from Project TALENT's 11-yr. follow-up (Wise, et al., 1979), 8 graduate major groups were formed for males, 6 for females. Three of the female groups were supplemented by undergraduate majors to increase sample size. Descriptive canonical discriminant analyses were conducted in each sex, high school grade, and separate sets of cognitive and self-report predictors that were administered in high school. Finally, analyses of the combined predictors were conducted in both sexes for the 9th grade only. Two to four canonical discriminants were retained for interpretation on two criteria: lambda probabilities and size of eigenvalues. Discrimination among groups is more effective in males than in females and in later than in earlier grades. Self-report tests discriminate more effectively among graduate major groups than cognitive ones in the cognitively restricted range of talent. Gains in differentiating among groups from adding cognitive to self-report predictors in the 9th grade are relatively modest. The two sets of predictors define overlapping if not identical dimensions. The first two cognitive functions are important supplements to a measure of general intelligence. For both sexes, a combination of mechanical, spatial, and mathematical tests distinguish science majors from those in the humanities and social sciences who appear on a verbal function.