Prediction of Successive Terms Performance in College from Tests and Grades

Abstract

Colleges follow an almost universal practice of administering ability tests to new students. These tests are used initially for purposes such as admission, placement, and counseling. The test scores then remain in the files and undoubtedly continue to influence decisions as to whether a given student with marginal initial attainment should be retained in college, be admitted to upper school, and so forth. In a previous paper (Juola, 1963) the applicability of these freshmanlevel ability tests to decisions made in later terms was questioned. Here, correlations between ability test scores and attainment in the student's initial quarter in college were found to be in excess of .60.. owever, these correlations decreased to .30 and lower by the fourth, fifth, and sixth terms. Other data indicated that a portion of this decrease in predictive validity was due to restriction in the range of abilities for the sample of students available in later terms, but even beyond this, the correlations successively decreased term by term. This tendency is shown in Table 1, which presents correlations between test scores and term grades over a seven-quarter span for a sample of 290 male and 250 female students selected randomly from the total freshman class that entered Michigan State University in Fall, 1960. The test scores reported are for the MSU Reading Test and the three subscores and total score of the College Qualification Tests, (CQT) Form B (Bennett, et al.). This table shows that the correlations between the CQT-Total Score, for example, and the first term grade point average were .52 for males and .50 for females. These predictive validities decrease progressively by term to but .27 and .28 for males and females respectively, by the seventh term. In contrast to this general pattern of decrease in precision of prediction, Table 1 also provides data which show that the cumulative GPA immediately preceding each of these terms provides a considerably better basis for prediction of future grades. The cumulative six term GPA, for example, correlates .66 with the seventh-term GPA for men; the cumulative five-term GPA correlates .68 with the sixth-term GPA; and the cumulative fourth-term GPA correlates .69 with the fifth-term GPA. One can now ask an additional