skip to main content skip to footer

Evaluation and Presentation of Differential Predictions

French, John Winslow, 1918-
Publication Year:
Report Number:
ETS Research Memorandum
Document Type:
Page Count:
Subject/Key Words:
Majors (Students), Predictive Measurement, Predictive Validity, Student Evaluation


The ability of tests to provide differential prediction was explored with tenth grade students and college freshmen. Data included intercorrelations of aptitude and interest predictors; validities for four criteria (scores in English, history, algebra or physical science, and geometry or architecture); beta-weights; multiple validities; and evaluative tables. For over 400 high school boys and girls, data were collected on grades achieved in grades 9 to 12. For 63 to 140 college students, data were collected on major field grades during the junior and senior years. Predictors were from a battery of aptitude tests, interest scales, and personality scales. Intercorrelations were high among predicted criteria, especially between the two predicted verbal criteria and the two predicted mathematical criteria. Correlations were estimated between predicted and observed differences; similarities to test validity were noted. Graphs showed the multiple validity of the predictors for each of the four college criteria. A counselor might be able to use this information for comparative prediction, predicting grades in different courses, and differential prediction. Standard errors of differences for all possible pairs of the college criteria were computed. It was possible to find the proportion of cases in which significant differences would be found, as well as the proportions of differences not accountable for by error or chance. Similar computations with the high school data resulted in some useful differential prediction for the girls, but not for the boys. Paper presented at the Multivariate Conference, sponsored by Office of Naval Research at Fort Worth, November 17, 1961.

Read More