The Predictive Validity of Cognitive and Affective Measures in a Small Religiously Oriented Liberal Arts College

Abstract

For a total sample of 224 freshman students entering Westmont College in the fall of 1981 as well as for each of two subsamples differentiated by sex (144 females and 80 males), the degree of relationship was sought between each of two criterion measures-freshman grade point average (GPA) and number of units satisfactorily completed during the freshman year-and each of four cognitive and eleven affective variables. In addition, the amount of association was obtained between each criterion measure and optimally weighted composites of the cognitive variables alone, the affective measures themselves, and a mixture of both cognitive and affective predictors. The statistical data revealed that (1) the criterion measure represented by freshman year college GPA could be predicted far more accurately than could one indicating the number of units satisfactorily completed, (2) high school GPA was overwhelmingly the most valid predictor of college achievement despite the presence of modest validities for standardized scholastic aptitude and achievement tests, (3) optimally weighted composites of affective variables served to explain only 40 to 60% as much of the variance in the college GPA criterion measure as did the high school GPA variable itself, (4) the most promising single affective variable for the total sample as well as for the subsamples differentiated by sex would appear to be a self-report measure indicating the perceived level of preparedness in mathematics and natural sciences held by students, and (5) the addition of one or two affective measures to a regression equation already containing high school GPA as its first-entered independent variable could not be anticipated to augment the precentage of variance accounted for in the college GPA criterion measure by more than three to five percent.