Abstract
This study asked how well students' grades at the University of Pennsylvania could be predicted from linear combinations of high-school class rank (CLR), total scholastic-aptitude-test score (SAT), and average achievement-test score (ACH), all of which are available in applications for admission to selective institutions. Underrepresented minorities were not included in the main analysis. Consideration of multiple regression coefficients revealed that CLR and ACH added significantly to overall prediction, whereas SAT did not. This asymmetry in incremental validity was not easily explained by restricted range, self-selection, or nonlinear effects. SAT did add significantly to prediction of grades in some individual courses. Of the three variables, only CLR added significantly to prediction of attrition or number of uncompleted courses.
Get full access to this article
View all access options for this article.