Changes in Predictor Weights in GPA Regression Equations

Publisher Website
Google Scholar
Abstract
Multiple regression equations predicting first semester grade point average (GPA) from high school percentile rank (HSPR) and composite score on the American College Test (ACT) were examined for five successive fall freshman classes at the University of Illinois. Slopes were significantly different among the five equations. Further analysis indicated a systematic decline in the ACT regression coefficients over time, while the HSPR coefficients remained approximately constant. Possible explanations given for these findings are unequal test score intervals, a low GPA ceiling, and changes in the GPA criterion.

This publication has 2 references indexed in Scilit: