skip to main content skip to footer

An Exploration of the Stability of Freshman GPA, 1978-1985 SAT

Author(s):
Pomplun, Mark; Burton, Nancy W.; Lewis, Charles
Publication Year:
1991
Report Number:
RR-91-40
Source:
ETS Research Report
Document Type:
Report
Page Count:
42
Subject/Key Words:
College Freshmen, Grades (Scholastic), Predictive Validity, Scholastic Aptitude Test (SAT), Trend Analysis

Abstract

This study is one of several meant to explain a decline in the predictive validity of the SAT and high school grades between 1975 and 1988. This particular study asks whether the criterion, freshman grade point average (FGPA), has changed. There may have been changes in the mix of academic competencies that students use in earning freshman grades. Possible changes in general reasoning abilities, technical skills, or subject knowledge were considered. College Board tests that are usually thought of as predictors were used in this study as proxies for the concurrent academic competencies that comprise FGPA. The academic components or factors, defined by various combinations of SAT, TSWE items and scores, HSGPA, and Achievement Test scores, included verbal and mathematical reasoning ability; mathematics, language, and writing skills; and knowledge of science and history. Confirma-tory factor analysis was used to test the consistency between 1978 and 1985 of the relationship of the academic factors to FGPA. The primary results indicate that FGPA has been stable from 1978 to 1985 in relation to the abilities, skills, and subject knowledge measured by the SAT, TSWE, Achieve-ment Tests, and HSGPA. No evidence was found to suggest that a change in the meaning of FGPA has contributed to the predictive validity decline. This study was designed as an initial exploratory analysis, to be followed, if warranted, by a study with better proxies for the academic components of FGPA and with the addition of noncognitive variables such as study habits and interest. However, the results were so stable over a number of different proxies, and over a number of different ways of combining proxies, that the researchers conclude that no follow-up study is warranted. (42pp.)

Read More