Open AccessPREDICTIVE VALIDITY WITHIN CATEGORIZATIONS OF COLLEGE STUDENTS: 1978, 1981, AND 1985 1
Author(s) -
Morgan Rick
Publication year - 1990
Publication title -
ets research report series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.235
H-Index - 5
ISSN - 2330-8516
DOI - 10.1002/j.2333-8504.1990.tb01346.x
Subject(s) - psychology , test (biology) , ethnic group , academic achievement , correlation , mathematics education , class (philosophy) , predictive validity , standardized test , demography , developmental psychology , mathematics , paleontology , geometry , artificial intelligence , sociology , anthropology , computer science , biology
SAT scores, student‐reported high school grades, and scores for four other Admissions Testing Program tests were correlated with freshman grade point average for students at 198 colleges from the enrolling classes of 1978, 1981, and 1985. The purpose of the study was to determine whether observed changes in correlations between test scores and college grades were 1) restricted to the SAT, 2) due to changes in student subgroups, and 3) comparable at different ability levels within the freshman class. Declines were found in the correlation of freshman grades with three of the four achievement tests examined. The data were also analyzed for students classified according to sex, ethnic group, intended major, and SAT score level. Nearly all of these demographic subgroups had the pattern of a stronger relationship of SAT scores with freshman grades for the enrolling class of 1978 than for the enrolling class of 1985. A minimal decline in SAT regression weights was found for students scoring in the upper third of SAT test takers within their college. The recent decline in the relationship between SAT sores and freshman grades may be largely focused on students with SAT scores in the lower two‐thirds of their respective college classes.