skip to main content skip to footer

Differences Among Racial and Ethnic Groups in Mean Scores on the GRE and SAT: Cross-Sectional Comparisons GRE SAT

Author(s):
Pennock-Roman, Maria
Publication Year:
1993
Report Number:
RR-91-13
Source:
ETS Research Report
Document Type:
Report
Page Count:
16
Subject/Key Words:
Graduate Record Examinations Board, Ethnic Groups, Graduate Record Examinations (GRE), Racial Differences, Scholastic Aptitude Test (SAT)

Abstract

The objective here is to highlight the limitations of making longitudinal inferences about racial/ethnic group differences in test scores using cross-sectional, published data provided by the Scholastic Aptitude Test (SAT) and Graduate Record Examinations (GRE) programs. Published data from the two programs are not comparable for a variety of reasons. For example, they use different definitions of demographic group membership (e.g., citizenship status), and these differences have a large effect on the population statistics. Furthermore, one year of GRE data corresponds to many years of SAT data. The lack of comparability of the two sets of data is illustrated by calculating mean differences between the non-Hispanic White group and minority groups on test scores (in standard deviation units) under two conditions. First, standardized group mean differences are derived from statistics published for one year of GRE data and for one year of SAT data. Second, differences are derived after making corresponding groups in the two populations more comparable according to residency, citizenship status, and year of baccalaureate degree. Estimates of the relative distance between groups change under the two conditions. Even after making the aforementioned corrections, the two data sets are not comparable because of possible unequal distributions of undergraduate major in the two populations and possible variability across racial/ethnic groups in the characteristics of students who choose to apply to graduate school. Thus, longitudinal changes in group differences cannot be studied using cross-sectional comparisons of the two populations.

Read More