Racial Differences in the Psychometric Properties of Grades: Are the Grades of non-White Students More Variable?

Although college grades are used in the assessment of academic performance and in the determination of the criterion-related validity of tests designed to predict academic performance, there exists virtually no systematic empirical research investigating race differences in the psychometric properties of college grades. One reason to suspect that the psychometric properties of grades might display subgroup differences is that current theories of racial bias suggest some professors might grade non-White students more harshly, but others might grade them more leniently, meaning that grades for these students might be more variable and less intercorrelated throughout the college career than the grades of White students. In addition to the possibility of racial bias, there are a number of other race-related factors which could also increase grade variability for non-White students.

In the present study, I use a large educational data set, which includes the first 60 grades received by more than 150,000 students from 41 different colleges, to determine (1) whether White students? grades are more intercorrelated than non-White students? grades, (2) whether the average within-person standard deviation of grades is higher for non-White students than for White students, (3) whether the patterns of race differences in within-persons grade variability are the same for both public and private schools, (4) whether the increased variability of non-White student grades can be accounted for by other alternative explanatory factors (SAT scores, socioeconomic status, high school grade point average, course load, English as best language, and college GPA), and (5) whether race differences in within-persons grade variability is a persistent trend across all four years of college. I found that Asian, Black, and Hispanic students did tend to have more variable grades than White students; however, this greater variability was almost completely accounted for by race differences in course load, high school GPA, English as best language, SES, SAT scores, and cumulative college GPA. Although the present study does not provide definitive evidence for or against the possibility of bias in grading, it suggests that, if there is bias in grading, it is not expressing itself by increasing the variability of non-White students? grades.