International Journal of STEM Education (Mar 2022)

STEM courses are harder: evaluating inter-course grading disparities with a calibrated GPA model

  • Jonathan H. Tomkin,
  • Matthew West

DOI
https://doi.org/10.1186/s40594-022-00343-1
Journal volume & issue
Vol. 9, no. 1
pp. 1 – 17

Abstract

Read online

Abstract Background Grades in college and university STEM courses are an important determinant of student persistence in STEM fields. Recent studies have used the grade offset/grade penalty method to explore why students have lower grades in STEM courses than their GPAs would predict. The results of these studies are in doubt; however, as they use GPA as a reliable measure of academic performance, which is a disputed assumption. Using a predictive model of student performance, it is possible to produce a more accurate measure of academic performance than the observed GPA and discover if STEM courses are graded more stringently, and under which circumstances. Results A weighted logistic model of GPA better predicts academic performance than the observed GPA. Using this calibrated GPA it is found that the grade offset method predicts that STEM courses, departments, and programs grade significantly more stringently than non-STEM courses. The average grade difference between STEM and non-STEM course grades and GPAs is around four tenths of a grade point. An exception is general education courses offered by STEM departments, which are graded with the same leniency as non-STEM courses. Grade offset calculations that use the observed GPA systematically underestimate the negative offset in STEM grading relative to calculations that use the calibrated GPA. The calibrated GPA is much more highly correlated with standardized tests such as the ACT (r = 0.49) than the observed GPA is (r = 0.25). Conclusion Observed GPA is a systematically biased measure of academic performance, and should not be used as a basis for determining the presence of grading inequity. Logistic models of GPA provide a more reliable measure of academic performance. When comparing otherwise academically similar students, we find that STEM students have substantially lower grades and GPAs, and that this is the consequence of harder (more stringent) grading in STEM courses.

Keywords