posted on 1996-01-01, 00:00authored byJonathan P Caulkins, Patrick D. Larkey, Jifa Wei
The computation of Graduate Point Average (GPA) incorrectly assumes that grades are
comparable across courses and instructors. GPA overstates the performance of students who elect an
“easier” course of study relative to those who choose a more “difficult” course of study. This paper
proposes a method of adjusting GPA and applies it to data from one cohort of undergraduates at
Carnegie Mellon University. Adjusted GPAs are more highly correlated with students’ high school
Grade Point Average and with SAT scores than are the raw GPAs or GPAs adjusted using a prominent
alternative method, Item Response Theory. A survey of students finds that the new methods’ estimates
of relative course difficulty are consistent with students’ perceptions of relative course difficulty.