A high GPA opens doors, and a low GPA slams them shut. With so much at stake, it’s no wonder that students obsess over the number.
Yet all of this attention belies the fact that GPA is a seriously flawed statistic.
Not only is it incapable of meaningfully representing a student’s academic achievements, but our fixation on it discourages academic engagement and contributes to the well-documented phenomenon of grade inflation.
Bucknell University is introducing a way to address some key defects in the GPA statistic that may also free some students from a risk-averse mind-set that is antithetical to the core mission of any institution of higher learning.
Grading norms vary significantly from course to course. The median grades awarded in introductory level courses at the same university can differ by more than a letter grade. A student’s GPA can suffer if they consistently enroll in courses in which grades are typically lower.
Studies both at Bucknell and at other universities have shown that this uneven playing field provides a powerful incentive for students to avoid courses (especially electives) in which grades tend to be lower.
Bucknell’s Committee on Instruction studied the cumulative effect of different grading norms on every student (anonymously) in a recent graduating class. We used a statistic called the GPA of Medians to measure the grading norms experienced by a student.
The GPA of Medians, or GPAM, is the GPA a student would have if they had earned the median grade in each course they took.
The results of the study are stunning and illustrate how GPAs without context can be terribly misleading.
The GPAM, however, provides that context, and it’s quickly evident how such a measure can be used to benefit students.
Take for example the five students represented in the table below, first using GPA alone.
The data suggest that student five is the top academic performer. Now introduce the GPAM for each student.
The story is quite different with the addition of the GPAMs.
The grading norms vary significantly between these five students. Students 1 and 5, for instance, both have a GPA close to their GPAM, indicating that they earned typical grades across their courses. But they graduated with very different GPAs.
Student 2, whose graduating GPA did not qualify for a degree with distinction, far outperformed the median in the courses he or she actually took.
And the addition of GPAM indicates that Student 4 — whose GPA far exceeds the GPAM — is probably the strongest academically.
Clearly, any GPA cutoff is asking much more of some students than others. What can be done? Since the unfairness is caused by a disparity in grading norms, an institution could impose standards on faculty to develop better uniformity, or provide additional information about the grading norms experienced by students.
A variation of the first option was implemented by Princeton University in 2004 with a uniform target for the proportion of A grades awarded in courses throughout the university. But any approach that dictates uniformity fails to recognize that different pedagogical practices (including in grading) suit different disciplines and pedagogies. Princeton’s approach also resulted in a sudden downward spike in grades that disadvantaged its own students. The uniform target system at Princeton was abandoned in 2014.
Cornell University and Dartmouth College both adopted policies in which median grades for every course are published online and displayed on transcripts. But some professors who studied the issue at Cornell concluded that publishing median grades exacerbated problems with both grade inflation and student academic engagement because it provided a way for students to shop for higher grades.
A similar approach was rejected at Bucknell out of concern that it would effectively shame individual faculty members who are using pedagogically justifiable grading approaches.
This spring, the faculty at Bucknell adopted a measure to include cumulative GPAMs on academic progress reports immediately below a student’s cumulative GPA. The initiative received the full support of Bucknell’s student government.
We expect the GPA+GPAM approach to benefit our students in a number of ways. Some students (such as 1, 2 and 4 in the tables above) will be able to better market their academic achievements. Access to the GPAM may also mitigate the incentive to weigh the expected grading norms heavily when selecting courses.
Providing students with their GPAM data will not fix all of the problems with GPAs, but it will add meaning to an otherwise unreliable figure.
At the very least, we hope it provides a foot in the door — and keeps it propped open — for students whose academic career can’t be accurately shown in one three-digit number.
Tom Solomon is professor of physics and astronomy at Bucknell University. Adam Piggott is associate professor of mathematics at Bucknell University.