Grades and incentives: assessing competing grade point average measures and postgraduate outcomes

Abstract

In many educational settings, students may have an incentive to take courses where high grades are easier to achieve, potentially corroding student learning, evaluation of student achievement, and the fairness and efficiency of post-graduation labor outcomes. A grading system that takes into account heterogeneity of teacher standards and student ability could mitigate these problems. Using unique data from a major Canadian research university, we calculate student grade point averages (GPAs) net of course difficulty and find evidence that raw GPAs systematically distort student achievement across majors. We then link undergraduate performance and law school data. We find that adjusted GPAs better predict Law School Admissions Test scores, while the raw GPAs better predict admission to law school and grades in law school. These results suggest nuanced relationship between grades, incentives and subsequent academic outcomes. We conclude by discussing implications of our results for university leaders.

Keywords:

• grading
• universities
• labor markets
• grade point averages
• admissions

JEL Classification:

• I2
• I21
• I23
• J24
• J44

Acknowledgements

This paper has benefitted from comments from Ben Alarie, Tracey George, Helen Levy, Jide Nzelibe and anonymous reviewers. We are very grateful to Glenn Loney and Sinisa Markovic of the University of Toronto for providing us with student grade data. We are also grateful to the University of Toronto Registrar and the Faculty of Law for making their data available. All remaining errors are our own.

Notes

1. It is possible, of course, that low grades in math arise because the students are weaker. The model we use to adjust grades allows us to directly assess such a question as it measures course difficulty and student ability simultaneously.

2. Another 21% answered ‘Not applicable’ to the question even though it included a full range of responses from ‘no effect’ to ‘very significant effect.’ Presumably, selecting courses on basis the of expected grade is socially undesirable and underreported in surveys.

3. Swift et al. (2013) use GPAs relative to the high school or university average as an adjusted measure of achievement. Such a measure has difficulty distinguishing between grades earned at an excellent school where all high grades are earned in challenging courses versus grades earned at a weak school with low standards. Within schools, such an adjustment does not distinguish between a student with a high GPA who took hard classes and a student with a similar GPA who took easy classes.

4. Available at http://probability.ca/jeff/ftpdir/grademcmc.c.

5. We omitted courses with fewer than three students and students with fewer than three courses. We repeated grades for two-semester, full-year courses as these count twice as much as single-semester courses in raw GPAs.

6. We do not have data on the declared major of each student. Because we have students' entire transcripts we are able, however, to identify the field in which a student took the most classes. We refer to the department in which a student took at least three classes and more classes than any other department as his or her ‘major’; we believe we do not misclassify many students and for those that we do, using the modal discipline may be a better measure of assessing the influence on GPA measures than declared major.

7. We explored nonlinear specifications as well and did not find meaningful nonlinear effects.

8. An exception is that each student takes one small (15 person) section course which is graded based on paper assignments and an anonymized final exam.

9. We found a small rise in raw grades of around 1 point on the 100-point scale from 2000 to 2010. Some schools, such as Princeton, have taken measures to reign in grade inflation (Foderaro 2010).