Exam credits and exam performance in university economics courses − evidence from a regression discontinuity experiment

ABSTRACT

We estimate effects of nonfinancial incentives, combined with the notion of ‘loss aversion,’ on students’ exam performance in two introductory economics courses. Our experiment awarded five points to students who scored below the median score on Midterm I as an incentive to improve their performance on Midterm II − these points would be lost if they failed to improve their performance. Regression discontinuity analysis indicates that this incentive improved male students’ performance by 0.38 standard deviations on Midterm II.

KEYWORDS:

• Loss aversion
• regression discontinuity
• behavioral economics
• non-monetary incentives

JEL CLASSIFICATION:

• A22
• C93

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ A voluntary survey conducted after the final exam indicated that nearly half of the students are female and half had studied economics before. The vast majority of enrolled students came from four majors: Agriculture (36%), Pharmacy (25%), Polytechnique (16.5%), and other (21%). Other includes students from colleges of Liberal Arts, Health and Human Sciences, and Engineering.

² The distribution of majors by gender (Table A1) shows that there are more female students from Colleges of Agriculture and Pharmacy while more male students from Colleges of Polytechnique and ‘others’ (mainly Health and Human Sciences, Liberal arts, and Engineering). However, results estimated separately for females from Colleges Agriculture and Pharmacy and those from College of Polytechnique and ‘others’ are not significantly different.

³ We also performed a DID estimation pooling data for semesters with and without incentives. The coefficient of the interaction term between ‘experiment semesters’ and ‘low performers’ is positive for both male and female students, although not statistically significant (Table A2).

⁴ We thank an anonymous reviewer for suggesting this test.