The Effect of Grades on Student Performance: Evidence from a Quasi-Experiment

  • Thomas Gray (Ko-Autor*in)
  • Bunte, J. (Ko-Autor*in)

Aktivität: VortragWissenschaftlicher Vortrag (Science-to-Science)


Grades are an unavoidable part of most of collegiate educating. While instructors may hope that grades prove to be only an output of the educational process, they may also be an input, endogenously impacting things such as effort and enjoyment, altering subsequent learning. Students who receive positive feedback in the form of higher grades may take more joy from their efforts going forward. Alternatively, grade signals may encourage those who get lower grades to increase their effort to improve their outcomes. These possibilities are difficult to ethically analyze, as it is impossible to randomly assign grades. We pursue a second-best strategy and implement a quasi-experiment, leveraging the incongruence between continuous point-schemes (0-100 points) used by many instructors during the course itself and the threshold-based letter grades (A-F) used as the final evaluation reported to their universities. The performance of students that receive 90 points do not differ significantly from students receiving 89 points. The difference between an 89.4 and an 89.5 may even approach randomness. However, their letter grades differ significantly: the former would receive an A while the latter receives a B. We utilize this discontinuity: Given that students just above and just below grade cutoffs do not differ significantly, we can conceptualize receiving the higher possible letter grade as a treatment relative to the baseline of the lower available grade. We hypothesize that this difference in letter grades (without a corresponding difference in student quality) affects outcomes. We investigate the effect of two grades. First, we analyze how midterm grades affect student participation and performance in the remainder of the semester. Second, we analyze the impact of course grades in term 1 on subsequent course enrollment in term 2. Using a regression discontinuity design, we uncover unintentional effects of grading schemes.
Zeitraum30 Aug. 20182 Sep. 2018
EreignistitelAnnual Meeting of the American Political Science Association
VeranstaltungstypKeine Angaben

Österreichische Systematik der Wissenschaftszweige (ÖFOS)

  • 502027 Politische Ökonomie