A recent study by researchers at the University of Michigan reveals an unexpected grading bias based on the alphabetical order of students’ surnames. The analysis of over 30 million grading records from U-M’s Canvas system found that students with surnames starting with A, B, C, D, or E received grades 0.3 points higher out of 100 compared to those graded randomly. Conversely, students with surnames later in the alphabet received grades 0.3 points lower, creating a significant 0.6-point gap.
Jun Li, an associate professor at U-M’s Ross School of Business, co-authored the study with doctoral students Jiaxin Pei and Helen (Zhihan) Wang. They discovered that as graders evaluated more assignments, the quality of grading declined. Furthermore, students with later-alphabet surnames received more negative and less polite comments on their assignments.
The study, currently under review by the journal Management Science and available as a working paper, highlights the unintended social impact of this alphabetical bias. It negatively affects students’ course grade-point averages, potentially limiting their opportunities in their respective career paths.
The researchers attribute this bias to the default alphabetical order setting in online learning management systems like Canvas. They suggest that randomizing the grading order could mitigate this bias. Additionally, they propose academic institutions could hire more graders, distribute the workload, or train graders to minimize bias.
While the study focuses on data from U-M, the researchers believe the findings could be generalized across institutions and courses. The research has garnered positive reactions from academic communities, raising awareness of this overlooked issue in grading practices.