An analysis of over 30 million grading records by researchers at the University of Michigan reveals a significant bias related to the alphabetical order of names on Canvas, a widely used online learning system.
Illustrative image from Pixabay
Students whose names begin with the first letters of the alphabet achieve, on average, slightly higher scores, by 0.3 points on a scale of 100, compared to random grading. Conversely, those whose names start with a letter at the end of the alphabet experience an equivalent drop in scores. This creates a difference of 0.6 points, which may seem minimal but has a real impact on students' overall averages and, by extension, their future professional opportunities.
The researchers also noted that comments directed at students "alphabetically disadvantaged" were often more negative and less polite. The quality of grading gradually decreases as evaluators work through the stack of assignments, a phenomenon exacerbated by marker fatigue.
The study's officials, Jun Li, Jiaxin Pei, and Helen Wang, suggest simple adjustments to counteract this bias: opting for a random order in assignment grading, employing more graders for larger classes, or more evenly distributing the workload.
Their findings, although still in prepublication form, are already widely discussed in academic circles and have garnered general interest, confirming suspicions long held by many educators.