Who Makes the Grade? Research Comparing Self, Peer and Instructor Grades in College

Abstract

With the increase in large college classes and online education, student grading of their own work and that of peers is also increasing in frequency. This meta-analysis of 36 studies and 103 effect sizes examined several questions regarding the relationship between grades assigned by college students (either to themselves or peers) and those assigned by their instructors on the same assessment. On average, students graded themselves .41 standard deviations higher than their instructors. The grade distribution correlation between the two types of graders averaged r=.71. Inter-judge reliability estimates suggested that a range of 2-4 peer-graders are needed in order to attain the same level of reliability achieved by the instructor. Little research was found on the effect of student grading on subsequent student performance. Moderator analyses revealed that differences between graders appeared to be minimized when (a) students are grading a peer’s work rather then their own, (b) they are in their freshman versus sophomore, junior or senior year, (c) some form of training is given, (d) assessment has lower stakes, (e) more objective tests are given, and (f) course content is English, social science, or professional versus science or engineering. These results have implications for what contexts best facilitate the use of students as graders, and bring to light areas where future research is needed.