Peer Grades Research Articles

Moderation of peer assessment in group projects

It is shown here that a grade distribution scheme commonly used to moderate peer assessments where self assessment is excluded is based on a false premise and will give an erroneous ranking in the situation where the best performer in a student group ranks the second best performer much higher than the other group members. An alternative to normalisation is proposed based on the idea that the rank order of peer grades should be preserved and should as far as possible be consistent between assessors. It is shown that the method correctly recovers the rank order of students within the group for all cases examined, while still eliminating biases that can result from differences in marking standards in the group. It is suggested that the approach could also be used to check for bias when self assessment is included.