# Weighting Grades based on group size

I am having students complete a presentation as a group project. Each student's grade will have two components: an instructor grade, from me, and a group grade from their group.

The group grade is calculated as the sum of five elements, each graded out of five, for a maximum of 25 points from each member of the group. Each student's 'group grade' is then calculated as the average of scores from the group.

The problem I have is that some of the groups have 4 students, some 5, and some 6. This means that weight of the score for students in larger groups is less than for those in smaller groups.

I plan to compensate for this by weighting the instructor grade more heavily. But as I think about it, even this doesn't cancel out the discrepancy.

How can I weight, or adjust, my grading scheme to make sure it's as fair as possible?

• Are you open to increasing/decreasing the grades of individual members of a group, based on evidence that their contribution was higher/lesser than their peers? Apr 10, 2018 at 16:05
• I could do that, but I don't see how it would solve my problem. The 'student grades' of those in the groups with more people would still count for less per grade than those of students in smaller groups. Apr 10, 2018 at 16:27
• Why is more people means lesser weight? Have you considered the extra work needed to schedule meeting, reaching consensus, etc.? Also, before applying any weight, do the students i) willingly join a group with higher or lower number of members and ii) know that this decision will affect their grade? If you have not declared anything from the start, I'd suggest just take this as a lesson and give every group equal weight for the group grade. Once you have the first batch of data, then you can derive a lot of schemes according to within-group variance and such. Apr 10, 2018 at 17:21
• Well you're trying to reach two contradictory objectives. First you want the student's grade to be given by the grade of the group, then you want this grade to actually be more individualized. This is not a problem that can be solved.
– user9646
Apr 10, 2018 at 17:56