Assessing group work fairly: a practical guide

Most of the unfairness in group marks is designed in by accident: a single shared grade, decided after the fact, with no record of who did what. The good news is that the same decision can be designed the other way. This is a practical walkthrough.

1. Do not give everyone the same mark

Start here, because it is the decision that determines the rest. A review of the literature concludes that awarding a single shared mark encourages freeriding and dilutes the benefits of group work, ¹ and published guidance for academics notes it can both over-reward and under-reward individuals regardless of effort. ² If you take one thing from this guide: plan to individualise the mark from the outset.

2. Use more than one signal

No single measure captures contribution on its own. A defensible mark combines independent signals that cover each other’s blind spots:

• ·Objective contribution from the work itself: who actually wrote and built the artifact.
• ·Confidential peer assessment: how each person worked within the team, which the traces cannot see.
• ·Your own academic judgement of the final product.

3. Design the peer assessment carefully

Peer marks are useful, and peer assessment tracks teacher marks reasonably well under good conditions, especially for global judgements against clear criteria. ^{3 4} But it has known failure modes: scores inflate when a grade is attached, ⁵ and friendship bias is not solved by a rubric — in the study usually cited for this, the friendship effect was <em>larger</em> with the rubric than without it, and that was students rating one peer’s individual work rather than a teammate. ⁶ Mitigate them deliberately:

• 1.Keep the grade weight of peer scores modest rather than decisive.
• 2.Ask for a short written justification, not just a number.
• 3.Keep ratings confidential, so students rate honestly and cannot see their own scores.
• 4.Use clear, shared criteria and favour global judgements over many fine-grained dimensions.
• 5.Run the assessment more than once, early and mid-project, not only at hand-in ⁵ — while watching for overuse across a whole programme, which dilutes the accuracy of the result. ⁷
• 6.Decide deliberately whether students see their peer feedback between rounds. Releasing it is what triggers students to return the ratings they were given. ⁸
• 7.Ask about what people did rather than what they are like: task-oriented behaviours are easier to rate reliably than group-maintenance ones. ⁴

4. Keep a record you can defend

Students judge the fairness of group assessment on the process more than on the mark. Asked what drives their sense of fairness, students in one UK study ranked how the group ran and how they were treated <em>above</em> whether the grade matched their contribution — a student can get the mark they deserved and still consider the process unfair. ⁹ You can only show either if you kept the evidence. And it is worth being precise about what students complain of: in a survey across all faculties of an Australian university, the most common theme by a wide margin was not free-riding as such but <strong>group marks failing to reflect individual contribution</strong>. ¹⁰ They are not asking you to catch anyone. They are asking for a mark that fits. Freeze the submitted work and the contribution picture at hand-in, so that if a mark is challenged you are pointing at a fixed record rather than reconstructing events.

How Dwixel does this for you

Dwixel is built around exactly this sequence. Contribution is measured from the work as the group goes; peer assessment is collected confidentially; and at submission the artifacts and contribution totals are frozen into a locked record. The established practice of weighting a group mark by peer-assessed contribution, used by tools such as WebPA for years, ⁷ becomes the default rather than a manual chore.