Introduction to ArtificiaI Intelligence in Higher Education
2.6 Fair shares for all? The role of peer feedback in group assignments
1. Institute for Learning Innovation
and Development
Fair shares for all?
The role of peer feedback
in group assignments
Adam Warren
a.j.warren@soton.ac.uk
7. MANG1017 Business Skills
First year, second semester option
200 students
Assessment: group project report
12 supervision classes (16-20 students)
50 teams of 4 students?
7
8. First formative feedback
After two weeks – 154/202 (76%)
8
65.5
52.7
62.1
59.9
31.7
40.2
34.9
39.0
2.7
7.0
3.1
1.2
0 10 20 30 40 50 60 70 80 90 100
Participation in class discussions
Communication outside class
Organising and planning work
Being helpful to other team members
Excellent Good Needs improvement
9. Second formative feedback
After six weeks – 145/202 (72%)
9
51
42
5
3
51
39
7
2
54
38
5
3
0 10 20 30 40 50 60 70 80 90 100
Strongly Agree
Agree
Disagree
Strongly Disagree
This student is making a significant contribution to the project's deliverables.
This student is making a significant contribution to the leadership or organisation of the project
This student makes a significant contribution to the class discussions
10. Summative feedback
Immediately after submission of report
Anonymous
20% penalty for non-completion (9 students)
10
Please divide 100 points between all
members of your team in proportion to their
overall contribution to the work.
11. Group project report
80%
Individual & collective research 20%
Clarity of analytical thought 20%
Specific recommendations 20%
Referencing and presentation 20%
Attendance and contribution to in-class discussion:
tutor will award up to 15%
Using TEAMMATES to give feedback 3 times = 5%
40%
The individual grade
for each report will
be scaled by the
feedback from the
other members of
each group.
Grade calculations
60%
20% penalty for
anyone who didn’t
submit the
summative peer
feedback
12. How to distribute 100 points
Suppose your group has four members, including yourself.
If you thought that everyone did an equal share of the work then you would
award 25 points each.
Suppose you thought that yourself and one other person put in more effort and
gave yourself an extra 10 for a total of 35, then the points might be 35, 35, 15
and 15 – but is that fair? Did you really contribute more than twice as much as
someone who only got 15?
How about 30, 30, 20, 20? Does that seem fairer?
Did the other strong team member actually do a bit more than you?
So 33, 27, 20, 20?
Did the two weaker team members do the same?
How about 33, 27, 22, 18?
If you have awarded someone more than twice or less than half their equal
share of the points, please give your reasons
12
13. Impact of peer feedback on score
17
0
5
10
15
20
25
30
35
40
45
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 110
Scores by mark band
raw project scores individual modified scores
Average 57.6%
Median 60.0%
SD 16.7
14. Impact of peer feedback on score
18
0
5
10
15
20
25
30
35
40
45
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 110
Scores by mark band
raw project scores individual modified scores
Average 57.6%
Median 60.0%
SD 16.7
Average 57.8%
Median 61.6%
SD 21.2
15. Impact of peer feedback on grades
19
0
10
20
30
40
50
60
70
Zero Fail 3rd 2/2 2/1 First
Scores by grade
raw project scores individual modified scores
16. Were students fair?
20
Students awarding themselves: Average
more than 10 points more 10 13.1
5-10 points more 29 7.4
up to 5 points more 103 2.6
the same 17 0.0
up to 5 points less 38 -0.8
5-10 points less 0
more than 10 points less 0
Within 3 points of average 103 55%
2 of the 50 groups awarded equal shares
17. Dealing with unfairness
One student awarded herself 40 and her three
teammates 20 each
Her teammates all awarded her 20 or less
We adjusted her points to equal shares: 25
points each
21
18. Groups of one
Group of two – e.g. assignment score 57%
– Student A = 120, student B = 80
– Individual grades are 61% and 52%
Group of one – e.g. assignment score 57%
– No reward for challenge of solo project
We decided to use a factor of 133
– So this student’s grade is 65%
22
19. Next steps
Student evaluation survey:
– Did they think their grades were fair?
– Did they act on the formative feedback?
– Their thoughts on assigning 100 points
Further data analysis and correlation
Presentation at ALT-C in September
Fine-tune process for next year!
23
20. TEAMMATES doing its job!
24
Student to tutor:
Hi, I've given everyone an equal 20 in my group but
I fear other will favour themselves and give
themselves an extra 5-10, and I will get less marks.
This system favours people being slightly biased to
themselves, enough so that they don't get picked up
on. I fear people won’t be as fair as I have been.
21. TEAMMATES doing its job!
25
Tutor to student:
With feedback comes a level of anxiety for
everyone. The system we are using spots patterns
in groups so strong gaming is identified.
We of course also look at the feedback from class
tutors as well as feedback from students so there
are checks and balances in the system.
22. TEAMMATES doing its job!
26
His feedback scores:
[student 1] = 20 (12.0) – weak student
[student 2] = 4 (29.2) – team with 4
[self score] = 20 (11.6)
[student 4] = 4 (29.0) – team with 2
[student 5] = 10 (18.2)
23. TEAMMATES doing its job!
27
Student 2 feedback comment:
He showed up to one meeting outside class, and in
class was always late and distracting from the group
work. His part of the project was entirely rewritten
by myself and [student] as it was incoherent and
had not answered the question.
24. TEAMMATES doing its job!
28
Student 4 feedback comment:
He showed up to one meeting and his submission
had to be totally rewritten as it was incoherent and
had not addressed the question at hand, he did not
attend the group session to help right this, he
showed no effort what so ever.
Editor's Notes
Created teams too soon – during first two weeks students left and joined course or swapped classes due to ‘timetable clashes’
Some teams needed to be reconfigured as some students never attended, or not until later in the semester
But TEAMMATES did make this fairly easy to manage technically
After two weeks – but teams not yet settled
Decided not to make this feedback anonymous
- students should take ownership of their feedback
- anonymous poor feedback can erode trust
New focus on CONTRIBUTION to project
Also not anonymous
One week to complete feedback.
Students do not get to see the results of this
21 non-completers – but 12 of those had left the course, so 9 students penalised
Reminder emails sent to all non-completers mid-week and one-day before deadline. Individual reminders to a few.
How much should the peer feedback influence the overall grade?
For the pilot we guessed 40% and will carefully evaluate the impact
Normal distribution around upper second grade boundary
Students who gained points were more likely to cross a grade boundary than those who lost points