Peer evaluation form for a group project
Ten questions each student answers once about every teammate and once about themselves: five behaviour criteria, one overall comparison, two open items.
- 10questions
- about 4 minper teammate
- One passper person rated
- Every itemwritten for this page
One form per teammate, everyone included
On paper this is a grid. Online it becomes a repeat.
Each student opens the link once per teammate, and once about themselves. Question 2 asks who the form is about, so a team of four produces sixteen submissions from one link. There is no roster import and no way to repeat a block of questions in one pass, so tell the team the number up front.
It is not anonymous, and should not pretend to be. The rater name is question 1, because a rating that moves an individual mark has to be something a student can be asked about. The teacher who owns the survey sees every response; the rest of the team sees none.
Three near neighbours that are not this. When the teacher is the one being rated, use the teacher evaluation form. When the teams sit inside an organisation, the cross-team collaboration survey asks what people can answer. For a colleague rating a colleague, 360 feedback questions.
The ten peer evaluation form questions
Every item with its scale and what to do with the answers.
Who is rating whom
Two items before the instrument starts, because nothing after them can be answered until the form knows who it is about.
- What is your name?The rater. A form that feeds an individual grade has to be attributable, or a mark that moves cannot be explained to the student it moved.
- Which member of the team is this form about?The subject. One submission per teammate, and one about themselves, so a four-person team produces sixteen forms in total. Sort the results by this column before you read anything.
What they did
Five criteria, behaviour rather than character, on one 1 to 5 line anchored at Never and Always. Swap in the words from your own brief.
- Did the part of the work they agreed to take on.Scoped to the agreed division of work rather than to effort in general, so the answer is about something the team wrote down and can be shown.
- Had their part ready when the rest of the team needed it.Separate from the item above on purpose. Work that arrives complete but two days after the rest of the team needed it costs the group differently from work that never arrives.
- Handed over work the team could use without redoing it.The quality item, framed as what it cost the others rather than as a verdict on the work. A low score here is the one worth reading the open answers about.
- Was reachable when the team needed an answer from them.Here because a team stalls on a question nobody answers, not only on work nobody does. Swap in your own channel if the team agreed one.
- Brought ideas the team ended up using.The one item that credits contribution the finished artefact does not show. It asks whether ideas were USED, not whether they were offered, so it cannot be scored by talking.
The overall judgement, and two open items
One global comparison and two free-text items. Question 8 is the one to weight if you weight anything.
- Overall, how did this person's contribution compare with an equal share of the work?The number a peer-moderated mark uses. Because the options are worded relative to an equal share, a whole team answering "about an equal share" is a clean result rather than a flat one.
- What did this person do that the team most needed?Read these rather than counting them. They are where a quiet contributor gets credited and where a low rating gets its evidence.
- What is one thing you would ask them to do differently on the next project?Forward-facing on purpose. It asks for one change on the next project rather than a complaint about this one, which is the difference between a form students can hand to each other and one they cannot.
Two swaps before you send. Replace the five criteria with the words in your own project brief. If the ratings are formative rather than graded, cut question 1 and tell students the form is anonymous, because then it can be.
Turning ratings into an individual mark
The ratings are not the mark. This is what to do with them.
The usual method is a group mark adjusted per person. WebPA, the open-source system built at Loughborough University, puts it in one sentence: each student grades their team-mates and their own performance, and that grading is used with the overall group mark to give each student an individual grade.[3] Base the adjustment on question 8, the only item worded as a comparison with an equal share.
Read the criteria as columns, not as a total. Adding five ratings into a score invents a precision the form does not have. Low on one criterion from everybody is specific and fixable; low on all five from one teammate and high from the rest is a disagreement to settle before it becomes a mark.
The exercise has evidence behind it. A meta-analysis of fifty-four controlled studies found a small to medium effect on academic performance, with students who assessed each other outperforming both students who were not assessed and students assessed by their teacher.[1] An average across studies, not a promise about your class.
| What you collect | What it is for | What not to do with it |
|---|---|---|
| Questions 3 to 7, five behaviour criteria on a 1 to 5 line | Telling a specific problem apart from a general one, and evidencing a mark that moved | Do not add them into a score. Five ratings summed is a number with no unit. |
| Question 8, contribution against an equal share | The adjustment itself, and the only item that carries a direction | Do not read one form on its own. It means something only next to the other members of the same team. |
| Questions 9 and 10, two open items | Evidence for a low rating, and credit for work the finished artefact does not show | Do not count them or code them. There are sixteen of these in a four-person team; read them. |
Sending it to a class
One link, sixteen submissions from a team of four.
Send one link to the whole class, not one per team. Question 2 carries the name, so responses sort themselves afterwards. Say how many forms each student owes you, or you will get one each. All ten are loaded into the free form builder; the text below goes in front of question one.
For feedback on the teaching rather than the team, use the course evaluation form. When one named person ran a session, the facilitator evaluation questions fit better.
Do not release the ratings to the team. These are answers about named classmates written by named classmates. Summarise for the student who was rated if you want to, and keep the forms.
What to put in front of students before question one
Before you start This form is about how the work on [project name] was divided up. Fill it in once for every person on your team, and once about yourself. Ten questions, about four minutes each. It is not anonymous. Your name goes on question 1 and [TEACHER NAME] can see who wrote each form and who it is about. That is because the answers feed an individual mark, and a mark nobody can be asked to explain is not a mark anyone can appeal. Your teammates will not see your answers. Please finish all of them by [date].
Peer evaluation form FAQ
What should a peer evaluation form ask?
What each person did, in the terms the team agreed in advance: the share of work they took on, whether it arrived when needed, whether it could be used as handed over, whether they were reachable, and whether their ideas were used. Leave out attitude and effort: a teammate sees what somebody did, not what they meant.
How many criteria should a peer evaluation form have?
Fewer than feels right. Falchikov and Goldfinch compared peer marks with teacher marks across forty-eight studies and found peer assessments resembled teacher assessments more closely under a global judgement against well understood criteria than under marking of several separate dimensions.[2] Five criteria plus one overall item follows from that.
Should a peer evaluation form be anonymous?
If it feeds an individual grade, no: in a team of four, a rating attributed to one of your three teammates is not anonymous in any useful sense. Put the rater name on question 1, and have everyone rate themselves too, so each student reads against what the team said.
Rating the teammates you worked beside is a different job from rating the event that put you together. For the event itself, the hackathon feedback form asks about the brief, the time and the judging.
Methods and sources
What this template is based on
Ten questions: two that record who is rating whom, five behaviour criteria on one 1 to 5 line, one global comparison against an equal share, and two open items. The count is deliberately low. Falchikov and Goldfinch analysed forty-eight studies comparing peer marks with teacher marks and found peer assessments resembled teacher assessments more closely under global judgements against well understood criteria than under marking of several individual dimensions,[2] which is the argument for one overall item supported by a few criteria rather than a long rubric. The five criteria are all things a teammate can observe: what was taken on, when it arrived, whether it could be used, whether the person was reachable, and whether their ideas were used. Attitude, effort and ability are excluded because a teammate cannot observe them.
Question licensing
Nothing in this template is reprinted from anywhere. Every published peer evaluation form on this subject that we could reach is either licensed non-commercially or carries no licence at all: the Carnegie Mellon Eberly Center forms are CC BY-NC-SA 4.0,[4] the Penn State Schreyer Institute examples carry the same licence in the PDF itself, the AAC&U VALUE teamwork rubric requires prior written permission for commercial reproduction, and CATME is proprietary to Purdue under an end-user licence. A non-commercial licence does not reach a commercial site and silence is not permission, so all ten items here were written from scratch.
Limits and disclosure
The effect reported in the meta-analysis is an average across fifty-four controlled studies in primary, secondary and tertiary settings, not a prediction about one class.[1] It also found no significant difference between peer assessment and self-assessment, which is a null result rather than a tie. The template carries no score and computes nothing: it collects ratings, and the adjustment from a group mark to an individual one is done by whoever set the assignment. The platform has no roster import and no way to repeat a block of questions within one pass, so a team of four means sixteen submissions rather than four; that is stated on the page because it changes how the survey is set up.
SuperSurvey builds the editor this template opens in. Nothing here depends on using it.
References
- Double, K. S., McGrane, J. A., & Hopfenbeck, T. N. "The Impact of Peer Assessment on Academic Performance: A Meta-analysis of Control Group Studies." Educational Psychology Review 32 (2020): 481-509. Open access, CC BY. doi.org/10.1007/s10648-019-09510-3
- Falchikov, N., & Goldfinch, J. "Student Peer Assessment in Higher Education: A Meta-Analysis Comparing Peer and Teacher Marks." Review of Educational Research 70, no. 3 (2000): 287-322. doi.org/10.3102/00346543070003287
- WebPA. Centre for Engineering and Design Education, Loughborough University, funded by the JISC e-Learning Capital Programme. webpaproject.lboro.ac.uk
- Eberly Center for Teaching Excellence and Educational Innovation, Carnegie Mellon University. "Assess Teamwork and Group Projects." Licensed CC BY-NC-SA 4.0. cmu.edu
Put your own criteria in and send it
Ten questions, about four minutes per teammate, closing before the group mark goes out.
Use this template