Direct Answer: Peer assessment works when learners use explicit criteria to inspect another person’s work, explain what is strong or weak, justify the judgement, receive a different perspective on their own work, and then revise independently. The educational value is not that peers become substitute teachers. It is that applying criteria to someone else’s answer can make quality more visible, expose alternative approaches and improve the learner’s ability to evaluate their own performance. Peer assessment becomes unreliable when comments are vague, criteria are poorly understood, friendship or status replaces evidence, incorrect advice is accepted without checking, or the activity ends with comments instead of revision. Strong peer assessment therefore teaches the criteria first, models useful feedback, limits the reviewer to judgements they can reasonably make, requires reasons and examples, and returns the final decision to the learner with appropriate verification.
HOW LEARNING WORKS · PEER ASSESSMENT
Judging another answer can teach you what quality looks like in your own.
Peer assessment becomes learning when criteria move from the rubric into the learner’s own judgement.
The simplest definition
Peer assessment is a structured process in which learners evaluate one another’s work against explicit criteria and use that evaluation to improve judgement, feedback and revision.
It can include written peer review, oral discussion, comparing solutions, rubric-based checking, paired editing, critique of explanations or review of a draft before final submission.
The peer-assessment mechanism
CRITERIA CLARIFIED → SAMPLE JUDGED → PEER WORK INSPECTED → EVIDENCE LOCATED → JUDGEMENT JUSTIFIED → FEEDBACK GIVEN → FEEDBACK RECEIVED → COMMENTS CHECKED → REVISION CHOSEN → FRESH SELF-ASSESSMENT → INDEPENDENT PERFORMANCE
1. The criteria have to become usable before peer review begins
A rubric does not automatically create judgement.
Words such as “clear,” “well developed,” “accurate reasoning” or “effective evidence” can be interpreted differently by learners. Before reviewing peers, show examples and non-examples. Ask what feature makes one response stronger than another.
Cornell’s Center for Teaching Innovation recommends modelling constructive criticism and using clearly defined criteria before peer assessment. Cornell: Peer Assessment.
2. Reviewing another answer can reduce ownership bias
Students are often attached to the answer they produced. Its reasoning feels familiar because they know what they intended.
Another person’s work creates psychological distance. The learner can inspect the criterion without defending their own first attempt.
The useful question is then: “If I can see this problem in their work, is it present in mine too?”
3. Giving feedback can be as educational as receiving it
To give good feedback, the reviewer has to notice a feature, compare it with a standard and explain the gap.
This evaluative work can sharpen the reviewer’s own understanding of quality. The educational value therefore does not reside only in the comments received.
Ask reviewers to justify one judgement with evidence from the work and one criterion. This prevents comments from collapsing into preference.
4. Vague praise has low diagnostic value
“Good job,” “nice answer” and “make it better” do not tell the learner what to preserve or change.
Useful feedback identifies the feature: “Your claim is clear, but the quotation is not explained, so the evidence does not yet support the interpretation.”
The comment should create a next action.
5. Peer feedback is evidence, not authority
A peer can be wrong.
The recipient should compare the comment with the task, rubric, source, teacher guidance or subject rule. “My partner said so” is not enough.
Peer review should therefore train evaluation of feedback as well as acceptance of feedback.
6. Disagreement is useful when it returns to criteria
Two reviewers may disagree.
Instead of voting, ask each to identify the criterion and evidence supporting the judgement. Sometimes the disagreement reveals an ambiguous criterion. Sometimes one reviewer has overlooked a feature. Sometimes the task genuinely permits more than one defensible response.
The disagreement becomes a learning event when it is resolved by reasoning rather than status.
7. Anonymous review can change the social conditions
Names, friendships, reputation and prior attainment can influence judgement.
Anonymous review can reduce some social cues in appropriate tasks, though it does not remove every bias or misunderstanding. It is a design option rather than a universal requirement.
The main goal is to keep the judgement attached to observable features of the work.
8. Peer assessment needs task boundaries
Students should not be asked to evaluate things they are not yet equipped to judge reliably.
A learner may competently identify whether a paragraph has evidence but not whether a complex scientific claim is factually current. A peer can check algebraic working against a known answer while still needing teacher oversight for a new proof technique.
Match the review task to the learner’s knowledge.
9. Review a sample before reviewing a peer
Use an anonymous sample with known strengths and weaknesses.
Ask learners to apply the criteria individually, compare their judgements and inspect why they differ. This calibrates the reviewing process before social relationships enter.
Cornell similarly recommends practising with sample work and modelling appropriate feedback before the first review session. Teaching Students to Evaluate Each Other.
10. Peer assessment and peer tutoring are related but different
Peer tutoring involves one learner providing teaching support to another. Peer assessment centres on applying criteria to work and producing evaluative information.
The roles can overlap, but keeping them distinct helps prevent a reviewer from rewriting the whole answer instead of identifying the feature that needs improvement.
11. Peer assessment and feedback are also distinct
Feedback is information about performance relative to a goal. Peer assessment is one social route for generating that information while also training judgement.
The existing How Feedback Works in Learning page owns the broader feedback mechanism. This page owns the learner-to-learner evaluation process.
12. The receiver should not implement every comment
Good revision includes selection.
For each comment, ask: Is the reviewer identifying a real criterion? Is the evidence correct? Does the suggested change preserve my purpose? Do two comments conflict? What will I change and why?
The learner remains the decision-maker.
13. Revision is the conversion step
A pile of comments is not improvement.
Require a revision receipt: one change made, one comment rejected with reason, one criterion now understood better. This turns peer interaction into visible learning.
Then remove the peer comments and ask the learner to self-assess the revised work.
14. Mathematics peer assessment should inspect reasoning, not only answers
Two students can reach the same wrong answer by different routes or the same correct answer with different strengths.
Ask reviewers to identify interpretation, method, justification, working, units and checking. A peer should not simply mark a final number correct.
Comparing alternative correct routes can also improve method flexibility.
15. Science peer assessment should separate evidence from assertion
Review whether the conclusion follows from the stated observation, whether variables are controlled, whether the mechanism is explained and whether certainty exceeds the evidence.
Peers should flag factual uncertainty rather than confidently inventing a correction. Teacher or source verification remains appropriate where content accuracy is consequential.
16. English peer assessment should preserve author ownership
A reviewer can identify where the reader loses the argument, where tone shifts, where evidence is unexplained or where cohesion breaks.
They should not turn every sentence into their own style. The writer’s task is to decide how to repair the issue while preserving purpose and voice.
17. Receiving feedback teaches emotional regulation too
Critique can feel personal, especially when work represents substantial effort.
Structure helps. Keep comments attached to criteria and artefacts. Separate “this sentence does not establish the claim” from “you are bad at writing.”
The learner should be able to disagree respectfully and verify rather than defend automatically.
18. Giving feedback should not become doing the work
The reviewer can identify a missing link without writing the replacement paragraph. They can ask which variable changed without solving the experiment question. They can point out that a method requires justification without supplying every line.
Feedback should preserve learner production where production is the target.
19. Peer assessment should improve self-assessment
The long-term value appears when learners begin applying the same criteria before another person sees the work.
“If I were reviewing this, where would I ask for evidence?” “Which line would I mark as an unsupported jump?” “Does this paragraph satisfy the same standard I used on my partner’s?”
External judgement becomes internal monitoring.
20. The final receipt is better independent revision
After peer review, remove the partner.
Give the learner another piece of work or a fresh problem and ask them to apply the criteria independently. If they can diagnose and repair quality more accurately, peer assessment has become a learning mechanism rather than a social event.
What peer assessment is not
- Peer opinion is not automatically authority.
- A rubric is not useful until learners understand how to apply it.
- Vague praise is weak feedback.
- Receiving comments is not the same as revising.
- Reviewers should not rewrite the work for the learner.
- Friendship, confidence and reputation should not replace criteria.
- The long-term goal is stronger self-assessment.
A peer-assessment diagnostic map
| What adults see | Possible peer-assessment issue | Useful next move |
|---|---|---|
| Comments are “good” or “bad” only | Criteria not operational | Practise on sample work and require evidence |
| Students accept every suggestion | Peer feedback treated as authority | Require verification and revision rationale |
| Friends give only praise | Social relationship dominates judgement | Use anonymous/sample review and explicit criteria |
| Reviewer rewrites answer | Feedback becomes task completion | Identify gap and next question instead |
| Peer review happens but work does not improve | No conversion step | Require revision receipt and self-assessment |
| Different reviewers disagree sharply | Criterion ambiguity or knowledge gap | Return to standard and compare evidence |
A practical peer-assessment cycle
- Clarify the learning goal.
- Teach the criteria with examples.
- Calibrate on sample work.
- Review one peer artefact.
- Locate evidence for each judgement.
- Give one strength, one gap and one useful next question.
- Receive comments.
- Verify them against criteria or trusted sources.
- Choose and explain revisions.
- Self-assess the revised work.
- Apply the criteria independently on a fresh task.
For parents
- “What criterion were you using when you gave that comment?”
- “Did your friend’s comment identify a real issue or just a preference?”
- “Which feedback did you reject, and why?”
- “What did reviewing someone else’s work teach you about your own?”
- “Can you now check the same feature without a partner?”
For students
- Use criteria, not taste.
- Point to evidence in the work.
- Explain why a feature matters.
- Do not fix everything for the other person.
- Check peer comments before accepting them.
- Revise deliberately.
- Use what you learned from reviewing others to improve self-assessment.
How do we know peer assessment is working?
- Judgements refer increasingly to explicit criteria.
- Comments become more specific and actionable.
- Reviewers distinguish preference from error.
- Recipients evaluate feedback rather than accepting it blindly.
- Revisions target genuine weaknesses.
- Self-assessment improves after repeated peer review.
- Fresh independent work shows stronger quality control.
The complete peer-assessment chain
CRITERIA → CALIBRATE → INSPECT → JUDGE → JUSTIFY → FEEDBACK → VERIFY → REVISE → SELF-ASSESS → TRANSFER
Research and evidence boundary
Peer assessment is widely used as a structured learning process, but its quality depends on learner understanding of criteria, task design, reviewer competence and opportunities to revise. Cornell’s current teaching guidance emphasises explicit criteria, modelling feedback and preparing students to evaluate one another. The Education Endowment Foundation includes peer assessment within the wider peer-tutoring evidence base, which reports positive average effects while stressing careful implementation. These sources support structured peer judgement; they do not imply that every peer comment is accurate or that peer grading should replace expert assessment. Cornell Peer Assessment; EEF Peer Tutoring.
