How Peer Assessment Works | Why Judging Someone Else’s Work Can Improve Your Own

A student finishes an essay, hands it to a classmate and receives three comments:

“Good.” “Add more detail.” “Nice vocabulary.”

Technically, peer assessment happened.

Educationally, almost nothing may have happened.

The classmate looked at the work, but did not necessarily apply a standard. The comments may not identify a specific quality problem. The writer may not know what to change. The reviewer may not have learned anything about quality. The teacher may collect the papers, feel pleased that students gave feedback, and move on.

Peer assessment is often treated as a classroom activity.

It is more useful to treat it as a mechanism.

The mechanism begins when students have a usable model of quality, apply that model to someone else’s work, locate evidence, make a judgement, communicate it in a form the other learner can use, and then test whether the judgement changes subsequent work.

The surprising part is that the reviewer may learn as much as the receiver.

When students inspect a peer’s response, they are temporarily moved out of production mode. They are not asking, “What should I write next?” They are asking, “What makes this work effective? Where is the evidence? Which criterion has been met? What would strengthen it?”

That shift can build evaluative judgement: the ability to recognise quality and compare actual performance against a standard.

Peer assessment therefore has two learning routes.

Receiver route: I receive information that may help me improve my work.

Reviewer route: I practise seeing quality, weakness and possibility in work that is not emotionally fused to my own effort.

Both routes can fail.

A peer can be inaccurate. A criterion can be vague. Friendship can soften judgement. Status can make one student’s comments feel more authoritative than they deserve. Students can spend ten minutes writing feedback nobody uses.

The educational value is not “students mark each other”.

The value is in a designed cycle of criteria → evidence → judgement → feedback → revision → retest.

The 50-second route

If you only have a minute, keep this:

Peer assessment is not simply swapping work. It is students using explicit criteria or standards to examine another learner’s performance.
The reviewer learns by practising evaluative judgement: noticing what quality looks like, locating evidence and deciding what matters most.
The receiver benefits only if feedback is understandable, credible, actionable and followed by an opportunity to revise or try again.
Train students before expecting useful peer feedback. Show examples, model comments, calibrate judgements and discuss what counts as evidence.
Do not ask peers to grade what they are not equipped to judge. Peer assessment is often stronger for specific features than for high-stakes final marks.
Separate the person from the work. “This paragraph needs a clearer warrant” is different from “You are bad at arguing.”
Build in revision time. Feedback without a next action easily becomes classroom paperwork.
Check the giver as well as the receiver: “What did reviewing this work teach you about your own?”
The goal is not to replace the teacher. It is to make standards of quality less private.

What peer assessment actually owns

Peer assessment sits beside several neighbouring educational mechanisms.

Peer tutoring involves one learner helping another understand or practise content.

Peer feedback is information from peers about work or performance.

Comparative judgement can involve deciding which of two performances is stronger.

Formative assessment is the larger system of using evidence to decide what to do next in learning.

Feedback uptake concerns what happens when a learner receives, interprets and acts on feedback.

Peer assessment has a particular job:

students apply criteria to peers’ work in order to make quality judgements and generate information for improvement.

The assessment part matters.

Without judgement against a standard, students may simply react.

“I liked the introduction.” “This sounds weird.” “I would write it differently.”

Those reactions can occasionally help, but they are not yet disciplined evaluation.

Why reviewing another person’s work can teach the reviewer

Students are unusually close to their own work.

They remember what they intended. They know what they meant. They have already invested effort. They may read missing logic back into the page because it exists in their head.

Another person’s work creates distance.

The reviewer can ask:

Where is the claim? Which sentence provides evidence? Does the method actually answer the question? Is this conclusion justified by the data? Where does the reasoning jump? Which feature makes this example effective?

That distance makes criteria easier to see.

Then something useful can happen when the reviewer returns to their own work:

“I just told Maya that her evidence did not connect clearly to the claim. Mine does the same thing.”

Peer assessment can create a mirror indirectly.

The learner notices a quality feature first in someone else, then detects it in themselves.

That is one reason the reviewer role should not be treated as unpaid teacher labour.

It is a learning task in its own right.

Stage 1: Make the standard visible before students judge anything

Students cannot reliably assess against a criterion they do not understand.

“Give each other feedback on your writing” is too open.

What should they attend to?

Ideas? Organisation? Evidence? Grammar? Tone? Sentence fluency? Everything?

If everything is in play, novice reviewers often choose the easiest visible feature: spelling, handwriting, length, format.

The teacher should narrow the target.

For example:

Today, review only claim–evidence connection.

Criterion: “Each major claim is supported by evidence that is relevant and explained.”

Now the reviewer has an object.

The teacher can show two brief examples:

evidence inserted with no explanation;
evidence linked explicitly to the claim.

Ask: “Which better meets the criterion, and why?”

This calibration should happen before students touch one another’s work.

Stage 2: Teach students to find evidence before making a judgement

Good peer assessment should not begin with adjectives.

“Strong.” “Weak.” “Clear.” “Confusing.”

It should begin with location.

“Underline the claim.” “Box the evidence.” “Circle the step where the equation changes.” “Mark the sentence that explains the graph.” “Point to the part of the conclusion supported by the data.”

Now the judgement is anchored.

A useful sequence is:

Locate → describe → judge → suggest.

For an English paragraph:

Locate: “The main claim is in sentence one.”

Describe: “The quotation appears in sentence three, but the next sentence summarises the event rather than explaining the quotation.”

Judge: “So the evidence is relevant, but the explanation criterion is only partly met.”

Suggest: “Add a sentence explaining what the word ‘reluctantly’ shows about the character’s attitude.”

The comment has become inspectable.

Stage 3: Calibrate before releasing students into pairs

Imagine a class of thirty students using a four-level rubric.

One student thinks a “4” means no spelling mistakes. Another thinks it means sophisticated ideas. Another gives every friend a 4. Another gives everyone a 2 because “there is always room to improve”.

The numbers look systematic.

The standards are not shared.

Calibration means students practise judging the same sample and discuss why.

Show an anonymous or teacher-created response.

Ask students to judge one criterion independently.

Then compare.

Where judgements differ, return to the evidence: “What in the work makes you say Level 3 rather than Level 2?”

The teacher can clarify the criterion.

This process teaches students that assessment is not personal preference.

A shared standard must be interpreted.

Stage 4: Limit the scope enough for novices to succeed

Expert teachers can read a piece of writing and notice content, organisation, sentence control, misconceptions, tone, evidence and next steps almost simultaneously.

Students usually cannot.

Give them one or two dimensions.

In mathematics:

correctness of representation;
justification of one key step.

In science:

whether the evidence supports the claim;
whether alternative explanations are addressed.

In English:

clarity of thesis;
integration and explanation of evidence.

In oral presentations:

whether the central message is understandable;
whether evidence is used purposefully.

Scope protects quality.

A short accurate peer assessment is more valuable than a full rubric completed mechanically.

Stage 5: Separate diagnosis from advice

Students often jump too quickly into advice.

“Add more detail.” “Use better vocabulary.” “Explain more.”

Advice is weak when diagnosis is vague.

First ask: What is happening in the work?

Then: What matters most next?

Suppose a science conclusion says:

“The fertilised plants grew taller, so fertiliser definitely makes all plants grow faster.”

Diagnosis: The conclusion extends beyond the evidence. The investigation tested one plant type, one fertiliser and one set of conditions.

Advice: Narrow the claim to the tested conditions and acknowledge the limitation.

That is much more useful than “be more specific”.

Stage 6: Teach comments that are usable

A useful peer comment has at least three qualities.

It is specific

Not: “Improve explanation.”

But: “After the quotation, explain how the word ‘isolated’ supports your claim about the character.”

It is criterion-linked

Not: “I don’t like this graph.”

But: “The graph communicates the trend, but the axis label is missing a unit, so the quantity cannot be interpreted precisely.”

It is actionable

Not: “Needs to be better.”

But: “Add one comparison between the two data points before you state the trend.”

Students can be trained with a simple pattern:

I noticed…
According to the criterion…
Your next move could be…

This is scaffolding, not a permanent script.

Stage 7: Make the receiver evaluate the feedback

Receiving peer feedback should not mean obeying it.

Peers can be wrong.

The learner needs judgement too.

A receiver can classify comments:

Use — clearly improves the work and matches the criterion.

Check — plausible, but I need to verify with the rubric, text, data or teacher.

Reject with reason — does not fit the task or is based on a misunderstanding.

This protects learner agency.

It also teaches a mature feedback principle:

feedback is information to evaluate, not an instruction to surrender judgement.

The receiver can write:

“I used Comment 1 because my evidence was not explained.”

“I did not use Comment 2 because the requested example was outside the task scope.”

Now peer assessment becomes reasoning on both sides.

Stage 8: Build revision into the lesson

Peer assessment without revision is incomplete.

If students give comments and immediately submit the work, the feedback arrives after the learning decision has closed.

Give time to act.

This can be five minutes.

“Choose one peer comment that will improve the criterion we are working on. Revise that part now.”

Then ask: “What changed?”

The revision is the test of whether the feedback was usable.

If no one can act on the comments, the assessment design needs repair.

Stage 9: Return the reviewer to their own work

This is the stage that often produces the second learning route.

After students review a peer, ask:

“What did you notice in their work that you should now check in yours?”

A reviewer might write:

“I told my partner to connect evidence to the claim. I need to check that in my second paragraph.”

Or:

“I noticed their graph had a clear scale. Mine jumps unevenly.”

This turns peer assessment from an outward judgement task into self-regulation.

The learner has practised the criterion at one remove, then brings it home.

A concrete English example: evidence is present but not doing work

Student paragraph:

“Macbeth is increasingly controlled by fear. He says his mind is ‘full of scorpions’. This is a metaphor. Macbeth is worried about Banquo. Later he arranges Banquo’s murder.”

A peer reviewing “evidence explanation” might initially say:

“Good quote.”

The teacher-trained process changes the review.

Locate: Quote: “full of scorpions”.

Judge: Relevant to fear, but the explanation stops at identifying the technique.

Feedback: “Explain what the image of scorpions suggests about the kind of fear Macbeth experiences—painful, invasive and impossible to ignore—and connect that to why Banquo feels threatening.”

The receiver revises.

The reviewer then checks: “Have I also named techniques without explaining their effect?”

That is peer assessment doing two jobs.

A concrete mathematics example: correct answer, weak justification

Question: Prove that the sum of two odd numbers is even.

Student response:

“3 + 5 = 8, and 8 is even. So two odd numbers make an even number.”

The example is correct.

The proof is not general.

Peer criterion: “The reasoning must show why the claim holds for all numbers in the stated class.”

Reviewer: “You have shown one example, not all odd numbers. Represent the odd numbers generally, such as (2a+1) and (2b+1), then show what their sum becomes.”

Now the peer is assessing the scope of justification, not the arithmetic.

The reviewer may strengthen their own understanding of what a proof is.

A concrete science example: claim stronger than evidence

Investigation: Students measure dissolving time at three water temperatures.

Conclusion: “Hot water always dissolves substances faster.”

Peer criterion: “The conclusion should match the variables and evidence actually tested.”

Reviewer: “The experiment tested one substance under three temperatures. ‘Always’ and ‘substances’ make the claim wider than the evidence. Say that in this investigation, the tested sugar dissolved faster at the higher temperatures.”

This assessment requires epistemic judgement.

The reviewer has to distinguish data from general claim.

A concrete oral-presentation example

A learner gives a three-minute presentation.

Peer criterion: “The audience should understand the central claim and how each example supports it.”

Instead of: “Speak louder.”

the reviewer records: “Your main claim was clear in the opening. Example 2 was interesting, but I could not tell how it supported the claim. Add one sentence after the example explaining the connection.”

The feedback targets meaning.

Delivery feedback may also matter, but it is not allowed to displace the selected criterion.

Why peer assessment can support learning

High-authority guidance on formative assessment has long emphasised making learning intentions and success criteria clear, eliciting evidence, giving feedback that moves learning forward, and activating students as resources for one another.

The Education Endowment Foundation’s current materials continue this logic. Its Embedding Formative Assessment programme explicitly includes students supporting each other as learning resources. Its 2026 independent-learning guidance on feedback emphasises feedback that is tied to goals and can be acted on.

Recent meta-analytic work gives useful but appropriately qualified signals.

A 2026 meta-analysis of peer and self-assessment synthesised 75 studies and hundreds of effect sizes, reporting positive relationships with several self-regulated-learning and affective outcomes, alongside substantial heterogeneity.

A separate 2026 meta-analysis on peer feedback quality synthesised 49 papers, 53 samples and 241 effect sizes. It found a positive relationship between feedback quality and post-feedback outcomes, with stronger associations for feedback students provided than feedback they merely received. Training also strengthened the relationship.

That pattern fits the mechanism described here.

Producing feedback requires the reviewer to apply criteria.

Receiving feedback only helps if the learner interprets and uses it.

But these results do not mean any peer-marking activity automatically improves learning.

Design matters.

Caveat: peer assessment can reproduce classroom status

If one student is viewed as “the clever one”, their feedback may be accepted without scrutiny.

A lower-status student may give excellent feedback that is ignored.

Friendship can soften judgement. Conflict can make it harsh. Language confidence can be mistaken for subject knowledge. Social anxiety can make public evaluation costly.

Teachers should therefore design the interaction, not merely assign partners.

Possible protections:

anonymous samples for calibration;
rotating partners;
narrow criteria;
evidence-required comments;
no public ranking;
teacher moderation;
option to ask for verification.

Peer assessment happens inside a social system.

Ignoring that system does not make it disappear.

Caveat: novices may not yet possess the standard

A student cannot assess advanced mathematical proof if they do not understand what proof requires.

A young writer may not recognise coherence problems that an experienced teacher sees instantly.

This means peer assessment must be calibrated to what students can reasonably learn to judge.

Start with visible, teachable criteria.

For example: “Does every graph axis have a quantity and unit?” may be suitable early.

“Does the statistical model appropriately represent heteroscedastic uncertainty?” requires much more expertise.

Do not confuse student agency with abandonment.

Common failure mode 1: feedback sandwiches

“Say one nice thing, one improvement, one nice thing.”

This can make comments polite.

It can also make feedback ceremonial.

Students search for praise because the template requires it, then insert one generic criticism.

A stronger routine is criterion-based.

Praise when a feature genuinely meets the criterion. Identify the highest-value next move.

Kindness and accuracy should coexist.

Common failure mode 2: peers assign marks too early

Marks can dominate attention.

Students compare scores. Friendship affects generosity. Arguments shift toward “Why did you give me 7?”

Early peer assessment often works better with qualitative judgement: met / partly met / not yet; evidence present / missing; strongest feature / next move.

If peer scoring is used, calibrate carefully and keep teacher responsibility clear.

Common failure mode 3: the rubric is longer than the work

Students receive a two-page rubric with fifteen rows.

They tick boxes.

The review becomes clerical.

Select the one or two criteria relevant to the current learning goal.

Peer assessment should focus attention, not fragment it.

Common failure mode 4: comments focus on surface errors

Spelling and punctuation are visible.

Reasoning is harder.

Without guidance, students often correct what they can see easily.

If the lesson goal is argument, direct their attention to claim, evidence and warrant first.

Surface correction can happen later.

Common failure mode 5: “two stars and a wish” becomes automatic

Any routine can become empty.

Two compliments: “Good handwriting.” “Nice title.”

Wish: “Add more.”

The problem is not the number of comments.

The problem is absence of criteria and evidence.

A single high-quality comment can be more useful than three required ones.

Common failure mode 6: students review final products only

Feedback is most valuable when change is still possible.

Review a plan, first paragraph, first solution, rehearsal or draft.

Early peer assessment can redirect the work before errors become expensive.

Common failure mode 7: teacher disappears

Peer assessment is not a way to outsource marking.

The teacher still:

selects criteria;
models judgement;
calibrates;
monitors accuracy;
intervenes when misconceptions spread;
decides what evidence matters;
checks whether revision improved performance.

Students gain responsibility inside a designed system.

Common failure mode 8: feedback is collected but never used

The class spends twenty minutes commenting.

The bell rings.

Next lesson starts a new topic.

Students learn an unintended lesson: feedback is something school makes us produce, not something that changes work.

Always protect at least a small action window.

Common failure mode 9: the receiver must accept every suggestion

Peers are fallible.

If compliance is compulsory, students learn that feedback means obedience.

Require evaluation: “Which comment will you use and why?”

That creates judgement rather than submission.

Common failure mode 10: personal taste replaces disciplinary criteria

“I don’t like this opening.” “I prefer another method.” “This colour looks better.”

Ask: “Which criterion does that comment relate to?”

If none, it may be a preference rather than assessment.

Not every preference deserves revision.

The learner route: how to give feedback that is worth receiving

Use this sequence:

Read the criterion.
Find the exact place in the work where it appears—or should appear.
Describe what the work currently does.
Judge how well it meets the criterion.
Give one high-value next move.
Check whether your suggestion changes the work rather than merely adding more.

Then ask yourself: “What did I just notice that I should check in my own work?”

That final question turns reviewing into learning.

The learner route: how to receive peer feedback without surrendering your judgement

When someone comments on your work:

Understand it. Can you restate what they mean?

Verify it. Does the criterion support the comment?

Prioritise it. Would changing this meaningfully improve the work?

Act or reject. Use it, check it with another source, or reject it with a reason.

Retest. Read the revised part again against the criterion.

Do not defend every sentence simply because you wrote it.

Do not obey every comment simply because someone gave it.

Evaluate.

The parent route: peer assessment is not children teaching the class for free

Parents can be understandably sceptical.

“Why is another child marking my child’s work? Isn’t that the teacher’s job?”

The answer depends on design.

If students are simply exchanging final papers and assigning unreliable marks, the scepticism is justified.

If the teacher has taught a criterion, shown examples, calibrated judgement, asked students to inspect one feature and then revise, peer assessment is itself a learning task.

The child is practising:

recognising quality;
explaining a judgement;
receiving critique;
revising.

Ask your child: “What criterion were you using?” “What did you notice in your partner’s work?” “What did you change in yours?”

Those questions reveal whether the activity had an educational mechanism.

The teacher route: a twelve-minute peer-assessment cycle

Minute 1–2: Re-anchor the criterion.
Show or restate what quality means.

Minute 3: Individual read.
Students inspect the peer work silently.

Minute 4–6: Locate and judge.
Mark evidence and write one criterion-linked judgement.

Minute 7–8: Write one actionable comment.

Minute 9: Receiver reads and classifies the feedback.

Minute 10–11: Revise.

Minute 12: Reflection.
“What did reviewing another piece make you notice in your own?”

Short cycles used repeatedly can build more judgement than occasional elaborate peer-marking sessions.

The teacher route: use exemplars before live peer work

Exemplars reduce the social risk.

Students can practise assessment on anonymous work.

Choose examples that differ in one important dimension.

Ask:

Which better meets the criterion?
Where is the evidence?
What comment would improve the weaker one?
What would an unhelpful comment look like?

Only after students can judge the samples should they judge classmates.

This is rehearsal for the assessment act.

The teacher route: moderate the peer system

Sample the feedback students produce.

You do not need to read every comment.

Look for patterns:

Are comments too vague?
Are students over-correcting grammar?
Are they misreading the rubric?
Are high-status students dominating?
Are receivers ignoring accurate feedback?
Are misconceptions spreading?

Then teach the feedback process itself.

Peer assessment quality should improve over time just like writing quality or problem solving.

The school route: distinguish low-stakes learning from high-stakes certification

Peer assessment is well suited to many low-stakes formative purposes.

It is more complicated when marks carry high consequences.

Schools should be cautious about allowing peer scores to determine grades that affect progression, certification or selection unless reliability, moderation and policy requirements are addressed.

Students can still participate in standards-based review without owning the final high-stakes judgement.

This keeps the learning value while preserving accountability.

Peer assessment in mathematics

Mathematics peer assessment should not reduce to answer checking.

Useful review targets include:

choice of representation;
justification of a transformation;
whether a proof is general;
interpretation of a result;
efficiency of method;
clarity of notation.

A student may have the wrong answer for an interesting reason.

Another may have the correct answer through invalid reasoning.

Peer assessment becomes powerful when students inspect the reasoning object.

Peer assessment in science

Possible criteria:

claim matches evidence;
variables are identified accurately;
graph communicates data correctly;
explanation uses scientific mechanism;
limitation is relevant;
conclusion does not exceed the investigation.

This makes peer review part of scientific argumentation rather than mere editing.

Peer assessment in writing

Avoid “mark the whole essay”.

Focus on:

thesis;
paragraph coherence;
evidence integration;
development;
sentence control;
audience;
ending.

Different rounds can target different dimensions.

Students gradually build a richer internal model of writing quality.

Peer assessment in oral work

Peer assessment can support:

clarity of claim;
organisation;
evidence;
explanation;
pace;
audience awareness.

Use one or two dimensions per rehearsal.

Live performance already creates cognitive and social load.

Do not make the observer track fifteen criteria.

Digital peer assessment

Digital tools can make peer exchange easier.

They can also make feedback mechanical.

Automatic assignment, anonymous review and inline comments are useful features only if the criteria and revision cycle remain meaningful.

AI-generated feedback adds another layer.

Students can compare: peer feedback, AI feedback, teacher feedback, and the rubric.

Ask: “Which comment is best supported by the criterion?”

That can turn feedback abundance into evaluative judgement rather than passive consumption.

How to know whether peer assessment worked

Do not count comments.

Look for:

Reviewer evidence

more precise use of criteria;
better identification of strengths and weaknesses;
stronger explanations of judgement;
transfer of noticed features into own work.

Receiver evidence

feedback correctly interpreted;
high-value comments selected;
meaningful revision;
improved performance on a fresh task.

Class evidence

standards become easier to discuss;
quality language becomes shared;
students rely less on “Is this good?” and more on specific criteria.

The mechanism succeeds when quality becomes more visible and students become more able to act on that visibility.

Frequently asked questions

Is peer assessment the same as peer feedback?

They overlap. Peer feedback is information from one learner to another. Peer assessment specifically involves judging work against criteria or standards. Assessment can generate feedback.

Should peers give grades?

Sometimes, but low-stakes criterion judgements are often safer and more educationally useful, especially with novice reviewers. High-stakes grading should remain professionally accountable.

What if the peer feedback is wrong?

Teach receivers to verify comments. The teacher should monitor patterns and intervene when misconceptions spread.

Do high-attaining students benefit more?

Not necessarily. Reviewing can benefit students across attainment levels if the criteria are accessible and the task is calibrated. Pairing and scaffolding matter.

Should feedback be anonymous?

Anonymity can reduce friendship effects and social pressure. It can also remove opportunities for dialogue and accountability. Use it according to the task.

How many comments should students give?

Enough to identify a high-value next step. Quality matters more than a fixed number.

When should peer assessment happen?

While revision is still possible: plans, drafts, rehearsals, worked solutions and intermediate products are often better than final submitted work.

Can peer assessment save teacher marking time?

It may, but that should not be the primary design reason. Good peer assessment still requires teacher preparation, calibration and monitoring.

How do students learn to give better feedback?

By seeing examples, comparing comments, practising one criterion at a time, receiving feedback on their feedback and observing whether their suggestions improve work.

Does peer assessment replace self-assessment?

No. The strongest cycles often connect them: judge someone else’s work, receive a judgement, then inspect your own work with a sharper standard.

The deeper lesson: standards become useful when students can see them for themselves

A teacher can hold an excellent model of quality.

If that model remains entirely inside the teacher’s head, students depend on repeated external judgement.

“Is this good?” “Is this enough?” “What mark would this get?”

Peer assessment can help make the standard public.

Not because classmates become miniature teachers.

Because students practise the act that expert learners eventually need to perform themselves:

look at work, compare it with a standard, identify what matters, and decide what to do next.

That is evaluative judgement.

When students learn to see quality in another person’s work, they gain a chance to recognise it in their own.

When they learn to receive comments without either defending everything or obeying everything, they become more independent.

When revision follows judgement, assessment stops being a verdict.

It becomes a learning move.

The best peer assessment therefore ends somewhere surprising.

Not with the peer.

With the learner becoming increasingly able to answer their own question:

What does better look like, and what should I change next?

Sources

Education Endowment Foundation, Embedding Formative Assessment: https://educationendowmentfoundation.org.uk/projects-and-evaluation/promising-programmes/embedding-formative-assessment
Education Endowment Foundation, Developing Independent Learners — Feedback (24 August 2026): https://educationendowmentfoundation.org.uk/16-19/developing-independent-learners/feedback
Double, K. S. et al., Peer and self-assessment as mechanisms for supporting self-regulated learning and affective outcomes: a meta-analytic review (2026), Assessment in Education: https://doi.org/10.1080/0969594X.2026.2681630
Peer feedback quality matters for post-feedback outcomes: A meta-analysis (2026), Educational Research Review: https://doi.org/10.1016/j.edurev.2026.100848
Leibniz Institute for Science and Mathematics Education, publication summary for Peer feedback quality matters for post-feedback outcomes: https://www.leibniz-ipn.de/en/research/publications/peer-feedback-quality-matters-for-post-feedback-outcomes-a-meta-analysis