Peer assessment improves academic performance by a small average amount relative to no assessment or to teacher assessment¶

Claim · All claims
Evidence · 1 study · quant-synthesis r? · q4 · i1 small · n=54 studies (k=141 effect sizes)
Currency · newest study 2020 · checked against its sources 2026-10-20 (automated, 1 of 1 entry) · not reviewed by a person

When learners evaluate one another's work, their subsequent academic performance improves by a small average amount (g = 0.31 against no assessment in one meta-analysis of 54 control-group studies). The claim covers both directions of the exchange — assessing peers and being assessed by peers — and the recorded evidence does not separate them.

Subclaims¶

q4 i1 A meta-analysis of 54 experimental/quasi-experimental control-group studies (k = 141) finds peer assessment produces a small-to-medium improvement in students' academic performance (essay writing, tests, and practical skills) relative to no assessment, and a comparable improvement relative to teacher assessment. → Double et al. 2020

Evidence¶

Double et al. 2020¶

Double, K. S., McGrane, J. A., & Hopfenbeck, T. N. (2020). The impact of peer assessment on academic performance: A meta-analysis of control group studies. Educational Psychology Review, 32(2), 481–509. doi:10.1007/s10648-019-09510-3

q4 · meta-analysis of experimental/quasi-experimental control-group studies · i1 · small-to-medium effect, g=0.31 · n=54 studies (k=141 effect sizes) · quant-synthesis · r?

A meta-analysis of 54 control-group studies (141 effect sizes; the abstract's figures — the Methods report 55 articles with 143 effect sizes, two of them removed as outliers before analysis) spanning primary, secondary and tertiary students across subjects, comparing peer-assessment interventions (grading, peer dialogue, written feedback) against no-assessment and teacher-assessment control conditions on academic performance (test scores, essay/writing quality, practical skills). Peer assessment improved performance relative to no assessment (g = 0.31, p = .004) and relative to teacher assessment (g = 0.28, p = .007), and its effect did not differ significantly from self-assessment (g = 0.23, p = .209). Meta-regressions found no significant moderation by online/offline delivery, assessment frequency or education level, though the subgroup of studies applying peer assessment only once did not show a significant effect on its own (g = 0.19, p = .103; multiple occasions g = 0.37, p < .001).

Studies bearing on this claim¶

Should learners assess each other's work, and does it need a rubric or script? This table gathers the studies behind this claim and the 16 narrower claims that bear on it: 16 distinct studies (5 quant-synthesis, 5 causal, 4 associational, 2 review), q2–q4, 6 of 16 with an effect size coded. One row per study, codes as each entry records them; no effect is averaged and no verdict is drawn. Built by scripts/synthesize_claim.py from eval/synthesis/; a row that looks wrong is fixed on the claim it comes from.

Discussion¶

Mechanisms. Two routes plausibly drive any performance benefit. First, reviewing peers' work forces learners to apply criteria to authentic artifacts, which resembles the worked examples benefit: studying others' solutions with evaluative intent exposes quality differences without the full load of production. Second, receiving structured feedback from peers adds feedback episodes beyond what an instructor alone can provide, aligning with assessment for learning principles. Both routes also position peer assessment as a form of active learning: learners construct evaluative judgments rather than passively receive them.

Moderators to establish. The literature suggests the effect depends heavily on conditions: rubric quality and criterion specificity, whether assessors are trained, the domain (peer judgment is more reliable for structured products than for open-ended creative work), and learner expertise — novices may lack the standards knowledge to judge accurately, echoing the expertise-reversal concerns in cognitive load theory. Anonymity and formative (vs. summative) use also moderate both accuracy and acceptance of peer judgments. Rubric scaffolding is likely a precondition rather than an optional enhancement — see Self-assessing against a rubric improved the quality of student work in two non-randomised studies, one of which also found more stress and performance-avoidance.

Open questions. Whether the benefit comes primarily from giving assessment (judging peers) or receiving it (being judged) remains contested; the meta-analysis recorded here does not separate the roles, and the one assessor-side study on this wiki is correlational (the quality of feedback students give peers predicts gains in their own work). A larger assessor gain would be consistent with the generative-processing rationale behind annotating improves learning. A second open question is whether peer grades are reliable enough for summative use, or whether the claim should be restricted to formative deployments as in assessment for learning improves achievement..

Merged from “Peer Assessment Improves Academic Performance” (peer-assessment-improves-academic-performance): Mechanisms. Peer assessment plausibly improves performance through several routes: acting as an evaluator forces learners to internalize quality criteria, which supports self-regulation of their own work; receiving feedback from peers adds perspectives beyond instructor feedback; and the social accountability of being reviewed can increase engagement. These mechanisms connect to Assessment for Learning and Self-Regulated Learning. Notably, the "giving" side may be the more powerful lever: reviewing a peer's work requires comparing it against criteria and exemplars, which is precisely the evaluative judgment learners often fail to apply to their own drafts. This evaluative-judgment route is consistent with the broader formative-assessment evidence summarized in Assessment for Learning Improves Achievement.

Moderators and boundary conditions. The benefit likely depends on the quality and specificity of the criteria learners are given — unstructured peer review tends to produce vague, unreliable judgments, while rubrics and exemplars anchor evaluations. Training in how to give feedback, and anonymity where social dynamics would distort judgments, are common design responses. Peer grades should generally be treated as formative rather than summative, since agreement with instructor ratings is imperfect and varies by task complexity; complex or open-ended products (essays, designs) show lower peer–instructor agreement than more objective work. Because peer feedback is one delivery channel for feedback effects generally, gains here should be interpreted against the baseline documented in Feedback improves learning by a medium amount on average, but its effect varies widely with the information it carries, and a substantial share of feedback effects are negative — poorly specified peer comments may add little over no feedback at all.

Open questions. Whether the performance gains come primarily from giving feedback, receiving it, or both remains unresolved, as does the durability of gains across domains and learner ages. Younger learners and novices may lack the domain knowledge to judge quality accurately, which could limit both the reliability of their reviews and the learning benefit of reviewing. The one meta-analysis recorded above (Double et al. 2020) rates the average effect but does not address these questions.

Merged from “Peer Assessment Improves Achievement” (peer-assessment-improves-achievement): Two mechanisms, one claim. Peer assessment plausibly improves achievement through two distinct routes: (1) recipients receive timely, actionable feedback that informs revision, and (2) assessors engage deeply with criteria and exemplars while evaluating others' work, which can clarify quality standards. These routes have different boundary conditions, and evidence for one does not automatically support the other. The second route resembles the benefit of studying worked examples — critiquing others' work exposes learners to models of quality without the full cost of producing work themselves.

Quality of feedback matters. Peer feedback is only as good as the criteria and training learners receive. Unstructured peer commenting tends to produce vague praise; rubric-supported, criterion-referenced peer assessment is more likely to yield feedback that supports improvement. This connects to Assessment for learning improves achievement — peer assessment is one delivery mechanism for formative assessment, and its achievement effects are plausibly mediated by the same feedback-quality conditions. The feedback route also depends on the general conditions under which Feedback improves learning outcomes operates: feedback must be specific, actionable, and acted upon.

Reliability concerns. Peer ratings can diverge from instructor ratings, particularly for novices and for complex or subjective work. Divergence is not necessarily a problem when peer assessment is used formatively (to drive revision) rather than summatively (to assign grades). Where stakes are attached, aggregation across multiple assessors and calibration against instructor-anchored exemplars are common mitigations.

Boundary conditions. The claim should be read as conditional rather than universal. Achievement gains are most plausible when: assessors are trained on explicit criteria; feedback is actionable and tied to revision; recipients actually act on the feedback; and the task domain has articulable quality criteria rather than purely tacit standards. Peer assessment used summatively with untrained assessors, or on tasks where novices cannot recognize quality, is unlikely to improve achievement and may mislead learners.

Assessor benefits. The second mechanism — learning while judging — is conceptually adjacent to Self-assessment improves self-regulated learning : applying criteria to someone else's work builds the evaluative knowledge learners then apply to their own. This route may operate even when the assessor's feedback is never used by the recipient, which makes it worth isolating empirically from the feedback-delivery route. It also aligns with the broader pattern that Active learning improves exam performance — acting as an evaluator is a more active engagement with content than passively receiving instructor comments.

Open questions. The strength of the achievement effect likely depends on whether feedback is actually acted upon, whether assessors are trained, and whether the assessed work is revised. The meta-analysis recorded above rates the average effect but did not test these moderators; studies isolating them are still needed.

Merged from “Peer Feedback Improves Work Quality” (peer-feedback-improves-work-quality): Dual benefit. Peer feedback is often framed as benefiting the receiver, but the act of reviewing a peer's work — diagnosing problems against criteria and formulating actionable comments — is itself a form of elaborative practice and can improve the reviewer's own subsequent work. This "giver gains" effect suggests peer feedback activities should be designed as learning experiences for both roles, not merely as a scalable substitute for instructor feedback. Reviewing also builds evaluative judgement: the capacity to judge quality in one's own work, which is arguably the more durable outcome. This connects peer feedback to Assessment for learning improves achievement., where students act as agents rather than recipients of assessment.

Moderators and boundary conditions. The quality of peer feedback depends heavily on scaffolding: without rubrics, exemplars, or training in how to critique, peer comments tend to be vague ("good job") or purely affective, and receivers may ignore or misapply them. Structured protocols — specifying what to look for, requiring specific and actionable comments, and pairing review with revision — are what convert peer commentary into improved work. Self-assessing against a rubric improved the quality of student work in two non-randomised studies, one of which also found more stress and performance-avoidance partly because they supply the shared criteria that make peer commentary specific and actionable. Anonymity, trust, and psychological safety shape whether reviewers give honest critique and whether receivers act on it. Domain expertise matters — peers may lack the knowledge to detect substantive errors, so peer feedback is best suited to clarity, structure, and audience-fit dimensions of work rather than deep correctness. Draft-stage timing also matters: feedback on work-in-progress can still change the product, whereas feedback on finished work rarely does — a point that aligns peer feedback with the formative logic of Assessment for learning improves achievement. and with the audience effects described in Authentic audiences improve student work..

Constraints. Peer feedback fails or backfires under identifiable conditions. When reviewers lack domain expertise and the work's quality hinges on factual or technical correctness, peer comments can be confidently wrong and receivers may "fix" non-problems or introduce errors. When feedback is unscaffolded and purely affective, it consumes class time without improving the product — and can erode trust if reviews feel arbitrary or unfair. Low psychological safety (competitive climates, non-anonymous review of high-stakes work) suppresses honest critique, producing inflated praise that receivers cannot act on. Finally, feedback delivered after submission rarely improves the work it addresses; without a built-in revision cycle, the activity yields little product improvement even if learning occurs.

Relation to collaboration. Peer feedback is a structured, role-differentiated form of the interactions studied under Collaborative learning improves outcomes.; its distinctive feature is that the product being critiqued belongs to an individual, so accountability and revision pathways are clearer than in group work.

Open questions. How much training is needed before peer feedback reaches useful quality, how receiver uptake should be assessed, and how peer feedback compares to instructor feedback across domains remain open. The one meta-analysis recorded above measured academic performance (tests, writing, practical skills) rather than the quality of the revised work itself, and found no moderator it tested, including delivery mode, frequency and education level, changed the effect. The scaffolding moderators described above are not among those it tested.

Related Claims¶

• Assessment for learning improves achievement. — peer assessment is a formative-assessment practice; this claim supplies the broader evidence base
• Feedback improves learning outcomes. — peer feedback is one delivery channel for the feedback effect
• Collaborative learning improves outcomes. — peer assessment structures peer interaction around evaluative criteria
• Self-assessing against a rubric improved the quality of student work in two non-randomised studies, one of which also found more stress and performance-avoidance — rubrics are the criterion scaffolding that makes peer judgment reliable
• Self-Regulated Learning — evaluating peers' work against criteria builds the evaluative judgment that self-regulated learners apply to their own work
• Self-assessment improves self-regulated learning — judging quality criteria benefits the assessor's own regulation
• Active learning improves exam performance — evaluating peers' work is an active engagement with quality criteria
• Participants reported significant improvement in their ability to deliver, receive, and implement feedback between the seminar and the time of survey — related
• Online peer assessment shifts instructional assessment from teacher-controlled to a combination of teacher assessment and student peer assessment — related
• The quality of feedback students give peers predicts gains in their own work, but no recorded study isolates a causal benefit of the assessor role — related
• Structured criteria in peer assessment are not shown to improve its effect on academic performance over freeform review — related
• In two secondary-school writing studies, peer reviewers' own performance predicted how helpful their feedback was but not how accurately it identified problems — related
• Peer Feedback Improves Writing — a narrower finding that bears on this claim
• Training in text structure improves college students' summaries, but summarizing has not been shown here to improve recall or writing more than rereading or other active reading — related
• Authentic audiences improve student work. — a real peer audience raises the stakes and quality of work
• Process Writing Improves Writing Quality — related
• Effective implementation of formative assessment is hindered by its complexity, summative-assessment pressure, and teachers' negative early experiences — related
• Peer and self-assessment both show significant effects on academic performance with no significant difference between their impacts — a broader claim this one bears on
• Students given AI assistance in peer feedback tended to rely on the AI-generated feedback and struggled to perform the task when the assistance was removed — related
• Student assessors led the pilot assessment of the Centering Students' Diverse Lived Experiences badge, shifting evaluation power toward student voice — related
• Actionable and descriptive feedback moves student learning forward, and effective feedback differs from less-effective feedback on identifiable qualities — related
• Qualitatively, AI-peer integrated feedback was associated with a broader repertoire of revision operations and more agentive cognitive strategies than peer feedback alone — related
• A formative assessment script increases the quantity and quality of peer feedback vocational students generate on a peer's draft — a narrower finding that bears on this claim
• Peer assessment of oral presentations improves self-concept relative to control and improves presentation performance, but outcome studies remain few — a narrower finding that bears on this claim
• Using rubrics has a positive, moderate effect on students' academic performance (corrected g = 0.45; standardized mean change = 0.38) — related
• Formative feedback dominates authentic assessment designs while sustainable feedback is rare — related
• Quality control in peer assessment can be exercised proactively or reactively on students and their assessments — related
• Web-SPA peer review activities showed considerable consistency between instructor and peer marks and improved quality of work — a narrower finding that bears on this claim
• Cognitive feedback benefits assessors rather than assessees, and weak students do not benefit from receiving more feedback — reports the opposite
• Process-based and product-based feedback produce larger performance gains than no feedback, with rubric-based feedback only marginal — a narrower finding that bears on this claim