Peer Review

Version

v2 · 2026-08-30 · History

Prime #

1491

Origin domain

Library & Information Science

Subdomain

editorial and professional review → Library & Information Science

Also from

Organizational & Management Science, Public Administration & Policy, Medicine & Healthcare, Engineering & Design, Accounting & Auditing, Education & Pedagogy

Aliases

Peer evaluation, Review by peers

Related primes

Evaluation, review artifact, External Analytic Challenge, peer debriefing, Validation, Verification, Reproducibility & Replicability, Editorial Independence

Core Idea

Peer review evaluates bounded work, proposed work, or professional performance through one or more people whose competence is relevantly comparable to that required to produce or judge it. The reviewers interpret evidence against a declared or reconstructable evaluative frame and return a judgment, critique, recommendation, score, finding, or attestation to someone able to use it. The use may be formative—revision, learning, defect correction—or summative—publication, funding, progression, certification, compliance, or public accountability.

The portable relation is bounded object or performance → peer reference class → competence-matched reviewer selection → criterion-bearing examination → independently attributable judgment or report → response, aggregation, or authorized use. “Peer” does not mean friend, exact equal, same employer, or member of an undifferentiated crowd. It means that the reviewer's relevant standing or expertise is measured against the work and practice under examination rather than supplied only by popularity, customer status, or hierarchical authority. “Review” names the organized evaluative process, not merely the document it produces.

The same relation occurs in distinct institutional substrates. NIH scientific review groups evaluate grant applications for scientific and technical merit; NASA engineering peer reviews bring appropriately experienced experts to requirements, designs, analyses, processes, and data; AICPA peer review examines accounting firms' quality-management and assurance practices; OECD committees use agreed criteria and other governments' representatives to review national performance; and students assess one another's work against educational criteria.^{[1][2][3][4][5]} The objects, standards, consequences, and authority differ, but the peer-relative evaluator role and criterion-to-judgment pathway remain literal. That unchanged multi-domain recurrence clears the prime bar.

Peer review is a fallible arrangement for informed scrutiny, not a truth machine. Reviewer selection can improve relevance while narrowing viewpoints. Anonymity can reduce some status cues while weakening accountability. Several competent reviewers can disagree because criteria are incomplete, evidence is ambiguous, or judgment is noisy. A well-run process makes these dependencies inspectable; it does not abolish them.

Structural Signature

Recognition roles:

• the bounded object or performance — a manuscript, proposal, design, program, professional practice, policy, case, or learner-produced work delimited for review;
• the sponsoring or receiving practice — the journal, agency, project, profession, committee, classroom, or other setting that commissions or recognizes the review;
• the peer reference class — the community of competence relative to which a reviewer counts as a peer;
• the competence match — evidence that each reviewer can engage the relevant substance, methods, risks, or standards;
• the separation condition — enough distinction from authorship or the reviewed performance for another judgment to be possible, with conflicts disclosed or controlled;
• the evaluative frame — criteria, standards, objectives, questions, rubrics, or professional expectations that make observations count;
• the review operation — inspection, reading, discussion, testing of reasons, comparison with criteria, and identification of strengths, defects, uncertainty, or alternatives;
• the attributable output — comments, scores, findings, recommendations, attestations, or a report traceable to reviewer roles even when identities are masked from some parties;
• the response or use route — revision, rebuttal, aggregation, editorial decision, funding action, design closure, quality improvement, certification, learning, or follow-up;
• the governance envelope — rules for confidentiality, conflicts, anonymity or openness, appeals, timing, evidence access, and final authority.

All ten roles need not be embodied in ten separate actors or documents. A classroom pair can combine sponsor, recipient, and author roles; an engineering panel can discuss and record findings in one session; an open post-publication review can omit identity masking. The invariant is functional: a bounded target is evaluated through competence qualified relative to that target, using an evaluative frame, and the resulting peer judgment is routed into a consequential or formative process.

A competence match without an evaluation is consultation. An evaluation by a manager solely because the manager has authority is supervisory appraisal. A consumer star rating may be a review artifact without peer qualification. A crowd vote may aggregate preference without relevant competence or explicit criteria. Peer review begins only when peer-relative qualification matters to why the judgment should carry weight.

What It Is Not

• Not a guarantee of correctness, safety, novelty, or fitness. Reviewers assess material available under a frame. They can miss defects, share mistaken assumptions, or apply poor criteria.
• Not synonymous with scholarly journal refereeing. Manuscript review is one important species. Grant panels, technical engineering reviews, professional practice review, policy peer review, and student peer assessment retain the same abstract roles.
• Not every evaluation by several people. A popularity poll, customer rating, citizen jury, or generic committee becomes peer review only when relevant peer competence is constitutive of reviewer selection and authority.
• Not hierarchy-free evaluation. A peer reviewer may be senior, external, appointed, or institutionally powerful. The peer relation concerns relevant competence or practice standing, not equal organizational rank.
• Not necessarily anonymous. Single-anonymized, double-anonymized, open, signed, and partially disclosed models can all instantiate peer review. Identity policy is a governance variable, not the core.
• Not necessarily independent of the receiving institution. Internal technical peers may review a design. What is required is sufficient separation from producing the specific work to make another evaluative judgment possible, with conflicts controlled.
• Not necessarily pre-publication or pre-decision. Post-publication review, recurrent professional monitoring, and retrospective case review can qualify. Pre-commitment timing changes corrective leverage but not the broad identity.
• Not consensus. Reviewers may submit separate reports, minority findings, or incompatible scores. Aggregation and final decision are downstream arrangements.
• Not replication or reproduction. A reviewer can inspect claims and methods without independently repeating the work. Replication supplies different evidence.
• Not the final decision. Editors, funding officials, project managers, accrediting bodies, or instructors may use peer judgments while retaining decision authority and additional criteria.
• Not the persisted report alone. A report is an output artifact. The peer-review abstraction includes qualification, assignment, examination, criteria, and routing.

Broad Use

Scholarly manuscripts and research outputs. Editors obtain advice from subject-area experts, manage conflicts and confidentiality, and use reports in publication decisions. COPE's reviewer guidance makes expertise matching, objective constructive critique, confidentiality, conflicts, and accountable conduct explicit.^[6] Models vary in identity disclosure, publication of reports, interaction among reviewers, and whether review occurs before or after publication. The invariant is not secrecy but competence-qualified evaluation of the submitted scholarly object.

Research funding. NIH's first-level review uses scientific review groups composed primarily of non-federal scientists with relevant expertise. Applications are evaluated against established criteria, receive criterion and overall-impact judgments, and pass into a second level that considers mission relevance.^[7] This cleanly separates peer evaluation from final allocation: merit review informs but does not itself make every funding decision.

Engineering and software-intensive systems. NASA engineering peer reviews examine bounded technical material such as requirements, interfaces, designs, analyses, manufacturing plans, test data, and processes. Reviewers bring relevant skill and experience, surface defects and alternatives, and route issues toward closure.^[2] Here “peer” is not scholarly authorship; it is technical competence relative to the product and life-cycle issue.

Professional quality and assurance. AICPA's Peer Review Program uses qualified reviewers to examine accounting and auditing practices, including firms' systems of quality management. The object can be a practice system and selected engagements rather than a single document, and the output can carry monitoring and remedial consequences.^[3] Clinical and other professional reviews similarly assess conduct or practice under field standards, although local legal protections and credentialing rules differ.

Intergovernmental policy. OECD peer review treats a member country or policy performance as the object, uses agreed principles and criteria, designates peer actors, and follows procedures leading to findings and recommendations. The OECD's own methodological account identifies those recurring structural elements while noting that mechanisms differ across substantive areas.^[4] A government can therefore be a peer relative to another government without resembling an individual author.

Education and learning. Student peer assessment asks learners of comparable status to evaluate the quality, value, or level of one another's work. Rubrics, training, calibration, and teacher oversight can turn the exercise into both feedback for the recipient and evaluative learning for the reviewer. Topping's review documents substantial variation in objects, outputs, directionality, and purpose while retaining peer assessment as a coherent practice.^[5]

Organizational and professional development. Design critiques, code reviews, clinical case conferences, faculty review, and communities of practice can instantiate peer review when competence matching and criterion-bearing judgment are real. Casual colleague feedback is adjacent but falls short when the target, frame, reviewer role, or use route is not defined.

Clarity

The central diagnostic is not “Did another person comment?” but “Why does this person's judgment count as peer judgment about this target?” The answer should identify a reference class and competence match. A cardiologist may be a peer for a clinical reasoning case yet not for a cryptographic proof. A senior editor may have decision authority yet lack the specialist competence to replace a referee. A student can be a legitimate peer for applying a taught rubric to another student's essay even without professional mastery, because the peer relation and task are calibrated to the learning setting.

Peer review also separates four stages often conflated in everyday speech. Selection establishes who is qualified and sufficiently unconflicted. Evaluation applies criteria to the bounded object. Representation records or communicates the findings. Use converts those findings into revision, learning, certification, funding, publication, or another consequence. Failure at one stage does not prove failure at all others. Excellent reviewers can be assigned an invalid criterion; a sound report can be aggregated badly; an editor can reasonably depart from reports using information reviewers did not possess.

Masking vocabulary requires care. “Single-blind” and “double-blind” are historically common but ambiguous about whose identity is hidden, so “single-anonymized” and “double-anonymized” are often clearer. No identity policy eliminates every cue: subject matter, citations, writing style, project history, or community size can reveal likely authors. A controlled experiment at WSDM 2017 found that single-blind reviewers favored famous authors and high-prestige institutions relative to double-blind reviewers in that setting, but this result does not establish one universal effect size or prove that anonymization removes all bias.^[8]

The adjective “peer” therefore does two kinds of work. It supplies epistemic relevance—reviewers can understand and challenge the material—and social legitimacy—the practice treats their judgment as standing within a reference community. Those functions can diverge. A person can have deep expertise but an unmanaged conflict, or hold recognized standing while lacking the narrow competence needed for a particular object. Robust systems test both rather than treating reputation as a complete proxy.

Manages Complexity

Peer review distributes scrutiny when a decision-maker cannot personally reproduce every method, inspect every component, or master every specialty. Reviewer assignment decomposes a heterogeneous object into competence-matched perspectives. Criteria focus attention. Separate reports preserve disagreement. Aggregation and response procedures turn plural judgments into an operable recommendation or revision path. The arrangement thereby converts diffuse community knowledge into bounded, accountable evaluative work.

This compression is lossy. A score can erase why reviewers disagree. A consensus statement can hide minority warnings. A gate verdict can make a multidimensional evaluation appear binary. Good process design therefore preserves enough trace to distinguish object defects, criterion disputes, reviewer uncertainty, and downstream policy choices. The review artifact helps, but only if warrants remain connected to findings and if recipients can respond.

Peer review also manages the “unknown unknown” problem imperfectly. A producer is adapted to the assumptions and vocabulary used to build the work. A qualified second reader may recognize an omitted control, untested interface, unsupported inference, or professional-standard violation that is locally invisible. Yet reviewers drawn from the same network may share the blind spot. Competence matching and cognitive diversity are related but not identical selection goals.

Abstract Reasoning

Let (x) be a bounded object, (P(x)) a peer reference class defined relative to the expertise needed for (x), (C) the evaluative frame, (e) the available evidence, and (r_i=E_i(x,e,C)) reviewer (i)'s judgment. A peer-review system additionally specifies assignment (A), conflict and disclosure rules (G), communication (M), aggregation \(H(r_1,\ldots,r_n)\), and a use or response function (U). This is a descriptive role model, not a claim that all peer judgment is numerical.

The model licenses several deductions:

1. If (P(x)) is too broad, nominal peers may lack relevant competence; if too narrow, conflicts, shared assumptions, and reviewer scarcity increase.
2. If reviewers see different evidence or interpret different criteria, disagreement cannot be diagnosed as mere unreliability until those inputs are aligned.
3. If all reviewers share one training lineage or incentive, adding reviewers may multiply correlated judgment rather than independent evidence.
4. If (H) converts separate reports into one score, the aggregation rule can reverse priorities or suppress veto-like concerns.
5. If the recipient cannot revise, rebut, appeal, or document a response, the process becomes a one-way gate even when reviewers intended formative critique.
6. If the final authority uses additional mission, budget, legal, or portfolio criteria, departure from reviewer ranking is not automatically corruption; it becomes suspect when the extra basis is hidden or contradicts governing rules.
7. If identity masking changes cues but not criteria, conflicts, or calibration, it changes only part of the error structure.
8. If reviewers can infer author identity from the artifact, nominal double anonymization does not yield full informational blinding.
9. If a peer review occurs after publication or deployment, it can correct the record or future practice even though it cannot restore the original pre-commitment choice.
10. If review findings are never checked against later outcomes, the institution cannot learn whether reviewer confidence or criteria were well calibrated.

Reliability and validity must remain separate. Reviewer agreement concerns consistency; it does not show that the shared judgment tracks the right standard. Valid criteria applied inconsistently and invalid criteria applied consistently are different failures. Empirical reviews of editorial peer review have repeatedly cautioned that plausible institutional value should not be inflated into strong universal causal claims without comparative evidence.^[9] Research on bias likewise identifies multiple possible mechanisms—status, affiliation, nationality, gender, discipline, content, and confirmation effects—whose evidence and operationalization vary.^[10]

Knowledge Transfer

Peer review transfers as a design grammar rather than as one fixed procedure. When a field faces a difficult evaluation, ask: What exactly is the object? Relative to what practice or competence is someone a peer? What knowledge must be matched? What separation and conflict controls are necessary? Which criteria and evidence will reviewers receive? Are judgments independent, deliberated, or sequential? What record survives? Who can respond? Who has final authority? How will the system learn from disagreement and outcomes?

NASA engineering review teaches scholarly editors to match reviewers to affected interfaces rather than merely to broad topic labels. NIH grant review teaches design teams to distinguish merit judgment from portfolio or mission decision. OECD country review shows that peers can be institutional actors and that review can operate through recommendations and mutual learning rather than binary acceptance. Student peer assessment shows that producing a review can be an educational intervention for the reviewer, not only a service to the recipient.

Concrete mechanisms do not transfer automatically. Confidentiality suitable for an unpublished grant may conflict with open public-policy accountability. A classroom rubric may be inappropriate for exploratory research. A technical review's action-item closure can be too rigid for plural theoretical disagreement. The prime carries the role structure and design questions; each domain supplies the legitimate peer class, standards, evidence access, legal protections, authority, and consequences.

Examples

Formal / abstract

A sponsor has a bounded proposal (x) spanning statistics and field operations. It defines criteria (C=(c_1,c_2,c_3)) for methodological validity, feasibility, and expected value. Reviewer A is qualified in statistical design; reviewer B is qualified in operations; both disclose conflicts and receive the same evidence. They write separate reports and scores, then discuss a discrepancy: A treated an unmeasured confounder as fatal, while B assumed a planned field procedure controlled it. The sponsor records the unresolved assumption, requests clarification, and asks both reviewers to update their judgments. Peer review here is not the average score. It is the competence-matched, criterion-bearing, attributable evaluation-and-response process that exposes why judgments differed.

Mapped back: The proposal is the bounded object; statistics and operations define the peer reference class and competence matches; disclosures implement the separation condition; the three criteria supply the evaluative frame; separate reports are attributable outputs; clarification and re-evaluation form the response route; and the sponsor retains final-use authority.

Applied / practice

NASA convenes an engineering peer review of an autonomous fault-detection subsystem before a major project review. The reviewed object includes requirements, interface definitions, analysis, test evidence, and operating assumptions. Reviewers are selected for relevant system and life-cycle experience and sufficient independence from the executing team. During the focused examination they identify an interface assumption that makes a recovery mode unsafe under one timing condition. The team records the issue, revises the design and test plan, and tracks closure before the higher-level review. The peer process adds expert scrutiny and corrective options; it does not itself certify the whole mission or replace later readiness authority.^[2]

Mapped back: The subsystem package is the bounded object; experienced technical reviewers provide peer competence; the project rules provide governance; safety, interface, and performance obligations form the criteria; the timing defect is an evaluative finding; and design revision plus issue closure is the use route.

Contrast cases

• A shopping platform's star rating is a review artifact but not necessarily peer review: customers have experience, yet no competence reference class may be required.
• A chief executive's performance rating of a subordinate is an evaluation based on authority; it becomes peer review only if relevant peers rather than the hierarchy supply the constitutive judgment.
• Reproducing an experiment can provide powerful evidence without evaluating a manuscript through peers.
• An editor's desk rejection can be an editorial evaluation without peer review.
• A hallway comment from a colleague can be useful peer feedback but lacks a bounded commissioned review process unless roles, target, and use are made definite.

Structural Tensions

T1: Competence depth versus perspective breadth. Narrow specialization improves technical relevance but can concentrate assumptions and conflicts; broader panels increase viewpoints but may dilute object-specific understanding. Diagnostic: Does every material dimension have a competent reader, and are their blind spots sufficiently nonidentical?

T2: Independence versus contextual knowledge. Reviewers distant from production may challenge assumptions more freely, while insiders understand tacit constraints and history. Diagnostic: What context is needed to judge fairly, and what separation is needed to judge rather than defend the work?

T3: Anonymity versus accountability. Masking can reduce some status cues and retaliation risks, while openness can expose conflicts, improve civility, and make warrants attributable. Diagnostic: Which error is identity policy intended to reduce, and what new error does that policy create?

T4: Standardization versus expert discretion. Criteria and rubrics improve consistency and auditability, while novel or complex work can contain qualities the frame did not anticipate. Diagnostic: Which judgments are rule-bound, and how may a reviewer surface an important extra-criterion concern without silently changing the standard?

T5: Independent judgment versus deliberative correction. Separate reviews preserve independent evidence; panel discussion can resolve misunderstanding but also create conformity and status pressure. Diagnostic: Are initial judgments captured before discussion, and can a minority rationale survive aggregation?

T6: Formative improvement versus summative gatekeeping. Review can help improve work or decide whether it passes. When the same exchange does both, authors may conceal uncertainty and reviewers may write verdicts rather than useful diagnoses. Diagnostic: Is the current output meant to change the work, authorize it, or both—and are response rights appropriate to that purpose?

T7: Confidentiality versus public accountability. Confidentiality can protect unpublished ideas, sensitive cases, and candid reports; secrecy can conceal conflicts, weak reasoning, and inconsistent treatment. Diagnostic: Which information needs protection, from whom, for how long, and what process remains auditable?

T8: Review burden versus scrutiny value. Additional reviewers and rounds may find more problems but consume scarce expert attention and delay beneficial work. Diagnostic: Is another review expected to add independent information or merely repeat a saturated judgment?

T9: Community self-regulation versus community closure. Peers possess the competence required for meaningful judgment, but established peers can also reproduce orthodoxy, exclude outsiders, or protect shared interests. Diagnostic: Does peer qualification identify genuine task competence, or mainly institutional membership and reputation?

Structural–Framed Character

Peer Review is mixed-framed. Its roles are highly structural: bounded object, competence reference class, reviewer selection, evaluative frame, examination, judgment, record, response, and authority travel intact across professions and institutional scales. The vocabulary also travels literally; NASA, NIH, AICPA, OECD, journals, and educators independently call their processes peer review or peer assessment rather than borrowing a decorative metaphor.

The abstraction remains human-practice-bound and evaluatively weighted. Nature does not appoint peers, disclose conflicts, set review criteria, or authorize a report. Institutions define what competence counts, who may review, which standards govern, what is confidential, and what consequences follow. Those choices can encode professional values and power. Institutional origin is therefore substantial, but no single institution owns the pattern. Import-versus-recognize leans structural because the same functional arrangement is recognized across domains even as local rules change.

Its character: a portable social-evaluative structure whose peer qualification and consequence system must always be instantiated by a practice, but whose literal cross-domain recurrence is broad enough to constitute a prime.

Substrate Independence

Role preservation: 0.90. Manuscript, grant, design, quality-management system, national policy, clinical performance, and student work can replace one another while target, peer class, competence match, criteria, judgment, and use retain their functions.

Vocabulary independence: 0.90. “Peer,” “review,” “reviewer,” “criteria,” “report,” “conflict,” and “recommendation” are ordinary working terms in all major applications. Local vocabulary adds detail without replacing the core.

Intervention independence: 0.80. Reviewer matching, conflict control, independent initial judgments, criteria clarity, response rights, aggregation rules, and outcome calibration are portable levers. Exact confidentiality, scoring, authority, and appeal rules remain local.

Recognition versus import: 0.85. Cross-domain cases are institutionally attested instances under the same name and role structure, not analogies inferred only after abstraction.

Composite: 0.88. The pattern clears the prime threshold despite being limited to social and institutional evaluation. Substrate independence requires literal transfer across domains, not occurrence in nonhuman physical systems.

Relationships to Other Abstractions

Current abstraction Peer Review Prime

Parents (1) — more general patterns this builds on

• Peer Review is a kind of Evaluation Prime

  The accepted reference-grade review places Peer Review under Evaluation because the child instantiates or depends on the parent's broader structure while retaining its own constitutive identity.

Hierarchy path (1) — routes to 1 parentless root

• Peer Review → Evaluation → Comparison → Self Checking

Neighborhood in Abstraction Space

Peer Review sits in a sparse region of abstraction space (99^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely rather than landing on a neighbor.

Family — Verification, Screening & Reference Standards (18 primes)

Nearest neighbors

• Evaluation — 0.71
• Schismogenesis — 0.66
• Recognition Justice — 0.66
• Spaced Repetition — 0.64
• Coaxiality — 0.64

Computed from structural-signature embeddings · 2026-09-10

Not to Be Confused With

• Evaluation. Evaluation is the strict genus: apply a criterion-bearing frame to a bounded object and produce a judgment. Peer Review specializes it by requiring a socially recognized peer reference class, competence-matched reviewers distinct from producing the work, and a governed route for their judgments. Tell: would the evaluation retain its identity if performed by an unqualified manager, automated rule, customer, or solitary owner? If yes, it is evaluation but not necessarily peer review.
• Review / Review Artifact. The live domain-specific Review node is a persisted, addressable artifact binding evaluator, object, verdict, and warrant. A peer-review process may produce several reports, oral findings, scores, or no public artifact; a product review can be an artifact without peer qualification. Tell: are you identifying the competence-governed process or the record it leaves?
• External Analytic Challenge. This prime requires a competent challenger outside the authoring frame, explicit probes, a response, and pre-commitment revisability. Peer review may satisfy that architecture, but post-publication review, certification review, and summative practice monitoring need not. Tell: must the challenge arrive while the work can still change and receive a documented response, or is peer judgment being used for another purpose?
• Peer Debriefing. This domain-specific qualitative-research method uses a competent, removed peer to probe interpretation during analysis and build a trustworthiness trail. It is a narrow formative species, not the umbrella peer-review relation. Tell: is the target an evolving qualitative analysis under the peer-debriefing tradition, or any bounded work or performance reviewed by relevant peers?
• Delphi Method. Delphi repeatedly elicits and feeds back usually anonymous expert judgments to develop convergence or characterize disagreement. Peer review can consist of one pass and need not seek consensus. Tell: is iteration with controlled feedback among experts constitutive, or only competence-qualified evaluation of an object?
• Verification and Validation. Verification tests conformance to stated requirements; Validation asks whether an artifact serves its intended purpose. Peer reviewers may perform either, but they can also judge novelty, significance, pedagogy, professional quality, or policy performance. Tell: is the identity defined by the question asked, or by who performs the governed evaluation?
• Reproducibility and Replicability. These concern obtaining consistent results through repeated computation, measurement, or study. Reviewers normally inspect evidence and reasoning without repeating the full work. Tell: is an independent result being produced, or a peer judgment about the submitted object?
• Editorial Independence. Editorial Independence separates content decisions from improper owner, sponsor, or advertiser pressure. Peer advice can support it, but an editor can receive reviews and still lack independence, or maintain independence while making some decisions without peer review. Tell: is the problem reviewer competence or insulation of final editorial authority?
• Legitimacy-Yielding Inquiry Insulated from Decision. That prime protects a careful advisory inquiry by separating it from the authority making the consequential decision. Peer review may provide such inquiry, but formative classroom or design review need not yield public legitimacy or institutional separation. Tell: is protected advisory independence the essential mechanism, or peer-qualified evaluation more broadly?

Solution Archetypes

No catalogued solution archetypes reference this prime yet.

References

[1] National Institutes of Health, NIH Grants Policy Statement, §2.4, “The Peer Review Process,” revised March 2026, https://grants.nih.gov/grants/policy/nihgps/HTML5/section_2/2.4_the_peer_review_process.htm. registry ↩

[2] NASA, NASA Program/Project Management Handbook, NASA/SP-20220009501, rev. June 2024, §5.10, https://www.nasa.gov/wp-content/uploads/2024/09/pm-handbook-nasa-sp-2014-3705-2024jun.pdf. registry ↩a ↩b ↩c

[3] AICPA Peer Review Board, Peer Review Standards Update No. 2: Reviewing a Firm's System of Quality Management and Omnibus Technical Enhancements, approved November 2024, effective 2024–2025, https://www.aicpa-cima.com/resources/download/peer-review-standards-update-no-2-reviewing-a-firms-system-of-quality. registry ↩a ↩b

[4] OECD, Peer Review: An OECD Tool for Co-operation and Change, SG/LEG(2002)1, 2003, https://www.oecd.org/content/dam/oecd/en/publications/reports/2003/01/peer-review_g1gh3114/9789264099210-en-fr.pdf. registry ↩a ↩b

[5] Keith J. Topping, “Peer Assessment Between Students in Colleges and Universities,” Review of Educational Research 68(3), 1998, 249–276, https://doi.org/10.3102/00346543068003249. registry ↩a ↩b

[6] Committee on Publication Ethics, Ethical Guidelines for Peer Reviewers, version 2, 2017, https://doi.org/10.24318/cope.2019.1.9. registry ↩

[7] National Institutes of Health, “First Level: Peer Review,” NIH Grants & Funding, 2026, https://www.grants.nih.gov/grants-process/review/first-level. registry ↩

[8] Andrew Tomkins, Min Zhang, and William D. Heavlin, “Reviewer Bias in Single- versus Double-Blind Peer Review,” Proceedings of the National Academy of Sciences 114(48), 2017, 12708–12713, https://doi.org/10.1073/pnas.1707323114. registry ↩

[9] Tom Jefferson, Philip Alderson, Elizabeth Wager, and Frank Davidoff, “Effects of Editorial Peer Review: A Systematic Review,” JAMA 287(21), 2002, 2784–2786, https://doi.org/10.1001/jama.287.21.2784. registry ↩

[10] Carole J. Lee, Cassidy R. Sugimoto, Guo Zhang, and Blaise Cronin, “Bias in Peer Review,” Journal of the American Society for Information Science and Technology 64(1), 2013, 2–17, https://doi.org/10.1002/asi.22784. registry ↩

[11] “Peer review,” Wikipedia, frozen revision 1363704620, https://en.wikipedia.org/wiki/Peer_review. Discovery provenance only. registry