What is peer review? Meaning, purpose, and the process step by step

In scholarly publishing, peer review (also called refereeing) is the process by which a journal asks independent researchers in the same field to evaluate a submitted manuscript and advise the editor. The reviewers, the peers, assess whether the research question is worth answering, whether the design and methods can answer it, whether the statistics are sound, whether the conclusions follow from the results, and whether the report is complete enough for others to judge and reproduce it. They recommend; the editor decides.

A peer-reviewed article is one that passed this process before publication. The label says that qualified readers scrutinised the work and the journal acted on their advice. It does not say the findings are correct, and the evidence on this page shows why that distinction matters.

Peer review does two jobs. It screens: editors use reviewers' judgments to decide what a journal publishes, which is why acceptance rates at leading journals are in the single digits. And it improves: reviewers' comments change the manuscript before it is published. Researchers overwhelmingly believe in both functions, and the measured effect on manuscript quality is real, if modest.[1][2]

COPE's guideline describes peer review models along six dimensions: timing (pre-publication, post-publication, preprints), identifiability (double-anonymous, single-anonymous, open), mediation (editors mediate, or reviewers and authors interact openly), whether reviews are published, who facilitates the review, and who owns it.[29] The common combinations:

• Single-anonymous. Reviewers know the authors; authors do not know the reviewers. The most common model in medicine.
• Double-anonymous. Neither side is named. Intended to reduce status and affiliation bias, which randomized experiments have measured.
• Open and transparent. Identities are known and, in transparent models, the reviews are published alongside the article.
• Post-publication. Commentary and formal review after the article appears, on preprint servers and some journals.
• Registered reports. The protocol is reviewed and accepted in principle before results exist, removing the incentive to report only positive findings.

Peer review improves manuscripts and blocks obvious problems. It is also inconsistent, misses most inserted errors, is swayed by author status and by positive results, and cannot detect fabricated data from the text alone; most retractions are for misconduct that reading did not catch.[4][5][26]

1. Read the journal's policy page. Reputable journals state their peer review model on an About or Editorial Policies page, and many now publish decision times and acceptance rates; our journal pages quote them.
2. Use a database limiter. Library databases and Ulrichsweb flag journals as peer reviewed or refereed. Google Scholar does not label articles this way.
3. Check the article type. Original research, systematic reviews and most review articles are peer reviewed; editorials, news, commentaries and letters usually are not, even in peer-reviewed journals.
4. Look at the dates. Peer-reviewed articles typically show received, revised and accepted dates.

Accepted manuscripts move to production. Rejected ones usually go to another journal, and most published papers were accepted where they were first submitted; the ones that were not tend to end up in journals that publish more resubmissions.[27] If you are the author, the preparation that survives review is the same at every journal: the right reporting checklist, statistics that match the design, conclusions that match the results, and references that exist.