Scientific peer review: process and reviewer responses

Scientific peer review is expert evaluation of a manuscript before publication or, in some models, after it. A reviewer does not stamp a paper as “scientifically true.” The job is to assess whether the work meets the journal's and field's standards, whether methods and conclusions are justified, whether results are presented clearly, and whether important problems have been missed by the authors or editor.

Who chooses the reviewers

Usually an editor or associate editor selects specialists in the subject. Authors may sometimes suggest or exclude candidates, but the journal keeps control of the choice. Reviewers assess the manuscript and recommend an outcome; the editor makes the final decision.

A useful review separates critical flaws from preferences. If the design cannot answer the research question, that is fundamental. If a reviewer would simply prefer a different headline, that is a minor editorial point.

In single-blind review, reviewers know the authors' identities while authors do not know the reviewers. In double-blind review, the journal tries to hide both sides. Complete anonymity is not always possible: a narrow topic, self-citations, a distinctive dataset or a particular instrument may suggest who wrote the paper.

Open review is an umbrella term for models in which reviewer identities, review reports or the history of discussion become public in whole or in part. Implementations differ, so always read the journal's actual policy.

No model automatically produces better science. What matters more is the transparency of the process, the reviewers' competence and the editor's responsibility.

How long does peer review take?

Anywhere from several weeks to many months. Timing depends on the discipline, the availability of reviewers, the number of revision rounds and the speed of the authors' responses. A journal may publish median time to first decision, but this is not a promise for a specific manuscript.

A claim of “complete peer review in 24 hours” for a complex study deserves extra scrutiny. A rapid desk rejection, by contrast, is entirely plausible because an editor can decline a manuscript before external review.

How to read reviewer reports

Do not answer immediately after the first emotional reading. First divide comments into four groups: items required by the editor; methodological issues; clarity questions; and suggestions with which you can reasonably disagree.

If two reviewers contradict each other, do not silently try to satisfy both. Explain the conflict to the editor and propose a reasoned solution.

How to respond

Copy each comment and place your response under it. State what changed and where. “Thank you, corrected” is too short for a substantive issue. A stronger answer is: “We agree that the original wording could imply causation. We replaced ‘causes’ with ‘is associated with’ in the title and Discussion, and marked the changes in the revised manuscript.”

If you disagree, say why: “We retained the original analysis because…; we added an explanation in Methods and a limitation in Discussion.” Professional politeness does not require surrender, but disagreement must be evidence-based.

Revise and resubmit

A revise-and-resubmit decision usually means that the journal is not accepting the present manuscript but is willing to consider a substantially revised version. It is not a promise of future acceptance. The new version may undergo another full review round.

Create a separate revision plan. Update manuscript, tables, figures and supplementary material together and check that new numbers do not contradict the abstract or conclusion.

For students: how to tell whether an article is peer reviewed

Do not judge from the appearance of a PDF. Identify the journal and read its official policy: does it describe a peer-review process, who publishes it, and which content types are reviewed? A peer-reviewed journal may also publish editorials, news items, letters or corrections that follow a different process.

Next, use your university library database or catalogue. If the database labels the journal as peer reviewed, still check the type of the particular article. Dates such as “received”, “revised” and “accepted” are useful clues, but their absence does not prove there was no review.

If an assignment requires a peer-reviewed source, record the journal and article type, not just the title of the PDF.

Checklist for a reviewer-response letter

read the editor's decision letter in full;
answer every reviewer comment;
identify every change precisely;
support disagreements with method or evidence;
update the abstract if the results changed;
make sure tables and text use the same numbers;
expand limitations when review reveals a genuine issue;
keep the tone calm and professional.

What happens when two reviewers disagree

Peer review is not a vote of “two in favour, one against.” One reviewer may find the method sound while another identifies a serious sampling problem. The editor compares arguments rather than counting positive and negative phrases. Sometimes a third specialist is invited; sometimes the editor asks the authors to address the disputed point directly.

Do not try to guess which reviewer “matters more.” Respond to each substantive comment. If reports conflict, say so calmly: “Reviewer 1 suggests removing analysis X as secondary, while Reviewer 2 considers it necessary. We propose retaining the analysis in the supplement and shortening the discussion in the main text.” This is a solution to competing requirements, not a choice of the more convenient reviewer.

Three kinds of comment require three kinds of answer

Fixable clarity. “The exclusion criterion is not explained.” Usually the right response is to add the explanation and point to the location.

Request for a new analysis. “Show the result separately for two age groups.” First ask whether the analysis was planned and whether the data support it. If appropriate, perform it and describe it. If statistically unsound or outside the design, explain why and consider adding a limitation.

A request that changes the research question. “Add a second experiment in another population.” That may be a good idea for future work without being a reasonable requirement for the present paper. Explain the boundaries of the study and adjust the manuscript so those boundaries are not hidden.

Example of disagreement without conflict

Comment: “The authors should remove the results for group B because the group is too small.”

Response: “Thank you. We agree that the small size of group B limits precision. However, this group was defined in the protocol before analysis and addresses a secondary research question. We retained the result, added confidence intervals, strengthened the limitation in Discussion and changed the conclusion from categorical to cautious wording.”

The author does not write “the reviewer is wrong.” The response gives a reason and reduces the strength of the claim where the data are weaker.

In a narrow field, a topic, dataset, self-citations or a distinctive instrument may still reveal the laboratory. Double blind means that the journal removes explicit identifiers as part of its procedure; it does not guarantee mathematical anonymity.

Authors should still follow anonymisation instructions: remove names, acknowledgements, identifying file metadata and phrases such as “in our previous study we showed…” when these reveal identity. Open review also varies: some journals reveal both names, some publish review reports, and some let reviewers choose whether to sign. Read the journal's policy rather than relying on the label alone.

How a student can check one specific article

First identify the content type: research article, editorial, letter, news item, correction or another format. Then read the journal policy to see which categories receive external review. After that, consult the bibliographic record in an academic database if it marks peer-reviewed content.

Dates such as received, revised and accepted are useful evidence of editorial processing but not absolute proof of the quality of review. For coursework, combine several independent signs instead of looking for one icon.

What happens after acceptance

The work is not always finished at “accepted.” Next comes the proof stage. This is where authors correct typographical errors, names, affiliations, table labels and production mistakes, not where they redesign the study or rewrite the results.

Check authors, affiliations, numbers in tables and correspondence between figures and captions carefully. After publication, many corrections require a formal correction notice.

What a good review report looks like

A useful report separates serious issues from minor ones. It first considers the research question, method and whether conclusions follow from the data. Then it asks for the explanations or analyses needed to assess the work. Style and typographical comments come after scientific issues rather than replacing them.

For authors, this distinction is helpful. A review consisting only of stylistic preferences does not automatically prove that the science is reliable. Conversely, a long list of comments does not automatically mean the paper is poor; a careful reviewer may be showing a path toward a publishable version.

After receiving the decision, make a comment table before editing the manuscript. This prevents you from fixing the first two points and forgetting the sixth in a long report. It also lets co-authors divide responsibility for statistics, methods, figures and wording without making contradictory changes.

If a review irritates you, do not answer on the same day. Separate tone from scientific content. A harsh sentence may hide a fair methodological issue, and a very polite request may ask for an analysis that does not fit the design. The response should follow the science, not the emotional tone.

Where to explain the work to a wider audience

A peer-reviewed paper primarily speaks to a professional community. Readers outside the field may need a second text that explains method and result without the conventions of a research manuscript. That text must not be presented as a peer-reviewed publication.

What Columnist provides

Science Official in the Mosaic network is a popular-science publication, not a peer-reviewed scientific journal. Through Columnist, a researcher can publish a readable column about the study alongside academic communication. The article carries the author's name and appears on an author page, but it does not undergo scientific peer review and does not create academic status.

After the scientific paper, prepare a separate explanation for a broad audience through Columnist, at columnist.mib.news.

FAQ

Can a reviewer reject my paper?

A reviewer recommends an outcome, but the editor formally makes the decision. In most systems, a reviewer is not the sole judge of the manuscript.

Does double blind guarantee anonymity?

No. Explicit identifiers are hidden, but in a narrow field authorship may sometimes be guessed from the topic, data or citation pattern.

Do I have to agree with every reviewer comment?

No. You must respond to every substantive point. Agree with justified criticism and explain disagreement calmly with evidence.

How can I check whether a specific article was peer reviewed?

Check the journal's policy, the article type and information in an academic library database. Not every item inside a peer-reviewed journal necessarily follows the same process.

Reading is half of it. The other half is a column with your name on it, in a real publication.

Start writing

All guides in the handbook

Peer review: how it works