You read Reviewer 2’s comment and your blood pressure rises: they have misread the analysis, asked for something already in the paper, or corrected you with a fact that is not true. You compose a three-paragraph reply in your head that you will never send. It is the hardest part of a revision round, and not because of the science: because of the tone.
The good news is that you can almost always disagree and come out ahead. Editors have read hundreds of response letters and recognise at once the one that defends with reasons and the one that is angry. The first usually ends in acceptance; the second in another round or a rejection, even when you are right.
This guide helps you decide whether they are really wrong, write the response with four rules and phrases that work, know when to flag something to the editor, and avoid the mistakes that cost most.
Almost every comment that looks mistaken is one of these three things, and each calls for a different response. Spend half an hour sorting before you write a line.
This is the most frequent case. They ask for an analysis that is already in the results section, or read as a limitation something you presented as a design decision. Even if you are right, the fact that they misread it means the text did not explain it clearly enough. If an expert reviewer, reading attentively, did not see it, other readers will not either. The right answer is not “we already said so” but “thank you, we have made it more visible”, with a real change in the manuscript.
Two competent people can choose two different analyses for the same question. The reviewer prefers a mixed model and you ran a repeated-measures ANOVA; they want a multiple-comparison correction and you have a single preregistered primary outcome. There is no error here, there is judgement, and it is resolved with a justification and, if you can, a sensitivity analysis showing the conclusion does not depend on the choice.
It is the least frequent and the most delicate: the reviewer states something demonstrably false, such as a misquoted fact from the literature, a formula misapplied or a figure that is in the paper but in a different table. You answer with the evidence in front of you and no triumph: the page, the table, the reference. And if you can, with a change in the text that spares another reader the same stumble.
They work the same for a minor comment as for a substantive objection, and they are the ones that most often separate a response that persuades from one that irritates.
There are no magic formulas, but there is a register editors recognise as professional. These are the turns that recur in well-received responses:
Avoid “the reviewer is wrong”, “as we clearly explained”, “this is a misunderstanding” and any sentence that puts the blame on the other person’s reading. Also avoid defensive one-line replies (“Not applicable”) and disproportionate ones: three pages to rebut a minor comment convey more nervousness than confidence. The length of the response should match the weight of the comment.
Reviewer 2: “The sample size (n = 84) is clearly insufficient for a mediation analysis. The authors should acknowledge that the study lacks power.”
A response that does not work: “The reviewer overlooks that we ran an a priori power analysis. With respect, we believe the comment is unfounded.”
A response that does: “We thank the reviewer. The sample size was set by an a priori Monte Carlo simulation for the expected indirect effect (described in the Participants section, page 6), which gave power of .80 at n = 82. We understand that this calculation was not prominent in the previous version, so we have moved it to the first paragraph of the section and added to the discussion that power covers only medium-sized effects (page 14, paragraph 3).”
The second response gives a reason with data, locates the change, acknowledges where the misunderstanding came from and adds an honest limitation. It is much harder to rebut, and the reviewer does not need to feel wrong to close the point.
Almost never. The response is addressed to the reviewers and the editor reads all of it. There are three situations where it is worth saying something separately, in the cover letter to the editor and not in the response table: when a reviewer asks for something that changes the aim of the study, when two reviewers contradict each other on a central point and you cannot satisfy both, and when a comment has inappropriate tone or content. In all three, a clear sentence without drama is enough: the editor will know what to do.
Leave the letter overnight and reread only the tone. If there is a sentence you would not say in conversation with that reviewer sitting across from you, change it. Then check that every change you promise is really in the manuscript, that the numbers still agree wherever you touched something, and that you have answered every point, including the ones you consider obvious. An unanswered point weighs more than a well-argued disagreement.
You can disagree, but with a reason and without putting it that way. Write “we respectfully disagree because…” and give a data point or a reference. The word “wrong” puts the reviewer on the defensive and adds no information for the editor.
No. You have to respond to all of them, not change everything. You can keep your version if you give a convincing reason and, where possible, make an adjustment that prevents the misunderstanding. What does not work is ignoring a comment.
Explain why it does not apply and offer an alternative that addresses the underlying concern. Often the reviewer is worried about something specific (bias, power, collinearity) and is satisfied by seeing a different analysis that addresses it.
Say so plainly in the response, justify the option you chose and mention the disagreement in the letter to the editor. Do not try to please both with a halfway version: editors prefer a reasoned decision.
Yes, the editor decides and may tell you a specific comment need not be addressed. If they think it necessary, they will say so in their letter. You should still answer it with a reason: do not count on that decision from the editor.
Respond to the content and leave the tone. If it is inappropriate, flag it to the editor in the separate letter, in a sentence or two and without adjectives. The response you sign should be the calmest document in the file.