To respond to reviewer comments, the structure that works is always the same: copy each comment verbatim, answer right underneath it point by point, state exactly what you changed, and point to the page of the revised manuscript where the change now lives. Always thank the reviewer, never argue upfront, and when you genuinely disagree, explain why with arguments and references instead of a flat refusal. Below you have concrete templates for the statistical comments that come up most often: small sample, assumptions, statistical power, alternative analyses, and legitimate disagreements.
Responding to reviewers' statistical comments is an art that is learned through experience, but one that can also be systematized with a few concrete strategies. The difference between a response that satisfies the reviewer and one that generates a new round of objections usually lies in the tone, the thoroughness, and the transparency with which you address each point. I have seen excellent manuscripts rejected because of defensive or superficial responses, and mediocre manuscripts accepted thanks to impeccable responses that demonstrated rigor and willingness to improve the work.
The first thing you should do when you receive a review letter is separate your emotions from the task. It is normal to feel frustrated or even angry when someone criticizes your statistical work, especially if you believe the criticisms are unfair or uninformed. But the response to reviewers is not the place to express those emotions. Let the letter sit for a few days before you start responding, and when you do, adopt a professional and grateful tone, even in the face of comments you consider incorrect.
Strategies for the most common comments
There are certain statistical comments that appear over and over again in reviews, and having a prepared strategy for each one will save you time and anxiety. The most common is probably the request for additional analyses: "the authors should control for age and sex," "a sensitivity analysis excluding outliers would be advisable," "it is recommended to supplement with a Bayesian analysis." In response to these requests, the best approach is usually to carry them out. Even if you believe they are unnecessary, implementing them demonstrates good will and strengthens your manuscript. If the results do not change (which is most often the case), you can write something like: "Following the reviewer's recommendation, we repeated the analyses controlling for age and sex. The results are virtually identical to those originally reported (see Table S2 in the supplementary material), indicating that our conclusions are robust to the inclusion of these covariates."
Another frequent comment is the questioning of sample size: "the sample is small," "the authors do not justify the sample size." If you conducted an a priori power analysis, cite it and explain the parameters you used. If you did not (something that happens more often than desirable), you can perform a post-hoc power analysis or, better yet, a sensitivity analysis showing the minimum effect size your sample could detect with reasonable power. This does not substitute for a priori planning, but it at least provides the reviewer with information about your study's ability to detect relevant effects. Watch out for one nuance that slips into a lot of letters: post-hoc power computed from your own observed effect is circular, and some reviewers will call it out, so before you put it in the response make sure you know why post-hoc power is invalid and what to report instead.
When the reviewer suggests an analytical method different from the one you used, the key is not to become defensive. If their suggestion is reasonable, implement it and compare the results with those of the original analysis. If you believe their suggestion is not appropriate for your data (for example, if they suggest a test that requires assumptions your data do not meet), explain this in detail, citing methodological references that support your position. The key is to argue with evidence, not with opinions. If the method they are asking for is one you do not master, you have three reasonable ways out and none of them involves bluffing: I go through them in what to do when a reviewer asks for an analysis you do not know. And if the request means rerunning the analyses from scratch, here is when reanalyzing your data is legitimate and when it is not.
How to handle legitimate disagreements
You do not always have to do what the reviewer asks. If you genuinely believe that their suggestion is incorrect or inappropriate, you can (and should) explain why you are not implementing it. But do so with respect and with solid arguments. A formula that works well is: "We appreciate this suggestion from the reviewer. However, we believe that [the current approach is more appropriate / the suggested analysis is not applicable in this case] for the following reasons: [arguments with references]." Never respond simply "we disagree" without explaining why.
It is also important to distinguish between comments that come from the editor and those that come from reviewers. The editor's comments usually carry more weight because they reflect the editorial decision, while reviewers' comments are recommendations that you can follow or decline (with arguments). If the editor explicitly states that you must do something, do it. If a reviewer suggests it and you have well-founded reasons not to, explain them and let the editor decide.
The response letter as a strategic document
The response letter is not merely a list of changes made but a document that must convince the editor and reviewers that the manuscript has improved substantially. Each individual response should be clear, complete, and self-contained: the reviewer should be able to understand your response without having to search through the manuscript. If you have added a new analysis, include the main results in the response. If you have modified a paragraph, reproduce the new text in the response. If you have added a supplementary table or figure, describe it briefly.
Finally, check the consistency between your response letter and the revised manuscript before submitting. Make sure that all the changes you describe in the letter are effectively reflected in the manuscript, and that the page and section numbers you cite are correct. A meticulous response letter conveys professionalism and seriousness, and significantly increases the likelihood that your manuscript will be accepted in the next round.
Before you write the letter, know what they are going to object to: paste your manuscript into the free AI paper reviewer and it returns the critiques a Q1 reviewer would raise, each anchored to a line of your own text and with a concrete fix. It is more useful before you submit than after the rejection, but it also works here: the objections you are answering now are the same ones it flags.
If you are facing a revision with statistical critiques outside your comfort zone, my reviewer response service drafts the rebuttal letter with you and reruns whatever analyses the editor asks for (typical turnaround: 48-72 hours). And before you submit, anticipate the objections with the free AI Paper Reviewer, which simulates a peer review of your Methods and Results.