Reviewer guidelines

Before you accept

• Decline if you cannot assess the manuscript's core claims, or cannot return the review within the agreed period.
• Declare any conflict: shared institution, recent collaboration, funding relationship, commercial interest, or personal relationship with an author you can identify.
• Under Double-anonymized peer review, do not attempt to identify the authors, and tell the editor if you learn who they are.

What to assess

1. Claim and evidence. Does the evidence presented actually support the claim made, at the strength it is stated?
2. Method. Is the method described well enough that a competent reader could repeat it?
3. Prior work. Is the contribution located honestly against what already exists?
4. Limitations and harms. Are foreseeable harms, affected communities, and limits of generalisation stated?
5. Data, code, and reproducibility. Is what is needed to check the work available, or is its absence justified?
6. Citation integrity. Do the sources cited exist and say what they are said to say?

How to write it

Separate what is wrong from what you would have done differently. Rank your points: what must change for the work to be sound, and what would merely improve it. Quote the manuscript when you object to something specific. Write comments to authors that you would be willing to sign.

Recommendations available to you: accept, minor revision, major revision, reject. The editor weighs your recommendation against the other reviews and decides.

Confidentiality

A manuscript under review is a confidential document. Do not share it, quote it, cite it, use its ideas or data before publication, or upload any part of it to a third-party service, including a generative AI tool -- see the AI use policy. Destroy your copy after the decision.