Editorial Policies

STEM Review reports on research published elsewhere. It runs no manuscript submission system, accepts no unsolicited submissions, and does not enter primary research into the scholarly record, so what follows is written for the work it actually does: reporting on findings that are already public. The policies are informed by practice common across scientific publishing, without claiming membership of any body that issues that guidance.

These policies govern the articles STEM Review publishes and the standards applied to them. They also set the standard that would govern any assessment of unpublished work the publication commissions, and the people commissioned to carry it out. Anyone commissioned to assess work is additionally bound by the Ethical Guidelines for Reviewers, use of the site by the Terms of Service, and the handling of personal information by the Privacy Policy.

Where STEM Review commissions an assessment of unpublished work, what follows governs both the material and the person asked to assess it. It sets the standard for that case; it does not describe a queue in operation.

Work commissioned out for assessment stays the authors' to release. It is seen only by the people asked to assess it and by whoever is handling the commission, and it goes no further without the authors' agreement. Nothing in it is reported or otherwise acted on before the authors' own work appears, and an assessment goes back without the assessor's name attached. What becomes of a completed review is set out in the Ethical Guidelines for Reviewers.

Decisions about what STEM Review publishes, corrects, and withdraws rest with the publication and are made on the merits of the work. No funder, institution, government body, or commercial party has a say in coverage, and no article is placed, timed, or worded at the request of anyone with an interest in how it reads.

STEM Review carries no advertising and runs no advertising network. It publishes no sponsored, branded, or paid placement, and accepts none. There is no paywall and no subscription tier. This is a description of how the publication is currently funded and run, not of a wall between a commercial side and an editorial one. There is no commercial side. Should any part of the site become commercially supported, that arrangement will be stated on this page and any paid material will be labelled where it appears.

Technical claims must be checked against the primary literature rather than against secondary coverage of it. An article must name the journal, repository, or institution the underlying work came from and link to the source material where a public version exists. A quantitative claim must carry its source and the period it describes—a figure without a window is not a figure. Where a result is modelled rather than measured, the article must say so and identify what the model rests on. Estimates, projections, and preliminary findings are labelled as such, and a claim does not become settled by being repeated widely.

Nothing is invented. No number, quotation, date, or result may be altered, filled in, or rounded toward a better story, and units, magnitudes, and statistical claims are reported in the form the source states them. Material supplied under an embargo is held until the stated time. Where the same finding appears elsewhere before that time, STEM Review may report it and will say where it came from.

Every article is written for STEM Review. Another party's sentences, structure, or findings are never presented as ours. Where an article draws on a paper, a preprint, a press release, or an institutional announcement, that source is identified and any wording taken from it is quoted and attributed; paraphrasing supplied text does not make it original. Published reporting is not reissued as new work. Where an article revisits ground already covered, it links to the earlier piece rather than running it again under a fresh headline and date.

Images are held to the standard applied to text. Nothing is edited into or out of a picture, and a chart is not restyled in a way that changes what it shows. Supplied and licensed imagery must carry its credit, and an image that was generated rather than captured must be identified as such. The site has no field for either line today. Until it has one, no image on STEM Review stands as evidence of a result: images illustrate the subject, and the evidence is in the source the article links to.

A byline names who is accountable for an article, not who is owed credit for it. Articles produced through the publication's own editorial process carry a publication byline, and responsibility for every published sentence, figure, and image rests with STEM Review. No named byline is attached to work its holder did not do, and no name is added for standing, courtesy, or affiliation. Where an outside contributor writes for STEM Review, the byline is theirs and every disclosure in these policies applies to them.

The sourcing and accuracy standards set out above apply to everything published here without exception, whatever process produced it.

Anyone commissioned to assess work is separately barred from entering the material under assessment into any AI system, and must disclose AI assistance used in composing their own comments. Those obligations are set out in the Ethical Guidelines for Reviewers and are not relaxed by anything on this page.

An interest is anything that gives someone a stake in how a piece of work is reported or assessed—money paid, a job held, shares owned, a patent filed or pending, a seat on an advisory board, or a close working or personal tie to the people involved. Money is the clearest form of it and not the only one. What matters is not whether an interest did move a judgment, which is rarely knowable from outside, but whether a reader who learned of it afterwards would think it might have.

Anyone writing for STEM Review, assessing work for it, or deciding what it publishes must disclose such an interest before the work begins. An assessor holding an interest in the material must decline, or say so as soon as it becomes apparent, and no one takes part in a decision on work they have a stake in. Where a disclosed interest bears on what a published article says, the article must say so rather than leaving the disclosure somewhere the reader cannot see it.

The same principle covers the money behind the research reported. Where the funder of a study stands to gain from its outcome, the article must identify the funder and, where it is established, the funder's role in the work. Where who paid for a study cannot be determined, the article must say that rather than implying no one did.

A result that cannot be checked is not yet a result. Where an article reports a quantitative finding, it identifies the data that finding rests on and, where that data is public, where to get it. Where a paper carries a data availability statement, the article says what it permits and what it withholds. Where a paper carries none, the article says that too—an absent statement is itself information about a result.

The same reading applies to analysis code, models, and custom software wherever a reported result cannot be reproduced without them. Naming a language or a library does not make a result testable. A version, a repository, or an environment specification does, and an article reports which of these the work it covers actually supplies.

Materials are treated the same way. Where a result depends on reagents, samples, cell lines, instruments, or a fabrication process, what matters is whether another group could obtain them and on what terms. Restrictions arising from participant confidentiality, biosafety, regulated material, or a third party's rights are legitimate, and an article reports them as stated. A restriction that exists only to keep others from repeating the work is not legitimate, and where the record shows that is what it is, the article says so.

STEM Review corrects the record. A typographical slip or a broken link is fixed without a notice. Anything that changes what an article says—a number, a name, an attribution, a conclusion—calls for a correction that states what changed and when. Substantive claims are not amended silently, and an article is not quietly rewritten to match how events turned out. Where the basis of an article no longer stands and a correction cannot repair it, the article is withdrawn.

Carrying that note on the article, and leaving a withdrawal notice at the article's own address, are standards this publication holds itself to. The site cannot yet display either: today a correction is a change to the article text, and a withdrawn article stops resolving and returns a not-found page.

Concerns about anything published here—a factual error, a sourcing problem, an undisclosed interest, an image, or the conduct of the publication itself—go through Contact, which sets out what to include and what to expect. Where a report is acted on, the outcome is a correction, a withdrawal, or no change. Where a correction is made, the amended article is the outcome, and it can be seen there whether or not a reply is sent. The site carries no comments, so correspondence is the only channel.