Public resource · citation audit
Checking citations: a six-question test for whether a link really supports a page claim
A citation improves a page only when it supports the exact statement beside it. A link alone is not an evidence standard.
Ask what kind of source it is
Identify whether the link is a primary study, review, registry record, regulator page, laboratory report, or commentary. Each type can be useful, but each has a different relationship to a claim.
A page that calls a database record ‘proof’ or treats a narrative review as a trial has already weakened its own evidence trail, even if the link is legitimate.
Check the claim against the source
Compare the page sentence with the source’s population, model, outcome, and date. Does the paper study people, animals, cells, or a document record? Does it report the outcome the page names? Is the conclusion narrower than the page’s wording?
This is the core citation test. If the answer is no, change the sentence, add a limitation, or remove the claim. Do not rely on a citation’s prestige to bridge a mismatch.
Keep the audit trail visible
Useful pages give readers a named source, direct link, short note, and page date. Material corrections should appear in a change log. That lets an outside reviewer assess the work rather than relying on an author’s assurance.
This resource is an editorial method, not a health or legal assessment service.
Look for the source’s own qualifiers
A citation review should not stop after finding a paper that mentions the same topic. Read the authors’ stated population or model, endpoint, result language, limitations, funding disclosure, and conclusion. Qualifiers such as “associated,” “in this model,” “short-term,” “exploratory,” or “further study is needed” can materially change whether a secondary page’s sentence fairly represents the source.
If the supporting document is behind a paywall, incomplete, retracted, or unavailable, the page should say what was actually checked rather than implying full review. Citation quality comes from a transparent claim match, not from the prestige of a journal name or the length of a bibliography. Readers should be able to see where evidence ends and editorial interpretation begins.
Check for corrections, retractions, and replacement records
A citation can remain widely repeated after its publication record changes. Before relying on a paper for a central claim, look for linked correction notices, expressions of concern, retractions, errata, or newer versions. The presence of a correction does not automatically invalidate every result, but it can change the precise statement a secondary page should make.
The record should also be dated. A source that was the best available document at one time may later be joined or qualified by stronger evidence. Good citation practice is therefore ongoing maintenance: retain the source trail, check the publication status, and revise the surrounding language when the record materially changes.