A source is only as strong as the claim it must support
Credibility is not a single feature. An official page may establish a current rule but cannot replace an independent effectiveness study. A peer-reviewed article may still have serious methodological limits. A preprint may contain valuable new data but needs different treatment from a later journal version. Begin with the proposition: what evidence type is appropriate and what population, period or setting must it cover?
The core route checks the responsible author or organisation, venue, date, method, data, references, interests and correction status. Record not merely “suitable”, but the purpose for which it is suitable. One item may provide context while another carries the central empirical claim.
Interpret publication type and institutional setting
Dissertations, conference papers, reviews, government reports and think-tank studies pass through different selection and review processes. No category is automatically reliable or worthless. Examine authority, review, methodological transparency, available evidence and institutional interest. For a systematic review, also inspect how literature was searched, screened and synthesised.
Verify publication identity and trust signals
A DOI is a persistent identifier, not a quality certificate. Resolve it and compare title, authors, year and venue with the document. A source without a DOI may remain appropriate; stable bibliographic details and an institutional location then matter more. For an unfamiliar journal, use a bundle of verifiable features rather than one blacklist or badge.
Treat version, correction and retraction as evidence
Online-first and print records may carry different pages or metadata. A preprint may differ materially from its reviewed successor. An erratum changes identified content; a retraction warns of serious reliability problems without erasing historical existence. Cite the version actually used and repeat the status check before submission.
Handle paywalls, secondary citations and changing pages
Do not imply that you inspected an inaccessible original. Search lawful library access, repositories and authorised manuscripts. If the original remains unavailable, identify the secondary route. For changing webpages, responsible organisation, retrieval date, version and an archive snapshot can preserve the finding; the snapshot does not establish factual quality.
A six-step rapid check
- State the exact proposition requiring support.
- Identify an evidence type suited to that proposition.
- Verify identity and the version actually read.
- Examine method, data, limitations and interests.
- Search for corrections, retraction and credible disagreement.
- Record the precise location and scope in your notes.
A central claim requires more than this rapid check. Read the relevant method and findings, compare independent evidence and disclose uncertainty. Source evaluation then becomes a traceable part of reasoning rather than a badge attached to the bibliography.
Keep an evaluation record
For each central source, keep a compact row containing claim, source type, verified identifier, version, correction status, methodological limit and intended use. Date the check. If a later version changes the conclusion, preserve the earlier row and add the update rather than silently replacing history. This record makes it possible to revisit crucial evidence before submission and prevents a reference manager’s metadata from being mistaken for substantive appraisal.