A research dataset changed after you cited it: preserve the release identity
Keep the exact data release behind an earlier reading separate from a later repository version and any unperformed reanalysis.
No Vape · Updated

Start from the earlier material you actually inspected
Find the version identifier, date, file names and citation in your existing reading note. Do not replace them with the newest page simply to make the reference appear current. If the original version was never recorded, say that rather than assigning a version retrospectively from an unfamiliar download folder.
Zenodo's version-management guidance describes a new version as a separately identified record linked to earlier and later versions when published files are updated. That is a named repository's model. Use the actual repository's rules for the dataset in question; not every file replacement or metadata edit follows the same procedure.
Compare release identities before comparing conclusions
Write two rows: earlier inspected release, and newly available release. Include each release-specific identifier and the documented relationship between them. Where a general record groups versions, keep that family-level link separate from the exact release citation. The aim is knowing which material a statement refers to, rather than making every reference point to the latest page.
In a fictional example, a dataset deposit adds corrected labels in a second release after a reader cited the first. The reader preserves the first release's identity and reads the change note. They do not silently claim that their earlier inspection covered the corrected files. No real dataset or error is being described here.
Read what changed without inventing an analysis
Use the release notes and associated article notices to establish what the authors say changed. A modified file can matter greatly, slightly or in a way you cannot judge from its name. Do not infer that all paper conclusions changed merely because a version number increased. Likewise, a familiar-looking summary does not prove the underlying analysis was rerun.
If a public result depends on the altered material, ask which analysis and reported version apply. Reproduction requires the relevant data, code and methods; downloading a new release is not reproduction. Keep participant restrictions intact and avoid seeking private records to satisfy a public version question.
Update the reference at the level you reached
If you inspected the newer release, state that and preserve the earlier citation where it describes earlier work. If you have not inspected it, note that a newer version exists and that its effect on your reading is unresolved. A corrected citation should not imply new calculations, a resolved methodological concern or an endorsement you did not perform.
This guide handles a change after earlier use, rather than general access to research data. Its output is a traceable relationship between releases and the statements actually checked. It is not a dataset audit or a treatment decision. When interpretation affects care, bring the actual report and its version question to a qualified professional.
Sources and scope
Sources inform the guidance. The worksheets and examples are No Vape editorial suggestions, not tested treatments.
Written with AI assistance and checked against the cited sources. No clinical review is claimed. This guide covers practical planning, not individual medical treatment. How we research and write