Start with identity, then compare content

A preprint and a journal article can be versions of the same research, but a similar title is not proof. Begin with explicit evidence: a DOI on the arXiv record, an arXiv identifier in the journal metadata, a relationship supplied by a trusted index, or a statement on the authors’ pages. Author overlap supports the check but is not sufficient by itself.

Once identity is established, keep both links. The preprint page gives an early public date and version history; the DOI resolves to the registered publication. Neither record should silently overwrite the other because readers may need to cite, audit, or access different forms.

Read the version history before the abstract

The date attached to a preprint version tells you which document a contemporary discussion could have seen. Later uploads can add data, repair an analysis, change authors, or align the manuscript with a journal decision. A current abstract can summarize the newest version even when an older claim circulated first.

Record the version number and date when a difference matters. For a major claim, compare the relevant section, figures, and stated limitations rather than relying on title changes. A revised manuscript is not automatically more reliable, but the revision record tells you where to look.

Peer review is meaningful but not a shortcut

A journal publication normally indicates that the manuscript passed that venue’s editorial and peer-review process. It does not guarantee correctness, and review practices vary. Conversely, a preprint has not necessarily lacked all scrutiny: authors may have presented the work, shared it with peers, or submitted it, but those activities are not the same as a documented journal decision.

Describe the status precisely. Use preprint for the arXiv manuscript and journal article for the DOI-linked record. Avoid saying peer reviewed when the source does not establish it, and check for corrections, retractions, or expressions of concern on the publisher page.

  • Confirm the DOI and arXiv identifier rather than matching on title alone.
  • Note which version you read and the date it became public.
  • Compare the methods and result sections when the conclusion matters.
  • Check the publisher page for corrections or retractions.
  • Cite the version that supports the statement you make.

Use the observatory without losing the source

The observatory groups explicitly linked versions so they count as one research work. The detail view still lists the source records separately. Use that group as a map, then follow the original links for the full manuscript, license, supplementary information, and publication status.

If two likely versions are unlinked, do not assume the index has proved they are separate. It may mean that the upstream metadata lacks a relationship. A correction request with both source URLs can be reviewed, and a documented override can connect them without weakening the rule for every other record.