Published Data

“Published data” is a narrower, more specific claim than “shared data” or “open data.” Sharing describes access — who can get a copy. Openness describes licence terms — what they’re allowed to do with it. Publication describes a formal act of release: the dataset has been deposited somewhere that assigns it a persistent identifier, describes it with structured metadata, and treats it as a first-class, independently citable research output in its own right, the same way a journal article is.

The three components that make data ‘published’

• Persistent identifier. Most commonly a DOI, minted through a registration agency such as DataCite. This is what FAIR’s F1 principle requires: “(Meta)data are assigned a globally unique and persistent identifier.” Without it, a dataset’s location can change or disappear, breaking any citation that points to it.
• Structured metadata. DataCite’s schema treats six properties as mandatory before a DOI can even be minted: Identifier, Creator, Title, Publisher, PublicationYear, and ResourceType. This is the practical floor for a dataset to be independently discoverable and citable — it is what lets a search engine, a repository, or a citation manager resolve ‘this dataset’ as a distinct, described thing rather than an anonymous file.
• Citation form. Publication gives a dataset a fixed citation format — Creator (PublicationYear): Title. Publisher. ResourceType. Identifier — the same structural role a journal citation plays for an article. The Joint Declaration of Data Citation Principles formalises the expectation that this citation should be treated with the same standing as a citation to any other research output.

Published data vs. related concepts

These terms overlap in everyday use but describe different things:

• Open data is about licence terms and access conditions — a dataset can be openly licensed without being formally published (e.g. code shared on a personal website with a CC-BY notice but no persistent identifier), and conversely a published dataset can sit behind controlled access rather than being fully open, particularly for sensitive human-subjects data.
• Data paper is a specific peer-reviewed article genre (venues such as Scientific Data, Earth System Science Data, Data in Brief) that describes a published dataset’s collection methods and quality, giving scholarly credit for the curation work — it is one route to publishing data, not a synonym for it.
• Data publication platform refers to the venue or infrastructure — a data journal, a curation-heavy repository like PANGAEA, or a general-purpose repository with DOI minting such as Dryad or figshare — that carries out the act of publication. Published data is the resulting output; the platform is where the act happens.
• Data availability statement is the manuscript-level disclosure (in a journal article) of where and how a study’s underlying data can be obtained. It typically points to published data, but the statement itself is prose in a paper, not the dataset’s formal publication record.

Why the distinction matters in practice

Funder and journal data-sharing policies increasingly specify ‘deposit in a repository and make it citable’ rather than just ‘make it available on request.’ Treating an informally shared file as equivalent to published data understates what a policy actually requires: reviewers, funders, and repositories checking compliance are typically looking for a resolvable identifier and a metadata record, not just confirmation that a copy exists somewhere. For a researcher writing a data management plan or a data availability statement, naming the specific repository and identifier scheme intended for eventual publication — not just describing a general intent to ‘share data’ — is the detail that satisfies this bar.

Frequently asked questions

Does data have to go through peer review to count as published?

No. A dataset deposited directly in a general-purpose or domain repository with a DOI and full metadata is published in the sense used here, without any accompanying peer-reviewed article. A data paper adds an extra layer of independent review of the data’s quality and documentation, but it is not a requirement for the underlying deposit to be considered published.

Can published data still be restricted-access?

Yes. Sensitive data — clinical, genomic, or otherwise disclosive — can be published with a persistent identifier and full metadata while the underlying files sit behind a controlled-access request process managed by the repository. The metadata record and citation exist and resolve publicly even when the data files themselves do not.

Is a dataset in a journal’s supplementary files ‘published’?

Only if that supplementary deposit itself carries an independent, resolvable identifier and standalone metadata record. Many journals now push authors toward a dedicated repository deposit for exactly this reason — supplementary-file datasets are frequently unindexed and effectively invisible outside the article itself.

Related terms

See also Data paper, Data publication platform, Open data, Data citation principle, and Data availability statement.

References

• Wilkinson et al., ‘The FAIR Guiding Principles for scientific data management and stewardship’, Scientific Data, 2016.
• DataCite Metadata Schema documentation (mandatory properties, citation format), datacite-metadata-schema.readthedocs.io.
• Joint Declaration of Data Citation Principles, FORCE11, 2014.