Data Availability Statement

A data availability statement is a section in a published research article that describes where the underlying data supporting the study's findings can be accessed, what restrictions apply to access, and under what conditions the data may be reused. It has become a standard component of journal manuscripts, often appearing at the end of the article before the references. Many major publishers now require or encourage one: PLOS requires a data availability statement with every submission, Springer Nature's policy requires one in all original research articles and is being rolled out across its journals in stages, and at Elsevier the requirement depends on the research data option each journal has adopted.

Transparency about data access is a cornerstone of reproducible science. When other researchers can obtain the data behind a study, they can verify results, conduct alternative analyses, and build on the work in ways that advance the field more rapidly. Funding agencies including the National Institutes of Health, the Wellcome Trust, and the European Research Council increasingly require researchers to plan for data sharing and to share the data behind their findings as a condition of grant support; the NIH Data Management and Sharing Policy, for example, has required a Data Management and Sharing Plan for NIH funded research that generates scientific data since 25 January 2023. In one study of more than half a million PLOS and BMC articles, papers whose data availability statements linked to data in a repository received up to about 25% more citations on average, although an association of this kind does not show that sharing data causes the extra citations. Conversely, the absence of a data availability statement can raise concerns about the reproducibility and trustworthiness of reported findings.

Authors draft the statement during manuscript preparation, specifying one of several standard categories of availability. Data may be publicly available without restriction, in which case the statement provides the repository name, a persistent identifier such as a DOI, and any relevant accession numbers. Data may be available upon reasonable request, meaning interested researchers must contact the corresponding author and meet certain conditions before gaining access. Data may be restricted due to privacy protections, ethical constraints, proprietary ownership, or legal limitations, in which case the statement explains the nature of the restriction and describes any mechanism for requesting access, such as a data access committee or a formal application process.

Some journals provide standardized templates with predefined categories, while others allow free form statements. The level of detail varies, but best practice is to be as specific as possible: naming the repository, providing the DOI or URL, describing the file formats available, and noting any embargo periods. Some statements also describe derived data or code availability, linking to statistical scripts or analysis pipelines that accompany the dataset.

Reviewers and editors increasingly verify whether the stated data access matches reality. Studies have found that most "available upon reasonable request" statements do not result in actual data sharing when requests are made: in one study of 1,792 BioMed Central articles whose statements said the authors were willing to share their data, only 6.8% of authors provided the data when asked. This raises questions about the effectiveness of this category. Some journals have responded by requiring repository deposit as a condition of publication, moving away from the "upon request" option.

A data availability statement should be included in every research article that involves empirical data, regardless of whether the data can be fully shared. Even when data cannot be made available, the statement serves a transparency function by explaining why. It is also appropriate for systematic reviews (where the extracted data can be shared), secondary analyses (where the statement should reference the original data source), and qualitative studies (where anonymized transcripts or codebooks may be shareable). A purely theoretical or mathematical paper that generates or analyzes no empirical data has nothing to share, but journals that require a data availability statement may still ask for one that says so.

A common error is providing vague statements such as "data are available from the authors" without specifying conditions, timelines, or contact information. This makes it difficult for other researchers to actually obtain the data. Another mistake is claiming data are available in a repository without actually depositing them, sometimes discovered only when reviewers or other researchers attempt to access the files. Researchers also sometimes overlook the need to address code and software availability alongside data, leaving other researchers unable to reproduce analyses even with the raw data in hand. Finally, overly restrictive access conditions, such as requiring co authorship on any resulting publication, discourage legitimate secondary use and undermine the spirit of open science.

A randomized trial of a behavioral intervention for smoking cessation could include the following data availability statement (a hypothetical example): "De identified individual participant data, the study protocol, and the statistical analysis plan are available in the NIMH Data Archive (DOI: 10.xxxxx). Access requires registration and approval of a data use agreement. Code for all analyses is available at https://github.com/example/smoking-trial. Requests for additional data elements should be directed to the corresponding author."