Key Takeaways:
- A data availability statement (DAS) tells readers which data support your results, where those data are held, and what conditions apply to access.
- All major publishers and journals (e.g., Springer Nature, Wiley, Elsevier, and BMJ) expect a statement in research articles.
- Match the statement to reality by choosing among 5 scenarios: open repository deposit, access on request, third-party restriction, data within the article, or no new data.
- Reviews, case reports, qualitative studies, and methods papers all need a statement; “not applicable” is rarely the most accurate choice.
What Is a Data Availability Statement?
A data availability statement is a short section that tells readers where the data supporting your results are held and how they can be accessed. Springer Nature describes it as the link between a paper’s results and the supporting evidence.
The statement is meant to be an accurate description of the current status of the underlying material, including cases in which nothing can be released.
- It names the data that underpin the figures, tables, and conclusions.
- It gives a location, ideally a repository plus a persistent identifier.
- It sets out access conditions such as licenses, embargoes, or approval processes.
- It explains, in specific terms, any restriction that prevents open sharing.
- It covers secondary datasets reused from other sources, not only new data.
Why Do Journals Require a Data Availability Statement?
Journals require the statement because it connects published claims to the evidence behind them, which supports verification, reuse, and trust. BMJ notes that greater transparency allows results to be independently verified.
- Reproducibility: readers and reviewers can check whether the analysis can be rerun.
- Research waste: shared data reduce duplicated data collection.
- Funder compliance: many funders now mandate archiving as a grant condition.
- Credit and visibility: several studies report a citation advantage for articles with available data.
- Editorial use: BMJ editors may use the statement to inform editorial decisions, and Wiley checks that links resolve to the intended data.
- Legal and ethical clarity: a well-drafted restriction protects participants and license holders.
What Should a Data Availability Statement Include?
It should describe how the data supporting your results can be accessed, naming the repository, the persistent identifier, the access conditions, and any restriction. Vague wording is the most common reason editors send the statement back.
Core elements to cover
| Element | What to state | Example phrasing |
| Scope | Which datasets underpin the results, including reused secondary data | The raw and processed data supporting Figures 1-4 … |
| Location | Repository name with a DOI, accession number, or other persistent identifier | … are openly available in Zenodo at https://doi.org/10.5281/zenodo.0000000. |
| Conditions | License, embargo, registration, or approval requirements | … under a CC BY 4.0 license, with no registration required. |
| Restrictions | The specific reason data cannot be released | … are not publicly available because participant consent did not cover secondary use. |
| Contact route | Who to approach, using a durable address rather than a personal one | Requests should be directed to datarequests@example.ac.uk. |
| Code and materials | Whether scripts, protocols, or materials are also available | Analysis code is archived at release v1.2 in Zenodo. |
Common mistakes to avoid
Editors return statements for a small set of recurring problems. Check yours against the 8 items below.
- Mismatched policy tier: writing “available on request” when the journal mandates deposit. Confirm the tier in the guide for authors before drafting, because a Wiley or BMJ Tier 1 journal will not accept a request-only statement for non-sensitive data.
- Missing identifiers: naming a repository without a DOI or accession number. Readers cannot locate a dataset from a repository name alone, and Wiley checks that links resolve to the intended record.
- Fragile contacts: giving a personal email that expires when the author moves institution. Use a departmental address, a generic lab inbox, or an ORCID-linked identifier instead.
- Wrong “not applicable”: claiming no data for a review that produced search strings, screening decisions, and extraction files. Synthesis-based articles almost always generate shareable outputs.
- Forgotten secondary data: describing only the data you generated. Datasets obtained from other groups, agencies, or public archives also support your conclusions and belong in the statement.
- Home-page links: pointing to a repository landing page rather than the specific dataset record. Deep-link to the versioned record, and paste every URL into a browser before submission.
- Internal contradictions: a statement that conflicts with the methods section, the funder data management plan, or the trial registry entry. Reviewers cross-check these, and inconsistency raises questions about the whole manuscript.
- Uncited datasets: omitting the dataset from the reference list. Springer Nature, Wiley, and BMJ all expect publicly available data to be cited, with the [dataset] prefix where required.
Data Availability Statement Requirements from Major Publishers
Every publisher below requires or invites a statement, but they differ on 3 things: whether deposit is mandatory, whether the data are peer reviewed, and whether “available on request” is accepted. Check the journal, not just the publisher, because policies vary title by title.
| Publisher or journal | What is required | Point to note |
| Springer Nature | A statement in original research articles across the portfolio | Guidance stresses hyperlinks, persistent identifiers, and an explanation when data cannot be shared |
| Nature Portfolio | A statement in every published manuscript reporting original research | Access conditions for the “minimum dataset” must be transparent; a separate code availability statement is also required |
| Wiley | A statement for all research- and synthesis-based articles, shared or not | Each journal sits in 1 of 4 policy tiers; standard templates are offered at submission |
| Elsevier | Many journals invite a data statement during the submission flow | The statement is displayed in the Research Data section of the article on ScienceDirect |
| BMJ | A statement for every submitted research article | Authors select from 6 standardized options and add detail; journals sit in 1 of 3 tiers |
| Sage | A statement wherever relevant, in a set format | A statement must take 1 of 6 prescribed forms, or a combination of them |
| Taylor & Francis | Varies by journal across 5 standardized policies | Over 2,000 journals carry a policy, from basic encouragement to open and FAIR |
| JAMA Network | A data sharing statement for all research reports | Trial reports must follow the ICMJE format, and “undecided” is not acceptable |
| Science journals | All data must be in the paper, in the supplement, or in a public repository | Access details, preferably a DOI, are cited in the acknowledgments |
| PLOS | A statement describing compliance with the data policy, for every article | Publication is conditional on compliance, and an author may not be the sole contact for restricted data |
Springer Nature and Nature Portfolio
Springer Nature bases its policy on transparency and publishes worked examples by research area. Nature Portfolio journals apply a stricter version of the same framework, and the 2 policies are compatible.
- Nature Portfolio requires a statement in every published original research manuscript, covering both primary and referenced datasets.
- Large datasets in supplementary information are discouraged; repositories are preferred.
- Deposit is mandatory for defined data types, including sequences, structures, and gene expression data, with accession numbers in the paper.
- Controlled access statements must give reasons, precise conditions, contact details, response timeframes, and any data use agreement restrictions.
- Editors may decline a manuscript if restrictions are found to be unduly prohibitive.
- Custom code central to the conclusions needs its own code availability statement, placed after the data statement and before the references.
Wiley
Wiley operates 4 policy tiers, and the tier determines what your statement must do. A statement is required for all research- and synthesis-based articles even when nothing is shared.
| Policy tier | What the statement must do | Data sharing |
| Encourages data sharing | Confirm the presence or absence of data | Optional |
| Expects data sharing | Describe availability or explain absence | Expected |
| Mandates data sharing | Describe the availability of data | Required |
| Mandates sharing and peer reviews data | Describe the availability of data | Required, and the data are peer reviewed |
Wiley checks that links resolve to the intended data and requires shared data to be cited. Where data are peer reviewed, reviewers may confirm that sample sizes and variables match the article, or that the reported analysis reproduces.
Elsevier
Elsevier journals encourage a data statement explaining what data were used and why any of it may be unavailable. The option is built into the submission flow of most journals.
- Check the Research Data section of the journal’s guide for authors before drafting.
- Research data include raw and processed files, software, code, models, algorithms, protocols, and methods.
- Manuscript text and published tables do not count as research data.
- Consider a linked data article or a Mendeley Data deposit to obtain a citable DOI.
BMJ
BMJ requires a statement for every submitted research article and asks authors to select at least 1 of 6 standardized options, supplemented with detail. More than 1 option may be selected when datasets sit under different conditions.
| Tier | What it means for authors |
| Tier 1 | Data must be openly and publicly available on publication; sensitive data go into a controlled access repository |
| Tier 2 | Sharing is strongly encouraged; trial data must be available on reasonable request; trials enrolling from 1 January 2019 need a data sharing plan in the registry record |
| Tier 3 | Sharing is strongly encouraged wherever it is legally and ethically possible |
The 6 options are: data in a public, open access repository; data on reasonable request; data from a third party; all relevant data in the article or supplement; data sharing not applicable; and no data available.
Sage
Sage supports the FAIR principles and asks for a statement wherever relevant, regardless of whether the individual journal requires one. The statement must take 1 of 6 prescribed forms, or a combination when different data types apply.
- Repository deposit, with the repository name and a persistent link.
- Data not publicly available for a stated reason, but available from the corresponding author.
- Data available from the corresponding author on request.
- Data sharing not applicable because nothing was generated or analyzed.
- All data included in the published article and its supplementary files.
- Third-party data used under license, available with the owner’s permission.
Sage also runs Open Science Framework badges on some journals for open data, open materials, and preregistration, and it advises anonymized private sharing links for double-anonymized peer review.
Taylor & Francis
Taylor & Francis applies 5 standardized policies across more than 2,000 journals, so the applicable rules are set at journal level in the instructions for authors.
| Policy | What you must do |
| Basic | Sharing, citation, and a statement are encouraged, not required |
| Share upon reasonable request | You agree to supply data on request, and you judge what is reasonable |
| Publicly available | Data must be freely available under a license of your choice |
| Open data | Data must be freely available under a license permitting any lawful reuse |
| Open and FAIR | Open data, plus compliance with FAIR standards in the subject area |
JAMA Network
JAMA Network journals require a data sharing statement for all research reports, including Original Investigations, Brief Reports, Research Letters, and every listed observational and trial design. Statements are entered through the submission system and published as an online supplement.
- Trial reports must follow the ICMJE format, answering whether deidentified individual participant data will be shared.
- The statement must specify which data, which related documents, when, for how long, and by what access criteria.
- “Undecided” is not an acceptable answer under the ICMJE requirement.
- Protocols and statistical analysis plans are required for all randomized and nonrandomized trials and are published as supplements.
Science journals
Science journals generally require all data underlying published results to be publicly and immediately available. Where everything sits in the article, the accepted wording states that all data are available in the manuscript or the supplementary materials.
- Data may be in the main text, in the supplementary materials, or deposited in a public repository and cited.
- Access details, preferably a DOI or other persistent identifier, are cited in the acknowledgments.
- Data, code, and materials must be available to any researcher for reproducing or extending the analysis.
- A separate materials availability statement is expected in every paper, even when no new materials were created.
PLOS
PLOS requires all data needed to replicate the findings to be publicly available without restriction at publication, and publication is conditional on compliance. Authors must share the “minimal data set,” meaning the values behind means, graphs, and reported measures.
- Repository deposit is strongly recommended; supporting information files are accepted, each under 20 MB.
- References to “data not shown” are not permitted.
- For restricted data, an author may not be the sole named contact; requests must go to a committee or institutional body.
- Third-party restrictions require the source, permission details, contact information, and any known access limits.
- Studies whose conclusions rest solely on inaccessible proprietary data are not considered.
- Qualitative studies should share relevant transcript excerpts, or explain precisely why even excerpts cannot be released.
Which policy applies if your funder and journal disagree?
Follow the more progressive requirement, meaning the one that results in greater sharing. Taylor & Francis states this explicitly, and it is a safe default across publishers, because meeting the stricter standard automatically satisfies the looser one.
- Compare the journal policy, the funder mandate, and any institutional or ethics condition before submission.
- Where a funder mandates deposit and the journal only encourages it, deposit the data.
- Where consent or law limits sharing, that limit governs, and the statement should explain it.
- Record the resolution in your data management plan so that coauthors apply the same rule.
How to Choose the Right Type of Data Availability
Pick from 5 scenarios: open deposit in a repository, access on reasonable request, third-party restriction, data contained in the article, or no new data. Most manuscripts combine 2 of these.
| Scenario | Choose when | Statement must include |
| Open repository deposit | Data can be shared without legal or ethical barriers | Repository name, persistent identifier, license, and version |
| Available on reasonable request | Sharing is possible but must be screened, for example for participant data | What the data are, who holds them, a durable contact, and the review conditions |
| Third-party or licensed data | Another organization owns the data | The owner, the license, and how others can apply for the same access |
| Within the article or supplementary files | The dataset is small and fully reported | A pointer to the specific tables or files, plus a route to raw formats |
| No new data | The work is theoretical, or nothing was generated or analyzed | A plain statement that no datasets were generated or analyzed |
Data openly available in a repository
This is the strongest option and the one most funders prefer. Name the repository, give the identifier, and state the license so that reuse conditions are unambiguous.
Example: The datasets generated and analyzed during the current study are openly available in the Dryad Digital Repository at https://doi.org/10.5061/dryad.000000, released under a CC0 public domain waiver. Analysis code is archived at release v2.1 in Zenodo at https://doi.org/10.5281/zenodo.0000000.
Data available on reasonable request
Use this only when a genuine screening step exists. State what the data are, who evaluates requests, and what conditions apply, because a bare promise of availability is increasingly viewed as insufficient.
Example: Deidentified participant data, the study protocol, and the statistical analysis plan are available from the study data access committee at dataaccess@example.ac.uk. Proposals will be reviewed within 60 days, and data will be released under a signed data access agreement for research uses consistent with the original consent.
Data from a third party or under license
When another body owns the data, say so and explain how a reader could obtain the same access. This preserves reproducibility even though you cannot redistribute the files.
Example: The linked administrative records analyzed in this study were used under license from the National Statistics Agency of Morgbanetria and cannot be redistributed by the authors. Researchers may apply for equivalent access at https://mgbnationalstatistics.gov/data-access. All code required to reproduce the tables and figures is available at https://doi.org/10.5281/zenodo.0000000.
Data within the article or supplementary files
Small datasets may be fully contained in the manuscript. Point readers to the exact location, and offer raw formats on request so that machine-readable versions remain obtainable.
Example: All data supporting the findings of this study are included in the article and its supplementary information files. Source values for Figures 2 and 3 are provided in Supplementary Tables S1 and S2. Raw instrument files in alternative formats are available from the corresponding author on request.
No new data generated or analyzed
Purely theoretical, conceptual, or commentary work may have no underlying dataset. State this plainly rather than leaving the section blank, and confirm that no reused datasets support the argument.
Example: Data sharing is not applicable to this article, as no datasets were generated or analyzed during the current study. The work proceeds within a theoretical and mathematical framework.
Where Does the Data Availability Statement Go?
Most journals place the statement after the discussion and before the references, under its own heading. BMJ publishes it in the footnotes of the final article, and Elsevier displays it in the Research Data section on ScienceDirect.
- Check the guide for authors before formatting, because placement is journal-specific.
- Enter the statement in the submission system when prompted, not only in the manuscript file.
- Keep the manuscript version and the submission-system version identical.
- Do not bury the statement inside the methods section; editors expect a labeled section.
- Anonymize the statement for double-blind review if repository links would reveal the authors, then restore it at revision.
Data Availability Statements by Article Type
Systematic review
A systematic review generates data even though it collects no new primary observations. Wiley requires a statement for synthesis-based articles, so “not applicable” is usually the wrong answer here.
Treat the following as shareable outputs of the review:
- Full search strategies for every database, with dates and hit counts.
- Title, abstract, and full-text screening decisions, including reasons for exclusion.
- The extraction template and the completed extraction dataset.
- Risk-of-bias or quality appraisal judgments at the domain level.
- Analysis code for meta-analysis, sensitivity analyses, and figures.
- The protocol registration record, for example in PROSPERO.
Example: No new primary data were collected. The search strategies for all 6 databases, the screening decisions with reasons for exclusion, the extraction dataset, the risk-of-bias assessments, and the R code used for the meta-analysis are openly available in the Open Science Framework at https://doi.org/10.17605/OSF.IO/00000. The review protocol is registered with PROSPERO (CRD42024000000). All included studies are listed in the reference list.
Narrative review
A narrative review often generates no dataset, so a “no new data” statement is usually accurate. If you ran a structured search or built a summary table, describe those outputs instead of defaulting to “not applicable”.
Example: Data sharing is not applicable to this article, as no new data were created or analyzed in this review. All sources discussed are cited in the reference list.
Example: This review did not generate new primary data. The literature search strings, the list of 214 screened records, and the summary table underlying Figure 1 are available in Figshare at https://doi.org/10.6084/m9.figshare.00000000.
Case report
A case report describes 1 patient, so the underlying record cannot be shared without breaching confidentiality. State that clinical details are in the article, explain the privacy restriction, and note the consent that was obtained.
- Never offer to share raw clinical records, images, or genomic data without an explicit consent and approval pathway.
- Confirm that images and timelines in the article have been checked for identifiers.
- Keep the availability statement separate from the patient consent statement.
- If a deposited variant or pathogen sequence exists, give the accession number.
Example: All data relevant to this case are included in the article. Additional clinical information from the medical record is not publicly available because it could compromise the privacy of the individual described. Reasonable requests for further deidentified detail may be directed to the corresponding author and will be considered subject to institutional approval and the written consent of the patient.
Qualitative study
Qualitative data are rarely shareable in full, because transcripts resist deidentification and consent often does not cover redistribution. Share what you can, and be precise about why the rest is withheld.
Materials that can usually be shared appear below.
- Topic guides, interview schedules, and recruitment materials.
- The coding framework or codebook, including definitions and examples.
- Anonymized illustrative extracts beyond those quoted in the article.
- Analytic memos or an audit trail describing how themes were developed.
- Fully anonymized transcripts, where consent and ethical approval permit, ideally through a controlled access repository.
Example: The interview topic guide, the coding framework, and anonymized illustrative extracts are available in the Open Science Framework at https://doi.org/10.17605/OSF.IO/00000. Full transcripts are not publicly available because participants consented only to the use of anonymized excerpts, and the small sample from a single clinical service means that complete transcripts could not be deidentified to an acceptable standard. Requests from qualified researchers may be made to the research ethics office at ethics@example.ac.uk.
Methodological paper
For a methods, algorithm, or software paper, the code is the primary data. Archive a versioned release with a DOI rather than pointing only to a repository branch that may change.
- Deposit a tagged release in Zenodo, Software Heritage, or an equivalent archive.
- State the license for both code and data, because they may differ.
- Include worked example data or a small test dataset so that the method can be run.
- List dependencies, version numbers, and the computing environment, for example a container image.
- Where a new instrument or protocol is described, deposit calibration files and standard operating procedures.
Example: The software described in this paper is openly available at https://github.com/example/toolkit and archived at release v3.0.2 in Zenodo at https://doi.org/10.5281/zenodo.0000000 under an MIT license. The benchmark datasets, simulation scripts, parameter files, and a Docker image reproducing the computational environment are archived in the same record. No human or animal data were generated in this study.
How Do You Write a Statement for Sensitive or Patient Data?
State that access is restricted, give the specific reason, name the body that controls access, and describe a route for legitimate requests. Vague privacy claims are the single most common weakness editors flag.
- Be specific: cite the consent wording, the ethics approval condition, or the applicable data protection law rather than saying only “privacy reasons”.
- Prefer controlled access repositories over informal sharing; BMJ recommends services such as Vivli, the YODA Project, or clinicalstudydatarequest.com for individual participant data.
- Give a durable contact, for example a departmental address or an ORCID-linked identifier, not a personal inbox.
- Say what an approved applicant would receive, such as a deidentified analysis dataset with a data dictionary.
- Explain the review process and the expected timeline.
- Share the elements that carry no risk, including protocols, statistical analysis plans, and code.
Example: The dataset analyzed in this study contains potentially identifying clinical information and is therefore not publicly available. Data are held in controlled access storage at the sponsoring institution under the conditions of ethics approval 2024/0000, which restricts secondary use to projects approved by the institutional review board. Deidentified data may be requested through the institutional data access committee at datacommittee@example.ac.uk. The protocol, statistical analysis plan, and analysis code are openly available at https://doi.org/10.5281/zenodo.0000000.
How Do You Choose a Repository?
Choose a certified subject repository first, and fall back to a general repository when no domain option exists. Registries such as re3data.org and FAIRsharing.org list certified options by discipline, and both Wiley and BMJ point authors to them.
| Repository type | Examples | Best for |
| Subject-specific | GEO, GenBank, PANGAEA, CCDC, ICPSR | Data with community standards and established metadata formats |
| General-purpose | Zenodo, Dryad, Figshare, Open Science Framework, Mendeley Data | Datasets with no obvious domain home, and mixed-format research outputs |
| Controlled access | Vivli, the YODA Project, clinicalstudydatarequest.com, national secure labs | Individual participant data and other sensitive records |
| Institutional | University data repositories | Compliance with local retention policies, often alongside a domain deposit |
- Confirm that the repository issues persistent identifiers.
- Check preservation commitments and whether versioning is supported.
- Verify that the license you need is available, for example CC0 or CC BY.
- Deposit before submission so that a resolvable identifier exists for reviewers.
Frequently Asked Questions
What do I write if no data were generated or analyzed?
Say so plainly: data sharing is not applicable because no datasets were generated or analyzed during the study. Add a short reason, for example, that the work is theoretical or a commentary.
Can I just say data are available on request from the corresponding author?
Only when a real review process exists behind that promise. Research on data sharing shows low response rates, so name what the data are, who holds them, the durable contact, and the conditions under which access is granted.
What is the difference between a data availability statement and a data sharing plan?
A statement describes the current status of data for a published article. A data sharing plan is a forward-looking commitment made at trial registration or grant application, and ICMJE requires one for trials enrolling from 1 January 2019.
Do I need a DOI for my dataset?
A persistent identifier is strongly preferred, though an accession number or stable repository URL is acceptable where the repository does not issue DOIs. Never rely on a personal website, a cloud drive link, or an untagged code branch.
Comment