Make Your Research FAIR: A Practical Guide to Sharing Data Well
Sharing data is becoming the norm, not the exception. This guide turns the FAIR principles and funder mandates into a concrete, do-this checklist for authors — and clears up what FAIR does and does not mean.
Office of Research Integrity··8 min read
<p>Data sharing has moved from a nice gesture to a baseline expectation. The reason is simple: research that others can find, check, and reuse is research that lasts. The shared vocabulary for doing it well is <strong>FAIR</strong>.</p>
<h2>What FAIR actually means</h2>
<p>Introduced by Wilkinson and colleagues in <em>Scientific Data</em> in 2016, FAIR stands for <strong>Findable, Accessible, Interoperable, and Reusable</strong>, with a deliberate emphasis on data that machines, not just humans, can locate and use. One common misconception is worth correcting right away: <strong>FAIR does not mean "open."</strong> "Accessible" means the way to access the data, and any conditions on it, are clear and standardised — not that everything must be free to download. Funders summarise the balance neatly: <em>as open as possible, as closed as necessary.</em></p>
<h2>Why it is worth the effort</h2>
<p>Beyond good practice, mandates are arriving. The US National Institutes of Health Data Management and Sharing Policy took effect in January 2023 and applies to all applications that generate scientific data. Horizon Europe requires a data management plan and deposit in a trusted repository. cOAlition S strongly encourages FAIR data alongside its open-access rules for publications. (One thing to keep straight: the NIH data policy is separate from the NIH public-access policy for publications — they are easy to conflate.)</p>
<h2>A five-step checklist</h2>
<ol>
<li><strong>Plan early.</strong> Write a short data management plan when you design the study, not after. Decide what data you will keep, in what format, and where it will live.</li>
<li><strong>Choose the right repository.</strong> Prefer a recognised domain-specific repository where one exists; otherwise use a certified generalist such as Zenodo, Dryad, or Dataverse. Look for a trust signal like CoreTrustSeal certification.</li>
<li><strong>Deposit properly.</strong> Give your dataset a persistent identifier (a DataCite DOI), a clear licence (CC BY or CC0 for open data), a plain README, and rich metadata so others can understand it without emailing you.</li>
<li><strong>Write an honest data availability statement.</strong> Name the repository and the identifier. If access is restricted for legal or ethical reasons, say so and explain why. Avoid the vague "available on request," which research has repeatedly shown rarely results in data actually changing hands.</li>
<li><strong>Cite the dataset.</strong> Datasets are first-class research outputs. Cite them formally — author, year, title, identifier, repository — following the FORCE11 data-citation principles, so you and others get credit.</li>
</ol>
<p>Done this way, data sharing stops being a compliance chore and becomes part of the quality of your work: more reproducible, more reusable, and more likely to keep contributing long after publication. That is the standard we encourage across Xpertia journals.</p>
