Open data & open science
Best practices for using digital object identifiers to reliably reference datasets in scholarly work.
Researchers seeking durable, precise citations should adopt a structured approach to digital object identifiers, emphasizing persistent access, metadata completeness, versioning clarity, and transparent provenance to sustain reproducibility and trustworthy scholarship.
July 30, 2025 - 3 min Read
In modern scholarship, datasets function as primary evidence and as essential research materials across disciplines. Digital object identifiers, or DOIs, offer a stable pointer that persists beyond the lifespan of individual storage sites or institutional repositories. The reliability of a citation often hinges on the accompanying metadata—who created the data, when it was collected, what methods were used, and under what license it is released. By prioritizing DOIs, researchers enable readers to locate, access, and evaluate the exact data used in analyses, supporting both replication and secondary study. The practice also aligns with funder expectations and journal policies that emphasize data availability.
Effective use of DOIs begins with careful selection and verification. Authors should register datasets with reputable repositories that issue persistent identifiers and provide rich default metadata. When possible, the DOI should be cited alongside a stable URL, but not as its sole route to the data. Researchers must avoid personal or temporary links that could fail with time. Including a concise description of the dataset’s scope, its temporal and spatial coverage, and any preprocessing steps helps readers understand applicability. Regularly checking the DOI’s resolution status ensures that links remain functional, reducing the risk of “link rot” in publications.
Tracking provenance and licensing clarifies reuse and attribution rights.
A cornerstone of dependable data citation is standardized metadata. Repositories should supply core fields such as creator(s), title, publication year, version, spatial and temporal coverage, data format, and licensing terms. Where applicable, keywords and related identifiers enhance searchability across platforms. Authors can augment machine readability by including a machine-actionable schema or schema.org markup in accompanying landing pages. Transparent licensing clarifies reuse rights and restrictions, which in turn informs potential reuse and attribution practices. When researchers document data provenance, including lineage and any transformations performed, they provide readers with a trustworthy lineage that supports rigorous evaluation and reuse in future work.
Versioning is a critical, sometimes overlooked, aspect of DOI-based data citation. Datasets are often updated; each version should have a distinct identifier or a versioned landing page. When presenting results, authors must specify which version was used to generate figures, tables, or models. This precision allows others to reproduce analyses accurately, or to understand how subsequent changes might affect conclusions. Repositories that track revisions can reveal a clear, auditable history of edits. Keeping version notes that outline changes to data, processing scripts, and any corrections helps maintain transparency and integrity across the research lifecycle.
Accessibility, documentation, and interoperability drive reuse.
Provenance information traces the data’s journey from collection to publication, detailing instrument settings, sampling methods, and any quality-control steps applied. Clear provenance reduces ambiguity about data reliability and suitability for secondary analyses. Licensing terms determine the boundaries for use, modification, and redistribution. Researchers should report whether data may be shared, whether attribution is required, and whether derivatives are allowed. When possible, place license statements on the dataset’s landing page or metadata record. Combining provenance with explicit licensing gives future readers a precise map for lawful and ethical reuse, thereby enhancing the dataset’s long-term value and impact.
Alongside technical details, practical considerations shape reuse potential. Data users benefit from accessible documentation, including data dictionaries, variable descriptions, and example workflows. Providing a reproducible example, such as a code snippet that downloads the dataset via its DOI and loads it into a standard environment, accelerates adoption. It is also helpful to provide guidance on data quality, potential biases, and known limitations. A thoughtful README or data card should accompany the dataset, outlining recommended preprocessing steps and any caveats that could influence downstream results or interpretations.
Ethical use and long-term stewardship sustain scholarly value.
Accessibility extends beyond merely hosting data in a repository. It encompasses persistent access, stable identifiers, and user-friendly interfaces that accommodate diverse researchers. To maximize accessibility, authors can select repositories with robust API support, downloadable data formats, and clear instructions for bulk access. Interoperability is strengthened when data are formatted using open, well-documented standards. Providing file formats that are machine-readable and widely supported reduces barriers to integration into varied workflows. In addition, offering programmatic access via APIs, coupled with rate limits and authentication guidelines, helps researchers incorporate the data into automated pipelines without compromising security or reliability.
Trustworthy datasets reveal their quality through transparent methodology. Describing how data were collected, processed, and validated gives readers confidence in the dataset’s integrity. When possible, include independent verification steps, such as cross-source comparisons or replication studies. If the dataset includes sensitive or restricted information, clearly state the access controls and de-identification measures. Researchers should document any data cleaning or normalization operations applied before sharing. Clear, thorough methodological notes reduce confusion and enable others to assess robustness, replicate results, and adapt analyses to similar contexts with confidence.
Practical adoption and community standards strengthen practice.
Ethical considerations must guide data citation practices, especially when datasets involve human subjects or sensitive material. Researchers should ensure that data sharing complies with consent agreements, privacy protections, and institutional review requirements. Anonymization or aggregation strategies should be described if applicable, along with any residual risks. DOIs help track provenance, but ethics statements in publications remain essential. Long-term stewardship requires selecting repositories that commit to data preservation, regular integrity checks, and disaster recovery planning. By choosing responsible stewards and communicating ethical safeguards, scholars demonstrate accountability and respect for participants, researchers, and communities impacted by the data.
Beyond ethics, editorial and institutional expectations increasingly mandate reproducibility and data availability. Journals may require DOIs for cited datasets and disclosure of access restrictions. Institutions can support researchers by providing centralized data management services, guidelines for metadata standards, and training on data citation practices. Collaboration across disciplines improves the quality and consistency of DOIs, as standardized practices reduce ambiguity when datasets are reused across fields. When authors align their work with community norms, they contribute to a culture where data are treated as durable scholarly assets rather than transient resources.
Achieving consistent DOI usage is a community effort that benefits from shared standards and guidance. Researchers should familiarize themselves with widely adopted metadata schemas, such as those recommended by data centers and libraries. Adopting discipline-specific conventions, when available, helps ensure that identifiers are meaningful within a given field while remaining interoperable with other domains. Training resources, checklists, and templates support researchers in implementing best practices without adding excessive overhead. When possible, departments and funders can incentivize rigorous data citation through recognition, funding criteria, and infrastructure investments that reward dependable data sharing and reuse.
Finally, persistent identifiers are most effective when embedded into the scholarly workflow. Integrating DOI citation into manuscript templates, reference managers, and data management plans reduces friction and improves consistency. Encouraging authors to review DOI links at multiple stages—from data deposition to manuscript submission—prevents broken references. Automated checks can flag incomplete metadata, missing licenses, or version mismatches before publication. By embedding DOIs deeply into the research lifecycle, the academic ecosystem strengthens reproducibility, accelerates discovery, and honors the collective effort that underpins scientific progress.
