How to Cite a Dataset in APA, MLA, and Chicago

Data-sharing requirements from journals and funders mean more students now cite the raw dataset behind a study, not only the paper that analyzes it. A dataset citation follows the same author-title-date logic as any other source, with two additions almost every style requires: a version number, since datasets get revised and reissued, and a descriptor marking it as data rather than a traditional publication. Most repositories assign a DOI to each dataset version, which is what makes a precise, permanent citation possible in the first place.

Why the Dataset Is a Separate Source

It’s tempting to treat “the dataset” and “the paper analyzing it” as one source, especially when you found both through the same search. They’re not the same citation. A research team can release a dataset years before publishing an analysis of it, other researchers can publish separate papers using the same dataset, and a dataset can outlive any single paper written about it. If you used the raw data directly – ran your own analysis on it, rather than citing someone else’s conclusions about it – cite the dataset itself, with its own DOI, alongside or instead of the paper.

The Example Dataset

We’ll cite this dataset in every style:

• Creator: Kwame Boateng
• Title: Urban Heat Island Temperature Readings for Accra, 2015-2024
• Version: 2
• Repository: Zenodo
• Year: 2024
• DOI: 10.5281/zenodo.1234567

APA 7th Edition

Boateng, K. (2024). Urban heat island temperature readings for Accra, 2015-2024 (Version 2) [Data set]. Zenodo. https://doi.org/10.5281/zenodo.1234567

Key rules:

• Creator formatted like any other author: last name, initials.
• Title italicized, sentence case, with the version in parentheses immediately after.
• [Data set] descriptor (two words) in square brackets after the version – APA’s term is “Data set,” not “Dataset.”
• Repository name (Zenodo) stands in for the publisher.

MLA 9th Edition

Boateng, Kwame. Urban Heat Island Temperature Readings for Accra, 2015-2024. Version 2, Zenodo, 2024, doi.org/10.5281/zenodo.1234567.

Key rules:

• Creator inverted (Last, First), same as an author entry.
• Title italicized as a standalone work, not placed in quotation marks.
• Version and repository act as the “version” and “publisher” core elements.
• MLA doesn’t require a bracketed descriptor – the container (Zenodo) signals the source type.

Chicago 18th Edition (Notes and Bibliography)

Footnote:

1. Kwame Boateng, “Urban Heat Island Temperature Readings for Accra, 2015-2024,” version 2, Zenodo, 2024, https://doi.org/10.5281/zenodo.1234567.

Bibliography entry:

Boateng, Kwame. “Urban Heat Island Temperature Readings for Accra, 2015-2024.” Version 2. Zenodo, 2024. https://doi.org/10.5281/zenodo.1234567.

Key rules:

• Title in quotation marks, not italicized – Chicago treats a dataset as a titled work within the Zenodo container.
• The repository and version identify the dataset without an added medium label.
• Full DOI URL at the end in both footnote and bibliography.

Harvard

Boateng, K. (2024) Urban heat island temperature readings for Accra, 2015-2024. Version 2. [Data set]. Zenodo. Available at: https://doi.org/10.5281/zenodo.1234567.

Key rules:

• Title italicized, sentence case.
• [Data set] descriptor (two words) in square brackets – Cite Them Right uses the same two-word term APA does.
• “Available at:” is required, but an access date isn’t – Cite Them Right only requires one when there’s no DOI or other stable identifier to point to instead.

IEEE

1 K. Boateng, “Urban heat island temperature readings for Accra, 2015-2024,” version 2, Zenodo, 2024. [Dataset]. Available: https://doi.org/10.5281/zenodo.1234567

Key rules:

• Initials before last name, matching every other IEEE author entry.
• Title in quotation marks, sentence case.
• [Dataset] descriptor placed after the year, before the availability statement.

Vancouver

1. Boateng K. Urban heat island temperature readings for Accra, 2015-2024 [Dataset]. Version 2. Zenodo; 2024. Available from: https://doi.org/10.5281/zenodo.1234567

Key rules:

• No periods in the author initial (Boateng K).
• [Dataset] descriptor immediately follows the title, before the version.
• Semicolon between repository and year, matching Vancouver’s journal-citation punctuation.

Quick Comparison Table

Special Cases

No Named Author (Organizational Datasets)

Many datasets are released by an institution rather than a named researcher – a government agency, a university lab, or a data consortium. Use the organization’s full name in the author position: “World Bank” or “Centers for Disease Control and Prevention,” treated the same way you’d treat a corporate author for a government report.

No DOI Available

Not every dataset repository issues a persistent identifier. DataCite, the nonprofit that mints most dataset DOIs, maintains search tools for locating one if the repository’s landing page doesn’t display it directly. If no DOI exists anywhere, cite the direct URL and add an access date in every style – this is exactly the situation Harvard’s access-date rule exists for, since without a DOI there’s no stable identifier to point to instead.

Citing a Specific Data Subset

If you only used a portion of a larger dataset – one region, one time range – note that in your in-text description of the source rather than altering the formal citation, unless the repository has assigned the subset its own separate DOI, in which case cite that one instead.

A Dataset with Multiple Creators or a Data Team

Large datasets are often credited to a research team or a named principal investigator rather than one individual. List every named creator the way you’d list multiple authors for any other source in your chosen style, and if the repository credits an organization rather than individuals, use the organization’s name in the author position instead of guessing at who specifically compiled the data.

Common Mistakes

1. Skipping the version number. A dataset revised after you accessed it can produce different results; the version pins your citation to the exact data you used.
2. Treating the DOI as the same one for the paper about the dataset. The dataset and the article analyzing it usually have separate DOIs – cite the one for the thing you actually used.
3. Dropping the bracketed descriptor – [Data set] in APA and Harvard, [Dataset] in IEEE and Vancouver. Without it, the citation reads like a book or report.
4. Citing a repository landing page instead of the specific dataset DOI. Link to the exact version you used, not the collection it belongs to.
5. Assuming a dataset’s most recent version is the one you actually analyzed. If you downloaded the data months before submitting, check that a later revision hasn’t replaced it.

Try It with CiteTools

Getting the version number, repository, and descriptor bracket right by hand across six styles takes real attention, and small differences between styles are easy to mix up when you’re formatting several dataset citations in one sitting. Drop a dataset’s DOI or URL into CiteTools.io for correct formatting – descriptor, bracket placement, and DOI style all handled – in whichever citation format your assignment requires.