How to Cite a Data Set in APA 7

How to Cite a Data Set in APA: Quick Answer

Cite the data set when you reanalyzed archived files or when you are depositing your own data for the first time. Cite the journal article, report, or webpage when you only used numbers already published there. Put the title in italics and sentence case. After the title, add any study number and version in roman parentheses, then [Data set] (or a close variant). End with the repository and a DOI written as https://doi.org/....

The APA Style data set references page, tied to §10.9 and still reprinted by university style centers in 2026, states the two cases in which a data-set reference belongs in the list:

That second sentence is the usual classroom error. A table you copied from the NCES 2010-040 First Look is a report citation. A Pew finding you read in the published guns report is a webpage or report. A file you downloaded from ICPSR, OSF, Zenodo, or data.census.gov and then analyzed is a data-set citation.

Anatomy of an APA 7 Data Set Reference

Section 10.9 uses the same four elements as every APA reference — author, date, title, source — plus two data-specific extras: a version or study number after the title, and a bracketed description of the object. The official page spells out the extras in four sentences that later examples on this page follow.

The official instructional aid Creating an APA Style reference list (last updated 23 February 2022) adds two locator rules students break on repository landing pages:

Author

Name the people or the group that produced the data, not the person who later wrote about it. APA 7 §9.11 tells you to use the most specific government agency as the group author, not the cabinet department above it. National Center for Education Statistics, not "U.S. Department of Education." If two agencies share credit, join two names with an ampersand. List individual authors the same way you would on a journal article: surnames and initials, up to 20 names in the reference. If the author is unknown, follow how to cite a source with no author rather than inventing a name.

A long agency name can take a familiar abbreviation after the first in-text citation. The official APA Style author–date table uses the National Institute of Mental Health as the model. The reference list still spells the name out; only the in-text citations shorten.

Date

Use the year the version you analyzed was published or posted, not the year the interviews were fielded and not the year the series began. Pew's American Trends Panel Wave 26 was fielded 4–18 April 2017 and released as a public file on 22 February 2018; the official APA example therefore uses 2018. If you cannot recover a date after a real search, write (n.d.) and see how to cite a source with no date. A continuously updated dashboard without a version number may also need a retrieval date (below).

Title, identifier, and version

Copy the official title from the repository landing page and put it in sentence case. Do not rewrite it into a methods-section description. If the record shows an ICPSR study number, a version tag, or both, they sit in one set of parentheses after the title, in roman type, separated by a semicolon: (ICPSR 36966; Version V1). That pair is how a reader gets the same files you used. If the record has no number and no version, omit the parentheses rather than inventing Version 1.0.

Bracketed description

APA wants a short description of the object, not the file format and not the academic field. The official examples use [Data set], [Data set and code book], and [Unpublished raw data]. You may write a close variant when the object is genuinely different, but do not build a private taxonomy ([Genomic data], [Time series data]) and present it as APA. File extensions such as .csv or .sav stay out of the reference.

Publisher and locator

The publisher is the archive that hosts the files — ICPSR, the National Archive of Data on Arts and Culture, Dryad, Zenodo, OSF, a university Dataverse. Omit it when the group author and the publisher are the same agency. Prefer a DOI over any other URL, written as https://doi.org/ with no period after it. If there is no DOI, use the landing-page URL, not a search-result or "download now" link that expires. See how to use DOIs and how to find a DOI.

Official Worked Examples

Every full example below is an APA Style or Publication Manual §10.9 example, the official APA Pew example, a labelled rebuild from a live landing page, or an agency's own suggested citation. None of the names, titles, study numbers, or DOIs on this page are invented. The ICPSR "Cite this study" line is the archive's own export, shown so you can see why it is not an APA reference.

Data set with a DOI: O'Donohue (ICPSR 36966)

William O'Donohue's ICPSR deposit is the primary worked data-set example in APA teaching materials. The live record is ICPSR 36966, Version V1. The DOI 10.3886/ICPSR36966.v1 resolves to that record (version date 20 December 2017). Rebuild the reference from that landing page: initials, a sentence-case italic title, (ICPSR 36966; Version V1), [Data set], the short publisher name ICPSR, and the https://doi.org/ form. In-text citations use the same author-date pair as any other APA source.

Do not paste the ICPSR "Cite this study" line

The same ICPSR 36966 landing page exports a distributor citation. It keeps the full given name, title-case title, Ann Arbor place, a [distributor] role, and a calendar date. That is useful metadata. It is not APA 7.

The O'Donohue files sit in ICPSR's Archives of Scientific Psychology collection. The public-use files are listed as available without an ICPSR-member login; one or more files in the collection are restricted. Cite the public DOI either way, and say in the methods how a reader requests restricted files. Do not paste a personal download link.

The 3 March 2026 handout prints ICPSR 21600 on O'Donohue

APA's Common Reference Examples Guide, last updated 3 March 2026, still uses O'Donohue as the §10.9 sample — but it prints study number ICPSR 21600 next to DOI 10.3886/ICPSR36966.v1 and spells the publisher out in full. University reprints opened in 2026, including the Alamo Colleges data-set page last updated 29 May 2026, copy that 21600 number. Do not.

ICPSR 21600 is a different study: the public-use National Longitudinal Study of Adolescent to Adult Health (Add Health), 1994-2025. The live Add Health DOI on 17 September 2026 is 10.3886/ICPSR21600.v26 (Version V26, posted 3 March 2026). O'Donohue's DOI is 10.3886/ICPSR36966.v1. Copy the study number from the DOI or the landing page you actually opened, not from a handout that mixed the two.

Cite the companion article only when you used the article

O'Donohue and Brendan Willis later published the analysis in Archives of Scientific Psychology. That paper is a journal-article citation. Use it when you discuss their findings or methods. Use the ICPSR deposit when you downloaded the files and ran your own counts. Citing both is normal. The article's own supplement (10.1037/arc0000040.supp) is extra material for that paper; it is not a substitute for ICPSR 36966 if you used the archived files.

Personal authors: D'Souza and Wiseheart (Publication Manual)

University reprints of Publication Manual page 346 use this two-author ICPSR deposit. The title, study number 37080, Version V1, and DOI match the ICPSR 37080 record (Annalise D'Souza and Melody Wiseheart, York University; version date 17 May 2018). Two authors take an ampersand in the reference and in a parenthetical citation; a narrative citation uses "and."

The companion open-access article is also real. Cite the article when you discuss the null training result; cite ICPSR 37080 when you reanalyze the files. The article's methods note points readers to the deposit, but that sentence is not itself the data-set reference.

If one paper cites both the ICPSR deposit and this article, they share an author–date pair. Order them alphabetically in the reference list and add a and b after 2018, as in same author, same year. O'Donohue's 2017 deposit and 2018 article do not need letters.

Group author and a flexible bracket: NCES (Publication Manual)

The same manual page uses a National Center for Education Statistics file archived at the National Archive of Data on Arts and Culture. The author is the specific statistical agency, not the Department of Education. The bracket is [Data set and code book] because the deposit includes both. The DOI 10.3886/ICPSR35531.v3 is Version V3 of ICPSR 35531, distributed 2 May 2016 — the version year is 2016, even though the survey asked about the 2008–2009 school year and collected responses from September 2008 through July 2009.

If you only copied a percentage from the First Look report, cite the report, not the ICPSR deposit. The PDF's own suggested citation is not APA 7 — it uses "and," title case, and a Washington place line. Rebuild it with an ampersand, sentence-case italics, the report number, the specific agency, and the stable PDF URL. The government-report guide owns the rest of §10.4.

No DOI: Pew Research Center Wave 26 (APA Style)

APA Style's other standard data-set example is a group author with a URL and no DOI. Pew released American Trends Panel Wave 26 on 22 February 2018 (field dates 4–18 April 2017; topic: guns). The official example omits a publisher because Pew is both author and host, and it does not add a retrieval date because the wave is a dated release, not a live dashboard. The URL in the official example is the older Social Trends path. Pew's current catalog lists the same wave at pewresearch.org/dataset/american-trends-panel-wave-26. Keep the official APA URL when you are copying the Style Center example; use the current catalog URL if that is the page you actually opened.

Unpublished Raw Data

Section 10.9 also covers files that were never archived. If the files have a title, italicize it and add [Unpublished raw data]. If they do not, describe the content in roman type inside the brackets and do not invent a title. Name the university or agency that holds the files when you know it. The two official examples below are the titled and untitled cases from the manual. University reprints note that the date for unpublished data is the year (or years) of collection, not a later deposit year.

Those two lines are teaching examples from the manual, not public downloads. Oregon Youth Authority now publishes recidivism tables — the FY 2009–FY 2021 outcomes PDF is a report, not unpublished raw data. If you used that PDF, cite the report. If a supervisor handed you unpublished files that a reader could request, use the Baer or Oregon Youth Authority pattern. Unpublished raw data is still a reference-list item when a reader could request the files. A patient chart is not. APA treats medical records as nonrecoverable; describe them in the text without a reference entry.

Cite the Data Set or the Publication?

Use the official APA test, not "I saw a number somewhere." If the number lives only inside an article, report, or page, cite that publication. If you downloaded the files and ran your own analysis, cite the data set — and cite the companion paper only when you also discuss its methods or findings. A data-availability sentence in the article ("data are on OSF") is not a substitute for the data-set reference if you actually used the files.

Failure Modes Students Actually Hit

Retrieval dates on versioned files

APA 7 dropped "Retrieved from" for ordinary URLs. A retrieval date is a special case, not the default. The official instructional aid Creating an APA Style Reference List (and §9.16) limits it to unarchived works that are built to change:

A versioned ICPSR deposit, a DOI, or a numbered Pew wave does not get a retrieval date. A live indicators dashboard with no version tag and no archive does. When you need the date, the APA Style elements of reference list entries page uses this shape: Retrieved September 30, 2020, from https://xxxxx. Never pair a retrieval date with a DOI.

The fieldwork year is not the reference year

NCES asked teachers about the 2008–2009 school year; the Version V3 deposit is 2016. Pew fielded Wave 26 in April 2017; the public file is 2018. Write the version year. If you used more than one version of the same series, cite each version you actually analyzed. If you used one slice of a large deposit, cite the deposit and name the slice in the methods, unless that slice has its own DOI.

The parent department is the wrong author

"U.S. Government" and "Department of Education" hide the agency a reader would search. APA wants National Center for Education Statistics, Bureau of Labor Statistics, or U.S. Census Bureau. If the agency is also the publisher, omit the duplicate publisher element — the Pew example above is the model.

A DOI beats a download URL

Repositories love "Cite this dataset" buttons. Use them as a checklist, then rebuild the reference in APA 7. Keep the DOI; drop a proxy session URL, a bare doi:10... string, and a second copy of the landing page. Do not add a period after the DOI or URL. And do not let a handout overwrite the study number encoded in that DOI.

Software is §10.10, not a data set

The analysis program is not the file you analyzed. APA's reference examples index puts data sets in §10.9 and software in §10.10. SPSS, R, and Excel usually stay in the methods sentence. University reprints of the manual say reference entries are not required for standard software and programming languages such as Microsoft Word or Excel, Java, Adobe Photoshop, or SPSS. A specialized program that is not common knowledge in your field gets a software reference. The official Publication Manual sample (p. 339) is Comprehensive Meta-Analysis, not a data set:

Journal supplements are usually part of the article

If the only files you used are the article's own supplementary tables, cite the article and mention the supplement in the text. Give the supplement its own data-set reference only when it has been deposited separately with its own DOI. O'Donohue's Archives supplement and D'Souza's curriculum PDFs belong with those articles. ICPSR 36966 and ICPSR 37080 are separate deposits.

Restricted, registered, or proprietary files

Cite a restricted ICPSR or agency file the same way you cite an open one: author, version year, title, study number, repository, DOI. Then tell readers in the text how access works. Do not paste a personal download link. If the files cannot be requested by anyone else, treat them as unpublished raw data or describe them in the methods without pretending they are a public archive.

Data you collected for this paper

If the files exist only in your methods section, you do not give yourself a data-set reference. Describe collection, software, and exclusions in the text. If you deposited the files so others can reuse them, APA's first official sentence applies: archive the data and cite that deposit as you would anyone else's.

Related Guides

• How to Cite Government Reports in APA
• How to Cite Websites in APA
• How to Cite Journal Articles in APA
• How to Use DOIs
• How to Find a DOI
• How to Cite a Source with No Date
• Psychology Citation Guide
• How to Cite a Dataset in APA (worked page)