PERRLA/Guides/APA 7/Data set

How to Cite a Data Set in APA 7

An APA 7 data set reference lists the author (the researchers or the organization), the year the version you used was published, the data set's title in italics and sentence case, any identifier and version number in parentheses, the description [Data set] in square brackets, the publisher or repository, and the DOI or URL. The in-text citation is the author and the year: (Hummer et al., 2026). Unpublished data take [Unpublished raw data] instead, with the year the data were collected.

The format

Author, A. A., & Author, B. B. (Year). Title of the data set (Identifier; Version number) [Data set]. Publisher. https://doi.org/xxxx

PERRLA APA Experts · Updated October 21, 2026
Checked against the Publication Manual of the American Psychological Association, 7th edition. Every example is formatted by PERRLA's reference engine.

First, the easy way: PERRLA

Have a DOI? Paste it into PERRLA's reference search. PERRLA reads data set DOIs, like the ones ICPSR gives every study version, and builds the reference as a data set: every principal investigator, the title in sentence case, the ICPSR number and version in parentheses, the publisher, and the DOI. Check the authors, the year, and the title against the study page before you save. Set Publication Year to the year of the version you used (ICPSR shows it as the Version Date), because the DOI record often carries the year the study first came out. And put back the capitals in names inside the title, like Add Health, and shorten the publisher to ICPSR.

Adding one by hand? Choose Datasets, then A published dataset. Add each person as an Author, or the agency as a Group Author (Organization) with its abbreviation. Type the title in sentence case in Title of the Dataset, put identifiers and the version in Version Information (like ICPSR 21600; Version V26), then fill in Publication Year, Publishers, and the DOI or URL. PERRLA adds the parentheses and [Data set] for you, and leaves the publisher out when it matches the author.

For data that were never published, pick Unpublished raw data. The form changes to Collection Year and Data Source (the university or organization that holds the data), adds a Retrieval Date box, and PERRLA prints [Unpublished raw data]. CDC WONDER and data.census.gov links read as webpages in the reference search, so add those data sets by hand under Datasets.

Not using PERRLA yet? Here is how to build the same reference by hand.

The parts of the reference

1. Author

The author is whoever created the data: the principal investigators of a study, or the agency that runs a survey or database. List people with their last name and initials, and spell out an agency's full name.

Hummer, R. A., Aiello, A. E., Harris, K. M., & Udry, J. R.

When a government agency is several layers deep, name the most specific one as the author (National Center for Health Statistics, not the whole department), and put the parent agency in the publisher spot.

2. Year

Give the year the version you used was published, in parentheses. That is often not the year the study began or the years the data cover. Add Health began in 1994, the data run to 2025, and Version 26 came out in 2026, so the year is 2026.

(2026).

For unpublished raw data, give the year or years the data were collected. No date anywhere? Use (n.d.).

3. Title

Italicize the title and write it in sentence case: capitalize the first word, proper nouns, and acronyms, and lowercase the rest. Keep years and table names that are part of the title.

Multiple cause of death, 2018–2024, single race

Unpublished data with no title? Write a short description in square brackets, not italicized, in place of the title: the description tells the reader what the data are.

4. Identifier and version

Right after the title, add any numerical identifier and the version number in parentheses, not italicized, separated by a semicolon. Archives like ICPSR give every study a number and every release a version.

(ICPSR 21600; Version V26)

Use this spot for a table number too, such as a Census table, so your reader can find the exact table. No version or identifier? Leave the parentheses out.

5. Description in brackets

After the title and version, add [Data set] in square brackets, with a capital D and no italics. It tells the reader the source is data, not a report or an article.

[Data set].

APA lets you adjust the wording when it helps, as in [Data set and code book]. For data that were never published, the description is [Unpublished raw data].

6. Publisher

Name the publisher or the archive that distributes the data, like ICPSR or the parent agency. If the publisher is the same as the author, leave it out: don't write the same organization twice.

Centers for Disease Control and Prevention.

For unpublished raw data, name the university or organization that holds the data instead.

7. DOI or URL

End with the DOI as a link when the data set has one. Without a DOI, give the URL of the page where the reader gets the data. Don't put a period after either one.

https://doi.org/10.3886/ICPSR21600.v26

Add a retrieval date only when the data are not archived and are designed to change, like a database that adds new records every week: Retrieved October 21, 2026, from and the URL. A versioned file with a DOI, or a final annual release, doesn't need one.

Examples

A data set from ICPSR with a version and DOI

Add Health, the national study that has followed a cohort of U.S. adolescents into midlife, on ICPSR. The study has four principal investigators, an ICPSR number, and a numbered version. The year is the Version Date of Version 26, not the year the study first went up.

  • Parenthetical citation(Hummer et al., 2026)
  • Narrative citationHummer et al. (2026)

Data from CDC WONDER

The final Multiple Cause of Death database on CDC WONDER, which CDC lists as released in 2026. The National Center for Health Statistics is the author, and its parent agency, CDC, is the publisher. Cite the database once, however many queries you ran.

  • Parenthetical citation(National Center for Health Statistics [NCHS], 2026)
  • Narrative citationNational Center for Health Statistics (NCHS, 2026)
  • Later citations(NCHS, 2026)

A U.S. Census Bureau table from data.census.gov

American Community Survey table S2701 on health insurance coverage, from the 2020–2024 5-year estimates the Census Bureau released on January 29, 2026. The survey years and table number go in the parentheses. The Census Bureau is both author and publisher, so its name appears once.

  • Parenthetical citation(U.S. Census Bureau, 2026)
  • Narrative citationU.S. Census Bureau (2026)

In-text citations

An APA 7 in-text citation for a data set gives the author and the year, the same as any other source.

  • Researchers as authors: one or two names, or the first name and et al. for three or more: (Hummer et al., 2026) or Hummer et al. (2026)
  • An agency as author: spell out the name the first time and give the abbreviation in brackets, then use the abbreviation: (National Center for Health Statistics [NCHS], 2026), then (NCHS, 2026). In a sentence: National Center for Health Statistics (NCHS, 2026), then NCHS (2026).
  • No abbreviation: if the agency has no familiar short form, write out the name every time: (U.S. Census Bureau, 2026).
  • Unpublished raw data: the author and the year or years collected, like any other source.
  • No author: use the title in italics and title case in place of the author.
  • No date: use n.d. in place of the year.

Data sets have no page numbers, so a citation never needs one. To point to a specific table or variable, name it in your sentence.

Common mistakes

  • Hummer, R. A., Aiello, A. E., Harris, K. M., & Udry, J. R. (2008).Hummer, R. A., Aiello, A. E., Harris, K. M., & Udry, J. R. (2026).The year is the publication year of the version you used. Add Health first went up on ICPSR in 2008, and that year is still in its DOI record, so generators often print it. Version 26 came out in 2026.
  • Multiple cause of death, 2018–2024, single race [Dataset].Multiple cause of death, 2018–2024, single race [Data set].APA writes data set as two words. Many DOI tools print [Dataset] as one word.
  • U.S. Census Bureau. (n.d.). Selected Characteristics of Health Insurance Coverage in the United States.U.S. Census Bureau. (2026). Selected characteristics of health insurance coverage in the United StatesThe Cite button on data.census.gov gives n.d. and a title-case, unitalicized title. The release year is known (the 2020–2024 estimates came out in 2026), and a data set title takes italics and sentence case.
  • [Data set]. U.S. Census Bureau. https://data.census.gov/table/ACSST5Y2024.S2701[Data set]. https://data.census.gov/table/ACSST5Y2024.S2701When the author and the publisher are the same organization, APA names it once, as the author.
  • Citing the data set for a statistic you read in a reportCiting the report, article, or webpage where you read the statisticCite a data set only when you analyzed the data yourself or are sharing data you archived. A number you read in a published source comes from that source.

Quick checklist

  • Cite the data set only if you worked with the data yourself
  • The year of the version you used, not the year the study started
  • Title in italics and sentence case, acronyms kept
  • Identifier and version in plain parentheses, separated by a semicolon
  • [Data set], or [Unpublished raw data] for unpublished data
  • Publisher left out when it matches the author
  • DOI as a link, or the URL; a retrieval date only for data that keep changing

Questions students ask

Do I cite the data set or the article that reported it?

If you downloaded the data and ran your own analysis, cite the data set. If you are using a number someone else reported, cite the article, report, or webpage where you found it. A paper can cite both: the data set you analyzed and the articles that used it before you.

What year do I use for a data set in APA 7?

The year the version you used was published. On ICPSR that is the Version Date; on CDC WONDER it is the release year in the suggested citation; for Census tables it is the year the estimates were released. For unpublished raw data, use the year or years the data were collected.

How do I cite CDC WONDER in APA 7?

Cite the database you queried as a data set. The National Center for Health Statistics is the author, the release year is the date, the database name is the italic title, CDC is the publisher, and the database's request page is the URL. If you used a provisional database, like Provisional Mortality Statistics, 2018 through Last Week, the data change as records come in, so APA calls for a retrieval date.

How do I cite Census data from data.census.gov?

Use the U.S. Census Bureau as the author, the year the estimates were released, the table title in italics and sentence case, and the survey, its years, and the table number in parentheses. End with [Data set] and the table's URL. Leave the publisher out, since it's the Census Bureau again.

How do I cite unpublished raw data?

Give the author, the year or years the data were collected, and the title in italics followed by [Unpublished raw data]. If the data have no title, write a description in square brackets in its place, like [Unpublished raw data on sleep quality among night-shift nurses]. End with the university or organization that holds the data, and add a URL if there is one.

Do I need a retrieval date for a data set?

Usually not. APA uses a retrieval date only for sources that aren't archived and are designed to change, like a database that keeps adding records. A versioned data set with a DOI, or a final annual release, stays the same, so it doesn't need one.

Is it data set or dataset in APA?

APA writes it as two words, data set, in the bracketed description and in its guidance. Keep a repository's own spelling when it's part of a title.

Rules checked against the Publication Manual of the American Psychological Association (7th ed.), Sections 8.17, 9.7–9.37 (including 9.16 on retrieval dates and 9.21 on bracketed descriptions), 10.8, and 10.9, and the APA Style website's data set reference page, elements of reference list entries page, and blog post on unpublished works. Example sources verified on ICPSR (study page, Version V26, and its DOI), CDC WONDER (database page and its suggested data source citation), and data.census.gov and the Census Bureau's 2020–2024 ACS release page.

PERRLA builds the data set reference from its DOI or your details, with the version, the [Data set] label, and the publisher rules handled for you.

Try PERRLA free for 7 days