A secondary data dissertation answers its question from data that someone else collected: a national survey, official statistics, an archive's records, or the files behind a published study. It saves fieldwork and moves the hard work elsewhere: finding data that fit, lawful access, and citing the exact version you analysed.
This guide, published by EdCitation, covers each step, with every rule read on 26 September 2026 at the university, archive or style publisher that sets it; what could not be confirmed is said. It belongs with our guide to how to write a dissertation.
Give EdCitation's free Cite a source a dataset's DOI and it builds the entry from the registry's record. The last section shows what it returned for four archives' datasets.
What counts as secondary data in a dissertation?
Data collected for another purpose and reused to answer your question. The University of Bath (n.d.) gives census data and datasets from earlier projects as examples; Leeds University Business School (n.d.) adds databases, company records, archives and reports. The glossary defines secondary data, secondary data analysis and a data archive.
One word, two meanings
Universities do not all mean the same thing. In the University of Manchester (2023) handbook for two Geography master's degrees, the secondary data option may rest on published research alone and must show "added value" in how it organises, synthesises and evaluates that research: closer to an extended literature review. At Leeds University Business School they are files and records you analyse yourself. Ask which your programme means.
The dataset or the publication?
The American Psychological Association (2020) reserves a data set reference for data you analysed yourself from an archive; figures you take from someone else's article, report or web page are cited by that publication. A table of unemployment rates copied from a report is the report. A model you ran on the survey's own files is the dataset.
Do universities allow a dissertation built on secondary data?
Yes, where the programme provides for it. Manchester and Leeds University Business School name it as a route in their own guidance; Bath, St Andrews, Loughborough and Utah set the conditions for research on existing data. What differs is the paperwork, from a full application to a record that none was needed.
| University | What its own guidance says |
|---|---|
| Manchester (Geography MSc, 2023–24) | A dissertation option based on secondary data only, if it adds value |
| Leeds University Business School (undergraduate) | Every dissertation needs ethics approval; for secondary data, agree the sources with your supervisor, declare them lawful and append the approval email |
| Bath | All research using secondary data goes through ethics review; the level depends on how identifiable and sensitive the data are |
| St Andrews | Review is usually not needed for data already robustly anonymised when you receive them; personal data need it |
| Loughborough | Review may not be needed for fully anonymous or public-domain data; data that are not public need the owner's permission |
| Utah (United States) | Analysis of de-identified public-use datasets, including ICPSR's public-use files, is not human subjects research; merging datasets in a way that could identify people needs approval |
Sources: University of St Andrews (n.d.), Loughborough University (n.d.) and University of Utah IRB (2025). Leeds University Business School (n.d.) also rules out shared logins and any source whose origin cannot be verified.
Ethics, briefly
The rules for reused data, and where committees disagree, are set out in research ethics approval for a dissertation. Whatever your committee decides, record it in your methodology chapter, along with the dataset, its version and the anonymisation it had when you received it.
Where can I find datasets for a dissertation?
Start at a data archive: archives document their data and give each dataset a DOI and a citation.
| Source | Access | Its own citation rule |
|---|---|---|
| UK Data Service | Open; safeguarded (register and accept the End User Licence); controlled (SecureLab, accredited researchers) | A citation on every catalogue record, with study number (SN) and DOI |
| ICPSR | Public-use files to download; restricted data by application | A citation on every study page |
| Eurostat | Free reuse with the source acknowledged, with exceptions such as data from non-EU countries | "Source:" the dataset's DOI and the date accessed; a customised extract by its datacode link |
| World Bank Data Catalog | CC BY 4.0 by default; the Microdata Research License limits use to research | The citation given with each dataset |
| Office for National Statistics | Open Government Licence v3.0 | A set attribution statement |
| Dryad | CC0 for every dataset | Author(s). Year. Title [Dataset]. Publisher. DOI |
| Zenodo | The licence the depositor chose | A DOI for each version and one for all versions |
The UK Data Service (n.d.) says its controlled data are not suitable for undergraduates. On ICPSR, even the study APA uses as its model example is restricted, with access requested from APA's editorial office. For a discipline's own repository, re3data (n.d.) lists repositories worldwide, filtered by licence, identifier and other facets.
EdCitation's Find sources has a Dataset option in its kind-of-source filter. On 26 September 2026 a search for British Social Attitudes with that filter on returned 73 records, all registered with Crossref: 20 from ICPSR, among them the survey's releases from 1986 to 2001, 46 from APA's PsycTESTS and PsycEXTRA, and seven others. None came from the UK Data Service, Zenodo or Dryad, and a search for the exact title of the Dryad dataset below did not find it. Use the filter for data that have entered the published record, and search the archives for the rest.
What do data licences and access conditions allow?
Read the terms before you download; they decide what you may do and sometimes how you must cite.
- UK Data Service. Version 16.00 of its End User Licence (UK Data Service, 2026), dated 25 February 2026, bars "Online Data Tools", a term that includes generative AI, from use with the data without written permission. It requires you to cite the data "in the form specified in accompanying metadata", and to send the service the details of any published work that uses them. Safeguarded data may not be shared with anyone who is not registered.
- Dryad. CC0 means attribution is not a legal requirement, but Dryad (n.d.) strongly recommends citing any data you reuse, as scholarly courtesy.
- World Bank. Under CC BY 4.0 you credit the source and say what you changed; data from its Microdata Library must be cited in theses and dissertations that use them (World Bank, n.d.).
- Eurostat and the ONS. Both allow reuse with the source acknowledged. The ONS gives its wording: "Source: Office for National Statistics licensed under the Open Government Licence v.3.0" (Office for National Statistics, n.d.).
A Creative Commons licence or a data sharing agreement is part of your method: name it in the methodology chapter.
How do I cite a dataset in APA 7?
Start with the creator and the year of the version you used. The italic title follows, then the dataset's number and version together in round brackets with a semicolon between them, a bracketed description, the publisher and the DOI. The description is flexible: "[Data set]" or "[Data set and code book]" (American Psychological Association, 2020). APA's own example:
O’Donohue, W. (2017). Content analysis of undergraduate psychology textbooks (ICPSR 36966; Version V1) [Data set]. ICPSR. https://doi.org/10.3886/ICPSR36966.v1
In the text: (O’Donohue, 2017). A UK Data Service study follows the same pattern, with the study number and edition from its record:
NatCen Social Research. (2026). British Social Attitudes Survey, 2020 (SN 9005; 1st Edition) [Data set]. UK Data Service. https://doi.org/10.5255/UKDA-SN-9005-1
How do I cite a dataset in MLA, Chicago, Harvard, IEEE and Vancouver?
Each style keeps the same facts in its own order, and two of them leave the details unconfirmed.
| Style | Title | Label and version | Read at |
|---|---|---|---|
| APA 7 | Italics, sentence case | [Data set]; identifier and version in parentheses | APA Style |
| MLA 9 | Italics in MLA's example of a whole data set | No label shown | MLA Style Center; the ninth edition's handbook could not be read |
| Chicago 18 author-date | Could not be confirmed | Could not be confirmed | Chicago's free guide has no data set example |
| Harvard | Single quotation marks | No label shown | UCL's Harvard guide |
| IEEE | Quotation marks, version inside the title | None; "doi:" before the number | IEEE Reference Guide, version 3.28.2025 |
| Vancouver | Plain | [dataset], or [dataset on the Internet] | NLM samples and Citing Medicine |
MLA 9 and Chicago 18
The Modern Language Association (2018) cites a whole data set with the title in italics: MLA Language Map Data Center, then the publisher and the address. Data from a survey you ran yourself get no works-cited entry; the report explains how you collected them. That post dates from the eighth edition, and the ninth's handbook is not free to read. For Chicago, the free author-date guide covers no data sets and the manual is by subscription, so its form could not be confirmed; the University of Michigan Library (n.d.) italicises the title in its author-date example, but for the 16th edition.
Harvard, IEEE and Vancouver
Cite Them Right's dataset page needs a login. University College London (2026) gives the author, the year in brackets, the title in single quotation marks and the DOI, with no accessed date. IEEE (2025) follows the FORCE11 principles: Author, Date, "Title of Dataset," Source, doi: and the number, with any version inside the title. The National Library of Medicine (2026) puts "[dataset]" after the title, then the date, the date cited, the repository and the address; Citing Medicine (National Library of Medicine, 2018) gives the full form, with an ICPSR study number after the title.
Which version and which DOI should I cite?
The version you analysed, by its own DOI where it has one. DataCite, which registers most dataset DOIs, recommends Creator (PublicationYear): Title. Version. Publisher. (resourceTypeGeneral). Identifier, and prefers the DOI as a full https://doi.org/ link (DataCite Metadata Working Group, n.d.). See what a DOI is for how the link works.
- Zenodo registers a DOI for each version and one for all versions together, which resolves to the latest. Zenodo (n.d.) says to cite the specific version, normally.
- UK Data Service numbers editions. British Social Attitudes 2020 was first released on 14 September 2022 and re-released, still as the 1st Edition, on 12 August 2026; the archive's citation now says 2026. Which year APA wants for a re-release under the same number could not be confirmed: give the year the archive shows for your edition, and keep your download date.
- Eurostat changes its tables as figures are revised. A customised extract is cited by its datacode link with the date you accessed it (Eurostat, n.d.).
Keep a data log from the first download: the DOI, the edition or version, the date, and the file names. Your results chapter and your examiners will need them.
How does the archive's own citation compare with the styles?
It holds the right facts, in no student style. The UK Data Service gives "NatCen Social Research. (2026). British Social Attitudes Survey, 2020. [data collection]. UK Data Service. SN: 9005, DOI: http://doi.org/10.5255/UKDA-SN-9005-1". ICPSR adds Ann Arbor, "[distributor]" and the date 2017-12-20; Dryad uses "[Dataset]" (Wang et al., 2026).
Take the facts and set them in your style. One clash is real: the UK Data Service's licence asks for the form in its metadata, your university for its style. No page of the service we read settles it; our view is to keep every element the archive gives (SN, edition, DOI) in your style's order, and ask your supervisor.
How EdCitation helps with a secondary data dissertation
For citing data, EdCitation is the strongest choice we know of, and the reason is where its entries come from: each dataset is looked up in the DOI registry, and the entry is built from that record rather than composed, which a chatbot cannot do. Cite a source costs nothing and needs no account: paste a DOI and get the entry in six styles, screened for retraction. EdCitation never writes any part of your dissertation and never touches your data.
What Cite a source returned
On 27 September 2026 it gave, in APA, for ICPSR:
O’Donohue, W. (2017). Content analysis of undergraduate psychology textbooks [Data set]. Inter-University Consortium for Political and Social Research. https://doi.org/10.3886/ICPSR36966.v1
It set the title in sentence case, as APA 7 asks; against APA's example, add "(ICPSR 36966; Version V1)". For the UK Data Service it wrote "(Version 1)" and lowered the survey's name to "British social attitudes survey, 2020"; restore its capitals, and as the archive numbers editions, make it "(SN 9005; 1st Edition)". The Dryad entry, "(Version 5) [Data set]. Dryad.", needed nothing. For Eurostat's GDP table (10.2908/nama_10_gdp) it chose the English of the record's three titles, French, English and German. In the other styles, its MLA and Chicago entries put the title in quotation marks, its IEEE entry puts the date after the source, its Vancouver entry lacks "[dataset]", and its Harvard entry keeps "Available at:" before the DOI.
Checking the list
Verify references took a list of six datasets. Five came back verified, among them Eurostat's table under its English title and a World Bank page with no DOI, checked at its address. The entry with an invented Figshare DOI came back not found: "No registry holds the DOI 10.6084/m9.figshare.99999999."
When the full draft exists, References from a file, part of Pro at $8 a month, reads the dissertation itself, checks each entry and pairs every citation in the text with its entry; Mechanics QA, also Pro, checks the formatting against your style. Check your paper, free, reads your handbook's rules into a checklist.
Quick questions
Can a dissertation use only secondary data?
Yes, where the programme allows it, as Manchester's Geography master's and Leeds University Business School's undergraduate dissertations do. Check first what your programme means by secondary data.
Does a secondary data dissertation need ethics approval?
Often. Leeds University Business School and Bath review every such project; St Andrews and Loughborough may not for anonymous or public data. Follow your own university's rule.
How do I cite a dataset in APA 7?
Give the creator, the year of the version used, the italic title, the identifier and version in parentheses, "[Data set]", the publisher and the DOI. Cite the publication instead when you only use figures it published.
Should I cite the version DOI or the all-versions DOI?
The version DOI, normally, because it points to the files you analysed. Zenodo keeps the all-versions DOI for citing a work that keeps changing.
Can EdCitation build a dataset reference from its DOI?
Yes. Cite a source builds it free from the registry's record in six styles; then glance at any name in the title, and check the identifier, version and label.
References
- American Psychological Association. (2020). Data set references. APA Style. https://apastyle.apa.org/style-grammar-guidelines/references/examples/data-set-references
- DataCite Metadata Working Group. (n.d.). Citation. DataCite Metadata Schema documentation 4.7. https://datacite-metadata-schema.readthedocs.io/en/4.7/introduction/citation/
- Dryad. (n.d.). How to reuse Dryad data. https://datadryad.org/help/guides/reuse
- Eurostat. (n.d.). Copyright notice and free re-use of data. European Commission. https://ec.europa.eu/eurostat/help/copyright-notice
- IEEE. (2025). IEEE reference style guide for authors (Version 3.28.2025). IEEE Author Center. https://journals.ieeeauthorcenter.ieee.org/wp-content/uploads/sites/7/IEEE_Reference_Guide.pdf
- Leeds University Business School. (n.d.). UG dissertations: Ethics approval. LUBS Student Guide. https://students.business.leeds.ac.uk/assessment/dissertations-and-projects/ug-diss-ethics/
- Loughborough University. (n.d.). Secondary or pre-existing human data. https://www.lboro.ac.uk/internal/research-ethics-integrity/research-ethics/guidance-investigators/secondary_data/
- Modern Language Association. (2018). How do I cite raw data from a survey or poll I created? How do I cite published data? MLA Style Center. https://style.mla.org/citing-raw-data-from-a-survey/
- NatCen Social Research. (2026). British Social Attitudes Survey, 2020 (SN 9005; 1st Edition) [Data set]. UK Data Service. https://doi.org/10.5255/UKDA-SN-9005-1
- National Library of Medicine. (2018). Databases/retrieval systems/datasets on the Internet. In Citing medicine: The NLM style guide for authors, editors, and publishers (2nd ed.). https://www.ncbi.nlm.nih.gov/books/NBK7273/
- National Library of Medicine. (2026). Samples of formatted references for authors of journal articles. https://www.nlm.nih.gov/bsd/uniform_requirements.html
- O’Donohue, W. (2017). Content analysis of undergraduate psychology textbooks (ICPSR 36966; Version V1) [Data set]. ICPSR. https://doi.org/10.3886/ICPSR36966.v1
- Office for National Statistics. (n.d.). Licences. https://www.ons.gov.uk/methodology/geography/licences
- re3data. (n.d.). About. Registry of Research Data Repositories. https://www.re3data.org/about
- UK Data Service. (n.d.). Types of data access. https://ukdataservice.ac.uk/help/access-policy/types-of-data-access/
- UK Data Service. (2026). End User Licence Agreement (Version 16.00). University of Essex. https://dam.ukdataservice.ac.uk/media/455131/cd137-enduserlicence.pdf
- University College London. (2026). Scientific dataset. UCL Library Harvard referencing guide. https://library-guides.ucl.ac.uk/harvard/scientific-dataset
- University of Bath. (n.d.). Working with secondary data. https://www.bath.ac.uk/guides/working-with-secondary-data/
- University of Manchester. (2023). Taught master's programmes in Geography: Notes of guidance on dissertation preparation and submission, 2023–24. https://documents.manchester.ac.uk/display.aspx?DocID=72050
- University of Michigan Library. (n.d.). Citing data & statistics. https://guides.lib.umich.edu/c.php?g=282964&p=3285995
- University of St Andrews. (n.d.). Secondary data. https://www.st-andrews.ac.uk/research/integrity-ethics/humans/ethical-guidance/secondary-data/
- University of Utah IRB. (2025). Secondary analysis of public use datasets. https://irb.utah.edu/guidance-series/secondary-analysis-of-public-use-datasets/
- Wang, H., Zou, Q., Zhang, J., Gao, J.-H., & Liu, Y. (2026). Data from: Human brain-wide activation of sleep rhythms (Version 5) [Data set]. Dryad. https://doi.org/10.5061/dryad.2fqz612x0
- World Bank. (n.d.). Public licenses. World Bank Data Catalog. https://datacatalog.worldbank.org/public-licenses
- Zenodo. (n.d.). FAQ: Versioning. https://zenodo.org/help/versioning