How journalists cite scientific studies in a news story

A news story cites a scientific study in its own sentences, not in a reference list: where the work appeared and when, who did it, what kind of study it was and how large, and a link to the paper. Then it says what the study cannot show. Every press code and newsroom guide we read asks for some part of that, and none asks for a citation style.

EdCitation publishes this guide, built on 27 September 2026 from press codes, newsroom and science-journalism guidance, a publisher's embargo policy, a preprint server's warning and five studies of science news. The AP Stylebook is sold, so no Stylebook rule is claimed; two sites that blocked our tools were read in archived copies.

EdCitation never writes a word of anyone's story; it looks studies up. Verify references confirms that each study you link exists, matches its record and has not been retracted; Cite a source builds a clean reference from a DOI for a fact box; Find sources finds the published research behind a claim. All three are free.

What do press codes say about citing research?

They ask for the source to be named clearly and checked, and the science guides go further: the study's size, stage, design and risks.

The general codes

The SPJ Code of Ethics, revised in 2014, asks journalists to verify information, use original sources whenever possible, and "identify sources clearly" so the public can judge their reliability (Society of Professional Journalists, 2014). For research, the original source is the paper, not the press release about it. It also warns against oversimplifying a story in its summary, which covers the headline.

The UK Editors' Code, in force from 1 January 2025 and enforced by IPSO, puts accuracy first: no misleading or distorted information, no headline the text does not support, comment kept apart from fact, and a significant inaccuracy corrected promptly and with due prominence (Editors' Code of Practice Committee, 2025).

The Associated Press's public statement of news values asks for anything that could reasonably be disputed to be attributed, and for a story to say so when a quotation comes from a written document such as a news release (Associated Press, 2026).

The science and health guidance

The Science Media Centre drew up its ten guidelines in 2012 with scientists, reporters and editors after the Leveson Inquiry, and Lord Justice Leveson called them "commendable for their utility as well as their succinctness" (Science Media Centre, 2012). The BBC note quotes its Editorial Guidelines: peer review is an indication of reliability but no guarantee (BBC, n.d.). The Association of Health Care Journalists (n.d.) adds: association is not cause.

How do you name a study and link to it in a news article?

Name it in the sentence where you use it, and link to the paper by its DOI. Linking to studies in news articles is now routine: in 100 stories about COVID-19 preprints, hyperlinks were the usual way of citing them (Fleerackers et al., 2022). The link replaces a reference list, so each element below must be in your prose.

A worked example

This paragraph is ours, written about a real study to show the elements in order:

Health news that exaggerates a study usually repeats an exaggeration already in the university's press release, according to research published in The BMJ on 9 December 2014. Petroc Sumner's team at Cardiff University analysed 462 releases on health research put out by 20 leading UK universities in 2011, setting each against its journal paper and the 668 news stories that followed. Because the study was observational, it cannot show that the releases caused the exaggerated news, the authors wrote.

Source, date, team, design, size and caveat, in three sentences. For a fact box or further-reading list, Cite a source builds a full reference from the DOI in any of six styles, and what is a DOI explains why a DOI link outlasts a journal's web address.

How do you read a study as a journalist before you report it?

Read the paper, not only the release, and put to it the five questions the BBC asks of any statistic (BBC, n.d.):

1. Who produced it, and how reliable is that source?
2. Why now, and does the source have an interest in the answer?
3. How was it done: what was counted, in whom, and how many?
4. What does it really show: does it answer the question it set, and what is left out, such as negative or failed results?
5. Where are the underlying data, and can you see them?

Then check the three traps the guides name. Association is not cause: an observational study shows a link, not that one thing drives another. Relative risk alone misleads: the BBC's own illustration is a 20% rise on a lifetime risk of 5%, which takes it to 6%, one extra case in every 100 people. And stage matters: a result in cells or mice is not a result in people, and an early drug trial mainly tests safety, not whether the drug works (Association of Health Care Journalists, n.d.).

How to read an academic paper fast gives a reading order; Find sources turns up the earlier evidence to set a new finding against.

How should journalists report preprints in the news?

Say it is a preprint, say what that means, and treat its findings as provisional. medRxiv, the health sciences preprint server, puts the rule at the top of its own page: preprints have not been certified by peer review and "should not be reported in news media as established information" (medRxiv, n.d.). Nature Portfolio (n.d.) recommends that reporters covering a preprint say it has not been peer reviewed and that its claims may change.

What studies of preprint coverage found

Coverage has not caught up. Van Schalkwyk and Dudek (2022) examined 80 COVID-19 news articles from four South African online outlets, published between January and July 2020, which referred to preprints 114 times. In 59% of those references readers got no sign that the findings were provisional, and the work was often just "a study"; only 5% said clearly what a preprint is and why it matters. The authors call their sample small. Fleerackers et al. (2022), whose abstract we read, found about half the stories stressed uncertainty, and outlets were much less likely to call the work a preprint than simply "research".

Does the label cost trust? In an experiment with US adults, Ratcliff et al. (2025) found no overall effect; readers who noticed the disclosure found the reporting and the scientists more credible. We read the abstract only, which gives no sample size.

Check whether the preprint has been published since

Many preprints later appear in journals. The Santa Clara County antibody study, posted on medRxiv on 17 April 2020 (Bendavid et al., 2020), appeared in the International Journal of Epidemiology on 22 February 2021, according to its Crossref record. A later story should cite the reviewed version. Types of academic sources covers preprints as sources more generally.

How do embargoes and press releases work for a reporter?

An embargo hands you the paper in advance, and in return you hold your story until a stated moment. Nature Portfolio (n.d.) gives registered journalists the full text about a week ahead, encourages them to read the whole paper, and has removed embargo breakers from its press list. Use the time for the methods and an outside expert.

One caution comes from the codes. The Association of Health Care Journalists (n.d.) warns against agreements that limit editorial independence, such as close-hold embargoes, and asks reporters who accept one to tell readers their reporting may not be complete. The press officer's side of embargoes is in how to write a press release.

What do studies say about exaggeration in health news?

Much of it starts before the newsroom. In the releases Sumner et al. (2014) analysed, 40% went beyond the paper in advising readers, 33% drew causal claims from correlational data, and 36% applied animal findings to people. Stories repeated each kind far more often after an exaggerated release (58%, 81% and 86% of them) than after a faithful one (17%, 18% and 10%). The design shows an association only, as the authors say, in one country and one year; later work is in our press release guide.

Yavchitz et al. (2012) looked for spin, reporting that makes a treatment look better than its results, in 70 press releases on randomised trials posted to EurekAlert between December 2009 and March 2010. Spin appeared in 40% of the papers' abstract conclusions, 47% of the releases and 51% of the 41 news items. Spin in a release was tied to spin in the abstract, so read the results, not just the abstract.

What should you do if a study you cited is retracted?

Correct the story openly, as the Editors' Code requires of a significant inaccuracy. Retraction can be quick: a Lancet paper on hydroxychloroquine and chloroquine in hospital patients with COVID-19 went online on 22 May 2020 (Mehra et al., 2020), and its retraction notice followed on 5 June 2020, by its Europe PMC record. The DOI in an old story still works; the record now begins "RETRACTED". Check a study again before you link it in a follow-up. Retracted papers in your reference list explains the notices; fabrication scandals in journalism and science shows how invented work was caught.

Our checklist for reporting a study

This checklist is EdCitation's own, drawn from the codes and studies above; your newsroom's standards come first.

1. Read the paper, not only the release: sample, design, main result.
2. Name the journal or server and the date.
3. Say what kind of study it is and how many took part.
4. Say whether it is peer reviewed; for a preprint, say what that means.
5. Give absolute risk beside any relative change.
6. Keep association apart from cause, and finding apart from interpretation.
7. Ask a researcher who was not involved.
8. Note funding or interests that bear on the claim.
9. Link the paper by its DOI, after Verify references confirms it exists and has not been retracted.
10. Write a headline the story supports, and correct errors openly.

How can EdCitation check the studies in a story?

For the part of a science story with a right answer, whether the study you link is real and still stands, we think EdCitation is the strongest tool going, because it checks every entry against the record its publisher registered, something a chatbot that generates text cannot do, and it never shows "could not check" as "not found". Here is what it did on 27 September 2026.

Verify references on five studies

We gave Verify references five entries: the retracted Lancet paper, the Santa Clara preprint, and the studies by Sumner, van Schalkwyk and Yavchitz. All five came back verified: "The DOI resolves to this record and the title matches." The Lancet entry added: "This paper has since been retracted." Verified means the work exists as described; the tool did not mention the preprint's journal version, so that check stays yours.

Cite a source on the same two

From the DOI, Cite a source marked the Lancet paper retracted and listed its notices, among them the retraction notice (DOI ending 31324-6); its APA 7 entry left out APA's retraction bracket, which we added by hand below. For the preprint it named medRxiv and kept "Santa Clara County" capitalised in the sentence-case APA title, and its Vancouver entry was labelled "[Preprint]".

All of this is free, with no account. Pro, at $8 a month, adds References from a file for a long report with a reference list; see Pricing.

Quick questions

Should a news article link to the study?

Yes. The Science Media Centre asks for enough detail for readers to find the study, or a link, and a DOI link keeps working when a journal's website changes.

How do you cite a preprint in a news story?

Name the server and the date, say the work has not been peer reviewed and its findings may change, and link to it. Check first whether a journal version has appeared.

Do journalists use APA or another citation style?

Not in the story, where sources are named in the sentence. A fact box or further-reading list can use one, and EdCitation's free Cite a source builds it from a DOI in APA 7 or five other styles.

What is an embargo in science journalism?

A publisher's condition that a study sent to journalists in advance is not reported before a stated date and time. Nature Portfolio sends papers about a week early.

Does EdCitation write news stories?

No. EdCitation finds, cites and checks the research; the reporting and the writing stay yours.

References

• Associated Press. (2026). The Associated Press statement of news values and principles [PDF]. https://www.ap.org/wp-content/uploads/2026/02/THE-ASSOCIATED-PRESS-STATEMENT-OF-NEWS-VALUES-AND-PRINCIPLES-1.pdf
• Association of Health Care Journalists. (n.d.). Statement of principles of the Association of Health Care Journalists. https://healthjournalism.org/about/principles-and-policies/statement-of-principles-of-the-association-of-health-care-journalists/
• BBC. (n.d.). Guidance: Reporting statistics and data. BBC Editorial Guidelines. https://www.bbc.co.uk/editorialguidelines/guidance/reporting-statistics-and-data/
• Bendavid, E., Mulaney, B., Sood, N., Shah, S., Ling, E., Bromley-Dulfano, R., Lai, C., Weissberg, Z., Saavedra-Walker, R., Tedrow, J., Tversky, D., Bogan, A., Kupiec, T., Eichner, D., Gupta, R., Ioannidis, J. P. A., & Bhattacharya, J. (2020). COVID-19 antibody seroprevalence in Santa Clara County, California. medRxiv. https://doi.org/10.1101/2020.04.14.20062463
• Editors' Code of Practice Committee. (2025). The Editors' Code of Practice. https://editorscode.org.uk/the-code/
• Fleerackers, A., Riedlinger, M., Moorhead, L., Ahmed, R., & Alperin, J. P. (2022). Communicating scientific uncertainty in an age of COVID-19: An investigation into the use of preprints by digital media outlets. Health Communication, 37(6), 726–738. https://doi.org/10.1080/10410236.2020.1864892
• medRxiv. (n.d.). About medRxiv. Archived August 20, 2026, at http://web.archive.org/web/20260820140838/https://www.medrxiv.org/content/about-medrxiv
• Mehra, M. R., Desai, S. S., Ruschitzka, F., & Patel, A. N. (2020). Hydroxychloroquine or chloroquine with or without a macrolide for treatment of COVID-19: A multinational registry analysis. The Lancet. https://doi.org/10.1016/S0140-6736(20)31180-6 (Retraction published 2020, The Lancet, 395[10240], 1820)
• Nature Portfolio. (n.d.). Press and embargo policies. https://www.nature.com/nature-portfolio/editorial-policies/press-and-embargo-policies
• Ratcliff, C. L., Fleerackers, A., Wicke, R., King, A. J., & Jensen, J. D. (2025). Using preprint sources in science news: Do transparent disclosures enhance credibility? Science Communication. Advance online publication. https://doi.org/10.1177/10755470251363191
• Science Media Centre. (2012). 10 best practice guidelines for reporting science & health stories. https://www.sciencemediacentre.org/guidelines-for-reporting/
• Society of Professional Journalists. (2014). SPJ code of ethics. Archived September 17, 2026, at http://web.archive.org/web/20260917181518/https://www.spj.org/spj-code-of-ethics/
• Sumner, P., Vivian-Griffiths, S., Boivin, J., Williams, A., Venetis, C. A., Davies, A., Ogden, J., Whelan, L., Hughes, B., Dalton, B., Boy, F., & Chambers, C. D. (2014). The association between exaggeration in health related science news and academic press releases: Retrospective observational study. BMJ, 349, Article g7015. https://doi.org/10.1136/bmj.g7015
• van Schalkwyk, F., & Dudek, J. (2022). Reporting preprints in the media during the COVID-19 pandemic. Public Understanding of Science, 31(5), 608–616. https://doi.org/10.1177/09636625221077392
• Yavchitz, A., Boutron, I., Bafeta, A., Marroun, I., Charles, P., Mantz, J., & Ravaud, P. (2012). Misrepresentation of randomized controlled trials in press releases and news coverage: A cohort study. PLOS Medicine, 9(9), Article e1001308. https://doi.org/10.1371/journal.pmed.1001308