Skip to main content

Science & Research Reporting

New findings, expert claims and health scares reach every newsroom: how to read a paper critically, use the embargo system without being captured by it, get the statistics right, and report evidence — from preprints to retractions — without amplifying hype or harm.

Last reviewed: Next review due:

What is the science beat?

The science and research beat covers new findings and the system that produces them: universities and research institutes, the funding councils under UK Research and Innovation, the journals where results are published, and the agencies — from the Office for National Statistics to the UK Health Security Agency — whose data underpins public debate. It ranges from a genuinely important trial result to a cell study dressed up as a cure, and much of the job is telling those two apart at speed.

Its defining feature is that the raw material is technical: claims arrive wrapped in methods, statistics and press releases, and verification means reading the paper, interrogating the numbers and phoning independent experts — not balancing one quote against another. Done well, the beat is a public service twice over: it explains research that changes lives, and it stands between readers and the steady stream of overclaimed, misreported or occasionally fraudulent science that would otherwise reach them unfiltered.

Why this beat matters

  • 1Science and health stories change behaviour: what people believe about vaccines, screening, diet and medicines is shaped by how research is reported, and inaccurate coverage can do real public-health harm.
  • 2Public money is at stake: UK Research and Innovation describes investing more than £9 billion a year in research and innovation, and universities and health agencies spend billions more — all of it accountable to the public.
  • 3The pandemic made evidence literacy a newsroom-wide need: preprints, modelling and daily statistics moved from specialist journals to front pages, and audiences now expect journalists to weigh evidence, not just relay it.
  • 4Research integrity is a growing story in its own right: retractions, paper mills and manipulated data are tracked daily by outlets such as Retraction Watch, and the UK institutions handling them deserve scrutiny.
  • 5Hype has an industrial pipeline: press releases and thinly staffed newsdesks can move an overclaimed finding from lab to headline in hours — this beat exists to be the check on that pipeline.

Core legal and ethical risks

Harmful health misinformation and miracle cures

Stories about unproven treatments can cause direct harm to people who act on them. The canonical cautionary tale is the retracted 1998 Lancet paper claiming a link between the MMR vaccine and autism: the journal fully retracted it in 2010 and the lead author was struck off the UK medical register, but the coverage it generated damaged vaccine confidence for years. Never present an unproven remedy as a cure, and remember that advertising treatments for cancer to the public is restricted by the Cancer Act 1939 — a mark of how seriously the law treats this territory.

Defamation when reporting on named researchers

Saying a scientist fabricated data or committed misconduct is a serious defamatory allegation. Anchor such reporting in documented findings — a retraction notice, an upheld institutional investigation, a tribunal or regulator ruling — and distinguish honest error from fraud: many retractions correct mistakes, and the notice normally states the reason. Offer a right of reply, keep your language precise about what has actually been established, and run contentious pieces through the checklist at /law/defamation-risk-checklist.

Embargo etiquette and capture

The embargo system — journals and press offices giving accredited journalists advance access on condition of a fixed publication time — buys time to read papers and seek independent comment. Breaking an embargo can cost you and your outlet future access. The subtler risk is capture: embargoed press releases arrive pre-packaged, making it easy to publish PR-led science news without independent scrutiny. Use the time an embargo gives you to test the claim, not just to polish the write-up.

Statistical accuracy

Most corrections on this beat are numerical: relative risks reported without absolute numbers, correlation presented as causation, small or unrepresentative samples generalised, and p-values treated as proof. The Editors' Code requires care over accuracy — including headlines supported by the text — and broadcast news must be reported with due accuracy under the Ofcom Broadcasting Code. The Royal Statistical Society publishes resources for journalists on handling numbers; when in doubt, ask a statistician before publication, not after.

False balance and evidence quality

Impartiality does not mean giving a fringe position equal weight against a strong scientific consensus; broadcasters' editorial guidelines speak of due weight for exactly this reason. Distinguish the quality of evidence in play — a conference abstract, a preprint, a single small study and a systematic review of randomised trials are not equivalent — and tell readers which one your story rests on. Presenting weak and strong evidence as equals misleads audiences as surely as a wrong figure does.

See also: Defamation risk checklist | Health & NHS reporting | Ethics hub

UK data and analysis sources for science reporters

FOI and public-record ideas for science reporters

  • Research misconduct at universities — the number of formal investigations opened, upheld and leading to retractions or dismissals over a period (universities are subject to FOI; ask for aggregate figures)
  • University spending on article processing charges and journal subscriptions paid to major academic publishers
  • Clinical trial transparency — records held by a university or NHS trust on whether its completed trials have reported results to the trial registry
  • Correspondence between a university press office and researchers over a contested or later-corrected press release
  • UKHSA incident and outbreak documentation — minutes, situation reports and risk assessments, requested once an incident has concluded
  • Commercial influence on research — gifts, sponsorships and consultancies recorded in university registers of interest, and confidentiality clauses in research contracts
  • Grant funding patterns — UKRI and research council data on application numbers, success rates and awards by institution (much is published; FOI fills the gaps)

See the full FOI toolkit at FOI guides.

Key UK organisations and contacts

Science Media Centre
Independent press office connecting journalists with researchers — briefings, roundups and rapid expert reactions when science is in the news.
UK Research and Innovation (UKRI)
The national research funding agency, spanning nine councils including the Medical Research Council and Innovate UK; a primary source on funding and research policy.
UK Health Security Agency (UKHSA)
Executive agency of the Department of Health and Social Care responsible for health protection — infectious disease surveillance, outbreak response and environmental hazards.
Royal Statistical Society
Professional body for statisticians — journalist resources, training and a network of experts who can sanity-check numbers before you publish them.
Office for Statistics Regulation
The regulatory arm of the UK Statistics Authority — publicly challenges the misuse of official statistics, and its interventions are often stories in themselves.
Nuffield Council on Bioethics
Independent policy and research centre that identifies, analyses and advises on ethical issues in biomedicine and health — a measured voice on contested bioscience.
Association of British Science Writers (ABSW)
The professional association for science journalists and writers in the UK — training, awards and a community of specialist reporters.
Committee on Publication Ethics (COPE)
Sets widely used guidance for journals and publishers on handling suspected misconduct, corrections and retractions — useful for judging whether a journal has responded properly to a problem paper.

Interview question bank

For Scientists and researchers

  • What does your study actually show — and what does it not show?
  • Was this in humans, animals or cells, and how large was the sample?
  • Who funded the work, and do any of the authors have competing interests?
  • How does this fit with previous research — does anything contradict it?
  • What would change your mind, and what needs to happen before anyone acts on this?

For Press officers

  • Does the headline claim in the release appear, in those terms, in the paper itself?
  • Did the study authors approve the release, and can I speak to them directly?
  • Is the paper peer reviewed, a preprint, or a conference presentation?
  • Can you send the full paper, the data availability statement and the funding details?
  • Are independent experts available who were not involved in the work?

For Statisticians and methodologists

  • Is the effect size meaningful in absolute terms, or only relative ones?
  • Could confounding or bias explain this result?
  • Was the study large enough to detect the effect it claims, and was the analysis pre-registered?
  • How fragile is the finding — would a handful of different outcomes change the conclusion?
  • If you were editing this story, what number would you insist we include?

Jargon glossary

Peer review
Evaluation of a manuscript by independent researchers before journal publication. A quality filter, not a guarantee — peer-reviewed papers can still be wrong, and are sometimes retracted.
Preprint
A full draft paper posted openly (for example on bioRxiv, medRxiv or arXiv) before or without peer review. Always label preprints as not yet peer reviewed.
Randomised controlled trial (RCT)
A study that randomly assigns participants to an intervention or a comparison group — the strongest single-study design for testing whether a treatment works.
Systematic review / meta-analysis
A structured synthesis of all relevant studies on a question, sometimes pooling their data statistically. Generally stronger evidence than any single study.
P-value
A measure of how surprising the data would be if there were no real effect. It is not the probability the finding is true, and "statistically significant" does not mean large or important.
Confidence interval
The range within which the true value plausibly lies given the data. Wide intervals signal uncertainty — report them, not just the central estimate.
Absolute vs relative risk
Relative risk describes proportional change ("doubles the risk"); absolute risk describes real-world frequency (from 1 to 2 in 10,000). Report absolute numbers alongside relative ones.
Confounding
When a third factor influences both the exposure and the outcome, creating a misleading association — the core reason correlation is not causation.
Retraction / expression of concern
A retraction formally withdraws a paper; an expression of concern flags unresolved doubts. Reasons range from honest error to fabrication — read the notice before characterising it.
Embargo
An agreement giving journalists advance access to a paper or briefing on condition nothing is published before a set time. Breaching one risks losing access to the system.

Story ideas and angles

  • Follow the funding: use the UKRI Gateway to Research to map which local institutions win grants, for what, and how the pattern is shifting.
  • Audit clinical-trial reporting: which universities or NHS trusts in your patch have completed trials whose results were never reported?
  • Track retractions with a local angle: search the Retraction Watch database for papers from nearby institutions and ask what happened.
  • Test the press-release pipeline: compare a week of university press releases against the underlying papers and document where the claims grew in translation.
  • Investigate research culture: precarious contracts, bullying findings and misconduct-investigation numbers obtainable from universities under FOI.
  • Localise national statistics: take a new ONS or UKHSA release and report what it shows for your area, with help from the producer's own statisticians.
  • Examine commercial influence: registers of interest, industry-funded studies, and how funding sources are disclosed — or not — in coverage.
  • Revisit a hyped finding from five years ago: did the wonder material, miracle drug or breakthrough therapy ever replicate?

Pitch angles

Science pitches land when they interrogate evidence or follow public money — not when they relay a press release. Try:

  • Accountability: “This university opened a record number of research-misconduct investigations. We obtained the figures under FOI.”
  • Money: “Millions in public research funding went into this field. We follow the grants and ask what came back.”
  • Health: “The headline said an everyday food doubles your risk. The absolute risk barely moved. We show the working.”
  • Integrity: “A local scientist’s paper was quietly retracted. The notice says error; the documents suggest a longer story.”

Recommended tools

Journalists whose work shaped evidence-led reporting

Harold Evans

Sir Harold Matthew Evans (28 June 1928 – 23 September 2020) was a British-American journalist and writer. In his career in his native Britain, he was editor of The Sunday Times from 1967 to 1981, and its sister title The Times for a year from 1981, before being forced out of the latter post by Rupert Murdoch.

Summary adapted from the Wikipedia article on Harold Evans, licensed CC BY-SA 4.0.

Nick Davies

Nick Davies (born 28 March 1953) is a British investigative journalist, writer, and documentary maker. Davies has written extensively as a freelancer, as well as for The Guardian and The Observer, and been named Reporter of the Year, Journalist of the Year and Feature Writer of the Year at the British Press Awards.

Summary adapted from the Wikipedia article on Nick Davies, licensed CC BY-SA 4.0.

Heather Brooke

Heather Rose Brooke (born 1970) is a British-American journalist and freedom of information campaigner.

Summary adapted from the Wikipedia article on Heather Brooke, licensed CC BY-SA 4.0.

Paul Foot (journalist)

Paul Mackintosh Foot (8 November 1937 – 18 July 2004) was a British investigative journalist, political campaigner, author, and long-term member of the Socialist Workers Party (SWP).

Summary adapted from the Wikipedia article on Paul Foot (journalist), licensed CC BY-SA 4.0.

Related guides

Primary sources

Frequently asked questions

What is the Science Media Centre and how do its briefings work?
The Science Media Centre is an independent press office for science, working to ensure the public have access to the best evidence and expertise through the news media when science hits the headlines. It runs press briefings at which researchers present new work — often under embargo — and take questions from journalists, and it publishes roundups and rapid reactions: short comments from independent scientists on new papers and breaking stories. It is a valuable route to expert comment at speed, but treat its quotes as a starting point; the strongest stories still come from reading the paper and doing your own interviews.
What is the difference between peer review and a preprint?
Peer review is the process by which a journal asks independent researchers to scrutinise a manuscript — its methods, analysis and conclusions — before deciding whether to publish it. A preprint is a full draft posted openly on a server such as bioRxiv, medRxiv or arXiv before, or without, that scrutiny. Preprints became a routine part of UK news during the Covid-19 pandemic because they surface findings months earlier. Neither status settles whether a finding is true: peer review is a filter, not a guarantee, and peer-reviewed papers are sometimes retracted. If you report a preprint, say clearly that it has not yet been peer reviewed and seek independent expert comment on it.
What is the difference between relative and absolute risk?
Relative risk describes a proportional change; absolute risk describes how often something actually happens. If a condition affects 1 person in 10,000 and an exposure doubles the risk, that is a 100 per cent relative increase — but the absolute risk has moved from 1 to 2 in 10,000. Headlines built on relative figures alone make small effects sound alarming, which is why statistical guidance for journalists consistently recommends reporting the absolute numbers alongside, ideally as expected frequencies such as "2 in every 10,000 people". If a press release gives you only the relative figure, ask for the absolute one; its absence is itself informative.
Can I report claims that a named researcher committed fraud?
Only with great care. An allegation of fabricating or manipulating research is seriously defamatory of a named individual, so anchor reporting in what has formally been established: a retraction notice, the upheld findings of an institutional investigation, or a regulator or tribunal ruling. Be precise about the difference between error and misconduct — many retractions correct honest mistakes, and the notice usually states the reason. Put allegations to the researcher, carry their response, and take legal advice before publication. Reporting that a paper has been retracted, and what the notice says, is generally safe; asserting fraud that no investigation has found is not.
How should I report a single new study responsibly?
Establish what the study is before deciding what it means: was it in humans, animals or cells; was it a randomised trial or an observational study; how many participants did it have; and does the paper itself support the claim in the press release? Check the funding and competing-interest declarations that journals require authors to make. Then place it in context — a single study rarely settles a question, systematic reviews of the whole evidence base carry more weight, and correlational findings do not establish cause. Give absolute risks alongside relative ones, and make sure the headline does not outrun the data.

Related guides

Primary sources