Science Correspondent Toolkit
If you are a science or health correspondent working in UK journalism — print, digital or broadcast — this toolkit covers the statistical literacy, accuracy standards and regulatory knowledge that should underpin every piece. It is built around the practical demands of science reporting, not general newsroom practice.
Last reviewed: Next review due:
Start here
Science correspondents face a specific set of accuracy risks that differ from those facing general reporters: confusing correlation with causation, overstating the significance of a single study, misrepresenting what peer review means, and reporting health claims without the clinical and regulatory context that UK readers need. Understanding these risks and having systematic processes for catching them is more important than any individual piece of specialist knowledge.
The Data Journalism hub and Ethics hub are your two most important reference hubs for science reporting. The guides and tools below are the ones science correspondents reach for most often.
Core guides for you
Statistics for JournalistsP-values, confidence intervals, relative versus absolute risk and sample size — explained without maths.Regression and Correlation BasicsHow to read regression tables, interpret correlation coefficients and spot spurious associations.Accuracy: Systematic ApproachesStep-by-step fact-checking workflows for complex science and health claims before publication.Data Protection Journalism ExemptionHow the UK GDPR Schedule 2 exemption applies when handling patient data or research records.Health and NHS ReportingNICE guidance, MHRA approvals, NHS data sources and how to interpret clinical evidence for UK readers.UK Statistical BodiesONS, NHS Digital, PHE and UKSA — what each publishes and how to access and cite the data correctly.The Public Interest TestHow to apply the public interest test when reporting sensitive health or scientific data.Anonymous SourcesWhen whistleblower scientists or researchers can contribute anonymously and how to verify their claims.
Recommended tools
Tools you'll use weekly
Source verification and headline accuracy checks for science and health reporting.
Blog posts you should read
Science Communication for UK JournalistsHow to communicate complex scientific findings accurately to a general UK audience.Fact-Checking at Scale UKSystematic fact-checking workflows for science and data-heavy investigations.Data Journalism for Beginners UKA practical introduction to working with data sources relevant to UK science reporting.
Templates that save you time
FAQs for science correspondents
How should a science correspondent interpret a confidence interval?
A 95% confidence interval means that if the study were repeated many times, 95% of the resulting intervals would contain the true population parameter — it does not mean there is a 95% probability that the true value lies within this particular interval. In practical reporting terms, report the central estimate and the range, note whether the interval is wide (suggesting imprecision) or whether it crosses zero or one (suggesting the effect may not exist), and avoid treating any single estimate as a definitive finding. The Statistics for Journalists guide on this site explains confidence intervals and p-values without requiring a maths background.
What is the difference between correlation and causation, and why does it matter for science reporting?
Correlation describes an association between two variables; causation means one variable produces the other. Observational studies (surveys, cohort studies, case-control studies) can demonstrate correlation but cannot by themselves prove causation. Randomised controlled trials (RCTs) are the gold standard for establishing causation because randomisation removes the effect of confounding variables. When reporting an observational study, use language that reflects the design: “associated with” rather than “causes” or “leads to.” Peer reviewers and statisticians at the Science Media Centre are available to check your interpretation before publication.
What are Science Media Centre embargo rules and what happens if I break them?
The Science Media Centre issues press releases and expert reaction under embargo ahead of journal publication. Embargo times are set by the journal and communicated clearly on each release. Breaking an embargo — publishing ahead of the specified time — can result in the SMC removing your organisation from its distribution list and, in serious cases, the journal withdrawing your accreditation. In practice, embargo breaks also damage your relationship with research institutions and other journalists who relied on the embargo. The SMC's embargo policy permits you to prepare copy and speak to additional experts before the embargo lifts, provided nothing is published until the specified time.
How do I report health claims accurately without overstating risk?
Translate relative risk into absolute risk whenever possible. A drug that “halves the risk of a condition” sounds dramatic; if the baseline risk is 2%, halving it means the risk drops from 2% to 1% — an absolute reduction of one percentage point. Report both figures. Seek NICE guidance or BMJ best-practice summaries for the clinical context, check the intervention is approved or under review in the UK (MHRA and NHS data differ from FDA approvals), and note the quality of evidence (RCT, observational, case study). For cancer and cardiovascular claims, the NHS Behind the Headlines page is a useful sanity check on existing coverage.
What does the data-protection journalism exemption mean for science reporting?
The UK GDPR Schedule 2 paragraph 26 (Data Protection Act 2018) exempts from certain UK GDPR provisions the processing of personal data for journalistic purposes, where the data controller reasonably believes that publication is in the public interest and that compliance with the provision would be incompatible with that purpose. In science journalism, this most often applies when a journalist obtains identifiable patient data or research participant records. The exemption is not a blanket pass: you must be able to demonstrate the public interest and must not use more personal data than necessary for the journalistic purpose. The data protection journalism exemption guide covers the full conditions.
How should I handle a pre-print study that has not been peer reviewed?
Pre-prints (on bioRxiv, medRxiv, SSRN and similar repositories) have not been subject to formal peer review and should be reported as such. Use language that makes the status clear: “preliminary research that has not yet been peer reviewed” or “a pre-print study.” Seek independent expert comment from at least two scientists not involved in the study. Do not treat a pre-print finding as equivalent to a published RCT. Be aware that pre-prints are sometimes released strategically to influence policy or public debate before the peer-review process might moderate the claims.
What is the best process for fact-checking a complex scientific claim?
The most reliable process for fact-checking a scientific claim in journalism involves: reading the original paper (not just the press release or abstract); checking the methodology section for sample size, study design and conflicts of interest; seeking two independent expert comments from scientists with relevant expertise who are not authors of the paper; consulting the Science Media Centre's expert reaction database for existing comment; and checking whether the findings are consistent with the existing peer-reviewed literature or represent an outlier. The Systematic Accuracy guide on this site provides a step-by-step framework for complex fact-checking tasks.
Common pitfalls for science correspondents
- 1Confusing correlation with causation in health reporting. Observational studies show associations, not causes. Reporting that a dietary factor “causes” cancer based on a cohort study is inaccurate. Use “linked to” or “associated with” for observational data, and explain why the study design matters to readers.
- 2Breaking a Science Media Centre embargo. Embargo breaks result in removal from the SMC distribution list and damage to your relationship with research institutions. Check the embargo time and your time zone before filing, and do not post on social media until the embargo lifts — a social post is treated as publication.
- 3Misrepresenting what peer review means. Peer review indicates that independent scientists considered the methodology sound enough to publish; it does not guarantee the findings are correct or replicable. A peer-reviewed paper is the beginning of the scientific process, not the end. Where a finding is significant, seek independent expert comment rather than treating peer review as a quality seal.
- 4Health claims without NICE or BMJ context. A drug or treatment finding from a study published anywhere in the world may not be approved, recommended or available in the UK. Always check NICE guidance and MHRA approval status before reporting health interventions as available to UK patients. FDA approval or European EMA approval does not automatically mean UK availability.
Where to next
The Data Journalism hub covers statistics and data analysis in depth. For accuracy and ethics frameworks, the Ethics hub is your primary reference. Beat-specific guidance lives in the Health and NHS Reporting guide.
Go to Data Journalism hub →Primary sources
- National Union of Journalists— NUJ
- National Council for the Training of Journalists— NCTJ
- Reuters Institute for the Study of Journalism— University of Oxford
- Society of Editors— Society of Editors
- IPSO Editors' Code of Practice— IPSO
- Press Gazette— Press Gazette