Exaggerated science news usually starts before the newsroom, in the university press office. In a 2014 study in The BMJ, Petroc Sumner and colleagues at Cardiff University analyzed 462 press releases. The releases came from 20 leading UK universities, and exaggerated advice appeared in 40% of them. The design is retrospective and observational, so it maps patterns without proving that releases cause bad coverage.
The same skepticism pays off outside medicine, and it applies to fitness headlines too, as this guide to the new ACSM weight guidance shows. The numbers below explain why the habit matters. A headline is a claim about a paper, and the paper is always checkable.
Those numbers are the backbone of how to read science headlines: two Cardiff analyses, one of university releases and one of journal releases. Both are content analyses, which count and code text at scale. Regular coverage of new findings sits in the science news section. Both studies name the exact gaps they tracked.
Where Does the Exaggeration Begin?
Mostly upstream, in the press release. The Cardiff team ran a retrospective content analysis, comparing each peer reviewed paper with its release and its news coverage across 462 releases and 668 news stories from 2011. Exaggerated advice turned up in 40% of releases, and exaggerated causal claims drawn from correlational research in 33%.
A third gap involved animals. In 36% of releases, findings from animal research were restated with more confidence about humans than the paper itself carried. Each figure carried a wide confidence interval, and none of the three categories crossed 46%. The team also found little sign that exaggeration bought more coverage, a point that returns in the follow-up.
How Much Reaches the Headlines?
When a release exaggerated, the news usually followed. Coverage repeated the exaggerated advice in 58% of stories, and the causal exaggerations in 81%. Animal-to-human leaps passed through in 86% of cases. Without an exaggerated release, pass-through rates sat between 10% and 18%. For related coverage, see Science News for Students: How Newsrooms Make Papers Kid-Readable.
A 2016 follow-up in PLOS ONE by the same lead author extended the test to 534 press releases from leading journals. Exaggeration in a release still predicted exaggeration in the news, with odds ratios of 2.4 and 10.9, yet it brought no extra coverage. Both odds ratios carried wide intervals. Pooling both datasets reached 996 press releases and 1,250 news stories. Caveats behaved the opposite way: releases that carried caveats produced news that carried them too.
What Should a Reader Check First?
The documented failure points make a short checklist, and every item maps to a measured gap rather than to taste.
- Sort the claim: causal wording on a correlational study was the most common tracked gap.
- Watch the species: animal findings restated as human advice formed the second gap.
- Look for caveats, since the 2016 analysis found caveats in releases reach the news.
- Compare headline numbers with the numbers in the paper itself.
The checklist rests on correlational designs, so its limits are worth naming. An odds ratio describes association rather than cause. The university sample is UK releases from 2011, and press offices are not the only source of distortion. The 2016 finding that exaggeration bought no extra coverage also suggests the incentives are mixed.
Conclusion: Read the Release, Then the Paper
The evidence says distortion usually enters early, in the release, and then propagates into news. That makes the reader’s move simple. Find the paper, check the design, and match every causal verb to it. Both Cardiff studies are observational counts of wording, strong enough to map the pattern but not to promise that cleaner releases would end misleading news.
The habit still pays. The paper holds the absolute numbers, the sample, and the caveats that the headline left out. Checking it takes minutes.
