The public is often oblivious when popular, attention-grabbing research articles are retracted, an?analysis suggests, raising further doubts over the scientific system¡¯s ability to self-correct when errors are discovered.

The findings add to concerns that retractions are failing to cut through not only among scientists, but in social and traditional media, too.

After looking at about 3,000 retracted articles published between 2010 and 2015, a team from Stanford University ¨C including John Ioannidis, a prominent expert on the failings of the research system ¨C concluded that retracted articles received more media attention than a control group of un-retracted papers.

Their analysis, published on 12 May in , focuses on a particularly troubling subset of ¡°popular¡± retracted articles.

³ÉÈËVRÊÓƵ

These articles ¨C about one in 20 ¨C garnered substantially more social and traditional media attention than others and were most commonly published in the high-profile generalist journals Science and Nature.

Despite being retracted, these ¡°popular¡± articles drew far more attention than their retraction notices, even when researchers adjusted for the fact that the articles might have been viewed precisely because they were retracted. They were more likely than average to have been retracted because of problems with their results or data, as opposed to other reasons, such as peer-review issues.

³ÉÈËVRÊÓƵ

¡°Worryingly, popular articles receive additional attention upon retraction and this attention does not always reflect their retraction, but may perpetuate the original misconception,¡± the paper finds.

In one example, an article that claimed putting branding on school lunch boxes could induce children to eat more healthily received four ¡°non-negative¡± tweets after it was published.

But after it was retracted, the study¡¯s findings were further tweeted more than 700 times, perpetuating the idea that stickers on lunch boxes can get children to pick fruit over cookies.

¡°There may be some reluctance of publishers to publish a retraction notice, to publish clear and informative notices and to make all potential readers of a retracted article aware that this article has been retracted,¡± the authors write.

³ÉÈËVRÊÓƵ

Professor Ioannidis told ³ÉÈËVRÊÓƵ that ¡°some papers are mostly curiosities that are of interest only to a few specialists, while others have major implications for health or other major outcomes that matter¡±. ¡°Retraction of papers that have implications of this sort should be clearly visible and properly communicated,¡± he said.

david.matthews@timeshighereducation.com