The study, published yesterday in the Proceedings of the National Academy of Sciences, concludes that just over 67 per cent of retractions in recent decades are attributable to misconduct or suspected misconduct. This amounts to three quarters of retractions whose cause is known.
More than 43 per cent of all retractions were attributable to fraud or suspected fraud, while duplicate publication accounted for 14 per cent and plagiarism for 10 per cent.
Only 21 per cent of retractions were attributable to error.
The study was carried out by Arturo Casadevell, a professor at the Albert Einstein College of Medicine, New York; Ferric Fang, professor of microbiology at the University of Washington, and medical communications consultant Grant Steen. The former two authors are both journal editors.
The study is based on more than 2,000 papers listed in the PubMed database as having been retracted since 1977.
Rather than relying on the retraction notices, the authors studied reports by the US Office of Research Integrity, as well as press articles and the Retraction Watch blog. They found that a significant proportion of retractions whose stated cause was scientific error were actually the result of misconduct.
“Incomplete, uninformative or misleading retraction announcements have led to a previous underestimation of the role of fraud in the ongoing retraction epidemic,” they say.
They also found that the proportion of articles retracted because of fraud has increased tenfold since 1975, with a particularly “dramatic rise” occurring in the last decade.
They found that most of the papers retracted for fraud were published in higher impact factor journals and originated from the US, Germany and Japan. Most of the papers retracted for plagiarism and duplicate publication came from the US, India and China, and were published in lower impact factor journals.
“This finding may reflect the greater scrutiny accorded to articles in high-impact journals and the greater uncertainty associated with cutting-edge research. Alternatively, the disproportionately high payoffs to scientists for publication in prestigious venues can be an incentive to perform work with excessive haste or to engage in unethical practices,” the paper says.
The highest number of retractions - 70 - appeared in Science, with other high-impact journals such as Nature, Cell and the New England Journal of Medicine all figuring in the top ten. However, Science was only third for retractions due to fraud or suspected fraud: higher numbers were recorded by the Journal of Biological Chemistry and Anesthesia & Analgesia.
The authors say that although only around 1 in 28,000 of the 25 million articles referenced in PubMed have been retracted for fraud, only a “fraction” of fraudulent articles are ever retracted.