Quality not quantity to tackle fraud

The pressure to secure grants can force scientists to publish work that offers little value – and can lead to misconduct

SCIENCE IS A human activity, and it is not immune to misconduct. The latest review of misconduct, by FC Fang, RG Steen and A Casadevall, has been published online this month in the Proceedings of the National Academy of Sciences. The review highlights an increase in misconduct by scientists, a worrying trend that calls for urgent reforms in the system to reduce the pressures that tempt some scientists to misbehave.

Fang and colleagues analysed the number and frequency of retraction of flawed publications from scientific journals in the biomedical and life-science fields. They used the PubMed database that references more than 25 million articles published since the 1940s. PubMed is maintained by the US Library of Medicine. They analysed the 2,047 articles indexed by PubMed as retracted on May 3rd, 2012. Retracted articles were classified under the headings fraud (data fabrication or falsification), suspected fraud, plagiarism, duplicate publication (publishing the same data in two or more journals), error, unknown, or other reasons.

Fraud or suspected fraud is the most common reason for retraction (43.4 per cent). Duplicate publication accounts for 14.2 per cent and plagiarism for 9.8 per cent. Where the reason for retraction is known, 75 per cent of articles are retracted for misconduct or suspected misconduct and only 25 per cent for error.

A marked recent increase in retraction frequency was noted, mainly due to misconduct. Retraction for fraud stands out like a sore thumb: the number under this heading, as a percentage of total articles published, has increased nearly 10-fold since 1975. The overall rate of retraction is still very low – about 0.08 per cent of papers are now retracted annually. America, Germany, Japan and China account for 75 per cent of retractions for fraud or suspected fraud.

Science is expensive and is mostly funded from the public purse. Scientific dishonesty could damage public trust in science, leading to diminished support from the public purse. Also, fraud slows down scientific progress by leading other researchers up false trails. Science is self-correcting and false trails are detected in time, but this correction can waste resources. Although scientific fraud can never be eliminated, it is critical to reduce it to the absolute minimum.

How is this to be done? Journal reviewers must become more proficient in sniffing out suspect data in papers submitted for publication. Scientists should receive more formal training in ethics; as far as I know, ethical training is offered to students only at postgraduate level: this should be extended to undergraduate training.

But the most serious cause of fraud is the pressure in the scientific-research system. A scientist’s career depends largely on the success of their research endeavours. Research must be funded, research grants are awarded by open competition, and money is available to fund only a relatively small fraction of the applicants. Grants are awarded largely on the basis of success in research to date, as evidenced by publications in scientific journals.

If you have no publications, you have no hope; the fewer your publications, the lower your chances of getting a grant. In all cases, career promotion is heavily dependent on success in winning research grants, and your job may depend on winning another grant to begin in six months’ time when your current grant runs out. So, the need to have a good number of publications is obvious, as is the temptation to misbehave if you are finding it difficult to maintain a high publication rate.

Steps could be taken to reduce stress in the system used to award grants. Researchers are perennially dogged by a shortage of research-funding opportunities. More stable and sustainable sources of funding should be developed. Flexible career pathways for scientists should be developed.

More attention should be paid to the quality of the papers previously published by grant applicants. The present system rewards numbers of publications out of proportion to the intrinsic value of the papers. Only a minority of scientific papers contribute anything of real value to the subject, but there is pressure to publish anyway, even if you have little to say. However, you can never eliminate pressure from the grant system and this pressure will always tempt some to cut corners.

William Reville is an emeritus professor of biochemistry and public awareness of science officer at UCC; understandingscience.ucc.ie