One of the longstanding merits of scientific disciplines is that when a mistake is made or a study yields a fluke result, another researcher can easily correct the mistake. Because the scientific method emphasizes replicability, the scientific literature can self-correct. Scientists are trained to be skeptical of results that have not been replicated by other studies and to question findings not consistent with existing literature. Prior to publication, a paper must make it past scrupulous editorial boards and peer reviewers in order to make it into any major journal.
To the credit of this rigorous bureaucracy, and to the strict standards of the scientific method, the number of retracted papers has historically been very low. At the beginning of the 2000s, there were only about 30 retraction notices on an annual basis. But as the decade continued, this number began to grow drastically: in 2011, the online database Web of Science had more than 400 retraction notices, which is an increase disproportionate to the additional number of papers published over that time span.
As this trend became apparent, the scientific community applauded the increased number of retractions. Many believed the scientific community was getting better at weeding out fraudulent papers — that the self-correcting mechanisms of the scientific community were hard at work debunking bad results.
An article published last year in the popular scientific journal Nature first called mainstream attention to the increase in retractions, casting doubt on whether the trend of increasing retractions should be celebrated or not. He viewed the root cause to be not a better system for seeking out fraudulent papers, but a culture shift where scientists were more willing to submit data for publication that may be erroneous or even false.
The author especially focuses on the difficulties of discussing retractions within the scientific community. A retraction not only removes a paper from the existing literature, but is a huge embarrassment to the paper’s author and to the journal that published it. With scientists and editors often unwilling to discuss why retracted papers get written and published in the first place, it becomes much harder to understand why the number of retracted papers are on the rise — something that has continually challenged scientific ethicists who study this trend.
Unwillingness to discuss retractions also makes it difficult to discern whether papers are incorrect due to “honest mistakes” or whether intentional fraud has taken place. While many critics have been optimistic that honest mistakes are largely to blame, a recent study of publications in the biomedical and life sciences showed something totally different. Two scientists reviewed a total of 2,047 papers published over the past 10 years. Their analysis was unique because it went beyond the papers themselves and instead examined the publication histories of the authors, looking at information on watchdog websites, the news media, and other public records. This study even led to the retractions of 118 papers to be reclassified from error to fraud.
The findings from this paper showed that, contrary to previous beliefs, suspected misconduct was the number one cause of paper retraction, accounting for over three-quarters of all cases. Of these, the most common reason for the misconduct by far was suspected fraud — the deliberate manipulation or falsification of figures and data. Other causes of misconduct included plagiarism and republication of previously published work.
To take a step back, instances of fraudulent and plagiarized papers are still very uncommon in the scientific community. We’re only talking about 400 papers being retracted annually for any reason, compared with the 27,000 per week that are published. And the problem is more widespread in some fields than others — the biomedical sciences have had a serious problem with fraudulent papers entering the literature, while no other field seems to have as persistent a problem with retracted papers.
However, just because the number of instances is small does not make the problem insignificant, or one that can be put off until later. Many critics of the way retractions are currently handled note that many databases will mark a paper as retracted but do not note the reason for the retraction. Some will not even specify if there have been claims of fraud or plagiarism, or whether it was just an honest statistical mistake that was made.
Even when a paper is retracted, it is not gone from printed copies or large, online databases. Websites like Web of Knowledge will “red flag” a retracted paper so that users are aware of the demerit, but previous studies of retracted papers have shown that it can be a slow process for other scientists to stop using retracted studies as citations for their own papers, and in some cases, the retracted paper can continue to be cited frequently even after it has been flagged for containing a serious issue. Although it is supposedly taken out of the literature, bad research can continue to have an influence even after there has been a retraction.
And of course, there are those few scientists who stand out for having committed fraud on so many occasions that they have built their entire careers around it. The most infamous of these is the anesthesiologist Yoshitaka Fujii, who was found earlier this year to have fabricated data in at least 172 scientific papers. All these papers were eventually retracted on the grounds of fraud.
The rise in bad scientific practices is likely the effects of a number of factors. Certainly a change in the scientific culture cannot be ignored. Scientists are under more pressure than ever to publish and publish often. There is a widespread expectation that scientists will do as much as possible to bring in grant money for their institution, especially as universities and research facilities invest more money in new research buildings and equipment.
The increase in publishing is also overwhelming for scientific journals, as they have more and more material to evaluate. Not only are scientists with long histories of scientific research publishing more, but a new wave of researchers in countries around the world create even more of a workload. The governments of countries such as China, South Korea and Turkey even offer cash rewards to scientists who get published in big-name journals, increasing the volume of submissions sent to these journals, even though only a few are published.
Online databases themselves are teeming with information and are still learning how to manage thousands of new publications and citations each week. Keeping content updated is quite a task for even the biggest of online databases, as it entails keeping tabs on papers with errors that have been corrected, papers that have been retracted and papers that cite retracted papers.
It will take a concerted effort by scientists and their publishers to remove the temptations of publishing fraudulent work before this problem becomes too severe. It may require not only a change in the culture of research in the sciences, but also a change in the demands of institutions and governments that currently reward scientists who publish often and punish those who put more time into their papers before sending them off to a publisher. Yet, all in the scientific community will feel the problems associated with fraud if it is allowed to increase unchecked. Scientists should feel urgency to begin addressing issues of fraud head-on and to reverse this trend before it permeates any more into the scientific community.