AMSTERDAM—In 2012 Richard Mann, a mathematician at the University of Leeds in the United Kingdom, received some very bad news from a friend and colleague. Because of a coding error, the friend explained, Mann had included only 1/100 of his data in a modeling paper on the collective motion of glass prawns, published earlier that year in PLOS Computational Biology. As a result, the paper was deeply flawed.
Mann wanted to set the record straight, but as he began researching his options, despair set in. Retractions are strongly associated with research misconduct. "I became worried about public shaming," Mann said last week at the fifth World Conference on Research Integrity here. He went ahead, but only after many sleepless nights.
His story and others like it have inspired two recent attempts to develop new terms for retractions that would make it easier for researchers, universities, and journals to admit errors. One would retire the dreaded r-word altogether. "You have to change the language," says Nicholas Steneck, who heads the Research Ethics and Integrity Program of the Michigan Institute for Clinical and Health Research in Ann Arbor.
Stigma is one reason to rethink the retraction system; another is expedience. Universities and journals are often slow to retract a paper, waiting for the outcome of lengthy investigations. Journals sometimes add an "editorial expression of concern" to a paper in the meantime. But such notes can be stigmatizing, too, and if it's clear the data are wrong, some argue it's better to pull a paper and report the causes later.
Some journals want more options for a troubled paper than either a correction for minor errors or a wholesale retraction. The EMBO Journal, headquartered in Heidelberg, Germany, has introduced the "partial retraction" for cases when, say, one figure is erroneous but the conclusions of a paper stand. "With a full retraction, you take a down a whole chunk from the scientific literature," says Editor-in-Chief Bernd Pulverer. JAMA Psychiatry in 2015 introduced the concept of "retracting and replacing," for a paper about a clinical trial that had pervasive errors but, once corrected, was still worth publishing. Editors at The Lancet and The Lancet Respiratory Medicine have instituted "retraction and republication."
But with a sharp rise in retractions over the past 15 years—to 664 in fiscal year 2016, according to the database MEDLINE—some feel a more comprehensive approach is in order. At a workshop at Stanford University's Meta-Research Innovation Center (METRICS) in Palo Alto, California, last December, a group of journal editors and other experts devised a more granular system for both corrections and retractions that has 14 solutions for different situations. Retractions would remain for misconduct, says Pulverer, who attended the workshop. But five other terms would cover papers withdrawn for other reasons, METRICS researcher Daniele Fanelli told the meeting last week (see list, below). For example, honest mistakes like Mann's would now come under the moniker "self-retraction." (An article describing the taxonomy is under review, and Fanelli stresses that it's likely to change.)
But the complexity of Fanelli's plan could create confusion, says Virginia Barbour of the Queensland University of Technology in Brisbane, Australia. Barbour is a member of a working group of the Committee on Publication Ethics (COPE) that posted a simpler plan on bioRxiv in March that would simply retire the word "retraction." Instead, the group introduced neutral-sounding "amendments," which would come in three categories—insubstantial, substantial, or wholesale—covering anything from a typographical error to made-up data. Amendments could be made even before misconduct investigations finish.
"That proposal goes a bit too far, in my view," says Pulverer, who thinks the word retraction sends a clear signal in cases of misconduct. Leonid Schneider, a former stem cell scientist who now blogs about scientific integrity from Erlensee, Germany, castigated the COPE group for making life easier for editors, who would no longer have to worry whether a paper involved misconduct or not. "Too many journal editors would sure prefer to hang some amendment note [on] a paper and be done with it." Lowering the bar for authors to "amend" papers could also allow fraudsters to clean up before they get caught, says Ana Marušić of the University of Split School of Medicine in Croatia, one of three editors-in-chief at the Journal of Global Health. "You can imagine a situation where, before I am officially accused of misconduct, I will make a correction," she says.
But a journal's primary responsibility is keeping the literature accurate—not finding or punishing those who engage in misconduct, says Elizabeth Moylan, a senior editor for scientific integrity at BioMed Central in London and a member of the COPE working group. "That's for institutions," Moylan says.
Pulverer recommends that journals experiment with the various categories in the two proposals. "Hopefully a consensus will crystallize," he says. In the meantime, the retraction will stay. For those who do have to retract a paper because of an honest mistake, a few studies offer encouragement: They suggest that such retractions aren't so bad for one's career. That was certainly Mann's experience: His supervisor was supportive, colleagues complimented him on owning his error, and after shelling out another $2500 in article processing fees, he published a new paper with the same title in the same journal. "The community didn't scorn me," he says.