A Tale of Two Retractions – It’s Complicated
Some mistakes are hard to correct. Sometimes, it’s even hard to figure out where the mistake lies. Thus, retractions can be quite messy when a problem arises with a scientific publication. Two recent examples illustrate just how complicated retractions can become.
Do Women Need Male Mentors?
Back in November, a paper in Nature Communications started quite a stir with a study on gender and mentors. They found an association. Women with male mentors tended to publish papers with higher impact. But the problem came in the paper’s conclusions. A male mentor might “increase the impact of women” in science, wrote the authors. Then came a swift and furious reaction:
The recent @NatureComms paper doesn’t tell us much about the impact of gender on mentorship but it sure does tell us that the statistics community needs to do a better job teaching scientists about correlation, causation, and confounding
— Dr. Daniela Witten (@daniela_witten) November 19, 2020
Cause and effect seems hard to handle. Finding a correlation does not prove it, but is sure does feed bias. Not only that, it also feeds the fury of people who are the target of that bias. So naturally, the uproar about this clumsy assertion of causality led to an investigation and a retraction in little more than a month.
The authors all agreed to retract the paper, though they insist that they had some valid findings. So naturally, there’s grumbling about subjective aspects of such a decision.
Nonetheless, the point here is quite important. Unwarranted conclusions about causality is not just a problem for questions about mentors and gender. For example, we recently wrote about two other papers that take associations with obesity too far – one on intelligence and one on honesty. Neither journal has yet corrected either problem. The wheels grind slowly.
Legal Threats
A paper about fatal liver failure associated with a Herbalife product presents a very different case. Cyriak Abby Philips published this paper in the Journal of Clinical and Experimental Hepatology. It is a case report of a patient death due to liver failure after taking a weight loss product. The paper included strong language that goes beyond a typical case report by criticizing the company harshly. So of course, the company did not like it and demanded a retraction. After intense discussions between the study authors, representatives of the company, and the journal, a retraction by the journal was the result.
At first, the retraction notice cited “legal reasons” for the retraction, notes Retraction Watch. But now the journal asserts that the methods, analysis, and interpretation of the results are inadequate. The company seems to agree, but the authors do not. So now those authors have threatened legal action of their own for defamation.
In short, this has been a mess.
Subjectivity in Publications and Retractions
One observation ties these very different situations together. Both decisions to publish and to retract a paper require subjective judgments. Would the liver failure paper have become a retraction if it simply reported the facts of a patient death? Would the authors of the mentorship paper have needed to retract their paper if they had not leapt to an inference of causality?
Restraint is often necessary to ensure scientific rigor. It can also save everyone from a lot of bother.
Click here for the paper on mentorship, here for the retraction notice, and here for the journal’s editorial about it. For details on the liver failure case report retraction, click here.
Subscribe by email to follow the accumulating evidence and observations that shape our view of health, obesity, and policy.
December 26, 2020
December 26, 2020 at 4:35 pm, Chester Draws said:
The issue with the mentoring paper wasn’t an unwarranted leap of causality. It was that their findings were unsuitable.
If they had found the reverse, that female mentors were more effective, then the paper would have had no problems.
December 27, 2020 at 4:42 am, Ted said:
Chester, you have a point. I often see papers published that make an unwarranted inference of causality and when people like the finding, they don’t object. But either way, it’s a serious mistake.
January 01, 2021 at 9:41 am, Christopher Lynch said:
Nice job on this. I really like the ending…”Restraint is often necessary to ensure scientific rigor.” Classic!