The Bad Retraction

Using retractions as censorship undermines scientific integrity.

Posted Jan 11, 2021

Just before Christmas, the prestigious journal Nature Communications retracted an article that examined the informal mentorship of junior scientists. Among other findings, the paper concluded junior female scientists benefitted from male mentors more than they did female mentors. This set off a firestorm of moral indignation on social media, putting pressure on the journal which then retracted the article following a second (and very unusual) round of peer-review. 

Is such a retraction warranted? Or are we seeing retraction increasingly being used by journals as censorship of unpopular conclusions in the “cancel culture” age? Wired magazine recently documented that retractions of controversial science seem to be on the increase. The Wired article made an unironic comparison to The Purge movie, acknowledging that politically charged papers are judged differently than those that are not. 

Obviously, any research paper with fatal flaws should be retracted. But this shouldn’t be based on popularity or Twitter mobs. Indeed, the argument that controversial papers should get more scrutiny is an odd one, scientifically. Such an argument is a recipe for disturbing scientific inquiry, which obviously distorts the scientific record: you better find the right things with your data or else you’ll be canceled. Further, papers that support popular moral narratives are arguably as likely to do harm as those that contradict them. Bad science leads to bad policy, bad medical decisions, bad use of grant funding, promotion of societal myths, and distracts from real solutions to real problems. That’s true whether the finding is popular or not.

Disputes about the validity of scientific articles are actually very common and a normal part of science. Retraction is mainly for outright errors or fraud. In an editorial on the Nature Communications controversy, the journal says the problem seemed to hinge on whether the authors had truly tapped into “informal mentorship” using co-authorships between junior and senior scholars on research articles. That’s a fair critique but, to my mind, not a fatal one. Issues such as this are typically handled in a comment and response format, where critics of the article publish their critiques and the authors can respond. The problems with this article did not warrant a retraction and it’s also not clear that the investigation, following a massive social media storm, could have been fair to the authors. The retraction notice says the authors agreed to the retraction, though defended their work and conclusions; an odd combination that, to my mind, speaks to the enormous pressure they must have been under.

The editorial regarding the retraction is not very transparent. It fails to note whether any scholars contacted the journal arguing against retraction (it portrays concerns about the paper as a consensus, though an argument to consensus is itself a logical fallacy). The investigation peer-review is not made public. The investigation involving another round of peer reviews is itself unusual (in an email, the editor Dr. Elisa De Ranieri declined to disclose how many other articles received such investigations in the past two years). 

In the editorial, the journal states “As part of these guidelines, we recognise that it is essential to ensure that such studies are considered from multiple perspectives including from groups concerned by the findings. We believe that this will help us ensure that the review process takes into account the dimension of potential harm...” It’s not clear what this statement means. Do special interest groups or moral advocates get veto power over science? If a paper suggests that conservatives have lower cognitive abilities (and such papers do exist) should conservative scholars be specifically recruited to review such papers, to avoid stereotyping conservatives as idiots? Good luck with that given how few conservative academics there are.    

Seeking diversity in reviewers is important. But statements such as these raise the specter of censorship, not based on diversity, but far-left ideology. It’s worth noting that the lead author of the retracted paper was, herself, a woman of color. Scientific publishers have a particular obligation to shield science from the whims of the masses (even masses of academics). Increasingly journal editors are failing at this important task, Nature Communications being only one recent example. We’re now at a state where retraction may be used, not for computation errors or fraudulent data, but as an ideological weapon against any unpopular science.