An editorial written Tuesday by Virginia Barbour and Kasturi Haldar at PLoS regarding a specific retraction and a general stance relative to retractions generated a great deal of discussion this week. The sentence that spurred the controversy was:
If a paper’s major conclusions are shown to be wrong we will retract the paper.
Retraction Watch was quick to cover the posting, of course, and a good deal of the discussion took place on that blog, including what nowadays we nicely call a “walk back” by Barbour yesterday afternoon.
Comments within the controversy followed a common thread — science is necessarily a hunt-and-peck adventure, with many dead ends; those dead ends may often fall into the category of “wrong conclusions,” and would by this new logic require retraction; retracting findings that are just misinterpreted or currently viewed as uninteresting doesn’t seem helpful, appropriate, or scientific; barring clear problems in the underlying data, this seems likely to create big holes in the literature for no good reason; therefore, this new policy creates some real concerns about PLoS’ retraction practices going forward.
Having read the editorial and then Barbour’s clarification, I felt myself going into Jon Stewart mode, partly because her retreat from the editorial begins with a little bit of an accusation of reader error:
The phrase “If a paper’s major conclusions are shown to be wrong we will retract the paper” has been pulled out of the blog and over/misinterpreted.
This immediately struck me as disingenuous. If the catalyzing sentence Barbour notes were a toss-off line in the editorial, and not the punch line to a long and involved argument, Barbour might have a point. But here is the entire context of the statement she accuses readers of various stripes of “over/misinterpret[ing]”:
At PLOS our mission is to accelerate progress in science and medicine by leading a transformation in research communication. We firmly believe that acceleration also requires being open about correcting the literature as needed so that research can be built on a solid foundation. Hence as editors and as a publisher we encourage the publication of studies that replicate or refute work we have previously published. We work with authors (through communication with the corresponding author) to publish corrections if we find parts of articles to be inaccurate. If a paper’s major conclusions are shown to be wrong we will retract the paper. By doing so, and by being open about our motives, we hope to clarify once and for all that there is no shame in correcting the literature.
Barbour and her co-author are, in essence, arguing that because PLoS wants to accelerate the publication of research findings, they’ve effectively raised the stakes, making retraction more necessary and something that, in this new environment, we shouldn’t be ashamed of any longer. The logic is bolstered by strong writing, declarative sentences, and punchy punctuation. Did we overinterpret? Or did they overwrite?
Within 48 hours, Barbour was apparently so ashamed of this line of argument that she retreated from it in a significant way. In her note on Retraction Watch, she even refers back to COPE guidelines to illustrate what she portrays as their “intention,” which was apparently to comport more with the status quo notion as expressed in the COPE guideline she quotes:
Journal editors should consider retracting a publication if: they have clear evidence that the findings are unreliable, either as a result of misconduct (e.g. data fabrication) or honest error (e.g. miscalculation or experimental error).
Barbour is right to regroup. There is a wide gulf between “wrong conclusions” and “unreliable findings.” The former is mostly a logical error — we once thought that ulcers were caused by stress, because we believed that the highly acidic stomach environment prevented bacterial growth. This was a “wrong conclusion,” not an unreliable finding. The stomach lining is highly acidic. We just hadn’t discovered H. pylori in the ulcers because we hadn’t thought that bacteria might survive, which kept us from applying Koch’s postulate. Once applied, our view changed, but none of the papers measuring stomach acidity should have been retracted just because this new finding revealed some of them reached a “wrong conclusion.” Unreliable findings are, as the statement notes, the result of data errors, not logical errors of interpretation. As one commenter on the PLoS editorial wrote:
No, no, no, no, no. This is a terrible policy. To use Feynman’s term, this is “cargo-cult” editing. Going through the motions of checking the scientific literature but without the intellectual integrity to acknowledge the limits of what is known. The conclusions of a scientific paper are the least important part of it. There is no reason to retract papers with wrong conclusions. It would be a bad thing to retract all papers with wrong conclusions. I come across many papers with wrong conclusions all the time. I often don’t even look at the conclusions, just at the data and then make my own conclusions.
In any event, perhaps the most heartening aspect of this hiccup is that the scientific community showed that when it comes to “wrong conclusions” about how to apply retractions, it can quickly generate enough concern for a de facto retraction to occur. And that seems profoundly appropriate.