The role of retractions in correcting the scientific literature
Last week PLOS Pathogens retracted an article published in 2006 that claimed to show the identification of a new gammaretrovirus XMRV, and its association with prostate cancer. This was a high profile paper , – highly cited and widely discussed in the scientific community and beyond as it was believed to be a cornerstone of early research of the link between XMRV and a number of diseases whose etiologies were unknown or poorly understood.
Over the years since it was published, however, an increasing body of literature has cast doubt on the nature of many of these associations as well as the early link to prostate cancer. This culminated in the publication of a paper in PLOS ONE by Charles Chiu from the University of San Francisco along with collaborators, including some of the original authors , which showed conclusively that the XMRV detected was an inadvertent laboratory contaminant of the tissue samples from the original study and has no etiological relationship with prostate cancer. However, the discovery of the virus and methodologies employed in the original study remained valid.
What is the role of retractions in situations like this? There is much misunderstanding about retractions. Authors and editors have been notoriously unwilling to use them, for the perceived shame that they bring upon authors, editors, and journals. Journalists regularly note the fact that retractions are increasing and ask whether the scientific literature is thus becoming less reliable. Websites such as Retraction Watch  list and dissect retractions – an extra exposure at what is already a difficult time for authors and editors. In addition there is much confusion about how to effect retractions practically. In an effort to bring some clarity to this issue in 2009 the Committee on Publication Ethics of which PLOS Pathogens is a member and one of us (VB) is currently Chair, issued guidelines on retractions, which explicitly state that retractions are appropriate when findings are unreliable, either as a result of misconduct (e.g. data fabrication) or honest error. 
Scientific discovery is accelerating. Science is more complex, there is more of it, and the methods of dissemination have fundamentally changed in the past 10 years into something that is much more open, via the web, and thus more scrutinized. But just as the pace of science accelerates, so inevitably will the discovery of errors in papers since more mistakes will happen and as the papers are scrutinized more errors will be found. The acceleration of science and the additional scrutiny are to be welcomed, but in order to ensure that errors (from whatever means – unintentional or intentional) are not simply incorporated uncritically into the scientific literature at an accelerated rate themselves, so the mechanisms and the attitudes concerning corrections and retractions will need to be rethought.
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. Despite the best of efforts, errors occur and their timely and effective remedy should be considered the mark of responsible authors, editors and publishers. We welcome further discussion of this important topic.
Virginia Barbour, Medicine Editorial Director, PLOS
Kasturi Haldar, Editor-In-Chief, PLOS Pathogens
1] Urisman A, Molinaro RJ, Fischer N, Plummer SJ, Casey G, et al. (2006) Identification of a Novel Gammaretrovirus in Prostate Tumors of Patients Homozygous for R462Q RNASEL Variant. PLoS Pathog 2(3): e25. doi:10.1371/journal.ppat.0020025
2] Stieler K, Schindler S, Schlomm T, Hohn O, Bannert N, et al. (2011) No Detection of XMRV in Blood Samples and Tissue Sections from Prostate Cancer Patients in Northern Europe. PLoS ONE 6(10): e25592. doi:10.1371/journal.pone.0025592
I think you are in danger of making a fundamental mistake here. The guidelines call for retraction in the case of fraud/misconduct or honest error that has lead to an unreliable result – and I think the contamination issue for the XMRV paper might count (though this needs to weighed against the benefit to the community of the paper trial of this story – but that is another issue).
But there is a huge leap to stating that papers should be retracted “if a paper’s major conclusions are shown to be wrong”. Should the first estimates of the mass of the electron be retracted because later estimates converged on a different number? Should papers that showed potentially significant relationships in small samples that didn’t hold up in larger trials be retracted – even if potentially that data was used as part of the meta-study? This cannot be justified.
The literature is a record of the progress of science – and that progress is not linear. By erasing by-ways and diversions that end up not contributing directly to the ‘final’ answer (whatever that is), you are imposing a view of science that is not true to itself and is perhaps sanitised beyond the point of usefulness.
If you erase from scientific history the merely mistaken (as opposed to the fraudulent or compromised) you are doing a disservice to the notion of science itself.
While I agree with the aim of correcting the scientific literature as rapidly as possible, I’m not sure retraction is the way to go.
If being wrong is sufficient to merit retraction, you should expect to retract at least half of the papers published in PLoS. If not then you should retract this one:
Ioannidis JPA (2005) Why Most Published Research Findings Are False. PLoS Med 2(8): e124. doi:10.1371/journal.pmed.0020124
Seriously, though, if retraction is the sanction for being proven wrong even for honest errors, you’ll discourage scientists from following up on their own research. You’ll also penalise researchers who try to be open about their methods and data.
In any case, science is all about being wrong. As a scientist, the best I can hope for is to be wrong in an interesting way. Progress comes through demonstrating how we were wrong before and, critically, understanding why.
I’m a big fan of the PLoS ethos, but rather than striking out papers that are found to be wrong, you’d be better served by flagging replications, failures to replicate, and critiques associated with each paper.
I hope you do not really mean to retract papers found to be incorrect. This is absolutely wrongheaded and detrimental to science. Retractions are for *fraud*. And even then the papers should not be disappeared. They should be marked clearly but not hidden.
I don’t see how the position of the editor is being supported here. If the concern is ‘the correction of the literature,’ then retraction is not needed. In the case highlighted here, the literature already has been corrected – through the later publication of papers that show the flaws in the original work. Anyone who works in this field has long been aware of the criticisms of the original paper, and all have been skeptical. Now, the verdict is in, and ‘the literature’ has moved on. A retraction now can only serve the interests of the journal – not the scientific community.
Retraction should be reserved for malfeasance or gross incompetence, and should be used to name and shame, not simply to correct a record that has already been corrected in the peer review literature.
I feel that there is a fundamental problem with PLoS ONE article processing by your editorial board. I am not sure how the members were chosen for the editorial board. it has been my experience that they do not handle manuscript properly – biased approach. Some even don’t have the credentials of being the honest editorial board members (PLoS One espoecially).
If a paper’s major conclusions are shown to be wrong we will retract the paper.
This is a terrible policy and I hope you will retract (!) it right away.
As you yourselves point out, there’s no shame in being wrong — that’s how science works! There is, however, enormous shame in e.g. fabricating data, so if you are going to start retracting articles for reasons other than fraud, the onus is on you to distinguish very clearly between “kinds” of retractions.
You seem to be arguing that “retraction is not a dirty word”, but that’s simply not the way it’s understood by the research community. If you want to flag a paper as having had its major conclusions overturned by later work, add a comment and link to the relevant literature — an erratum or an editorial. There is no need for the blunt instrument overkill of a retraction.
Regardless of the above, how are you going to decide that a paper’s major conclusions are wrong? What happens when new evidence shows up to indicate that the paper was right all along? Will you start retracting retractions? Of retracted retractions? Etc. The growth of experimental knowledge is a slow, messy process — it’s just not possible to put neat little “right” and “wrong” labels on it.
Finally, it simply beggars belief that you would take such drastic action without adequate consultation with the authors. The only thing that could possibly give you as editors that right is clear evidence of misconduct.
Tell me: as an author, why would I submit my work to a journal wherein I might at any time find it emblazoned with the dreaded scarlet R by unilateral editorial fiat?
[…] has just put out an editiorial titled “The role of retractions in correcting the scientific literature” where they make a number of suggestions around retracting papers once the consensus of […]
Some even don’t have the credentials of being the honest editorial board members
And those credentials are….?
This might be one of the least thought out and most damaging policies to be enacted in scientific publishing.
Porting some of the discussion from Google+ to here, I find that I have good company. If conclusions being wrong is sufficient cause for retraction, over half of the literature needs retraction, and using the retraction mechanism like this provides a HUGE disincentive for people to publish self-corrections (which works directly against the mission of PLOS as an organization, as well as a publisher.)
My G+ comment is below:
The major difference between my point of view and that of PLOS (and other publishers, too, I reckon) is where the decision to retract is extended to honest mistakes as well as deliberate deception. As a scientist and as a student of altmetrics, I know that the majority of the literature consists of honest, well-meaning but mistaken conclusions. The literature is not a solid foundation upon which to build, and by it’s very nature cannot be. There is no scientist I know which assumes that any result in the published literature is necessarily replicable, so in the absence of partial retraction, I probably would have used some other mechanism which doesn’t penalize self-reporting of mistakes so strongly.
So the main issue is how the extension of retraction to cover honest mistakes disincentivizes self-reporting of mistakes, and it’s unfair given that so much of the literature is mistakes that just haven’t been uncovered yet. The high-profile nature of the case is just why it got uncovered in this case.
I really did think PLOS had begun to move beyond thinking of themselves as the gatekeepers of truth.
Dreadful idea. Once something is said, then it shouldn’t be unsaid.
Extending papers to point forward to work that has happened in the future would be a better idea, and something that is possible outside the tree based publication industry. We should be seeing more of this.
Are you returning the publication fees to authors, following retraction?
Our blog yesterday has generated a huge amount of comment on the issue of retractions – an unexpected but great outcome if only because it raises the issue of retractions and their function. We’ve pulled out a few issues (of the many that were raised):
• 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. Obviously, we have no intention of attempting (even if we could) to change how science works –by replication, by building on one’s own and the work of others, by testing and refining theories. Anyone who knows PLOS would hopefully understand our intention here. Our intention is not that retractions are used indiscriminately but to make it clear (as has been said many times before by us and others) that retractions are a useful tool in correcting the literature and do have a place when a piece of research is so unreliable that readers need to be alerted. To refer back to the COPE guidelines:
“ 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)”
• Authors clearly hate the word retraction– no matter what the intention, we don’t seem to be able to get away from the idea that somehow there is shame in the name. Do we need another term? “Correction” is not enough and “Expression of Concern” should be used when there is a degree of uncertainty (again see the COPE guidelines). There is clearly a need for a term that marks a paper as not to be relied on. Maybe we need to come up with another name, but we can’t back away from the idea.
• Can editors act unilaterally? Yes, though obviously they should not if possible. Editors have a key part to play in correcting the literature and though ideally everyone will agree with a course of action, sometimes they won’t. The COPE guidelines say this: “Who should issue the retraction? “Articles may be retracted by their author(s) or by the journal editor… responsibility for the journal’s content rests with the editor s/he should always have the final decision about retracting material. Journal editors may retract publications (or issue expressions of concern) even if all or some of the authors refuse to retract the publication themselves.”
• The role of the “Corresponding author” as the major conduit for communication between authors and the journal about issues such as this is important and deserves discussion and clarification.
It’s fantastic that the issue of the reliability of the scientific record and the mechanisms for correcting it has raised such a passionate response – let’s continue the debate about that.
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. If the absence of wrong conclusions was essential for non-retraction, then large fractions of the literature should be retracted.
What is much worse than wrong conclusions, are wrong assumptions. Unfortunately wrong assumptions are much harder to find because many researchers and reviewers share those wrong assumptions and papers questioning or refuting those wrong assumptions get rejected unexamined during peer review. Those are the paradigms that Kuhn talked about. Everything is obvious in hindsight to those able and willing to question their implicit assumptions.
An example of a perniciously wrong assumption is the myth of homeostasis.
There is no such thing as homeostasis. If you want to retract papers because of wrong conclusions, papers that start out with the wrong assumption of homeostasis should be retracted first. Compel researchers who use the term homeostasis to justify it.
The most essential part of a scientific paper is the data. If the data are wrong, then the paper must be corrected or retracted. It matters not a whit if the conclusion is correct but is based on wrong data.
Data can be wrong for multiple reasons, transcribing error, arithmetic, contaminated reagents, unknown unknowns, use of the wrong value for the viscosity of air (which gave Millikan the wrong value for the charge on the electron). The correct response is to issue a correction. When the data is wrong because of fraud, then retraction is the only option.
Science is a process. The process is what is important. That means going from correct data, using valid logic to reach tentative conclusions (which might be overturned with more or better data and/or with more or better logic). The reason methods sections are so important, is not so the results can be replicated (but that is important), but so that the process of going from instrument readings of something to data is laid out in complete detail so that the process can be assessed for reliability.
If you can’t look at data and reach your own conclusions, then you should not be reading scientific papers. Scientific papers are communications from one scientific peer to another scientific peer. If you don’t understand the underlying background, then you cannot understand a scientific paper and have no business pretending that you do.
The whole point of publication is so that others can see the data, analyze the conclusions for themselves and build on it. A paper with correct data but a wrong conclusion is a perfect opportunity for post-publication peer review. Someone could read the paper and from the correct data reach a correct conclusion. That would be a good thing. If the paper has good data, there is no reason other researchers should have to replicate that data. Why would anyone think that is a good idea? More hoops for other researchers to jump through? There is not a shortage of things that research is needed on. Putting more barriers in the way will only slow down progress.
The problem isn’t that there are wrong conclusions in the scientific literature, the problem is that we don’t know which conclusions are right because there are many implicit assumptions that are wrong. There are too many implicit assumptions to test them all, or even to list them.
Stop passing the buck to the COPE guidelines and take responsibility for your actions. If your duty is to the scientific community, then answer to the scientific community, not to the community of “editors and publishers of peer-reviewed journals”.
You are demanding that the scientific community uncritically accept the judgement of editors and publishers. Why should we trust you? Are editors better judges of scientifi merit than the specialists in the field? (Presumably not, or there would be no need for peer reivews.) Is there any conflict of interest here? (Yes, of course there is.) Have editors of scientific journals proved universally reliable and trustworthy in the past? (No, of course not; the vast majority probably are, but there have been many cases of sloppy, or downright abusive, editors.)
“Can editors act unilaterally? Yes, though obviously they should not if possible.” And yet your actions here were appalling. If reporting is accurate, you sent one email, to one author (to which address? The one he used 6 years ago, when he submitted?) — then unilaterally retracted the paper when there was no reply after 3 weeks, which included a major US holiday. Obviously you made no serious effort to avoid this unilateral action. This shows at best horrible judgement at a time when you are demanding we must uncritically accept your judgement on the much more complex issue of scientific accuracy.
This is a terrible policy, decided apparently without consulting the scientific community, and implemented in a terrible way.
This is a great discussion, because it touches on may distinct issues that are all important for improving the peer review process. I have commented on these issues in a blog:
but I want to highlight a few key points:
1. Post-publication peer review may be a more appropriate corrective measure than a “retraction”, because it would involve a dialogue between the authors, editors and the scientific community. Even if one cannot achieve a consensus, a transparent presentation of this post-publication peer review would allow all readers to understand the issues raised and learn from the discussion.
2. I think that one concerning point was the unilateral editorial decision to retract a paper without the consensus of the authors. Retractions are a pretty big deal, and in the absence of overt fraud, which is why the scientific community needs to be sure that such a major editorial decision was not just due to a mere scientific disagreement between the authors and the editors/peer reviewers.
I am quite sceptical about this.
What will be the ‘selection criteria’ for a ‘wrong conclusion’ ?. As said by others, retraction should be for clear violations of ethics (fraud/misconduct/fabricated data) or in extreme cases like the XMVR virus or the arsenic loving bacteria.
But again, what is a ‘wrong conclusion’ ? Just unreproducible by other labs ? How many times do you read in a paper something like : ‘previous studies, resulting in several compelling, but somewhat contradictory findings’, ‘our methods are slightly different from….’, ‘cell-specifc effect….’.
Overall conclusions can very well be “wrong” just because that’s how science works.
Correcting the literature by publishing contradictory papers happens everyday. When papers which seemingly contradict each other, which one should be retracted ??? Which one is wright and which one is wrong ?
Look, dear analytical type professors: Biology is not at the point where one can analyze it correctly. I say this as a younger analytical chemist that would actually like to publish in PLoS ONE, as it currently is seems like the organizational force in biomedical science, for a “working journal”.
I am more accustomed to reviews from JACS , J.Chem.Phys.B. and Biochemistry, whose editors would always correct me if I am being wrong. This is what I desire as an “author”: an independent opinion on the work my students have done from established scientists both outside and within of my field.
I am deeply disturbed by the implications here.
Retraction should be reserved for cases of fraud. Period. Or errors so gross that not noticing them constitutes a kind of fraud on the part of the referees and the editor.
This is a terrible case for retraction. There was no fraud. No one knew there was a “lab contaminant.”. The possibility arose during the process of trying to duplicate the research that is the essence of the scientific method. All trails were followed. Nobody did anything “wrong”.
To the casual reader – and you can tell this by reading media responses – the word “retraction” implies some sort of fraud. But there was neither fraud nor intellectual dishonesty. Someone thought they had demonstrated the proof of a new thesis; someone else convincingly demonstrated it was NOT proof and the new thesis remains conjecture. That happens all the time in research.
Why on earth would you “retract” it? There was nothing wrong with publishing this. That’s part of the process. I think it IS fair to ask, historically, what would have been “retracted” had these standards been in effect through the 20th century. Because this is clearly a new standard.
But maybe I am naive. Maybe I am assuming we are talking about scholarship, universities, the scientific process. Maybe what’s really going on is economics. Patents, money-to-be-made.
As a field, medical science has already made itself a little ridiculous (frankly) in the eyes of scholars from other fields with the “embargo” rule – that if you discuss a finding before it is published, it can’t be published. There appears to be no place for “manuscript in progress” or even “paper presented at xxx” in medical science. That seems to have more to do with the media circus that attends some findings – and this happened to be one of those findings.
So I suspect the “retraction” idea is more for the new Economics of Scientific Discovery (patents included) and the Media’s love of Controversy, than anything having to do with the scientific process or discovery or scholarship.
And that is a very unsettling trend.
[…] Blogs: “The role of retractions in correcting the scientific literature” (Virginia Barbour and Kasturi […]
Wow, just wow. I seriously question this interpetation of the COPE guidelines. The guidelines advocate correcting the literature and seemed to be on the path to encourage authors to come forward when they have made an honest mistake. I think most people see the “honest mistake” part of the COPE retraction guidelines as what to do when an author tells you that they made an honest mistake and the entire paper turns out to be false. This situation seems to be intepreting that guideline a whole other way.
I have a hard time believing that journal editors would agree to this as they are the champions of the scientific process. It’s usually just the lawyers who don’t get that science is self-correcting.
I still want to know:
1. How are you going to decide which papers are “wrong enough” to warrant retraction? What are you going to do when later work overturns the initial decision (as it frequently will)?
2. How will you distinguish between retractions for fraud and retractions for being “wrong”?
3. Why did you not consult with the authors of the retracted paper? As others have pointed out, the paper had been out for years, there was no need for haste.
There’s a simple way to vent your disapproval of this decision….either the editors reverse this ludicrous proposal or we send our work elsewhere. We (the authors, the scientific community) decide on which editorial decisions are a success and which will be laughed about and derided in the months to come.
Do not underestimate the magnitude of what is being proposed here.
[…] An editorial written Tuesday by Virginia Barbour and Kasturi Haldar at PLoS regarding a specific ret… 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. […]
As others have stated above – “Who will retract the retractions?…”
The late George Orwell gave us a glimpse of where this kind of thinking leads to, see http://en.wikipedia.org/wiki/Ministry_of_Truth
Please retract this insane policy
Just wanted to add one more name to the call to unretract the XMRV paper. It’s the primary citation for the discovery of XMRV.
By convention, we do not knowingly cite retracted papers but in this case we would have to. The scientific record has been violated.
Use retraction routinely for fraud but very, very, sparingly for reasons of error. It’s the internet age and verbiage is cheap – encourage authors to place comments, corrections, reappraisals, if it would be useful. PLOS can begin to get their house in order by unretracting this paper and inviting these authors to attach a comment putting the paper in its current perspective.
Please be careful as to what opinions you attribute to scientific publishers as a whole. I am employed by a different publisher than PLoS, and I do not share their views on this issue at all.
Something else I want to know: is this policy PLOS-wide, or limited to PLOS Medicine? Specifically: is this policy going to be adopted by PLOS ONE, where I serve as an AE?
I tend to agree with most of the repeated points which have been raised by others-
1. How can you tell what is an ‘ultimate’ finding? With the scientific literature getting longer and longer as time goes on with more and more findings refuted then replicated then refuted again then replicated again as well, what will happen when you need to unretract the retractions? And then re-retract the previously unretracted retractions? After a while such a situation could easily get ridiculous.
2. If science is truly a self-correcting process then why is it necessary for editors to stick their thumbs in the process?
3. If errors will lead to retractions then this seems like a potentially serious example of a ‘chilling effect’. Will the original authors of a work go through the lengths that the PLoS One XMRV authors did to answer a question if it means that their original work will be retracted? Will friends and/or colleagues of an individual or group go through the effort of debunking a finding if it means that their friend’s/colleague’s work will be retracted as a result?
4. Was the PLoS One paper even incorrect? A novel gammaretrovirus was in fact found in samples of prostate cancer tissue. It got there by a round-a-bout way but the paper was correct in this regard, was it not?
5. The Science XMRV/CFS paper was retracted as a result of a ‘final straw’ regarding a mislabelled gel which qualified as false data, wasn’t it? Not just erroneous data but false data. Again, there was no such false data in the PLoS One XMRV/prostate cancer paper, only contaminated and/or erroneous data. Now what happens to the rest of the XMRV papers that were published? Are the authors of those papers going to go through the lengths that the PLoS One XMRV authors did to find out what (if anything) went wrong just to ensure that their papers get retracted?
6. It could be seen that the retraction of the PLoS One XMRV paper was an attempt by PLoS One to save face. But if mistakes are made then should those mistakes be left in place to serve as reminders that findings can be erroneous and not whitewashed away? Is this simply a window dressing solution to a larger problem or set of problems, ie journals not requiring inclusion of complete data sets as a pre-requisite to publication, peer reviews not being made public, etc, ie what the whole ‘open science’ ethos is trying to bring about?
7. As the XMRV story shows, there’s a fine line between listening to what the public has to say and identifying legitimate concerns and/or complaints and being bullied and/or harassed by a bunch of people who don’t necessarily even know what they’re talking about.
[…] the editors at PLoS Pathogens chose to retract the paper. This retraction was accompanied by an editorial from PLoS Pathogens editor Kasturi Haldar and PLoS Medicine editor Ginny Barbour on the role of […]
‘+1. Those of us who volunteer our time and energy for PLOS deserve a quick and clear answer to this question.
I wanted to add myself to the steady stream of voices arguing that this is a major overreach that will be to the detriment of science. To repeat my comment from Michael Eisen’s excellent blog post:
I worry about what will happen to the speed of scientific progress if eventually being shown to be wrong is judged so strongly. In my experience the only way to make sure that you are “right” in the long run is to do simplistic and generally uninteresting science. Pushing the envelope is hard enough already. If we discourage it further by having serious consequences for being wrong, I fear that the already limited number of major steps forward will be reduced even further.
Please note we have done a further post on this topic aiming to clarify the points raised.
In addition as we say there – we’d like to invite longer responses for a constructive discussion on these issues at this blog. If you would like to contribute a blog post please email us on email@example.com and we will publish a selection of opinions over the next few weeks. If you have specific questions on PLOS policy please also contact us on firstname.lastname@example.org or email@example.com
[…] Three Charts from Different Omics » XMRV Paper from DeRisi Lab Retracted, Mike Eisen Argues ‘some people are more equal than others’ An XMRV-related paper titled Identification of a Novel Gammaretrovirus in Prostate Tumors of Patients Homozygous for R462Q RNASEL Variant from Joe Derisi’s lab was pulled out by the journal PLOS Pathology. The journal explained its decision in a separate blog post. […]
Mike Eisen is right about the chilling effect of the stated policy, whatever the good intentions. Interpreted in the most straightforward way, a policy of retracting papers whose “major conclusions are shown to be wrong” both overreaches the role of retraction and likely chills good science (or encourages authors to send it to other journals without such a policy). Not to mention the practical problem that if papers can get through peer review with wrong conclusions, how can we be sure that the retraction is based on certain knowledge that will not itself be refuted later on? That would start a cycle of retracting retractions that would quickly become a big distraction from real science.
As a colleague has pointed out to me, the problem of how a newcomer to an area of the literature can understand the main findings has become harder and harder. We would all benefit from a way to signpost big advances, unproductive directions, and reinterpretations of the data. Nonetheless, retraction is too blunt an instrument to fix this problem, and will drain vast amounts of time into defending turf and reputation rather than seeking truth by doing more and better studies, which is what most people in this business still prefer to be doing!
I think the discussion of Publication Ethics should frame this debate because, in my opinion, being wrong in interpretation is not unethical.
Reporting experiments that were not performed, or have been manipulated by addition or removal of specific data points, is unethical and always was. Those papers that can be shown to be based on fabricated data should be retracted. Perhaps editors and boards have a stronger role here but the burden of proof is not insubstantial.
Mistakes in experiments are a grey area. It would be ethical for an author to correct a paper if it became clear that there was, for example, a key experimental reagent was incorrect – taking the wrong bottle off the shelf is not unethical (it may be unwise) but reporting that mistake, if it can be traced, is the ethical behavior. There is a boundary here, including issues such as underpowered studies, population effects and reagents that are unstable or unavailable, but nonetheless it would seem in general that corrections should be preferred to retractions. It should be reasonable that the role of journals should be to publish rebuttals from others in the field if there is a concern from other parties but the original lab cannot identify a specific problem.
However – being “wrong” in interpretation is not unethical. Science is one of the few areas where being wrong should be celebrated on a daily basis. The word may be used in different ways, and that may be inflaming this debate, but having interpreted data in the light of current literature but coming to the wrong conclusion is a reasonable thing to do. Admitting it is, of course, reasonable but if the underlying data is honestly collected and not prone to gross artifact, then no ethical infraction has occurred.
What happens if a paper is retracted after a conclusion is proven wrong by a later paper, but then the second paper is proven wrong, thus proving the original paper correct again? Will the original paper be unretracted?
[…] retraction had already prompted Haldar and PLoS editorial director for medicine Virginia Barbour to write a post explaining their retraction policy. Some of the language in that post drew heavy criticism from […]
[…] « An Email that Changed my Life (and the World of Retractions) Sucker Deal for Particle Physicists » XMRV Paper from DeRisi Lab Retracted, Mike Eisen Argues ‘some people are more equal than others’ An XMRV-related paper titled Identification of a Novel Gammaretrovirus in Prostate Tumors of Patients Homozygous for R462Q RNASEL Variant from Joe Derisi’s lab was pulled out by the journal PLOS Pathology. The journal explained its decision in a separate blog post. […]
[…] « An Email that Changed my Life (and the World of Retractions) Three Charts from Different Omics » XMRV Paper from DeRisi Lab Retracted, Mike Eisen Argues ‘some people are more equal than others’ An XMRV-related paper titled Identification of a Novel Gammaretrovirus in Prostate Tumors of Patients Homozygous for R462Q RNASEL Variant from Joe Derisi’s lab was pulled out by the journal PLOS Pathology. The journal explained its decision in a separate blog post. […]