In an interesting letter printed in today’s Nature, biologists Sophien Kamoun and Cyril Zipfel suggest that “failure by authors to correct their mistakes should be classified as scientific misconduct.”
They note that this policy is already in place at their institute, The Sainsbury Laboratory (TSL).
We contacted Kamoun to ask what constituted a mistake, given that numerous papers have received queries, such as on sites like PubPeer, but it’s not clear whether those are legitimate mistakes. He told us:
The main point is that when someone points to an error or even a potential error it should be the author’s responsibility to address it promptly. I’m not sure how else to define an “error” or a “mistake” beyond the dictionary’s definition. The author may feel that the error/mistake is not valid or is irrelevant. But it remains the author’s responsibility to respond and not to ignore the criticism. The analogy would be the way authors would to pre-publication peer reviews. As we wrote, snubbing criticism by not addressing it promptly goes counter to our fundamental ethos as scientists and threatens to erode society’s trust in the scientific community.
I don’t think the author should wait until the error is independently validated before they react. But seeking independent validation could be part of the response of the author.
What do you think? Take our poll below.
Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up on our homepage for an email every time there’s a new post, or subscribe to our new daily digest. Click here to review our Comments Policy.
The authors fail to address a crucial issue: what if the report is made anonymously? Do they still feel the same way?
Is that a crucial issue? The report is the same, whether it bears a name or not.
In most cases, referees of articles are anonymous too.
As I was preparing to react to your comment, I realised that I am without the means to locate your CV. Without this, I simply cannot judge if your insightful and well-phrased inquiry is worth an iota of my time.
However, I will speculate that there are many authors with an utter disdain for criticism of any kind and that refusing to engage concerns b/c they are unsigned may be more of an excuse than a legitimate reason not to address well-founded critiques. Such authors, obviously, are in contrast to the numerous responsible researchers that will respond to honest concerns.
The answer is here http://pubpeer.com/publications/4C70FCE2AC5F1737A1C7560C1C3590#fb47875
Unfortunately the discussion drifts from error/mistake via _report_ of error to criticism.
What if the report/criticism is clearly from a crackpot? “You didn’t cite my proof of the Riemann Hypothesis, which shows that the world was created 6000 years ago, so obviously your interpretation of those fossils is wrong.” Or even worse, a crackpot with an agenda? “Everyone knows that Zika was created and spread by the CIA, and your claimed DNA sequences are just part of the conspiracy.”
It’s a simple question with a “Yes” or “No” answer. Given the harm that can be caused for many by erred or fraudulent publications in medical and scientific journals, the answer seems obvious: Yes. Of course anyone who knows for certain that they’ve published error/scientific fraud and does not move to correct it is practicing misconduct.
Years ago I have defined (intentional) missing citation (of very relevant publication) and (persistent) lacking response (by corresponding author) as new types of misconduct besides the conventional FFP (fabrication, falsification and plagiarism) misconduct.
I answered “Yes” based upon general principals but the truth is that many (most?) journals don’t have a ready way to handle such things post publication. I know this from first hand experience.
Then of course there’s the trickier question of what actually constitutes a condemnation-worthy “mistake”? Is a failure to correct (for example) a mislabeled axis of a graph the same as a failure to correct an erroneous data presented in the graph?–they are sometimes comparable but the former will usually be easier to sniff out than the latter.
I am aware of several cases that even when errors are pointed out to the authors, that the editors and the publishers have stone-walled the correction of the literature, so this complicates the issue even further. Some editors and journals are not willing to spend time and issue errata unless there is clear misconduct. In other words, even if some authors are willing to correct errors, even apparently menial ones, sometimes the problem lies with the editors and publishers.
I should note that Kamoun and Zipfel have stated their clear position regarding anonymous queries and have distanced themselves from Blatt’s perspectives:
https://pubpeer.com/publications/4C70FCE2AC5F1737A1C7560C1C3590
What about an error in interpretation? A conclusion that turns out to be untrue? Are these mistakes? Because if so, I suspect that the majority of research papers are going to have to be annotated somehow. And what if you have two groups with different, incompatible theories, each supported by some data? Are the papers presenting this material mistakes in the making or part of the normal give-and-take at the cutting edge of science?
Mistake seems to me to be too imprecise a word, on its own, to bear the weight of an institutional policy.
What if the authors are no longer active researchers? What if they have moved on from the postdoctoral position they held when they published that particular paper? Are they expected to address (potential?) mistakes in their own time, unpaid? I can think of quite a few problems with making a sweeping generalization like this.
That was my thought too… and it’s not just postdoctoral researchers. It’s any researcher, at any stage in their non-tenured career, who finishes the contract associated with the research.
It’s also possible – even probable? – that a researcher who no longer has affiliation to the project/institution will also lose legal access to raw data. While each project may have its own agreements regarding data ownership, it’s standard in Australian universities for the data (and all products/outputs of the research) to be owned by the institution, not the individual researchers.
Tom, this is very closely related to the issue of deceased scientists and how to hold them accountable. I share your concern, especially about such cases. Maybe my recent paper may be of use to you:
Teixeira da Silva, J.A., Dobránszki, J. (2015) The authorship of deceased scientists and their posthumous responsibilities. Science Editor (CSE) 38(3/4): 98-100.
http://www.councilscienceeditors.org/wp-content/uploads/v38n3_4p98-100.pdf
Also the affilation of such an author (often a university) has a responsibilty to act when it turns out that there are issues with one or more papers of such an author. See for example the postings about Diederik Stapel and about Jens Förster at Retractionwatch for backgrounds.
I guess mistakes are only mistakes if the one who made the mistake admits it is a mistake.
I would love to see the peer review for this piece in Nature.
There are many types of errors, ranging from incorrect calculations to incorrect summarizing of the information that is being cited. The latter case is particularly troublesome because once the error is made it becomes “set in stone”. For example, a review article makes a statement and cites an article to substantiate the statement. Some time later, another author wants to include that “fact” in their article and rather than go to the primary literature they cite the reference without having looked at the primary literature. The original error is then perpetuated. Why is this a big deal? If somebody wants to learn about a new area, they will assume that everything that they read is correct and not know that certain “facts” are in fact, fiction. I have several examples of such cases.
As mentioned, there are several paths for alerting readers to these errors, including Letters to the Editor or comment in PubMed Commons. There are different sanctions that can be taken against the authors, whether they are in academia or industry. However, what can you do when the error is made by a Health Authority reviewer and is included the review document that is published on their website? I have examples of such errors in in Summary Basis of Approval (FDA) and an AusPAR (TGA).
While I generally agree that these problems can arise with reviews, one also should acknowledge that part of the problem is poor science on the part of the users of these reviews.
Assuming that everything that you read is fact is poor science. Whether you are new to a certain area or not.
Citing a review without knowing the primary literature behind it is also poor science. The biggest, but certainly very common, mistake is to actually cite the primary literature while only having read the review that cites the primary literature. Like: Review A claims Fact B has been shown in Paper C while one writes in a paper “Fact B has been shown in Paper C” without having read paper C but only review A. Yes, that can go wrong. It does not make the error in review A any better, but the process of producing “good science” is a multi-step process and everyone involved should apply rigorous standards.
As to the problems of a “you must correct any mistake”, a lot of good issues have been raised in earlier posts.
Especially in cases of old papers, this will be difficult. There is a huge amount of research where not only one main researcher like a post-doc or PhD student has left the institution, but where the institution has completely scrapped the field. Not sure about every jurisdiction, but in a lot of places raw data don’t need to be kept any longer than 10 years. What are you going to do if there is no primary data left anymore? Retract “just in case”? Which raises the interesting question of whether it is ethically acceptable to retract sound data without good proof of error or misconduct.
And as some have also pointed out, there is a grey area of mistake vs. scientific dispute. If you don’t allow scientific dispute, science itself will lose. If everyone who loses a scientific dispute has to retract what s/he published regarding the dispute, even less scientists will be willing to leave paved ground for something new.
Genetics – I agree that we should always do our own due diligence on any cited literature, whether it is a review article or not. I agree that the end-user should share some “blame” but the main responsibility lies elsewhere. It definitely lies with the author but also lies with the reviewer and the publisher. There certainly needs to be a process that could readily correct these errors and take advantage of the “wisdom of the crowd”. Other than making a statement on PubMed Commons, I’m not aware of any simple way of raising awareness of these errors. However, that doesn’t fix the issue in the original article and doesn’t help with cases in which Health Authorities make the mistake.
Where does one look to find complaint documents which have caused journal publications to be retracted?