Adventures in Ethics and Science

Over at DrugMonkey, PhysioProf notes a recent retraction of an article from the Journal of Neuroscience. What’s interesting about this case is that the authors retract the whole article without any explanation for the retraction. As PhysioProf writes:

There is absolutely no mention of why the paper is being retracted. People who have relied on the retracted manuscript to develop their own research conceptually and/or methodologically have been given no guidance whatsoever on what aspects of the manuscript are considered unreliable, and/or why.

So, asks PhysioProf, have these authors behaved ethically?

I think in order to get clear on what obligations the authors have to the scientific community, it may be useful to start with the question of what this kind of retraction communicates to the scientific community.


Withdrawing an article that has already been published, obviously, communicates that there’s something wrong with the article. When there is something wrong with the article, sharing this information is a good thing — along the lines of identifying the wonky pH-meter for your labmates so they don’t waste their time using it and thinking they’ll get reliable data from it.

In this case, however, the retraction essentially says, something is wrong with what we reported initially, but we’re not going to tell you what.

Either the authors don’t know what’s wrong with the article they’re retracting, or they do know what’s wrong and have decided to keep that information to themselves. In the case that they don’t know what is wrong with the article, presumably there was something tangible that convinced them that something was wrong with it — and they have chosen not to share their clues that a problem exists with the article, even if they can’t precisely locate the problem.

There are lots of ways to goof when conducting (and accurately reporting) scientific research, so it’s easy to imagine what some of the problems that led to this retraction might be. Maybe after the article was published, the authors discovered that they could no longer reproduce their own results. Maybe they made an error in analyzing their data. Maybe they decided their interpretation of the results was wrong. Maybe they discovered a fundamental flaw in their experimental design or measurement technique.

Maybe they discovered that someone on the research team fabricated or falsified results, inappropriately manipulated an image, or really didn’t know what he or she was doing and screwed up the results.

All of these are possibilities that will occur pretty quickly to the scientists reading the retraction. And by not explaining the reason for the retraction, the authors are presenting a situation where readers can’t tell is the problem was an honest mistake or cheating, a result that shouldn’t have been reported in the first place or one that just looks problematic in the light of further knowledge gained since then. The authors are essentially letting their fellow scientists fill in the blank with an explanation for the retraction.

Letting your fellow scientists make up their own minds is usually a good thing. However, it’s customary to present them with enough information to provide the basis for making a judgment. Seeing as how science is an activity based on collecting and communicating (and scrutinizing) information about the objects of study, scientists have a duty to share information with each other. Asking a fellow scientist to render a reasonable judgment when you’re holding back the crucial details needed to render that judgment is not playing fair.

Some might argue that scientists hold back details from each other all the time — that this is how one avoids getting scooped in the battle for priority. However, publishing scientific claims makes what you’re reporting the property of the scientific community. Once it’s in the journal, your obligation to be forthcoming about the details (including answering questions raised by your fellow scientists) is turned on, full strength.

The original article put forward claims with the clear expectation that other scientists might make use of them — using them as a basis to evaluate their own results, or as a starting point for developing and conducting new research, or what have you. In the event that the original article is withdrawn, the obvious question another scientist would ask is, “How does this effect what I was using that knowledge for?”

Having made your claims public (to secure your priority for them), you’re on the hook to share the relevant information around them — even if that information ends up being why you are no longer willing to stand behind those claims. You’d have this obligation even if your fellow scientists hadn’t invested time and resources in trying to build on the results you’re now retracting.

Indeed, a retraction — and the details it includes or omits — does not just communicate a message about how to regard the results reported in the original paper. It also communicates a message about how the scientific community should regard the scientists who produced, reported, and ultimately retracted these results. Are they reliable producers of scientific knowledge or not? Careful or sloppy in their experimental work and data analysis? Honest or dishonest?

Once again, with a retraction so short on details, the scientists reading it have to fill in the blank with their own assessment of the authors as scientists. In the absence of relevant information, they may well err on the side of caution and decide that the authors are as reliable as a wonky pH-meter.

Here, I think there’s an important question about what obligations the authors of the paper (and its retraction) have to each other. If the paper is being retracted owing to mistakes the authors made together, standing together to take responsibility for those mistakes is fine. However, the low-content retraction is ambiguous enough that some readers might wonder if one of the authors in particular make the mistake (or worse, committed the fraud). Removing the ambiguity and taking that possible interpretation seems important, especially so a graduate student or post-doc on the project doesn’t get saddled with being the prime suspect.

On the other hand, if blame for the collapse of the original paper can be localized — especially if it resulted from fraud or gross incompetence — that information is relevant to other scientists, who want to know whose work is trustworthy and who might be a good collaborator in the future. It’s not the case that identifying the author at fault completely absolves the other authors of responsibility — after all, they were all supposed to sign off on and stand behind the results reported in the original article. However, it’s less problematic to trust someone’s skill (or integrity) and discover later you were wrong than it is to conceal this information from the rest of the scientific community.

Comments

  1. #1 RM
    November 26, 2008

    The only semi-valid reason I can see for issuing an detail-less retraction is the case where one of the authors has committed fraud (e.g. fabricated data), but the investigational board hasn’t ruled yet. Here you’re balancing the ethical requirement not to waste the time of other researchers with potentially fraudulent data with the ethical requirement of not dragging the suspect researcher’s name through the mud.

    You don’t want to issue a retraction effectively stating “We’re retracting the article because of investigator misconduct”, only to have the review board clear Janet First Author of all wrongdoing. (c.f. the Dwyer/Hellinga saga). However, given the time necessary for the review boards to come to a thoughtful, measured conclusion, you don’t want to leave other researchers wasting time thinking the paper is still valid while the wheels of justice grind slowly.

    I think, though, that a better way to handle such a situation is with a “white lie by omission”. Don’t come out and say “Janet faked the data in Figure 1″, but rather do a weasel-worded “We are unable to repeat the data for Figure 1″, and then let everyone know later (after the board has rendered an official decisions) that the reason you can’t repeat it was that Janet massaged the numbers.

  2. #2 jrhs
    November 26, 2008

    If I know I would be subject to ridicule no matter what reasons I offer, I would just say nothing.

  3. #3 Martijn
    November 26, 2008

    Maybe they already know something is terribly wrong – they cannot reproduce the results – but they haven’t quite found out what went wrong and are afraid it may take a while before they find out.

  4. #4 Comrade PhysioProf
    November 26, 2008

    Beautiful analysis, SG. I concur in every particular. And I would amplify the point that there are ways of providing the scientific community with enough information to know what of the content of the original article is considered unreliable, but while remaining completely agnostic as to “blame” or “fault”.

  5. #5 Jonathan Vos Post
    November 26, 2008

    Please excuse me for saying something politically incorrect, but I’ve actually heard someone remark, with indignation on reading an unlikely published paper: “I wonder who you have to sleep with to be published in THIS journal?”

    I suppose that one can be forced to withdraw a paper if such a comment turns out to be based in fact…

    And I will swear under penalty of perjury that the former Dean of Arts and Science at one place that I taught managed to appoint not only that person’s spouse, but also that person’s homosexual lover to faculty and administrative positions.

    In both cases, the appointments were eventually undone by the executive VP, in one case demoting from Assistant Dean to mere professor, and in the other case by non-renewal of teaching contract.

    I have no basis for associating that with the paper retraction story, but it does indicate an area of ethical concern in the scientific/academic community, which (for legal reasons) is rarely discussed openly.

    Disclaimer: as I’ve mentioned in this blog before, both my wife and I are scientists, have been professors, and have several times worked for the same employers. But neither of us hired the other!

  6. #6 Joshua Zelinsky
    November 26, 2008

    They may also retract because they realized that there’s some serious problem but haven’t pinned down precisely what that is yet? Presumably one would then want to see if in a few months there is a follow-up note or something similar?

  7. #7 Pinko Punko
    November 27, 2008

    I wouldn’t be surprised if the authors wanted to do right by the literature in terms of some shady behavior but also not wanting to invite possibly the endless and terrible institutional review of something fraudulent. So this is like trying to be good but also using the rug and the broom.

    I have no idea, this is merely a hypothetical and has nothing to do with the case at hand where we are data-free.

  8. #8 andre
    November 27, 2008

    This brought to mind a recent incident that I find similar in a lot of ways. In May an employee was killed in an explosion at one of Boulder Scientific’s chemical manufacturing plants. To this day they haven’t made public (to my knowledge, and I’ve been looking) the cause of the explosion, or any of the chemicals that may have been involved.

    This concerns me in that rumor has reached me that it may have involved chemicals similar to the ones that I am currently synthesizing.

    I think these two cases parallel each other, but the Boulder Scientific incident in my mind clearly crosses the ethical line that the infoless retraction is tiptoeing, since lives could be in danger.

    I of course think it’s especially ethically problematic because it’s my life.

  9. #9 S. Rivlin
    November 29, 2008

    Janet,

    You and the commenters letting the J. Neurosci. off the hook very easily.

    I find it disturbing that when you submit a manuscript for publication, not only that they run it through two, three and, at times, more reviewers; that they also require the submitting author(s) to sign different documents regarding the ethical conduct of the research, conflict of interest and, of course, copyrights, however, these very standards are not being followed by the journal and its editors when dealing with a retraction. They owe such due diligence to their readers, to the editorial board and to the public in general, who, in most cases, pays the tab for the published/retracted paper.

    By not informing their readers about the reason(s) for the retraction, the editors of the J. Neurosci. neglecting their responsibility and simultaneously making themselves suspect of having an ulterior motive in not revealing the reason for the retraction.

  10. #10 Michael Pyshnov
    December 4, 2008

    First, I disagree with Dr. Stemwedel when she says: “publishing scientific claims makes what you’re reporting the property of the scientific community” I believe science is nobody’s “property”. This property thing is repeated again and again, but it’s wrong. The author has authorship which is not a property, but a mere recognition of a fact. All others have the knowledge, that’s all.
    Now, the publishers have policies (practically identical) which list the points that must be included in retractions. They include reasons for retraction. This is very important when retraction is made because of plagiarism, to name the real author.
    In the case discussed now, a curious possibility arises when the reasons are not given: someone can claim results published and retracted by the authors, in a “new” paper, saying that the retracted paper sort of doesn’t exist. Just another reason why the reasons must be published in a retraction.
    I have now copied on my site the exchange of letters with the Editor whom I asked to retract a paper. See it at http://ca.geocities.com/uoftfraud/ Some legal points and some horrendous difficulties with this retraction can be instructive.

The site is currently under maintenance and will be back shortly. New comments have been disabled during this time, please check back soon.