How To Read A Retraction Eleventeen

Apr 30 2009 Published by under Conduct of Science, Ethics, Scientific Misconduct

There are two retractions in PNAS this week. Here is the body of the first one:

The authors wish to note the following: "After thorough efforts by K. Berry in the laboratory of J. A. Doudna to reproduce our reported IRES-specific translational inhibition were unsuccessful, we initiated an extensive effort to reproduce the IRES-specific peptide binding. These experiments, carried out with the assistance of Y. Guillen in the laboratory of J. W. Szostak, also failed to confirm our previously published results. Therefore, we retract the paper. We sincerely apologize for any confusion that the publication of this study may have caused."

Translation: "There is a fraud investigation going on right now."
Here is the second:

The undersigned authors wish to note the following: "We have discovered errors in some of the figures in this paper. Therefore, the undersigned authors regretfully retract the paper."

Translation: "There is a fraud investigation going on right now."

19 responses so far

  • Physiogroupie IV says:

    Also, in that 2nd one, the first author is not part of the retraction. Hmm!

  • neurowoman says:

    Maybe I'm naive, but on the first paper, how do you know that Litovchick, who "carried out the research", didn't just screw up? Without knowing exactly what is inconsistent, why leap to intentional fraud?

  • Eric Lund says:

    @neurowoman #2: I'm also naive (I'm not in any biomedical field), but there is a difference between a fraud investigation and a finding of fraud. We cannot have the latter until the former is completed. The fact that Litovchick, unlike the lead author of the second retracted paper, signed his name to the retraction suggests that he is cooperating with the investigation.
    Again, I'm an outsider, so take this comment with a large grain of salt: If I were Litovchick, and were only guilty of error rather than fraud, I would welcome a fraud investigation. Presumably if it were only an error without intent to defraud, the fraud investigation would conclude thus, which strikes me as a better outcome than having the cloud of a possible fraud that was never investigated. It may be possible though difficult to recover from the former; the latter would almost certainly abort Litovchick's career.

  • If there were no investigation going on, then it is highly likely that the retraction would have explained exactly what the authors thought had happened that led to the erroneous result. Both of these retraction contain absolutely no information at all about what happened to lead to the retraction, which is exactly what strongly implies that there are investigations going on.

  • DrugMonkey says:
    Well, well. Nature strengthens their anti-fraud policies by requiring contribution statements and the senior author of each group to take some explicit responsibility for data preservation and quality.

  • juniorprof says:

    Holy Crap, we covered some of the work in that second paper in my molecular pharmacology class when talking about nutrient sensing in the hypothalamus. The last day of class was today, WTF am I supposed to do now? I suppose a mass email is in order...

  • microfool says:

    Note that J. Doudna originally served as editor for the first article that was retracted, in addition to trying and failing to replicate the work. That is an involved editor!

  • Heraclides says:

    The first does not sound like a fraud investigation. It sounds more like an experiment couldn't be reproduced. It's awkward, it's not good science and it's not "supposed" to happen, but it does sometimes happen. They have clearly tried to reproduce their own results and found—no doubt to their horror and embarrassment—that they couldn't. Although not as detailed as it might be, they do say what it is that could not be reproduced both in the other lab and then later by themselves.
    The second might be anything from incompetence (perhaps mixed with language issues) to something more serious, they simply don't say. If the first author is a student, it's possible that poor supervision may be a part of the issue.
    Personally, I think it's poor taste to assume fraud ahead of more ordinary reasons. (Or making assumptions at all, for that matter.) Does no-one like Occam's Razor around here...?
    I would second the point that it's not fraud until it's shown to be intentional deceit and until then stay off the accusations. Mere accusations can affect people, even if they were never right.
    Eric: My reading of it is that they have already done the checking themselves.

  • Samia says:

    Heraclides +1.

  • Reading this, I don't assume fraud. As an outsider, though, I just wouldn't be surprised if it really does mean that an investigation is taking place. It makes sense that a science community recently embarrassed by high-profile cases of fraud would launch investigations of "everything", including cases in which honest mistakes are as likely as or even more likely than fraud to account for the errors. I have no inkling whether or not this is one of those cases. I'd rather not be totally stupid.

  • msphd says:

    These things are interesting to me, if only because I think either side of the case is horrifying. Have you read The Baltimore Case?
    I've had people get mad at me when they couldn't reproduce my experiments. In every single case, they weren't not following the explicit instructions I gave them (detailed protocols). In the cases where we went over it all again, they got it working and everything was hunky-dory. Some people just moved on and found some other way to do it, they were too lazy to try again. But I shudder to think about the ones who were convinced they were doing everything right when they weren't. Because I'm still the one who gets blamed, aren't I? They can go around saying they couldn't reproduce my results, and strictly speaking, it's true that they couldn't. How much of that is my fault, and how much of it is not?
    These things are not always as simple as fraud. And I've personally witnessed other cases where the evidence clearly showed that the results were artifactual, but everyone had so convinced themselves it was true that they went ahead and published it anyway. Still waiting to see if that catches up with them...

  • MBench says:

    What seems fishy to me about the Litovchick/Szostak retraction is that the peptide discovered was demonstrated to have activity in three separate assays, two different in vitro binding assays and an in vitro translation assay. There are really only three explanations, 1. fraud, 2. monumental mistakes or misinterpretation of data (requiring three separate mistakes/misinterpretations), or 3. problems with peptide prep (ie they had a different version or mixture of versions of the peptide than they thought they did). The peptide itself is quite complex so at first I thought 3 was the most likely. But the original paper includes a series of peptides used to uncover the minimal active peptide structure. Thus the same set of mistakes (three separate assays) were made for the whole series of peptides. A systematic problem in peptide prep? Maybe, if they were sloppy, but less likely...
    I still don't think it's fraud, I am probably naive but broad misinterpretation is still more likely. Of course, the difficulty in telling the difference between the two is well-documented (Baltimore Case and others).

  • bill says:

    Heraclides: "I would second the point that it's not fraud until it's shown to be intentional deceit and until then stay off the accusations. Mere accusations can affect people, even if they were never right."
    Didn't SRivlin get a hard time on this very blog for unsubstantiated accusations of fraud...? Granted that UAO being the subject of a fraud investigation are not the same as UAO having committed fraud, still a reputation is a delicate thing.

  • S. Rivlin says:

    Fraud or not, it seems that there are more retractions these days than ever before. Irreproducibility of your own results in your own lab smells much worth than unsuccessful reproduction in someone else's lab. The latter could be explained, as mentioned by msphd, through faulty, lazy following of the methods or such. When the results cannot be repeated in your own lab, it is usually a sign of things not kosher, especially considering that some time elapsed from experiment performance to publication.
    Thus, I tend to agree with PP that fraud investigation is involved.

  • neurlover says:

    Has the rate of real "retractions" increased? I only find out about them here, on the whole, so I don't know if retractions have increased, or if I'm just hearing about them more.
    I work in a field where retractions never occur because everything is a bit too fuzzy to ever justify a straight out retraction and where the reasons for lack of reproduction are too far flung to justify retractions. I do wonder how we'd ever detect fraud. But, I'm guessing that it would be nearly impossible.

  • Luigi says:

    I work in a field where retractions never occur because everything is a bit too fuzzy to ever justify a straight out retraction and where the reasons for lack of reproduction are too far flung to justify retractions. I do wonder how we'd ever detect fraud. But, I'm guessing that it would be nearly impossible.

    So you're not a scientist? In science, sort of by definition, reproducibility is usual and hypotheses can be disproven.
    In science, one is either right or wrong. That's why cheating baffles me. A Nobel prize can't make you right if you're not. And the spittle of a thousand colleagues won't make you wrong if you're right. Sure I can understand the pressures of a scientific career. But if you just want a job, go get a job. Don't go into science, because as far as plain old jobs go, a science career has a lot of disadvantages. In my opinion, fraudsters should be banned forever. Not because I don't think they should be forgiven. Rather, I just don't understand how the idea of cheating would even occur to anyone honestly interested in figuring things out.

  • Lab Lemming says:

    "Has the rate of real "retractions" increased?"
    Almost certainly, but the rate of publications has increased as well, so who knows what the normalized rate is doing.
    Out of curiosity, did all authors sign both of these retractions? I reckon that partial retractions are way more fun than complete ones...

  • S. Rivlin says:

    Here's a study that could shed some light on the possible magnitude of the problem.

  • Pinko Punko says:

    Heraclides plus a zillion.
    First one suggests grave error/sloppiness/unknown reagent issue.
    Second one=ultra shady. "Problems" could very well have been specified.
    In the case of the first retraction, they stated they repeated the experiments and the results were not reproducible.

Leave a Reply