1

I have a paper that is accepted and published in a journal. After it was accepted I realized that some of the errors were calculated in a wrong manner and hence they appeared around 1.5 times smaller in the paper (this was due to a wrong advice that my advisor gave me). I was busy at the moment however later, when the paper was published, I managed to recalculate the errors and realize their difference. I told my supervisor about it and sent the new results to him. Although he seems to admit that his advice was wrong he is opposing my idea to send an erratum to the journal. Here is his argument:

  • The differences are minimal and do not change the conclusion or the main point of the paper. (Although the actual errors are 1.5 times the reported ones, they are still small because the reported ones are small too)
  • We have to change many parts of the paper and not only one place.
  • It will put us and the journal in a bad situation.
  • I am neither the first person nor the last person who has done a mistake in his paper.
  • Our code and dataset are public.

I have had serious arguments with my supervisor in the past. The last time it got to the point that he ordered me to do as he says. This makes me worried if I push him harder I will ruin my relationship with him.

What are your thoughts about this situation? What is the right thing to do in your opinion?

  • 2
    Could you just post an update on your webpage? In this kind of situation (Eg sign errors that don’t affect the results) I would fix the arxiv version and not submit an erratum, but in non-arxiv fields I’m not sure what to suggest. – Noah Snyder Dec 10 '19 at 19:23
  • These concerns read like an intellectual bancruptcy on your advisors side. Admitting an error and publishing an erratum makes you look good. Bad case of impostor syndrome, afraid that people notice he`s not a genius after all? – Karl Dec 10 '19 at 23:50
  • @Karl Could be like that. It could be also that he is worried that he will have less chance in publishing in the journal in the future. However for me the important thing is to know what I am supposed to do in this situation. – user74315 Dec 11 '19 at 8:25
  • That would only make some sense if this is the third paper in a row from your advisor that needs an erratum. Complete nobrainer, but nevermind. Is the derivation of the error calculation written down in there, or just the results? In the latter case, I would indeed say forget it. No sense in writing an erratum just to make a few error bars a bit longer. – Karl Dec 11 '19 at 19:36
  • @Karl No, it is just the final numbers. In fact we have not detailed how exactly the errors are calculated. – user74315 Dec 12 '19 at 8:49
3

You said the error is small. Therefore the need for an erratum is small as well. Ideally you would publish the erratum, but if you have more important things to do, that is understandable.

  • As I said in the question the actual errors are 1.5 times the reported ones. However their absolute value is still small. – user74315 Dec 11 '19 at 8:27
2

Don't argue with your supervisor. It's not worth it.

  • Your supervisor is likely a much more experienced academic than you, so his judgment of the situation is more likely to be accurate than yours.
  • As mentioned, the differences are minimal.
  • A good relationship with your supervisor will be very, very helpful in graduating.

Given that you have little to gain, there's no reason to hurt your relationship even more.

  • 4
    "Your supervisor is likely a much more experienced academic than you, so his judgment of the situation is more likely to be accurate than yours." Clearly not true in this case. – Anonymous Physicist Dec 10 '19 at 19:50
  • I admit that he might be considering things that I might not know about. – user74315 Dec 11 '19 at 8:38
  • @AnonymousPhysicist I think it's courageous to draw that conclusion based on a secondhand description. Not saying you're wrong, but I'd say the null hypothesis is that it's true. – Allure Dec 11 '19 at 9:13
  • @user74315 True, but then he should better explain these things to you. – lighthouse keeper Dec 11 '19 at 10:11
  • 1
    I think this advice is good in general, but you should probably have a talk (just a talk, not a whole process) with someone else experienced in your field to confirm that these errors are indeed not such a big matter. Trust your supervisor, but tactfully verify. – ObscureOwl Dec 11 '19 at 10:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.