How bad is this? And what is my obligation here?

[Taeyers]

I need some straight-forward informed advice on a moral dilemma, because I want to avoid a situation where I show up with a flamethrower to a knife fight, but I also want to address this ASAP if it's a serious problem. 

 

Background: I worked with a PI for a couple years before starting grad school, and he did a handful of studies for a paper. He asked me to do some work and generate data for a minor contribution, which now makes up 2 of several figures in the paper overall. However, in my biomedical science field, the norm is to list the PI as last author, and no one else worked on the paper, so by default I'm listed as first author. He's not my thesis advisor and probably won't be (still rotating at this time, but I plan to choose another lab). 

 

Dilemma: right now, the manuscript includes a graph that shows 2 conditions across 3 doses of a drug. X0, X1, and X2 are one condition in 3 dose groups, and Y0, Y1, and Y2 are another condition in the same doses. Symbol * is on the graph, showing that X1 and X2 are statistically different from X0, and symbol # shows that Y1 and Y2 are significantly different from Y0. The legend says that symbol $ would indicate if there was a statistical difference between X and Y in any dose group, but this symbol doesn't appear on the graph. There does seem to be a difference between X2 and Y2, but it must be just short of statistically significant. 

However, in the written results section, he wrote that X and Y are different with reference to that graph. I pointed this out and told him that I don't see that in the graph, and asked if maybe the symbols got mixed up (like maybe instead of * or #, he meant to use $ to show the difference). He didn't respond to my email (which is normal, he usually doesn't directly respond to comments on written things he sends to me), and I got an email a few hours later from the journal thanking us for submitting the manuscript. I downloaded the file that was submitted, and while he did change most of the things I commented on, he left that statement and the graph exactly as they were, claiming that it shows something it doesn't actually show. 

 

So what do I do about this? While it's mostly not my work and not my writing, it still looks like it is because my name happens to be in the first author position. I don't feel comfortable asking the PI why he submitted it like that because he likes to be argumentative and defensive by default, but I do want it to be corrected. I'm hoping at least one of the reviewers will catch it and he will have to fix it, but what if they don't? I asked a fellow grad student what she would do, and she says go around him and contact the editor to say it was an accident and needs to be corrected, but I'm worried that's a bit more aggressive than necessary. 

 

Help please! Thanks y'all. 

[ss2player]

This is tricky: does he say "statistically significant" or just a "trend" in the results? Precise language is important here, because if they claimed there is, for example a P<0.05 when there isn't, that's lying...BUT he can say there was a possible difference based on the graph and it merits further investigation. 

Love to hear other's thoughts! I was always taught the discussion is the place for philosophizing but the results should reflect only what was found.

[Taeyers]

This is tricky: does he say "statistically significant" or just a "trend" in the results? Precise language is important here, because if they claimed there is, for example a P<0.05 when there isn't, that's lying...BUT he can say there was a possible difference based on the graph and it merits further investigation. 

Love to hear other's thoughts! I was always taught the discussion is the place for philosophizing but the results should reflect only what was found.

Right, that's what I always learned too! 

 

The statement doesn't specifically include "statistically", but the statement is unambiguous in indicating a "significant" difference between the two conditions. I hoped that my comment would have inspired him to say something more appropriate about a trend or potential difference instead, but that's just not what happened...

[TakeruK]

It's a little weird (based on the norms in my field) that the first author is not the same person that actually submits the paper to the journal.

 

Here is my advice:

 

1. Don't go around him and contact the editor. It's already submitted, let that part run its course.

 

2. When you get a chance to sit down with the prof, bring up this sentence again and ask him to explain why X and Y are different with respect to that graph (i.e. do it politely, frame it as a chance for him to educate you). Bring up your concerns as questions of understanding and see what happens. Maybe you were wrong and his statement was valid. Hopefully, if he was wrong, he would see your arguments against his statement. At some point here, you should reach an agreement on the wording (more meetings if necessary).

 

3. You will get a response back from the referees that may or may not address this point. Work with your prof to implement the change so that the statement is something that you, the prof, and the referee(s) agree with.

 

4. Revise and resubmit, and then...hopefully acceptance and publication!!

 

Again, to reiterate: first, don't go around your supervisor!! second, the original submitted version is not the final version of the paper--you have a ton of time between the referee's report and the resubmit deadline to work this statement out with your supervisor. There's no need to be hasty and try to contact the editor etc. I sympathize with the frustration and the alarmed reaction (I would too, if I was first author on a statement I felt was wrong), but there is time to work it all out in a way that benefits everyone

[Taeyers]

If it's relevant, this is already the revised version, but this statement appears in a newly added paragraph in the revision, and the figure was also changed per reviewer request. This makes me worry that since we're already making changes that reviewers asked for previously, they won't be as thorough in looking for additional revisions. 

 

And truth be told, I'm not naive or uninformed about his work in the paper at all, so I don't think I could convincingly pass off my inquiry as asking him to educate me on something I don't know much about. He and I have butted heads before because I stand up to him on things that he wants to gloss over but really shouldn't, so that's why I expect him to be readily defensive and argumentative if I bring it up. Not to mention that there were a couple other things in the paper that he tried to misrepresent, but changed when I pointed them out, so he knows that I'm not just a confused baby scientist, and I know that gently bending the truth when describing data is something he does try to do. 

 

I'm glad you agree that going behind his back is extreme. That was my gut reaction too, but since I've never been in a remotely similar position before, I don't have any reference point for an appropriate course of action. 

Edited November 4, 2014 by Taeyers

[Eigen]

Just to weigh in, for biomedical sciences/chemistry/biology it's common for the first author to have little say in the final content of the paper. The PI is always last, and everyone considers them ultimately responsible for the content.

They are also likely the ones making the final submission and the final call in the conclusions the paper and data draw.

[TakeruK]

Good point -- I should say that my advice comes from my field's perspective where the author order for a student-led project is usually:

 

Student, then Supervisor(s), then Collaborators in order of their contribution to the work (usually; sometimes in ABC order if the list is really long).

 

In addition, the first author / student is listed as the corresponding author, does all of the direct communication with referees etc and is ultimately responsible for implementing (or not implementing) their coauthors' suggestions. However, in reality, the student should seek approval of their supervisor for all content. When I say "responsible for", I mean that it would not be acceptable for a student who is first author to say something like "I don't know" or "I don't agree with the reasoning" when questioned about the content of the paper. This is slightly tempered by the fact that people know that academics don't agree 100% on everything and they would know the student's writing would be strongly influenced by their supervisor.

 

So, please frame my above advice with that description of how my field operates. And also here is further advice, based on the extra information (i.e. this is a resubmit and it's already very clear you and him do not agree on the wording of this statement):

 

First, I think you should make sure you know where you stand scientifically on this statement. You must absolutely be certain that you are right. If you are 100% certain, then you now really have two choices:

 

1. You can accept your advisor's wording and accept his statement as something that is true and acceptable to you. In the end, paper writing and data interpreting is subjective and sometimes there are more than one way to word something so that it's not wrong. It might not be the interpretation you agree with, but sometimes you will be "outranked" on this, and I think the student should yield to the PI in cases like this.

 

2. Or, if you find this statement so reprehensible that you cannot bear to put your name on this paper, then you should exercise your right to remove yourself from this paper. The professional thing to do would be to let the prof know that you don't want to be on this paper any more and give him permission to publish your work without your name on it. 

 

However, this second option is basically almost as extreme as going behind their back and would probably hurt you in the future. It's a terrible thing to do but technically your right if you feel that strongly against putting your name on this. And if this is a case where you feel the statement is academic dishonesty that will get you into more trouble later on then this might be a viable option. This second option is still less extreme than going behind your PI's back because at least in this case, it's a disagreement between you and him, instead of including the journal editors/staff. I would reserve this option for use only if you are somehow certain that keeping your name on this paper is a bad thing.

 

In your case, it just sounds like you and your PI have different interpretations of the meaning of the data and you both want to express your interpretation. Collaborating is a compromise and it sounds like he does consider your thoughts and implemented some changes. It also sounds like, from Eigen's post, that your field is one where the last author is expected to be the "voice" behind the manuscript so that you might not have to be expected to agree with every statement. In my opinion, I think this is one case where you have to recognize that there are more than one way to present things and sometimes you have to compromise and accept what your coauthor/PI prefers.

[fuzzylogician]

If you are in a field where the final word and the responsibility for the findings go to the PI, not the first author, then I think you will ultimately have to defer to his interpretation of the results, or else ask to have your name removed from the paper if you really think something terrible is going on (but you'd want to be extra sure because nothing good can come of telling your advisor you want off a paper because you think there is a misrepresentation of facts in it). At this point, as others have said, the paper has been resubmitted and I would just wait to see what the reviewers say before doing anything else. In the meanwhile, though, I think it's legitimate to ask your advisor for an explanation of the graph and its interpretation. You don't need to feign ignorance, but this is a point where you clearly are disagreeing with him or missing something and it's important that you understand what that is. I'd bring it up as exactly that - a comprehension question about something you are currently confused about. 

[St Andrews Lynx]

Is the data from this particular figure central to the main conclusions of the paper? Does it prove or disprove the central hypothesis, or merely act as one of several pieces of evidence? That determines how serious the misrepresentation of the data is.

 

I agree that this isn't particularly ethical...but I think that over-interpreting a graph is less unethical than say, fabricating a graph with made-up data or photoshopping Western blots. At least anybody who reads the paper will be able to see for themselves that the conclusion drawn is not well founded. 

 

It's a tricky situation for you. My advice would be to echo what others have said - don't complain to the editors, wait to hear back from the reviewers. If the PI is just going to get defensive when you talk to him about it...avoid doing so. Archive the "query" email you sent him. At least that way you will be on record as having disagreed with what he did. Find another more morally-scrupulous PI to do your PhD with, and publish lots of good papers.

[Taeyers]

Thanks everyone, that's all very helpful. 

 

I will clarify just a bit further that in my specific subfield, I have always understood the claims to be an equal responsibility of both the first author and the last (the PI), but a paper is typically assumed to be primarily written by the first author and then scrutinized and edited by the PI. This assumption is easy enough to correct, and I don't think anyone would be surprised to find out that a first author didn't do much writing but the majority of the lab work, but the default assumption is that the first author is heavily responsible for the writing. 

 

Anecdotally, this PI's postdoc tells me that she writes most of the paper, but then he rewrites basically everything she wrote. She also writes the drafts of the revisions and rebuttal letters. The PI whose lab I will probably join after rotations says that you have to have written at least the first solid draft to be a first author on a paper in his lab. 

 

Is the data from this particular figure central to the main conclusions of the paper? Does it prove or disprove the central hypothesis, or merely act as one of several pieces of evidence? That determines how serious the misrepresentation of the data is.

 

I agree that this isn't particularly ethical...but I think that over-interpreting a graph is less unethical than say, fabricating a graph with made-up data or photoshopping Western blots. At least anybody who reads the paper will be able to see for themselves that the conclusion drawn is not well founded. 

This is especially helpful, and "over-interpreting a graph" is the wording I was looking for. This graph is not a central point in the paper, and is more along the lines of supporting observations. I think I would have freaked out a little if he "over-interpreted" something critically important...

 

Everyone's advice basically reinforced what I already suspected: that it's not good, but not abhorrently bad either, and I should get a grip and accept that as his (somewhat questionable) choice. I will still cross my fingers that reviewers will catch it, and I will probably bring it up to him at some point, but I'm not going to burn bridges over it. Thanks guys. 

[Eigen]

I think especially given that you are not planning on this being a long-term relationship (i.e., you don't plan on them being your thesis advisor), I'd just let it go. 

 

My experience has been that the PI will likely be the corresponding author, not you- is that correct for this manuscript?

 

I have had friend who have been in the position of either staying quiet or asking to have their names removed from the paper, and neither is a good option to have. In your case, I think the issue is a lot less egregious than theirs were with respect to research ethics, so I'd just let their interpretation as last and corresponding author stand, as they're the ones who will have to defend it primarily. 

[Taeyers]

My experience has been that the PI will likely be the corresponding author, not you- is that correct for this manuscript?

 

That's right, the PI is the corresponding author. That does remove a good chunk of the responsibility from my lap. Thanks!