Should I report this error on a published paper to my PI?

[ScienceGiraffe]

So, I am a second year student in science. I am taking over a project from a senior student who is graduating this year. over the course of learning and training on the particular setup for this project, i noticed one of the testing parameters was set wrong. the parameter could be based on reference 1 or reference 2. the senior graduate student picked the wrong reference to set the parameter for the test. I would very much like to give the senior graduate student the benefit of the doubt - maybe she was right to pick the reference that she did, however, i am 90% convinced that is not the case.

 

for argument sake though, let's just assume i was right and she was wrong. will you bring this up to the senior student and the PI? before you all jump on me saying that i must speak up, there are couple papers publish with this. it won't disprove anything, however, the performance will not be as good as it claims. Also the senior student is pretty much half way out the door and she is generally a very nice person. i hate to tarnish her work...

 

i hope i am wrong as well...but for a situation like this, what would you do?

Edited December 19, 2014 by ScienceGiraffe

[dr. t]

I would bring it up to the student, and I wouldn't start off with the assertion they're wrong. Try the soft sell - "Hey, why did you pick X, wouldn't Y work better?" The student's reaction will determine how to proceed from there.

[TakeruK]

I agree with telkanuru. First, talk to the student and asked them why they chose to do things the way they did. If their answer satisfies you, then great. If not, and asking for more clarification does not work, you could consider asking the PI about it. But frame it as "I'm confused why X was done" instead of "I think your student did something wrong by doing X". 

 

You should also keep in mind that there might not be a "right" or "wrong" answer. I know that in my field, there are many cases where the field is not yet in full agreement of which value to use for some parameters. So, some people choose X and some choose Y. It doesn't matter as long as the scientist is explicit about their assumptions. 

 

Also, a very important part of scientific research is the ability to adapt to change. If this paper was written awhile ago, perhaps back then using X was the right/popular thing but in recent years, more evidence has shown Y to be the better choice. This doesn't mean all the old papers that used X are completely wrong and it doesn't always mean that the work put into the old papers are wasted/tarnished. Scientific/logical arguments are still sound arguments even if the premises (assumptions) are later proven wrong. 

[ScienceGiraffe]

Thank you for the respond guys, but please allow me to think couple steps ahead here: i will ask the senior graduate student about why she picked reference 1 instead of reference 2 in the most casual way possible. we will have a small civilized discussion about it, and then in a 90% odd, i will win the discussion. then there will be a silent awkward momemnt because she realized she needs to publish correction for her papers now. once i bring it up, the PI will eventually have to hear about it. he may not go as far as to stop the senior graduate student's graduation plan (i hope not). but his recommendation letter for her postdoc will probably reflect this in some way. also deep down, the senior graduate student hates my gut now....

 

i am more than happy to accept i was wrong, but it's a very cut and dried mistake. once you see it, you will know reference 2 is right. if you don't see it, it's an honest human mistake. i hold intellectual integrity high on my priority list, it has been bothering me very much for not speaking up. it's just once i said it...i can't unsaid it. the ramification is not small for the senior student (her graduation plan, her postdoc, her chance of academic career)...so let me ask this instead, if i find an error on YOUR paper, what would you like me to do?

Edited December 19, 2014 by ScienceGiraffe

[TakeruK]

First of all, it is not about "winning" the discussion. Even if you are right, I would not see it as a "win" because this is not really a contest where you can win anything.

 

Second, to answer your question, if it were my paper, I would want you to ask me why I used the parameter from reference 1 instead of reference 2. If I was wrong, I would try to figure out if it makes a difference in the end result. If it does, then I would be sad, but I would rather redo it and have a corrected paper than keep something wrong in the literature. If I was right or it doesn't change any results, then it doesn't matter. Again, my old paper would still stand, because the paper argues "If X=ref 1, then my results".

 

Finally, scientific papers are sometimes wrong and it is really not a big deal. Someone else might point it out in their next paper, or perhaps the student herself will write a followup paper with more analysis along with a reanalysis of the old paper showing that changing X=ref 2 makes very little difference. 

 

So my advice is yes, you should ask the graduate student about it. But your attitude and tone will make all the difference. You should think about this as bringing it up as a fellow colleague and you're on their team, not because you want to prove that you are right or that you are smarter (which is what it sounds like when you use the term "win the discussion"; however, perhaps you just chose the wrong word there!). 

 

Intellectual integrity is very important in academia. But ultimately, I think of us as humans doing research, and our "humanity" is the most important. It is important to find out if a mistake has been made, but you can always do this in a supportive and good way, instead of an attacking way!

[ScienceGiraffe]

Thanks for the good reply TakeruK. I am most certainly not trying to win any discussion over the senior graduate student. that was a poor choice of words. I would very much like myself to be wrong. for a lack of better words, i want to "look out" for the senior student.

 

unfortunately the mistake will change the results of the paper. it's like "if pH=7, results will be this". however, the pH meter was actually not calibrated and reading 2 pH off. it will make the results not as dazzling as it claimed. And props to anyone who has the courage to own up to your mistake, that is really the right way to do it. but am i really going to do that to her while she is half way out the door?

[bsharpe269]

I think that it is very important to discuss this with the student. If I made a big mistake then I would much rather realize it when I was half way out the door than after I already left! If you know a mistake was made then I think it would be unethical to pretend you dont realize it.

[Vene]

Thanks for the good reply TakeruK. I am most certainly not trying to win any discussion over the senior graduate student. that was a poor choice of words. I would very much like myself to be wrong. for a lack of better words, i want to "look out" for the senior student.

 

unfortunately the mistake will change the results of the paper. it's like "if pH=7, results will be this". however, the pH meter was actually not calibrated and reading 2 pH off. it will make the results not as dazzling as it claimed. And props to anyone who has the courage to own up to your mistake, that is really the right way to do it. but am i really going to do that to her while she is half way out the door?

I would be shocked if such a thing prevented her from graduating. There are tons of mistakes in the literature, it happens.

[fuzzylogician]

First of all, mistakes happen all the time. They shouldn't stop someone from graduating. What you are describing is a situation where the result is not as pretty, not where her entire thesis is refuted. (And even then, people still graduate, they just won't get publications out of their thesis later on.) Second, even if there is a mistake, it's not clear that it needs to entail a retraction of the paper or even an erratum. If the error makes the result less pretty but the basic result still stands, it may be the case that no action is necessary. I think that in some cases, this is a fair decision. Third, if a correction is necessary, there are different ways to go about it; I might imagine that instead of retracting the paper, the student might be able to redo the analysis with the new reference and write a new paper based on that, and possibly some other expansion/addition/discussion. In any event, these are decisions that are basically none of your business--it's between the student and her PI. If she chooses not to do anything with it, you might be able to step in and redo the analysis and write a new paper showing how the result changes if this error is controlled for. Science is all about this kind of thing. So all of this is to say that the best course of action is to bring it up with the student, and to phrase it as a question and a learning opportunity, not as an accusation or an argument that you are winning. What she does with it as a first step is her business, and only if she decides to do nothing would I *consider* following up on it myself, and even then I'd do it carefully. 

[St Andrews Lynx]

Do you know what the role of the PI was in supervising her project and writing up the paper? It is usual in most fields that the PI will (at minimum) look over the drafts of any manuscript before it is submitted to a journal. 

 

My point is, if the PI has been involved supervising this particular project/checking the manuscript, there is a chance that they are aware that the grad student chose Reference 1 for their parameters and is OK with it. It isn't uncommon for research groups to select a set of parameters/data sampling sets that "boosts" their results - it isn't completely ethical...but people do it quite a lot. 

 

Personally, I'd ask the grad student why they chose Reference 1 over Reference 2. It's possible that you are missing something - maybe they've adjusted for what you see as an error elsewhere in their protocol. Or perhaps they first ran the experiments using Reference 2 and the parameters didn't work at all for their dataset (or they lacked the resources to use the Reference 2 protocols). If the grad student gives you a response that doesn't satisfy you...don't push the matter. 

[eragon]

Better inform the grad student. If there is a error, probably the reviewer will find that as well while reviewing the paper. So earlier he/she knows, the better.