I am helping a STEM PhD student with a statistical approach they developed for a key thesis chapter. Unfortunately, the approach is fundamentally flawed and makes no statistical sense. (I'm a freelancer with a statistics PhD and first got this as a programming job before finding the issues.)

I've tried to explain why the approach won't work, but to no avail. The person keeps on looking for workarounds, which I know won't work either. The approach should be ditched entirely.

What's a reasonable solution in such cases? Should I suggest potential alternatives for the chapter—and I can think of a few? If so, how should I proceed ethically?

There is a related question, focusing on a midterm marking.

Added in response to comments:

The job began as statistical programming help, which I later discovered to be for the client’s PhD (as usual, the initial job details were scant). The student client is the one paying; I’m unaware of the funding source. My previous academic clients obtained university grants to cover the fees and noted my contributions, primarily code, in their thesis or paper acknowledgements. For me personally, correct execution is a matter of professional ethics. I also do not wish to abet violations of academic ethics.

  • 4
    Are you their advisor or otherwise have a say in accepting their thesis?
    – Buffy
    Commented Jun 1 at 14:10
  • 4
    You note that the student will need sign-off from their advisor for their wrong approach. So: what's the advisor's stance in this situation? Commented Jun 1 at 20:42
  • 8
    One possibility would be for them to post their approach at CrossValidated and solicit feedback. There are some statistical experts there. Perhaps they could sway your student. Commented Jun 1 at 20:44
  • 10
    The relationship s a little unclear. You're doing a statistical "programming job" for this person's PhD? Are you paid by the student, or someone else? Will you be an author? Is this a common arrangement? Is there a question of professional ethics if you execute on a flawed methodology? Commented Jun 2 at 1:35
  • 7
    – EvilSnack
    Commented Jun 4 at 2:50

5 Answers 5


Beyond suggesting that they speak with their advisor about it, you have probably done all that you can do. You can't force someone to take good advice.

If you suggest alternatives, I'd make the statements general. If you give minimal advice that they can follow up on it will benefit their learning. Don't do their work for them. Suggesting directions is the proper course, even for their advisor.


Edited to add: An advisor can and should certainly stop a student from taking a wrong direction. They have both the authority and responsibility to do so. They can also point to a better path, but they shouldn't provide solutions to the student's problems. They should let the student work out the solutions for themselves. That is also my advice to faculty in general when approached by students with questions. Let the student find their own insights.

The OP here doesn't have the authority, but should still respect the learning of the person by not doing their work for them.

  • 5
    +1 Though "Suggesting directions is the proper course, even for their advisor" is not enough for the advisor in case the student uses the wrong methodology. I am usually quite "laissez faire"-oriented, but it is the adviser's job to nip scientific malpractice in the bud. In the case here, as an advisor, I would make a very strong point not to do this. If that doesn't help, I would tell the student to keep me off the author list. If that is a persistent pattern, I would also suggest that the student gets an advisor whom they are prepared to listen to. OP, however, has none of these recourses. Commented Jun 1 at 15:48
  • @CaptainEmacs, I agree, actually, but saying "NO" is different from doing their work for them. Put them on the path, but don't walk them down it.
    – Buffy
    Commented Jun 1 at 16:25
  • I understand you now. Perhaps make that more unambiguous? Commented Jun 1 at 16:28
  • 4
    A slight complication: the student’s PhD is not in statistics, but that thesis part requires intricate and specialist stats. In any case, except for the most basic outlines of alternatives, I’ll require approval of their PhD supervisor. Commented Jun 1 at 19:34
  • 8
    It doesn't matter that the thesis isn't in statistics. If the error brings the results into question it is a serious issue.
    – Buffy
    Commented Jun 1 at 19:49

After a certain amount of pushback, the next natural question is, at what point will you remove yourself - and most importantly, your name - from this work? It appears as if this person is disinterested in taking the proper steps to rectify this matter. If that is the case, then I would say it is obviously time to part ways.

  • 4
    My current instinct is that the person created and got wedded to what they thought was a brilliant theory. Now it’s hard to come to terms that it’s not the case. And it’s also big efforts to change course. Commented Jun 2 at 17:06
  • 7
    Step away as soon as you're convinced beyond a reasonable doubt that you won't be able to change their mind (or that their supervisor won't force them to follow your advice) (you could tell the student that you're happy to resume the work if they change their mind, e.g. if their thesis examiners force them to drop the improper approach — although by that time it will probably be quite painful for the student ...)
    – Ben Bolker
    Commented Jun 2 at 21:34

It sounds like you have already done what you can to help the situation. It is not your job or responsibility to force a correction. If the method is flawed to the extent you say then the PhD advisor and/or committee should catch it and force a correction. Get paid for your work this far and disengage.


The simple approach is 'the customer is always right'. You gave them advise but they chose to ignore it. They are paying you so it is their problem, not yours.

Now you said you see it as problematic for your professional ethics if this methology is used as is. In that case I would recommend talking to the student and telling them you would like to talk to their advisor about the issue. A three-way talk is fine, a one-on-one with the advisor works as well.

First, it seems a bit unusual to me that a PhD student is paying someone external for statistics help. This could be fine, depends on the area but then the PhD advisor should definitely be aware of this and probably will also be the source of the money.

If the advisor is not involved with this this would be a major red flag for me.

If the advisor knows about your involvement they can respond to your advise and the students argument. Even if they don't have the statistics knowledge to decide the case they know that you have that technical knowledge, that is what you where hired for. So let the advisor arbitrate (and depending on the results decide whether you want to continue working with them).


The proper venues for expressing your doubts about a student's work is in supervision /progress meetings. If you're not part of those meetings, then all you can do is try to convince your PhD student client, or ask them to let you make your case in their supervision meetings.

Of course, there is always a venue for a scientific critique of their work: you could publish an article or commentary stating your case. If the consequences of the student publishing a flawed article are high, you could consider doing that.

One important thing to consider is that you might be wrong (I doubt it, but it's not impossible). If you believe that the approach is flawed, then you can withdraw as a coauthor or named contributor.

FYI, under the CRediT authorship model, programming software qualifies you for authorship, so check if the journal subscribes to that model!

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .