Close Home Forum Sign up / Log in

Tricky situation

D

Some of the statistics I'm doing are beyond me. Time and again I've asked my supervisor for support (they designed the project), I get it but don't feel they're entirely sure about what they are doing. Some of the work has been picked up as wrong (after I've followed my supervisors advice) and when I've checked with my supervisor after revising things and got their OK it's still flawed.

There's no-one I can approach in my department. I'm reluctant to be associated with research I'm embarrassed by. Thankfully, I'll soon be able to break ties with my supervisor. I wish I'd dropped out just prior to submitting but obviously have to follow through on the corrections or else I'll fail.

Any advice?

L

Hi Delta,

Sorry to hear you're having problems with your stats, do you have a stats/mathematics department at your uni? Is it worth pinging them and asking for some advice? You're so close to the end now, so don't give up! The end is definitely in sight and you will be free of your supervisor before you know it!

We have a stats advisory service at our university (they pretty much disapprove of anyone doing their own stats) who are accessed through the graduate centre, I don't know if it's the same at your university?

Actually... having re-read your post... when you say you're embarrassed by your research and don't want to be associated with it, is this why you feel you can't approach anyone and ask for help?

Finally, and I know I probably shouldn't condone this approach as we'll all get smacked wrists but never underestimate what you can find through a good search engine! I'm sure there will be bods on this forum who will be able to help you out too, I'm not much of a statistician (I'm taking a refresher course at the moment as I've not done stats since undergrad)

Please don't give up - you are so nearly there!

B

If this has been listed as a correction in your thesis, then could you ask your internal examiner to be more precise about exactly what they want you to change and what analysis they want instead? If it's been phrased at all vaguely, it would strike me as a very legitimate question to ask. That might give you something more concrete to go looking for help with at the places LindaLou suggested.

L

I agree with Bewildered too, I think you would be justified in asking your examiner for clarification.

D

Many thanks to you both for responding. I'm not giving up, and I'm in good form, but while I have faith in my ability (aside from the statistics issue) I've no real faith in my supervisor. There's no stats help available to me within the department apart from my supervisor. It would be great to have a stats department to approach but there isn't one. I can't go into detail on the forum but I can't approach my internal as I feel they'll tell my supervisor. I feel terrible for saying this but I don't rate my supervisor as a researcher and they designed the project. I'll get something sorted. Thanks for the kind words and support.

======= Date Modified 26 Jan 2012 20:29:33 =======

Quote From delta:

Some of the statistics I'm doing are beyond me. Time and again I've asked my supervisor for support (they designed the project), I get it but don't feel they're entirely sure about what they are doing. Some of the work has been picked up as wrong (after I've followed my supervisors advice) and when I've checked with my supervisor after revising things and got their OK it's still flawed.

There's no-one I can approach in my department. I'm reluctant to be associated with research I'm embarrassed by. Thankfully, I'll soon be able to break ties with my supervisor. I wish I'd dropped out just prior to submitting but obviously have to follow through on the corrections or else I'll fail.

Any advice?


What type of analysis are they asking for? If it's probability of populations being different, are we talking normal distribution, student 't', analysis of variance, etc. I'm talking methodology, BTW.

Translated, I may be able to help a little as I've done some statistical techniques.

Ian (Mackem_Beefy)

F

Hi Delta, what field are you in? My PhD field is ecological statistics, and I also teach statistics in medicine + psych. I might be able to help, or at least tell you if what you have done is statistically valid...send me a PM if you like?

I'm in a somewhat similar situation to you, in that we don't have strong statisticians in my department, my advisor isn't particularly quantitative (plus I don't trust her because she makes stuff up when she doesn't know the answer! So. Frustrating!). We also don't have a statistics department. When I get stuck, I ask everyone I know whether they know anyone who is a stats person. Everyone knows *someone*, the *someone* usually can't help but in turn knows someone else that can. You'll get there!

Worst case...pay a consultant for an hour or two. So worth it. If you count your time as money, it works out kind of inexpensive :)

D

======= Date Modified 26 Jan 2012 21:10:42 =======
Thanks everyone. Yes, I want to learn and understand these (and don't have confidence in my supervisors statistical ability) and so I may consider getting someone to teach me so that I can be sure I understand what I'm supposed to be doing. Furry, you have your first star as a thanks. Ian, thank you for the offer, at some point in the past I must have contributed towards your stars.

D

It's only because my supervisors bands his PhD students names about that I have to finish this, otherwise I would leave it and accept a fail (I wouldn't accept an MPhil).

hey delta, I'd be happy to help - I teach stats too. Although prob mid-feb might be better. I self-taught all my stats (after falling asleep in lessons and not taking anything in) I got Andy Field's book and then from there have taught myself more advanced stuff. I find that if you have to do it and its YOUR project (rather than hypothetical data/scenarios) then its much easier to get to grips with.

D

Everyone has been really fantastic. I've knuckled down (working harder now than at any point during my PhD) and am getting to grips with it a bit more but may have questions.

Sneaks, that's a very kind offer but you've more important things to attend to and I wish you all the best.(up)

G

I had a less severe case of supervisors who are not experts on stats. I took an advanced stat course to refresh my knowledge on that topic. The key is to then look for papers with similar research problems and inspect their methodology. In my case I even found the 'holy grail': a set of methodology papers for my domain. When a peer reviewer has any stats remarks on my paper now, I find it's because he's not expert on stats and I can simply quote from and refer to state-of-the-art methodology papers. I really believe every research group should have a tailored 'methodology package' it gives to all who join it. 

21550