Then you'd be better off looking on www.findaphd.com than posting on the forums.

Good luck :)

That sounds horrible. I don't really know what to say other than I think you did the right thing with the fellowship application under the circumstances.

Though it must have been unpleasant, at least you had witnesses in the car to see how unreasonable she was being. Not that it solves anything but sometimes it's reassuring to have that.

I hope that your time with family refreshes you and gives you some respite from your supervisor.

======= Date Modified 12 Jul 2011 23:26:01 =======
I agree with Keenbean. Better to ask a clear question, measure the variables well, and use a simple (and appropriate) analysis, than go for something fancy which the data doesn't actually support and makes you lose the question in a see of stats. Your choice of methods should be determined entirely by the nature of the question you are asking and the characteristics of the key variables, so asking "what are the more complex stats that can be used with data" isn't going to get you much useful advice without further detail - and it's questionable whether you should do it anyway.

Have you sought any advice from your dissertation supervisor as to suggested analyses? If not, might be advisable. I'd also check the guidelines/marking scheme for the dissertation to make sure you're not concentrating your efforts in the wrong place.

Is there a course tutor/personal tutor you have access to? Arrange a meeting with them asap and raise your concerns about completing in time. They may be able to arrange an extension and/or identify a different project angle or supervisor. But don't leave it too long before doing this.

There are indeed ways to handle skews, and legitimate reasons for them occurring. But if, as you say, it's arisen because of your approach to data collection, and if all it would require you to do to rectify this is a bit of targeted marketing of your survey, then I can't really see a reason not to make the effort, unless you are very short of time. Otherwise you may be laying yourself open to avoidable criticism at a later point.

Unless you're asking a female-focussed question then such a massive gender imbalance in your responses will skew your results. I would make a concerted effort to redress this. Or change your research focus to just women. You may not succeed, but you should definitely try or else you're opening yourself up to too much methodological criticism.

I'm sorry to hear about the lack of support. I would recommend you stick to your plan and focus on the PhD. That's the key thing and dividing your attention without support from her is going to lead to too much stress.

I hope a good solution to this is found, but for now just make the PhD your priority.

All the best

Does your sup realise how close to going out the window the application is? I would hope given the circumstances she could rearrange things just this once. Is there anyone you could get to lean on her to give you a meeting (grad tutor etc)?

Good luck with everything

I was a massive Firefox enthusiast since converting from IE, but when I downloaded version 4 I was very disappointed with the changes, to the extent that I uninstalled and reverted to v 3.17. I don't know whether Firefox 5 is any better - I'm nervous about going there!

======= Date Modified 05 Jul 2011 08:34:51 =======
Excel:

First column have the age in whole years. Second column have the months of age that are in addition to the years. Third column write a formula to divide column 2 by 12. Fourth column write a formula to add column one to column 3.

So for someone who was 8 years 6 months, then someone who was 3 years 3 months it would read:

8 / 6 / 0.5 / 8.5
3 / 3 / 0.25 / 3.25

If you have a lot of a data you don't want to be entering the years and months manually - it's too time consuming and you might make errors. So do get advice from someone in your dept about handling the data.

Whether or not that is a big enough sample size depends on what you're measuring and how much of a difference you're hoping to find between the (controls and experimental) groups.

It's hard to give you exact advice via a forum. I suggest that you try to talk to a statistician in your department who will be able to look at the whole picture with you.

======= Date Modified 04 Jul 2011 22:35:36 =======
1. The months issue:

You are correct that it would be wrong to use 0.7 for 7 months. There are two possible ways of doing it:
(a) As you suggested, calculate everyone's age in just months, such someone who is exactly 15 would be 180 months. I would suggest that you convert back to years when reporting your findings.
or
(b) for the number of months beyond a certain year, divide the number of months by 12, then add the decimal on to the number of whole years. So for someone who is 8 years and 3 months old, calculate 3/12=0.25 and then add that to 8 i.e. that person is 8.25 years old. You can just do those month calcs once, and make a note of them so you can just add on the right decimal each time. But really you shouldn't be doing it manually - it would be easy to set up a batch calculation in Excel or another programme. Ask if you're not sure how to do this.

2. Missing data:
What percentage of your participants have no age data? If it's a couple you could consider excluding them from the analysis. If it's a lot then to do so could introduce errors such as bias in your findings. In such a case you might want to consider something called imputation - talk to a statistician about this. Here's a brief intro to some of the issues around missing data: http://en.wikipedia.org/wiki/Missing_data

It wouldn't be polite, no. It would leave the organisers with an empty slot to fill at short notice. Also, in giving you a slot they have probably turned other people down. So you really should use this opportunity.

But it's not uncommon for people to change from the abstract that they submitted to talk about something different. If you want to take a seriously different angle you should probably let them know in case they have a chance to change the conference materials before they are printed. But otherwise the suggestion others have made of presenting 'work in progress' or a methodological discussion sounds sensible.

I assumed all the Govt research councils had the same funding rates. The amount you've quoted is what the MRC pay outside London - hopefully the EPSRC has highger rates for inside London too:
http://www.mrc.ac.uk/Fundingopportunities/Applicanthandbook/Studentships/Stipend/index.htm