Stats query

Kevster

Yo,

I have some data that I want to analyse statistically. The data is related to cancer patients who were treated with a certain drug. I have data for each from before and after the treatment. Should I just use a simple 2-sample t-Test (assuming equal variances) for this?

There's another thing too, in that the time difference between sample taking is different between each patient. Could I use this as the 'random factor' in an ANOVA, right?

Don't worry about this if you're too busy.

Take care,
kevin

nothing

Hire a statistician. Even just to give you some proper advice. It's hard to give statistical advice without full copies of data, how it was collected etc etc.

Kevster

Thanks for the response. I used to be quite good at stats but i left all of my notes back home in Ireland. Plus, as you implied, it'd be difficult to give solid advice without seeing the data for yourself. there is actually a lot of variability.

As money is toght, hiring is out of the Q! lol - I'll continue at it and ask around in my lab.

Kevin

LeixlipRed

You need to use a paired sample t-test, not 2 sample. The samples here are not independent. Look here. I know nothing about ANOVA though, sorry!

Kevster

You're right - thanks dude. The time-lapsed samples I have are each from the same patient (before and after drug treatment), and are therefore not independent (being instead 'paired'). There is a whole host of other factors underlying the data though, so i need a good long think about it.

For example, age, tumour size, menopause status, cell-surface receptor status, etc.

Thanks guys,

Kevin

LeixlipRed

Maybe try posting in the researcher forum? Those guys would probably know a lot more about the specifics.

Kevster

I didn't even know there was one! I can see where it is now though. i've actually figured out the battery of stats tests I can run on my data, so i'm alright for now. I might hang around on the researcher's forum anyway though.

Thanks
Kevin