Hi,
I have a time series data for two variables and some confounding variables
and want to see the relationship between two variables by some sort of
modelling adjusting for the confounding variables. Just to explain a bit
more say I have daily data for variables X; Y and other counfounders for
five years (for example). I am interested in modelling Y on X (continuous)
adjsted for other confounders.
How do I calculate the power to detect any association between X and Y ?
This is basically to check that I have enough power with five years of data
to detect the effect of X on Y.
Most of power and sample size calculations that I have known is to compare
two (or more) different groups. Here I don't intend to divide data (five
years) into two (or more) groups rather the analysis will be done in a
single group to study the effect of X on Y.
Just to add a small question on the above, will the power to detect the
effect of X on Y depends on the values of X and Y. (I think no, but I may be
wrong)
Any direction to reference, suggestion is highly appreciated. Given the
recent discussion of the use of allstat, please send reply to me directly.
I'll summarize and post it to the group.
Thanks.
Biswas.
_________________________________________________________________
Hot chart ringtones and polyphonics. Go to
http://ninemsn.com.au/mobilemania/default.asp
|