Dear Andrea,
> Actually I am performing an experiment with fMRI, builds up of two
> different runs: Run1 (task A) and Run2 (task B), after having
> realigned, normalized and smoothed my data, I started with the
> statistical procedure:
> If I analyse the run 1 and the run2 separately, and than I define the
> contrast:
>
> activation - rest
>
> -1 1
>
> where trial 1 was the rest condition and trial 2 the activated
> condition.
>
> The results obtained are not the same as when I analyse both runs
> together and define the contrast:
>
> -1 1 0 0 for the first run
>
> and
>
> 0 0 -1 1 for the second run
>
> So my two questions are:
>
> 1) which is the right way to analyse my data?
Depends on what you're trying to do... If you want to analyze both runs
separately, one can analyze them separately. If you want to compare
conditions A and B or make some inference about all the data, you should
include both runs in one analysis.
>
> 2) how do you explain these differences that I got in my final
> results?
If you analyze both runs with one model, you are using all available
scans. This gives you more degrees of freedom for the estimate of the
error variance and changes your statistical results as compared to the
two single analyses.
Stefan
--
Stefan Kiebel
Functional Imaging Laboratory
Wellcome Dept. of Cognitive Neurology
12 Queen Square
WC1N 3BG London, UK
Tel.: +44-(0)20-7833-7478
FAX : -7813-1420
email: [log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|