Print

---

Hi there, 

I'm working with EEG data from 20 subjects, analysed with a 2*2*2 ANOVA. My main result is from one of the two-way interactions, where I averaged across the 'high' and 'low' conditions. 

Reviewers now asked that I compare the strength of that interaction for the 'high' condition and the 'low' condition. Although I can do this with a 3-way interaction, I wasn't sure this is the most robust way of testing the question. 

I re-analysed the data using two 2*2 ANOVAs, one for the 'high' and one for the 'low' condition, and produced the relevant interaction contrast in each. I can see that the interaction is significant in the 'high' condition but not in the 'low' condition. I planned to compare the values of these contrasts using a t-test outside SPM - does this sound sensible? 

But... how do I extract these values for each subject, namely, get a value for each subject from a specific location of an image file - ideally collapsed across all pixels in the cluster?

With fMRI data I used spm_get_data to extract beta values from a specific cluster by first defining a cluster VOI and then using a script, adapted from one Guillaume wrote, that looked something like this... 

vx=xSPM.iM*[xY.XYZmm;ones(1,size(xY.XYZmm,2))];
vx=vx(1:3,:);   

for n=1:20
for j=1:size(vx,2) %looping over all voxels in defied cluster
     my_vx(j)            = spm_get_data(SPM.xCon(1).Vcon,vx(:,j));
end
out(n)=mean(my_vx)
end

But I don't know how to adapt this to EEG because I no longer have a separate SPM for each subject. 

Thanks in advance for your time, 
Deborah