On 9 June 2011 15:30, Edward H Patzelt
<[log in to unmask]> wrote:
Eugene -
I selected it at the 1st level analysis which are the individual blocks for each subject. So you're saying for example that if I have 2 designs: one with z>0 checked at the 1st level, 1 without out; they will be the exact same analysis at the group level if z>0 is unchecked for both?
Yes - this only affects thresholding
So if I do check z>0 at the group level I will get activations only associated with positive deflections. Subsequently if I take this mask and pull the "pe" from the fixed effects subject mean it will be possible to get a negative "pe" because statistically it is doing a group mean at the group level analysis?
Yes.
Is there a way to look at only positive deflections when pulling the % signal change from the subject fixed effect means while masking with a contrast from a significant group result?
I can't really think of a scenario where this is a sensible thing to do - if you are interested showing the range of responses at the subject level, you want to show them all, positive or negative..
Cheers,
Eugene
Thanks,
E