Hi Folks
I have two separate questions about the use of 3dDeconvolve in the case of event related data (I swear I have tried to read the manual and the afni board a lot before posting this! sorry if the answers are trivial...).
1) First, as I am running my script (matured with love and time) on individual datasets, I get this frightening message:
** WARNING: Problems with the X matrix columns:
* Column 12 is all zeros: SVD on => will be kept
* Column 17 is all zeros: SVD on => will be kept
* Column 22 is all zeros: SVD on => will be kept
* Column 27 is all zeros: SVD on => will be kept
* Column 32 is all zeros: SVD on => will be kept
* Column 37 is all zeros: SVD on => will be kept
* Column 42 is all zeros: SVD on => will be kept
* Column 47 is all zeros: SVD on => will be kept
* Column 52 is all zeros: SVD on => will be kept
* Column 57 is all zeros: SVD on => will be kept
* Column 62 is all zeros: SVD on => will be kept
* Column 67 is all zeros: SVD on => will be kept
* Column 72 is all zeros: SVD on => will be kept
* Column 77 is all zeros: SVD on => will be kept
* Column 82 is all zeros: SVD on => will be kept
* Column 87 is all zeros: SVD on => will be kept
* Column 92 is all zeros: SVD on => will be kept
* Column 97 is all zeros: SVD on => will be kept
* Column 102 is all zeros: SVD on => will be kept
* Column 107 is all zeros: SVD on => will be kept
* Column 112 is all zeros: SVD on => will be kept
* Column 117 is all zeros: SVD on => will be kept
* Column 122 is all zeros: SVD on => will be kept
* Column 127 is all zeros: SVD on => will be kept
++ Matrix condition [X]: 1.21384e+08 ** BEWARE **
Do I need to be seriously concerned about those warnings? Are they telling me I didn’t design my analysis pathway correctly?
2) Also, I would like to run group analyses on my results, both within and between group. Thus I followed the logic of normalizing my data before running 3dDeconvolve, using commands like these:
3dTstat -prefix mean_EA-condA EA-condA+tlrc
3dcalc -a EA-condA+tlrc \
-b mean_EA-condA+tlrc \
-c EA_full_mask+tlrc \
-expr "((a/b)*100)*c" \
-prefix EA-condA_scaled
I believe the result is that after 3dDeconvolve I get scaled values directly usable in a ttest. Is this right?
Finally, should I run my ttests on the ‘coef’ subbrick of my contrasts of interest? Is that the right thing to do? (statistically speaking) or is there another strategy to adopt?
(for example, I am interested in regions activated in the group for the contrast stimulus A – stimulus B, and I am also interested in comparing activations in response to that contrast between two conditions or two groups of subjects).
Thank you very much for your input!!
Barbara