Several potential problems. It makes the statistics non-gaussian and the abs(ces) becomes noise dependent (and so there will be a noise dependent bias). This makes interpretation more difficult (eg, is a difference between groups a difference in ces or a difference in noise?). The non-gaussianity could be addressed with a permutation test.
doug
On 01/08/2014 07:11 PM, shahin@nmr.mgh.harvard.edu wrote:
Exactly! I need ces and cesvar files for a second order analysis. More specifically, I need to subtract this absolute map from another map. Would you be more specific about what I should check when I am doing this second order analysis?
You can do it to the sig file (the p-value already assumes a two-sided test). Why do you want to do it with the ces and cesvar files? If you are planning to use the abs(ces) for a higher level analysis then you need to be very careful as you can get some unintended effects doug
On 01/07/2014 03:46 PM, SHAHIN NASR wrote:
Hi, I have two related questions. I have two conditions (e.g. C1 and C2) and I want to generate a map that shows all voxels that can differentiate C1 from C2 _irrespective_ of whether C1>C2 or C1<C2. I think it can be done easily by:
mris_calc -o sig.nii abs_sig.nii abs
If I am right, then:
- Can I apply the same command to ces.nii file to have the absolute
value of the effect size (i.e. abs(C1-C2 effect size)) or should I consider other factors? 2) Do I need to apply the same routine to cesvar?
Thanks
Shahin Nasr
PhD in Cognitive Neuroscience Martinos Imaging Center, MGH Harvard Medical School
-- Douglas N. Greve, Ph.D. MGH-NMR Center greve@nmr.mgh.harvard.edu Phone Number: 617-724-2358 Fax: 617-726-7422
Bugs: surfer.nmr.mgh.harvard.edu/fswiki/BugReporting FileDrop: https://gate.nmr.mgh.harvard.edu/filedrop2 www.nmr.mgh.harvard.edu/facility/filedrop/index.html Outgoing: ftp://surfer.nmr.mgh.harvard.edu/transfer/outgoing/flat/greve/