Hi Xiaomin, I finally got around to looking at this, and this is a bug. The "cespct" is not actually a percent, rather it is a fraction of the mean functional. However, the "cesvarpct" is a percent, which makes the group ffx analysis with percent fail. I have fixed this in both dev and stable such that cespct is actually a percent, live on next update. Let me know if this does not fix your problem.

doug

xiaomin wrote:

Hi,

I am running the following command for group fixed factor and random factor
analysis with and without percent parameter on.

Here are the commands:
1) for the no percent:
isxconcat-sess -sf subjlist -df subjdir -analysis RotDepth -all-contrats
-hemis -o . -subject fsaverage
mri_glmfit --y rh.ces.nii --osgm --glmdir rh.ffx.osgm --nii --mask
../rh.mash.nii --yffxvar rh.cesvar.nii --ffxdofdat ../ffxdof.dat --surf
fsaverage rh

2) for the percent:
isxconcat-sess -sf subjlist -df subjdir -analysis RotDepth -all-contrats
-percent -hemis -o . -subject fsaverage
mri_glmfit --y rh.cespct.nii --osgm --glmdir pct.rh.ffx.osgm --nii --mask
../rh.mash.nii --yffxvar rh.cesvarpct.nii --ffxdofdat ../ffxdof.dat --surf
fsaverage rh

However, the significance between percent and no percent is inconsistent.
For testing if the data has problem, I run it with rfx design.  The
significance between percent and no percent is almost same. (see attached
graph)  Second, the significance with percent parameter for the fixed factor
analysis is very small, which is abnormal.  We need to use ffx significant
map to define a ROI.

The data is here:

/autofs/space/yneuron_001/users/xiaomin/projects/FaceRotation/subjects/Rot_D
epth.


Xiaomin Yue 
Research Fellow
NMR center, MGH

-- 
Douglas N. Greve, Ph.D.
MGH-NMR Center
greve@nmr.mgh.harvard.edu
Phone Number: 617-724-2358 
Fax: 617-726-7422

In order to help us help you, please follow the steps in:
surfer.nmr.mgh.harvard.edu/fswiki/BugReporting