Well, neither thickness nor surface area are gaussian to begin with, so volume is definitely non-gaussian. I'm not sure how much difference that will make in terms of parametric statistics. You can do permutation using mri_glmfit-sim if you have an orthogonal design. If not, you can use PALM by Anderson Winkler.
On 6/15/17 9:46 AM, Lim, Lena wrote:
Dear Freesurfer Experts,
A review questioned the validity of examining volume in addition to cortical thickness and surface area within the same study. In particular, he commented that “The volume is the product of the thickness and area ...As the product of two Gaussian-distributed variables is not Gaussian-distributed, the volume is likely to violate assumptions of the statistical methods.” Could you please kindly advice on this?
Many thanks,
Lena
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer