what is your contrast of interest? If you are interested in the age slopes, then you don't need to demean. If you want to use age as a nuisance variable, then you should first test for an interaction between age and group using the DODS model. If there is no interaction, then switch to a DOSS model (in which case you can/should remove the age mean). This is all the same regardless of VBM, thickness, etc
On 2/27/18 12:23 PM, C.P.E. Rollins wrote:
Dear Freesurfer developers,
I'm running a standard cortical thickness analysis between 2 patient groups (patients with schizophrenia with hallucinations (H) vs. patients with schizophrenia without hallucinations (NH)). I'm looking at the main effect of group (H vs. NH), while controlling for sex and age. I'm using default QDEC parameters (DODS, fwhm 10). I'm wondering whether I should demean age (ie., demeaned = age - mean age), as this has been recommended by FSL for VBM. I'm asking because whether or not I demean changes my results in QDEC. Moreover, I'm wondering whether it makes any difference to include age (demeaned or not) as a nuisance factor or a covariate. I've attached my qdec table (I'm not including both age and demeaned age in my model, I've just put them both in for viewing purposes. Age_MC is demeaned age). Thanks in advance, I hugely appreciate your insight.
Kindest regards, Colleen
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer