Hi Gabriel,
This should have been posted with the message contained the GLM images -- not sure why it wasn't:
Thank you very much -- that explanation cleared things right up.
I have a followup question regarding my GLM set-up and how to go about regressing out covariates (age, gender).
I have 32 surgical patients, (25 responders, 7 non-responders), and I would like to do a unpaired group comparison between them.
My first GLM (GLM #1 below) design matrix placed all 32 patients in the same 'group' and separated them (responder and non-responders) by using different events (as specified in the unpaired group comparison wiki).
However, in the VBM practical, it was specified to use the GLM Wizard. When I select '2 group unpaired', there are actually two groups (one for each patient population) in the design matrix, but everything else stayed the same as the GLM #1 matrix. This is shown in GLM#2 below.
I am wondering which of these two is correct. Additionally, how should I go about regressing out covariates with two separate groups (as per GLM #2) in my design matrix (opposed to one group separated by different events), as this is not specified in the GLM Wiki.
Look forward to hearing back.
Regards,
Hayden
|