 |
 |
Hi Jeanette, In this case While I was configuring the GLM model through the GLM GUI, the wizard provided options to choose the single group average model, or two groups, unpaired model. Since I have a patients group and a control group, Shouldn't I be choosing the two groups unpaired model and then add EV3 and EV4 for the age and gender. In this case the contrasts would become. Group A > Group B 1 -1 0 0 Group B > Group A -1 1 0 0 group A mean 1 0 0 0 group B mean 0 1 0 0 where my design matrix would be group ( 1; 2 ) EV1 ( 1 control, 0 patient ) EV2 ( 0 control; 1 patient ) EV3 ( age ) EV4 ( gender ) and then feed these design matrix and contrast matrix to the randomize script. Just want to confirm, before reaching a conclusion. Regards Aman ----- Original Message ----- From: "Jeanette Mumford" <[log in to unmask]> To: [log in to unmask] Sent: Thursday, June 28, 2012 1:12:25 PM Subject: Re: [FSL] add covariates to TBSS statistical comparison Hi, This is just a regular old multiple linear regression (any regression text should be very helpful), so all you do is add a column with your age values and a column for gender. If EV1=1 patients, 0 controls, EV2=0 patients, 1 controls, EV3=age and EV4=1 female, 0 male your [1 -1 0 0] contrast is patients-controls, controlling for age and gender. Cheers, Jeanette On Thu, Jun 28, 2012 at 2:52 PM, Amanmeet Garg < [log in to unmask] > wrote: Hi, I have a FA images dataset for the Patients + controls groups. I can conduct the regular group wise comparison through the instructions in the TBSS page, but there is no clear instruction to how to add covariates to the comparison. I want to add AGE and GENDER as covariates to see if they have any effect on the significant difference otherwise observed in the FA values in the group comparison without the covariates. Reading through the Forum has not helped me at all. Any guidance will be much appreciated. Regards Aman