 |
 |
On Wed, Jan 22, 2014 at 4:05 AM, Gabriel Go.Es. <[log in to unmask]>wrote: > Dear SPM experts, > > > Sorry for such basic questions. I'm new to the longitudinal design and > analysis, and I have some doubts I would like to clear. > > I have two groups of 12 subjects each and a follow-up MRI of two years for > each subject. > Factor 1 (group) : 2 levels (Gp1 and Gp2) > Factor 2 (longitudinal) : 2 levels (year1 and year2) > > A) As I undesrtood I have to set a full-factorial 2x2 anova (Group x > Follow-up MRI as above), but I've seen in some faqs that I actually have to > set up a third factor "subjects" which is the number of subjects in my > study, but as I have never set a full factorial design, I'm not sure on how > to set this third factor. > You need to use the flexible factorial design and add subject as a factor. It can't be done in the full factorial design. > > B) How and which would be the constrasts for the desing in A? I mean, to > test the effect of the groups and/or the differences in year1 and year2. > You can not test for the effect of group in a repeated-measures design using contrasts in the GLM. You can only test the difference in year1 and year2 or the interaction between group and year. > > C) Is it possible to introduce nuisance variables i.e. age in year1 and > age in year two? > Not in a repeated-measures framework using the GLM. > > D) how do I correlate any of the tested effects with a variable of > interest? > I would subtract year1 from year2, this will create a difference image and eliminate the repeated measurements. Then you can use a two-sample t-test with a covariate. > > > > Many thanks in advanced, > Gabriel >