 |
 |
Hi Katie, Yes, design and contrasts look right. All the best, Anderson On 5 May 2015 at 21:03, Katie Bessette <[log in to unmask]> wrote: > Hi all, > > I've read through all the archives and user guides and haven't found a > direct answer to this particular question: > > I'm analyzing white matter of a patient vs healthy population, looking at > a behavioral continuous covariate interaction of interest. I want to > control for age (as if a nuisance covariate). My assumptions are that my > FEAT design model would be a blend of both the Two-Group difference > adjusted for covariate and the Two groups with continuous covariate > interaction as described in the GLM user guide. So EV1 and EV2 would be my > two populations, EV3 and EV4 would be the all-subjects-demeaned behavioral > data separated by group, and EV5 would be the all-subjects-demeaned age. > Will this model correctly give me results that I can interpret as a > difference between the groups' white matter in the slopes of the behavioral > data, even after controlling for age? Below is an example of my design in > FEAT. > > EVs: > (patient) (healthy) (pt_bx) (hc_bx) (age) > EV1 EV2 EV3 EV4 EV5 > 1 0 .852 0 13 > 1 0 .359 0 16 > 1 0 .921 0 11 > 0 1 0 .933 15 > 0 1 0 .241 14 > 0 1 0 .542 18 > > Contrasts: > EV1 EV2 EV3 EV4 EV5 > C1 pt_slope > hc_slope 0 0 1 -1 0 > C2 pt_slope < hc_slope 0 0 -1 1 0 > > > Thank you for your help in advance! > > Katie Bessette > > *Clinical Neuroscience & Development Lab* > > *Olin Neuropsychiatry Research Center* > > *Institute of Living / Hartford Hospital* > > *200 Retreat Ave, Hartford, CT 06106* >