Dear FreeSurfer experts,

 I would appreciate a confirmation regarding correctness of my approach.

I am trying to run an LME model for three groups (group1=controls, group2 and group3) and two time points.

I followed the tutorial on https://surfer.nmr.mgh.harvard.edu/fswiki/LinearMixedEffectsModels

and adapted the four group example. However, I am not 100% sure if my adaption is correct and would appreciate your input.

 

The design matrix for the model contains the following columns:

 

1)   Intercept

2)   time (tij)

3)   one for Group2 and zero otherwise

4)   column 3) .* time

5)   one for Group3 and zero otherwise 

6)   column 5) .* time

 

To test the null hypothesis of no group differences in the rate of change over time among the three groups I applied the following contrast:

 0     0     0     1     0     0    

 0     0     0    -1     0     1  

Is this contrast correct?

In a second step, I would use the additional three contrasts to identify which groups were different:

0 0 0 1 0 0   and 0 0 0 0 0 1 and 0 0 0 -1 0 0 1

 

I really appreciate your help!

Best wishes


Sarah