 |
 |
Hello FSL list, We have two groups for our fMRI study, controls and patients. We found group differences, and also found a correlation for trauma scores and BOLD responses; and we also found some clinical measurements were not matched between the two groups for depression and anxiety. First, for higher analysis we want the main effects for trauma. We put all the subjects, patients and controls for EV1, and add all trauma scores for EV2, orthogonalise EV2 wrt EV1, which means demean EV2, then we can get negative(-1 for EV2) and positive (+1 for EV2) correlation between trauma and BOLD respone in all subjects. Is this correct? We also want the corrrelation between trauma and BOLD response controled by depression and anxiety. Most part we need to do is similar with the previous paragraph's correlation, this time we need to add two more EVs, EV3 for depression, EV4 for anxiety, we need orthogonalise EV3 and EV4 wrt EV2, then negative(-1 for EV2) and positive (+1 for EV2) correlation between trauma and BOLD respone has exclude the influence of depression and anxiety. Is this correct? Next we want the interaction between group and trauma. We put patients and controls for EV1, EV2 and add trauma scores for EV3, EV4 for the two groups, orthogonalise patients trauma wrt patients mean EV, orthogonalise controls trauma wrt controls mean EV. We add two F-tests. F1 for EV3+1 and EV4-1; F2 for EV3-1 and EV4+1; Now I a not sure if we will get two different results for the interaction. Maybe they will be same. Is this correct? Thanks for your advise, Hongyu