Dear FSL experts,
I am running a higher level analysis and wanted to look at the interaction between two variables on task-related activity. The variables are stress and genotype. I have one score of each per subject. I set up the design as follows. I multiplied stressXgenotype to produce a composite variable and entered that in one column. Then I mean centered stress and genotype and entered those mean centered values in columns 2 and 3. In column 4 I modeled the intercept (all 1's). My interpretation is as follows. voxels that show up for the contrast for column 1 are voxels in which the effect of genotype on task related activity is moderated by stress ( i know it can be reversed this is a theoretical decision). contrasts for columns 2 and 3 are interpreted as conditional effects (i.e. effect of stress at mean of genotype and vice versa).
Is there anything wrong with this method? I am familiar with SPSS so I applied the same methodology. I just wanted to check for problems with this approach. I know there may be different methods, but this is working. I just want to make sure it is a reasonable approach.
Thank you,
Noam
|