 |
 |
Dear experts, I would like to be sure that my approach to test interaction in this design is correct. I have two groups and want to test the effect of a continuous variable (independent variable, IV) on FA after controlling for age and gender. In this particular case my matrix should include as EV1 the first group, as EV2 the second group, and as EV3 and EV4 the two covariates. But if I wanted to test the effect of a potential interaction (i.e., age x gender, age x IV, gender x IV, or age x gender x IV), should I build a matrix with all possible interactions, or should I only include the group variables and the interaction of interest? In the latter case, if I wanted to test age x IV, the matrix should include EV1: group 1; EV2: group 2; EV3: age x IV (demeaned). I have learnt from another post (073075) how to proceed to test the different interactions, but if I include in the matrix the IV, the two covariates and all possible interactions, when performing the contrast of interest, the IV, covariates and remaining interactions will be set at zero, and consequently would be considered as confound factors. Am I correct? Thanks a lot in advance, M del Mar