Hi Qian,

There are several equivalent ways to specify this, the one I typically recommend is:

1) define two separate 2nd-level covariates characterizing the levels of your first factor (e.g. a MALE and a FEMALE covariate each having 0/1 values identifying the subjects within those groups)

2) define three separate 2nd-level covariates characterizing the levels of your second factor (e.g. a GROUP1, GROUP2, and GROUP3 covariates, each having 0/1 values identifying the subjects within each of those groups)

3) in Setup.Covariates (2nd-level), select all those 5 variables and then click on 'covariate tools' -> 'Create interaction of selected covariate(s)'. That will create 6 new variables defining the interaction of the above 2 factors (e.g. MALE*GROUP1, MALE*GROUP2, etc.)

4) in the Results (2nd-level) tab, define the interaction analyses as:

Subject effects: MALE*GROUP1, MALE*GROUP2, MALE*GROUP3, FEMALE*GROUP1, FEMALE*GROUP2, FEMALE*GROUP3
Between-subjects contrast: [-1 1 0 1 -1 0; 0 -1 1 0 1 -1]

Best
Alfonso



Originally posted by zhuang qian: