 |
 |
Hello FSLers, I am wondering if there is a way to run an unpaired t-test between two groups (e.g., patients and controls) with a continuous covariate (e.g., age) in the case where each group is assumed to have different cross-subject variance (two group memberships are specified in the "group" column in Feat). Simply adding an extra EV for the covariate would violate the requirement that only one of the groups has non-zero values, while splitting the covariate into separate EVs for each group seems like it would defeat the purpose of examining whether the covariate accounts for group activation differences. Many thanks, Michael