Hello

As I understand with this model, you can not look at the contrasts within the ANOVA, but have to go and do t-tests to examine the specific comparisons of interest.

In my study, I have 4 tasks and 2 groups of subjects. My main interest is to examine task differences between groups. In the typical (non-neuroimaging) ANOVA, you can not look at these types of comparisons unless you have significant main effects and interactions. Theoretically, the type 5 model is the correct one to use, but I find that I am looking at the t-test results more than those from the ANOVA. The ANOVA results tell me that there is a significant task (collapsed across diagnosis), diagnosis (collapsed across task) and diagnosisxtask effect, but it is only once I examine the specific contrasts between groups that I can understand the direction of the interaction effect.

Could you please comment on the proper way to report the main and interaction effects. Is it wrong to concentrate on the t-tests rather than the ANOVA results?