Hi Debbie,
Before getting tangled up in including covariates, you need to address your current approach. What you have, if I’ve understood it correctly, is a 2x2 within-subject design. If you want to model it at the group level it requires special treatment to take the repeated measurements into account. The more typical approach would be to model all 2x2 factors at the 1st-level, using a combination of different conditions and different sessions. Because your design is 2x2, you can use the contrasts you indicated to create single effects at the 1st level corresponding to the main effects and interactions for each subject. You can then use a one-sample t-test at the 2nd level for each effect. This will make it very easy to add any between-subject covariates. This is what I would recommend.
Instead, if you definitely want the full 2x2 design represented at the 2nd-level, you will need to add factors for subject and each subject X factor interaction for the within-subject effects. You then need to take care with the error term being used for each effect, meaning you’ll need multiple models to correctly calculate the main effects and interactions. This is not what I would recommend, but it is all explained in my paper: https://www.frontiersin.org/articles/10.3389/fnins.2019.00352/full
Best wishes,
Martyn
|