Dear Lena,
in SPM12, there is support for mixed effects models. However, this is computationally much more expensive than the summary statistics approach. Also I am not sure how SPM would allow to enter 2nd-level covariates, since the group level design is specified as an F-contrast without the ability to enter any subject specific information.
In repeated measures summary statistics, the subject constant models all effects that are constant within each subject (e.g. age, IQ). Modelling them at 2nd level will only work if one can do all within-subject calculations at 1st level and then bring a single difference image per subject to the 2nd level.
Best,
Volkmar
|