Dear all,
We're stuck with the following second level model for fMRI analysis and would appreciate any helpful comment.
We have here a longitudinal training paradigm and we like to reveal activation differences between two timepoints (T1, T2), but additionally, the model should only search for differences that are correlated by task performance. To this end, we set up a paired T-test design with an additional covariate of interest, introducing T1- and T2- related individual task performances. In order to gain comprehensive insight irrespective of the effects' directions (including T1>T2 and T1<T2), we defined the following f contrast [-1 1 1]; (column order from left to right: T1, T2, Performance). A screenshot of our design matrix can be seen here: http://goo.gl/gieHlX.
We wonder, whether we correctly tackle this issue and/or whether there is a better solution.
Thanks for you help and have a good start into week 30.
best wishes,
mathias
|