 |
 |
Dear SPM experts, I have 2 groups of subjects who each performed a task with 4 conditions. I would like to compare the shape and amplitude of the HRF in the regions activated by the task between these two groups of subjects. Approach 1: On the FFX level take the average of the 4 conditions (contrast 1 1 1 1) for all subjects in both groups and take this average to the RFX level in a one sample t-test (over the 2 groups). Then take the voxels significantly activated in this one-sample t-test and compare the shape and amplitude of the HRF between the 2 groups of subjects within these significantly activated voxels. The idea here is to make sure that I am comparing the 2 groups only in voxels that are actually implicated in the task. I am just not 100% sure whether the one-sample t-test and the following ROI analysis to compare the 2 groups are statistically orthogonal and the approach therefore statistically valid. Approach 2: On the FFX level calculate a contrast for each of the 4 conditions for all subjects in both groups (resulting in 4 con images per subject). Then on the RFX level perform a 2x4 ANOVA. Take the voxels significantly activated in the main effect of condition (over the 2 groups) to build the ROIs and then compare the shape and amplitude of the HRF between the 2 groups within these ROIs activated by the main effect of condition. The first approach is by far the simpler one and therefore the one that I would prefer. I think both approaches are valid, but am not 100% sure about the first approach and would really appreciate some feedback. Are both the first and the second approach valid? And would I be performing valid statistics by taking the first approach? Many thanks in advance for any thoughts on this matter, Bianca de Haan. -- Dr. Bianca de Haan Section Neuropsychology, Center of Neurology Hertie-Institute for Clinical Brain Research University of Tübingen +49 (0)7071 29 85661