Dear SPM’ers,
I need some advice on the second level analysis model for my current experimental design. I am reposting this message since I haven’t received any reply and I would really use some help/suggestions.
I conducted an event-related cued task/state switching paradigm with 3 types of cues indicating different types of trials, namely, “rest” (waiting) periods (cue1), task1 (cue2) (parity judgment on number stimuli) and task2 (cue3) (magnitude judgment on number stimuli) with 2 groups of participants (clinical and control). In this paradigm subjects have to alternate between the 2 task trials and also “rest” (waiting) periods. I am interested in first establishing the very cue-related effects, and also the switch or repeat related effects (with reference to n-1 trial).
The first idea was to use the flexible (2x3x3) factorial design with a group factor with 2 levels (clinical and control), cue type factor (cue1, cue2, cue3) and switch/repeat factor with three levels as well (cue(1,2,3) to cue(1,2,3)). However, I run into a problem here since the switch/repeat factor involves the qualitatively different repeats and switches, namely, the switches from “rest” to task (vice versa) and between tasks, same with the repeats (state switches and task switches/state repeats and task repeats), and the results become uninterpretable.
So, my question is, is it valid instead of having a model encompassing everything, to split it into a number of flex factorial models, involving only a few conditions, e.g. cue1 repeat (rest repeat) trials and cue2/3 switch to cue1 (rest trials preceded by task trials), for all the relevant switches and repeats. Or, maybe a more proper way to go is to do the t contrasts of interest (e.g. cue1cue1>cue2/3cue1) in the first level analysis, bring them to second level and do the 2 sample t test on them.
Thanks in advance for your suggestions.
Juste
|