 |
 |
Hi, My paradigm is a 2x2 design, and each trial contains 5 periods (cue-a, stimulus-a, cue-b, stimulus-b, response). In my first-level model, if I model all periods, it will contain 20 EVs in total. However, I'm only interested in hemodynamic responses during stimulus-a and stimulus-b, not interested in the rest three EVs. Can I model 11 EVs (1 for cue-a, cue-b, and response; 4 for stimulus-a and stimulus-b) in my model? My understanding is, if I model all 20 EVs, the fitting to the model will be better but with a high degree of freedom, which will result in reduced statistics in the effects of the regressors of interest. If I model 11 EVs, the statistics will be higher, and it would be more possible for the effects of the regressors of interest to survive multiple comparison thresholds. Am I right? Thanks in advance.