Any help on this statistical question would be greatly appreciated!
I am interested in comparing two studies (done with two different groups of subjects) that have different designs. Is it reasonable to do a group analysis to compare a contrast that is present in both studies when one is a block design (20 sec blocks, 4 conditions and fixation rest) and the other is an event-related design (same 4 conditions and fixation null events)? I know that the power of the designs are quite different, but am not familiar enough with the implementation of the statistics to know whether this would drive the results of a direct comparison in one direction or another (making them uninterpretable).
The strategy I had in mind was to do a 3 level analysis: first level contrast estimates, combine sessions for each subject with fixed effects, compare across studies using 3rd level random effects unpaired t-test). Does this seem reasonable? If not, is there any way to perform this comparison that is more desirable?
|