Hi,
I have 3 groups and I want to compare (A) mean group activations and (B) correlation with a behavior score among the 3 groups (which brain region showed different degree of correlation with the score among the three groups).
(1).
In the first approach, I selected "multiple regression" in SPM5 second-level analysis --> select all subjects for "scans" --> setup 6 covariates (group A/B/C; correlation A/B/C).
(2).
In the second approach, I selected "full factorial"--> setup 1 factor (group) with 3 levels --> specify cell 1/2/3 for group A/B/C by order --> in "covariates" I set up 3 covariates (correlation A/B/C).
For (A), I used a F contrast [1 -1 0 0 0 0; 0 1 -1 0 0 0] to get ANOVA results of mean group activations in (1). In (2), it automatically generates a "main effect of group" contrast that is equivalent to the F contrast [1 -1 0 0 0 0; 0 1 -1 0 0 0] in (1). HOWEVER, I noticed the results of these two approaches are slightly different. So which one should be the one I want?
For (B), I used a F contrast [0 0 0 1 -1 0; 0 0 0 0 1 -1] in both (1) and (2). Am I correct? AGAIN, the results of these two approaches are slightly different. So which one should be the one I want?
Thanks in advance.
Ming
|