Dear SPMers,
I have 3 groups of subjects (A and B are patient groups, and C is healthy control) and I want to know if brain activations to a sensory stimulation are different among the 3 groups, and if so, where the difference comes from (A vs. B, B vs. C or A vs. C). However, subjects had different perceptual ratings to the same sensory stimulus, and these ratings per se affect brain activation.
I think I should perform an ANCOVA analysis using perceptual rating as a covariate, and then do subsequent post-hoc between-group pair comparisons.
In SPM5, I chose “second-level analysis” and “Full factorial design.” The name of “Factor” was “Group,” with 3 levels. Then I specified 3 cells: scans from group A were labeled as level 1, B as level 2, C as level 3. Then, there is an “ANCOVA” under the menu “Factor”.
My first question is: should I make it “Yes?”
After doing that, under the main “Design” manu, I entered the “Vector”(individual perceptual rating scores) and “Name” (“Rating”) in the “Covariates” item. Then I run it and got results of “main effect of group” and pair comparison contrasts.
My second question is: is my procedure right?
Is what I’ve done called “one-way ANCOVA with post-hoc comparisons?”
Final question: The instruction for “ANCOVA” below the “Factor” manu says that “This option is only used for PET data”. Since my data is fMRI, I should not make this “YES”?
In conclusion, the following understanding for setup:
If I just enter values in "Factors" and their "Levels," this is a ANOVA analysis.
If I enter values in "Covariates" within the "Design" menu as well as "Factors" and their "Levels," then this is a ANCOVA analysis.
Are both correct?
Is any instruction for ANOVA/ANCOVA model setup in SPM?
Thanks a lot!
|