Dear Experts,
I am wondering whether I have set up my 2nd level tests correctly.
My experimental set-up looks like this:
-2 groups (patients and controls)
-A Drug challenge (drug or placebo) for a total of 4 groups (PT-Placebo, PT-Drug; HC-Placebo; HC-Drug)
-2 fMRI sessions (pre and post-scan), where participants perform a task at each fMRI run.
The Task has 2 levels - question 1 (emotion-recognition) or question 2 (emotion-inhibition)
Emotional Valence with 3 levels: (1) neutral (2) negative (3) positive.
For a total of 6 conditions per fMRI run.
In my 1st level t-test contrasts I have contrasts set up for:
-a main effect of scanning session,
-a main effect of question,
-2 main effects of emotional valence: valence1 = negative vs neutral; valence 2 = negative vs positive.
I also have 1st level contrasts set up for the following interactions:
QuestionXvalence1;
QuestionXvalence2;
Question1Xnegative; and Question2Xnegative;
Question1xpositive; and Question2xpositive;
Question1Xneutral; and Question2Xneutral.
In my 2nd level analysis I have a series of independent t-tests testing:
1) All PT vs All HC
2) All Placebo vs All Drug
Between-groups comparisons:
3) PT-Phys vs. HC-Phys
4) PT-Placebo vs. HC-Placebo;
Within-groups comparisons:
5) PT-Placebo vs. PT-Drug
6) HC-Placebo vs. HC-Drug
QUESTION: I am wondering if, traditionally, at the 2nd level, it would be more correct to run an overall F-test testing for an interaction between all 4 groups (PT-Placebo, PT-Drug; HC-Placebo; HC-Drug), and afterwards perform the between/within groups comparisons to understand what is driving the F-test results?
Or, if my 6 2nd level comparisons are adequate and adding an F-test would be redundant.
Thank you,
Leila
|