Dear FSLers
I would be very grateful for advice on a higher level Feat Design Matrix for experiments such as a 2 group study (case-control) where each participant is scanned under 2 conditions (active drug and placebo). I checked the manual but couldn't see a similar example.
Can someone please check my proposed design matrix for an example with 3 participants per diagnosis and one visit for active drug and one for placebo.
To model this, I tried 2 groups, 10 EVs (1 for each group, 1 for active drug > placebo in cases, 1 EV for active drug > placebo in controls, and 6 EVs (EVs 4 to 10) for the 6 participants which I have written below as S1 to S6).
Gp EV1 EV2 EV3 EV4 S1 S2 S3 S4 S5 S6
1 1 0 1 0 1 0 0 0 0 0
1 1 0 1 0 0 1 0 0 0 0
1 1 0 1 0 0 0 1 0 0 0
1 1 0 -1 0 1 0 0 0 0 0
1 1 0 -1 0 0 1 0 0 0 0
1 1 0 -1 0 0 0 1 0 0 0
2 0 1 0 1 0 0 0 1 0 0
2 0 1 0 1 0 0 0 0 1 0
2 0 1 0 1 0 0 0 0 0 1
2 0 1 0 -1 0 0 0 1 0 0
2 0 1 0 -1 0 0 0 0 1 0
2 0 1 0 -1 0 0 0 0 0 1
Then, I can use the following contrasts:
C1: Group 1 (patients) > Group 2 (controls): EV1 +1, EV2 -1
C2: Group 2 (controls) > Group 1 (patients): EV1 -1, EV2 +1
C3: Drug > placebo in patients: EV3 +1
C4: Placebo > Drug in patients: EV3 -1
C5: Drug > placebo in controls: EV4 +1
C6: Placebo > drug in controls: EV4 -1
C7: Drug effect > placebo overall: EV3 +1, EV4 +1
C8: Placebo > Drug overall: EV3 -1, EV4 -1
C9: Interaction drug x diagnosis 1 (drug effect greater in cases): EV3 +1, EV4 -1
C10: Interaction drug x diagnosis 2 (drug effect greater in controls): EV3 -1, EV4 -1
F tests: 3
1) Group effect: include C1 and C2
2) Drug effect: include C7 and C8
3) Interaction effect: include C8 and C9
However, I get the following error when I try to set this up:
Problem: Design Matrix is rank deficient. F Test 1 isn’t valid, each contract cannot be a linear combination of the others.
What did I do wrong and any suggestions to fix it?
Thanks a million
Graham
Dr Graham Murray
Brain Mapping Unit
Department of Psychiatry
University of Cambridge
|