Dear Matthew and FSL-Team,
I have a few questions about my TBSS-Analysis concerning my dissertation about white matter integrity (measured by Fractional Anisotropy; FA) in a group of patients versus healthy controls.
I already experimented a lot with different design matrices and contrast-files, now I am not sure whether the design matrices and contrast files adequately assess my intention.
So to explain my intention shortly:
1. FA-Group Comparison
First I want to perform TBSS group comparison (G1 and G2). I have a few Co-variates of no interest: age, gender and rating scale data.
Would the design matrix be
G1 G2 Age(G1) Age(G2)Gen(G1) Gen(G2)RS(G1) RS(G2)
1 0 78 0 2 0 11 0
1 0 65 0 2 0 24 0
1 0 73 0 1 0 3 0
0 1 0 78 0 1 0 13
0 1 0 68 0 1 0 25
0 1 0 60 0 2 0 12
and the contrast file:
1 -1 0 0 0 0 0 0
-1 1 0 0 0 0 0 0
I executed the procedure by using the FSL-Tool randomise (no –D Option, no mean centering)
2. FA-Correlation- and FA-Interaction- Analysis:
After the FA-Group comparison I also want to do both a within group correlation for RS and testing for interactions of RS. So RS would be the Co-variate of interest; age and gender the co-variate of no interest.
To the designmatrix listed above, the contrast file would be in this case:
0 0 0 0 0 0 1 -1
0 0 0 0 0 0 -1 1
0 0 0 0 0 0 1 0
0 0 0 0 0 0 -1 0
0 0 0 0 0 0 0 1
0 0 0 0 0 0 0 -1
I executed the procedure also by using randomise (without –D Option, and without mean centering).
Is it correct, that my first 2 contrasts represent the interaction and the following 4 contrasts the within group correlation?
How can I interpret whether the interaction 0 0 0 0 0 0 1 -1 and on the other hand whether the interaction 0 0 0 0 0 0 -1 1 is significant? The results would be in fact different, but I have never heard of anything like positive or negative interactions?
Also if you could refer to –D Option and mean centering the values in the design matrix, i would be very thankful. As far as I understand the explanation of Jeanette Mumford (http://mumford.fmripower.org/mean_centering/) I am right with my assumption no mean centering.
Thanks a lot! I really appreciate any answer!!!
Kind regards
Berel Weinhöppel
|