Dear FSL experts,
I have been using TBSS and randomise to look for significant differences in FA from DTI images among 4 groups or categories of subjects.
Based on the FSL website, I designed my contrast to look like below (since I wanted to do a one-way ANOVA between the four groups):
D=control group (no attacks)
A=1 attack/episode
B=2 attacks/episodes
C=3 attacks/episodes
1. A-D 0 1 0 0
2. B-D 0 0 1 0
3. C-D 0 0 0 1
4. Mean 1 0.25 0.25 0.25
5. A-mean 0 0.75 -0.25 -0.25
6. B-mean 0 -0.25 0.75 -0.25
7. C-mean 0 -0.25 -0.25 0.75
8. D-mean 0 -0.25 -0.25 -0.25
First off, I wanted to make sure that I designed the contrasts correctly. Second, when I ran this matrix, the only ones that showed significant regions were the mean contrast (contrast#4) and the D-mean contrast (contrast#8). I am getting confused on what a significant finding in the D-mean contrast mean. Does this mean that the overall one-way ANOVA is significant, but that my pairwise comparisons are not?
Thank you very much!
Jo Ann
|