Dear FSLers,
due to my specific study design, I am looking for an approach where I can evaluate interaction effects and/or main effects of group or time in my TBSS data.
I measured:
- 3 different groups of participants
(group 1 - patients with intervention, group 2 - patients without intervention, group 3 - healthy participants).
- at 2 time points – pre and post
(the first measurement took place before the intervention in group 1 started).
As covariates, I want to add:
- age
- objective disorder severity degree of the patients
- subjective disorder severity degree of the patients
What I want to do certainly (by using randomise):
- unpaired t-test at pretest: group of all patients and group of healthy controls
→ to evaluate the differences in FA that both groups show at the first measurement
- unpaired t-test: couple of two groups (group 1 and group2 / group 1 and group 3), using difference images (post-pre) of each participant
→ to see how FA change over time differs between the two groups
- paired t-tests: for each group, post vs pre
→ to show changes from post to pre-test
What I want to do but I don’t know how/what:
- run an approach where I can evaluate interaction effects and/or main effects of group or time
For this purpose, can I use randomise or has anybody experience with PALM?
For randomise, I found the 2-way mixed effect ANOVA (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/GLM#ANOVA:_2-groups.2C_2-levels_per_subject_.282-way_Mixed_Effect_ANOVA.29) on your wiki, which seems to be partly suitable for me.
But this approach is only suitable for 2 and not 3 groups. Furthermore, the wiki claims it is not usable in randomise ( or is there a way how I could use the presented GLM)?
I am glad to hear from you!
Thanks and all the best
Annika
|