Hi,
I have a question regarding the correct setup of a design matrix & contrasts in the following two cases:
Case 1: Repeated measures (1-factor, 5 levels)
I have a melodic component from 90 input files (18 subjects, 5 different timepoints per subject). I am interested to see whether there are differences in this melodic component between these 5 timepoints.
I'm hesitating what kind of design matrix & contrasts should I use. I've checked the repeated measures design
https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/GLM#Multi-Session_.26_Multi-Subject_.28Repeated_Measures_-_Three_Level_Analysis.29
With this design it seems that I'll get as output subject means. However, I wish to see the mean difference per timepoint average, and not differences between individual subjects.
On the other hand, a matrix described here does not seem to account for the repeated measures design, so it doesn't seem to be the answer either.
https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/GLM#ANOVA:_1-factor_4-levels__.281-way_between-subjects_ANOVA.29
Case 2: Repeated measures (within-subject factor: time with 5 levels, between-subject factor: condition)
Again, starting with the melodic component from 90 input files (18 subjects, 5 different timepoints per subject). 10 subjects belong to condition A, 8 subjects to condition B. I am interested to see whether there is a difference in the melodic componennt between the different conditions on different timepoints.
How do I setup the design matrix in this case - should I just indicate the between-subjects factor in the design.grp and otherwise keep the design similar to the Case 1? Or add one extra EV indicating whether the input belongs to condition A or B?
Any help is very much appreciated :)
Cheers,
Anu
|