Hello all,
I am trying to set up a flexible anova using SPM8 in order to compare 2 groups (22 and 25 subjects) who completed 4 conditions. I created 3 factors (subject, independence YES, variance Unequal; group, independence YES, variance Unequal; and condition, independence No, variance Equal). I used the option specify all and used an input of 188 scans and a 188x4 factor matrix
[1 1 1 1
1 1 1 2
1 1 1 3
1 1 1 4
2 2 1 1
2 2 1 2
2 2 1 3
2 2 1 4
3 3 1 1
3 3 1 2
3 3 1 3
3 3 1 4
......]
I want to model the main effects for group, condition and subject as well as the interaction for group x condition. So I chose three main factors (1, 2, 3 for subjects, group, condition) and the interaction [2 3] for group x condition. Even though, I place subjects first the design matrix puts subjects at the end, could someone please explain why this occurs?
SPM will estimate this model, however if I leave out the main effect 'subjects' I will get the error "Warning: Matrix is close to singular or
badly scaled." I am not sure why this sometimes happens and sometimes doesn't.
However, the main difficulty I currently have is the following: I am not sure what contrasts weights I have to use in order to look eg at G1C1 vs G2C1 or any of the main effects.
For the G1C1 vs G2C1 I tried using [1 -1 0 0 0 0 1 0 0 0 -1 0 0 0 zeros(1,47)] likewise I tried [ zeros(1,47) 1 -1 0 0 0 0 1 0 0 0 -1 0 0 0] as I am not sure where SPM models subjects in my matrix. For the main effect group I tried [1 -1 0 0 0 ...] as well as [zeros(1,47) 1 -1 0 0 0...] both of which did not work.
I am using thttp://www.sbirc.ed.ac.uk/cyril/download/Contrast_Weighting_Glascher_Gitelman_2008.pdf as reference for my contrast.
Could anyone please help me with any of my questions
Best regards
Ben
|