Dear Jason,
> I have two questions: We are doing fmri studies with analysis in
> SPM99b. First, With the random effects analysis does the following
> seem correct: I have two groups, each subject has two conditions,
> First set up a design matrix including all subjects and look at each
> subject individually. This will create a contrast image for each
> subject. Then using the basic stats use ANOVA with two groups. Then
> enter the contrast images from each group into their respective groups.
> Finally set a contrast of [-1 1] to look at the between group effects.
This sounds exactly right (assuming the contrasts test for a differece
between the two conditions).
> I have a study with three conditions, each having two factors, how
> would I look at the main effects and the interactions using the
> contrasts? So this is a 2x2x2 ANOVA.
A factor has different levels. I assume you have three factors
(condition with 3 levels and another two factors with two levels).
This gives a 3 x 2 x 2 design. These are generally treated as nested 2
x 2 designs and interactions assessed with the appropriate contrast.
For example (ignoring one of the factors with 2 levels)
-1 1 0 1 -1 0
Would test for the interaction between the first two conditions and the
other factor.
> Also how would I use reaction
> times as a covariate in the design matrix? What would the contrasts be
> to look at any effects of the covariates?
You would specify covariates of interest that were the product of the
mean-corrected contrast of cell-specific effects and the
(mean-corrected) RT covariate to look at interaction and the RT
covariate itself to model the main effect.
I hope this heslp - Karl
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|