On Tue, Aug 21, 2012 at 11:40 PM, Jaclyn Billington
<[log in to unmask]> wrote:
> Dear SPMers,
>
> I would like to clarify some of the output of a mixed ANOVA I have been running and can't seem to find any documentation related to this.
>
> I have two groups and one within subject condition (with three levels). I have set up a flexible design matrix in SPM which is something like this...
>
> G1 G2 C1 C2 C3 G1C1 G1C2 G1C3 G2C1 G2C2 G2C3 Sub1 sub2 sub3 .... sub26.
>
> I've determined my contrast weights as per Glascher and Gitelman and all seem to be working correctly.
> My two questions are.
>
> 1. Is it okay to look at the main effect of group using this design? I have read a few references to using an additional model to look at this effect but presume if you weight the subjects according you can look at group effects in this model?
>>>>You can not look at the main effect of group or any between-subject effect in this model because the error term is the within-subjects error term.
>
> 2. Glascher and Gitelman state to use [-1 0 1] for either main effect of condition (with weightings for interaction factors) or interactions.
>>>>>>The main effect contrast should be: [1 -1 0; 0 1 -1]/
Is this contrast statement looking at main effect of C3>C1 accounting
for C2, or looking for main effect across all conditions with beta
extraction required to determine the nature of the effect or
interaction? I ask because I get different results depending on
whether I use [-1 0 1], [0 -1 1] etc but also when I extract beta
values the interaction done quite look how I'd expect....
>
> Any pointers would be gratefully received.
>
> Thank you.
|