See answers below.

Best Regards, Donald McLaren
=================
D.G. McLaren, Ph.D.
Postdoctoral Research Fellow, GRECC, Bedford VA
Research Fellow, Department of Neurology, Massachusetts General Hospital and
Harvard Medical School
Office: (773) 406-2464
=====================
This e-mail contains CONFIDENTIAL INFORMATION which may contain PROTECTED
HEALTHCARE INFORMATION and may also be LEGALLY PRIVILEGED and which is
intended only for the use of the individual or entity named above. If the
reader of the e-mail is not the intended recipient or the employee or agent
responsible for delivering it to the intended recipient, you are hereby
notified that you are in possession of confidential and privileged
information. Any unauthorized use, disclosure, copying or the taking of any
action in reliance on the contents of this information is strictly
prohibited and may be unlawful. If you have received this e-mail
unintentionally, please immediately notify the sender via telephone at (773)
406-2464 or email.



On Fri, Jul 8, 2011 at 8:11 AM, Ben Becker <[log in to unmask]> wrote:
Dears SPMers,

I ran a repeated measures flexible factorial anova with the within subject factor condition (repetition) (3 levels) & the between subject factor group (2 levels) following main effects & interactions:

Main effect1 subject
Main effect 2 group
Main effect 3 condition (repetition)
Interaction 2 x 3 (group x repetition)

the F-contrast for the group x repetition effect revealed significant interaction effects.

Now: how can I further analyses the direction of the interaction effect? - My Ideas:

1. t-contrast to test for an directional effect

Yes. You can do post-hoc t-tests using the F-contrast as mask. This would determine which conditions are driving the interaction.


 
2. plot the responses

For plotting, I'd do the following:
(1 - using SPM only) create an omnibus F-contrast, 1 row for each group/condition combination. In your case you will have 6 rows. Each row represents the group/condition effect. 
An example contrast might be:
1 0 1 0 0 1 0 0 0 0 0 ones(1,n1)/n1 zeros(1,n2) where n1 is number of subjects in group1 and n2 is the number of subjects in group2. The order of columns is group, condition, group*condition, subjects. You'd have to plot in SPM at each cluster (using Plot --> Contrasts) and then save the results from SPM/MATLAB (I think the variable is contrast.con)

If subjects is NOT on the right of the design matrix, then your using an old version of SPM that has a bug in the computation of repeated measure designs with more than 2 levels of the within-subject factor.

(2 - using SPM and my peak_extract_nii program) create 6 t-tests, 1 for each group/condition combination. Then you can use peak_extract_nii with the 6 con_ images and the spmF to extract the regional values from every peak and cluster. The downside is that there is no variance available. However, if you use the input images; then you could use the mean function in MATLAB to create the means for each group/condition combination as well. Then you can just plot in MATLAB

If subjects is NOT on the right of the design matrix, then your using an old version of SPM that has a bug in the computation of repeated measure designs with more than 2 levels of the within-subject factor.

Hope this helps.
 

Would it be valid? And regarding plotting the responses: should I plot the F contrast or the t-contrast (or is it the same?)

Valid, yes. Plotting the F- will give you the same as plotting the 2 t-contrasts. This is because you are plotting each row of the contrasts and the plotting is independent of the statistic as it is derive from the contrast (Beta*contrastvector)


Thanks in advance & kind regards

ben