Hi Jian,
Just noticed your old apparently-unanswered post. I think you should
be able to use either the full or flexible factorial models (in SPM5)
and something similar in SPM2, to get a suitable design. You won't
need to worry about the F-test here, as the interaction is just a
single element with this 2-by-2 design. You need to select the con
images from the first level (for A1,A2,B1,B2) as the images for the
second level (not the betas).
I'm sorry I can't give more specific advice on SPM2, but I think you
should be able to work this out from the manual. I believe, you could
use a one-way (4-level) ANOVA with a design matrix that was block
diagonal, and then test the interaction with something like
[1 -1 -1 1].
I hope this helps,
Ged.
jian wrote:
> Dear SPMers,
> I have a block design experiment with two conditions (A B)each
> varying at two levels (1, 2) and I have estimated the activation effect
> of each type(A1 A2 B1 B2) using a GLM model for each subjects. Now I
> want to use the BETA images of these 4 types (A1 A2 B1 B2) to do a
> two-way ANOVA analysis to examine whether there is any interaction
> between the condition (A/B) and level(1/2). Is there any easy way to do
> in within SPM frame? The basic model in SPM did not support any two-way
> ANOVA analysis and F-test is somewhat confusing and I could not have my
> thought straight about it.
> Any suggestion would be really appreciated.
>
> Thanks
> Jian
>
> *Jian Li*
>
> *Division of Neuroscience*
>
> *Baylor college of Medicine*
>
> *713-798-1734*
>
> *www.hnl.bcm.tmc.edu*
>
> * *
>
>
|