Print

Print


Dear Mirko,

SPM5 uses a generic approach to create this kind of designs. There are two 
ways to go - full or flexible factorial designs. Anyway, you should think 
of your design in terms of factors and levels.

From what you said, I guess you have the following factor candidates:
- group (2 levels)
- subject (nsub1 levels in group 1, nsub2 levels in group 2)
- condition (2 levels) or replication (2 levels), depending on the meaning
   of your 2 scans per subject

You have to choose what kind of error covariance structure you expect - 
almost noone will be able to help you with this decision, but from an 
estimation point of view you should only assume unequal variance or 
dependent variance if there is a strong reason for it.

Usually, you want to treat "subject" as random factor. If you do this, you 
can use "Full factorial design" with your remaining factors and you will 
end up with 2x2 cells of group/condition level combinations. Probably, you 
would assume unequal variance for your group factor levels(, and perhaps 
dependent variance between your condition levels).

When using "Flexible factorial design", there is another decision to 
be made in design specification: which effects you want to test, i.e. 
which colums you want to appear in your design matrix. At least, you would 
want a group-by-condition interaction (which is just the same as a full 
factorial design), but perhaps also a group main effect (per-group mean) 
or even a group-by-subject interaction (per-subject mean).

Hope this helps,

Volkmar


On Wed, 14 Feb 2007, mirko wrote:

> Dear All,
>
> In SPM2 I used "multi-groups: conditions-covariates" but I can not find an 
> equivalent set up in the SPM5. I have two groups and in each group subjects 
> had two scans (paired scans). Could someone suggest to me how to set up this 
> calculation/comparison in SPM5?
>
> Many thanks,
>
> Mirko.
>

-- 
Volkmar Glauche
-
Department of Neurology		[log in to unmask]
Universitaetsklinikum Freiburg	Phone	49(0)761-270-5331
Breisacher Str. 64		Fax	49(0)761-270-5416
79106 Freiburg			http://fbi.uniklinik-freiburg.de/