See below.

On Tue, Oct 15, 2013 at 5:39 AM, Delphine Roussillon <[log in to unmask]> wrote:
Dear SPMers,

 I sent a mail last week and as I received a spam and no answer so I send my questions again.

I'm trying to run a full factorial design with one factor (factor group, 3 groups so 3 levels). Each cell I have entered corresponds to a group. Then I have 3 covariates. One of them is not the same across group. The idea is to show that group differences are not due to these covariates.

Here are my questions:

1) Is this design correct ? Is that ok to say that by adding a covariate in a full factorial design we assume that the effect of the covariate is the same in all groups ?

Yes. Yes.
 

2) When I run my design, everything is ok til the 11th suject in each group.
(so I have a unique factor). When I run the batch with 11 subjects per group, it's seems to be ok. When I add a 12th subject in each group, I'm out of memory (with two different PC), although it's not the case in a flexible design. Is it normal ? Is it very memory-consuming ? Or is my design wrong ?

This seems odd.

3) I understand that a full factorial is a fixed-effect analysis, as there is no subject factor. What is annoying for me is that I would like to take into account the between-subject variability, what is possible with a flex. However I can't add a covariate in a flex. So what would be the best model ?

You model only has between-subject variability. The flexible factorial model is used when you have multiple images/conditions per subject, which is not what you have in your design. The flex. factorial only has within-subject variance estimated when you have repeated measures.
 

I would be very grateful if someone could help me.
Delphine