On Wed, Nov 21, 2012 at 5:14 AM, <Catherine> <Jutzeler>
<[log in to unmask]> wrote:
> To whom it may concern,
>
> I am struggling in defining the contrast vectors for a 2x4 factorial design using the flexible factorial design in SPM.
>
> So far I defined three factors: subject, group and condition. I have 2 groups and 4 conditions. I am interested in the main effect of group and condition and in the interaction of group x condition.
>
> Thats what I found in the manual of Gläscher..... how to adapt that to the 2x4 factorial design (the example is a 2x3 factorial design).
>
>
>
> Main Effects and Interaction:
>
> 1. Main effect: 2 (A)
>>>> The main effect of group is not a valid contrast because it is a between-subject effect. You need to model each condition (or average of conditions) in a one-way ANOVA of group to get the effect of group. Alternatively, you can use GLM Flex.
>
> 2. Main effect: 3 (B)
>>>> Valid because it compares the conditions.
>
> 3. Interaction: 2 3 (A — B)
>>>> Valid because it compares the difference of conditions between groups.
>
> Contrast Weights
>
> 1. Main effect of A: 1 -1 0 0 0 ones(1,3)/3 -ones(1,3)/3
>
> 2. Main effect of B: 0 0 -1 0 1 [-1 0 1]*[1/2] [-1 0 1]*[1/2]
>
> 3. Interaction A — B: 0 0 0 0 0 -1 0 1 1 0 -1
>
> I am also interested in calculating t-contrasts in the meaning of post hoc tests (e.g., scans of group1_condition1 vs scans of group2_condition1). Is that also possible with the flexible factorial design?
>>>> No. Tests of a single condition are not valid. You will need an ANOVA to test for group differences of a single condition or average of conditions. The rule is that unless you comparing two conditions, then it can't be tested in the flexible factorial model. However, GLM Flex will allow the between-subject effects and the post-hoc tests to be evaluated accurately.
>
> Many thanks in advance.
>
> Kind regards,
>
> Catherine
|