Hi david - yes, this mostly looks ok, though you don't have a full
interaction contrast - I would have thought that something like [1 -1 -1
1] would be right?
Cheers, Steve.
On Sun, 13 Feb 2005, Glahn, David C wrote:
>
> Howdy,
>
> I am attempting to perform a higher-level analysis and would like to make sure I have set up the model correctly. The analysis includes 2 groups (PT and CT) and two tasks (A and B; from separate runs). I would like to conduct a 2 x 2 ANOVA where I can model group differences, task differences and group x task interactions. Unfortunately, I do not have a balanced model, with 11 PT on Task A; 12 CT on Task A; 10 PT on Task B; and 9 CT on Task B (n=42 total).
>
> The design I cam up with has 4 EV and the between group column. EV1 models the PT group on Task A; EV 2 the CT group on Task A ext. I include 13 contrasts as follows:
>
> 1. ALL 1 1 1 1 (to use for subsequent ROI analyses)
> 2. PT TaskA 1 0 0 0
> 3. CT TaskA 0 1 0 0
> 4. PT TaskB 0 0 1 0
> 5. CT TaskB 0 0 0 1
> 6. TaskA: PT - CT 1 -1 0 0
> 7. TaskA: CT - PT -1 1 0 0
> 8. TaskB: PT - CT 0 0 1 -1
> 9. TaskB: CT - PT 0 0 -1 1
> 10. PT: A-B 1 0 -1 0
> 11. PT: B-A -1 0 1 0
> 12. CT: A-B 0 1 0 -1
> 13. CT: B-A 0 -1 0 1
>
> I also include 3 F-tests:
> 1. Between group differences on Task A: Contrasts 1 & 2
> 2. Between group differences on Task B: Contrasts 3 & 4
> 3. Group x Task Interaction: Contrasts 10 & 12
>
> Does this design actually equate to a 2 x 2 ANOVA, or should I include a global mean EV? Also, is there a better way to model the interaction?
>
> Thanks for your help,
>
> David Glahn
>
--
Stephen M. Smith DPhil
Associate Director, FMRIB and Analysis Research Coordinator
Oxford University Centre for Functional MRI of the Brain
John Radcliffe Hospital, Headington, Oxford OX3 9DU, UK
+44 (0) 1865 222726 (fax 222717)
[log in to unmask] http://www.fmrib.ox.ac.uk/~steve
|