Print

---

We're looking to run a repeated measures ANOVA in SPM to test for interactions between our two experimental factors.  In our experiment, we have 17 subjects with 3 scans each (time factor), and each subject is assigned to 1 of 4 drug treatment groups (dose factor).  We're wanting to see if there's an interaction between time and dose.

We've started out using flexible factorial in SPM and have set up our design matrix to have 36 columns in the following order: subjects (17 columns), time (3 columns), dose (4 columns), time*dose (12 columns).  We're quite unsure about how to define the contrasts.  We've consulted the Glascher/Gitelman tutorial and other listserv posts, but we don't know how to expand what we find there to our case of 3x4.

Our first stab at it involved the following reasoning: a main effect of time modeled in those interaction terms might look something like:

M1 = zeros(3, 24) m1 m1 m1 m1

where m1 is:

1 −1/3 −1/3

-1/3 1 −1/3

-1/3 −1/3 1

 and a main effect of dose modeled similarly might look like:

M2 = zeros(3,24) m2 m2 m2

where m2 is:

1 −1/4 −1/4 −1/4

−1/4 1 −1/4 −1/4 

−1/4 −1/4 1 −1/4

−1/4 −1/4 −1/4 1

We thought of multiplying all pairs of rows of M1*M2 to give a 12x36 matrix but the contrast manager wouldn't accept it.

Our next attempt was just to use the single row: zeros(1,24) 3 0 −3 1 0 −1 −1 0 1 −3 0 3, but again we got the invalid contrast message.

Could anyone offer some advice to steer us in the right direction?

Thanks so much!!!