(1) the subject factor should be set to equal variance.
(2) see the attached file for an introduction to contrast weights. Although, in the examples there isn't the second repeated factor. It should be a good start though and hopefully provide you enough information to construct more complicated contrasts for the larger design.
ANOVAers:
I've taken the advice to specify a subject factor in the flexible factorial
model, and currently have a design matrix (see attached) with the following
factors:
1. subject (45 levels, indep, uneq var)
2. group (2 levels, indep, uneq var)...an across subject effect
3. region (2 levels, no indep, eq var)...a within subject effect
4. side (2 levels, no indep, eq var)...a within subject effect
When it has you specify main effects and interactions, the following are
specified (in this order):
1. main effect of factor 2
2. main effect of factor 3
3. main effect of factor 4
4. interaction of factors 2 and 3
5. main effect of factor 1
6. interaction of factors 2 and 4
7. interaction of factors 3 and 4
The result of all that is a design matrix such as the one attached.
Can someone take a look at the design matrix to see if it seems correct?
Also, how do I go about specifying contrasts using the flexible factorial
and with the design matrix attached in order to get F tests of main effects
and interactions, and corresponding t-tests? I've so far only dealt with the
full factorial approach, where it's all laid out for you.
many thanks!!!
Amit