I have a question on which statistical test to use for 2 separate problems:

1) I have 4 different groups, 1A, 1B, 2A and 2C and have 3 different response variables measured at several different times. However I am not interested in the behavior over time and want to treat them as different problems for each time increment. I also have 7 particpants in each of the 4 groups and each participant has those 3 response variables measured.

Within those 4 groups, they contain 2 independent variables for which I want to know the interaction. Namely, "Letter" has 2 types, A or B and "Number" also has 2 types, 1 and 2. I want know if Number 1 behaves the same way with letters A and B as does Number 2 with A and B.

I know I need to find main and interaction effects. Can I use a 2 way ANOVA by taking the means over the 4 groups from the 7 participants from each group. Or should I be using a MANOVA for the same? Is there one method better than the other? Is there another way of analyzing this problem?

2) This time I have the same 4 groups but I want to compare each group to on of the other 3 with no interest if each Number behaves the same way with each Letter.

Thoughts? Ideas?