Dear André,
Not sure what you want to test, as the pattern of the vectors doesn't look consistent to me.
visit A: + - - + ... - + - + ...
visit B: ... - + - + ... - + - + ... [at least I guess there should be a + at the end]
In any case, once you have the correct contrast vectors for visit A and B, to average across time just sum up the weighting factors of the two "visit" contrasts and divide by 2. Thus, in case the pattern above is correct, it would be:
visit (A+B)/2: + - - + ... - + - + ... ... - + - + ... - + - + ...
with 1/8 as the absolute value of the weighting factor. Foward these con images into a two-sample t-test, [0.5 0.5] shows this differential condition effect averaged across time and groups, [1 -1] shows the corresponding main effect group with regard to that "differential condition effect averaged across time".
To test for main effect time, you have to subtract the contrast vector for visit B from that for visit A, thus (again assuming the pattern for visit A & B contrast vectors is the correct one):
visit A-B: + - - + ... - + - + ... ... + - + - ... + - + - ...
with 1/4 as the absolute value of the weighting factors. Foward these con images into a two-sample t-test, [0.5 0.5] shows main effect time for the "differential condition effect averaged across group", [1 -1] shows the interaction group x time.
Hope this helps
Helmut
|