Hi,
I was wondering if it was appropriate to have as the 3 levels to the a factor in 3dANOVA2 be as follows:
A
B
A-B
Where the A and B are main effects and A-B are contrasts performed for each subject individually. In this particular case I would take the mean of A-B by way of -amean # and use that for further data analysis.
I have noticed that using pre-contrasted main effects (A-B) and averaging them, as I mentioned above ,gives me better results than having the ANOVA program contrast them via -adiff. Hence the reason for the question.
Any answer would be much appreciated.
Thank you for your time!