We are about to perform a 3DANOVA2 on the results of a go/no-go experiment in which we've further categorized the responses as correct vs. incorrect. This gives us factor A with four levels (correct go, incorrect go, correct no-go, incorrect no-go) and Factor B with seven levels (i.e. subjects.) The sampled time points are linked to the response such that, for example, if subject #1 had 10 errors of commission, subject #1 will have 10 sampled time points while subject #2 might have had only 2 errors and therefore only 2 sampled time points. This gives unequal weight to subject #1 over subject #2 in the analysis. We are wondering if anyone has dealt with this problem and could offer any suggestions as to which mathematical manipulation AFNI might best handle.

Thank you in advance.
Mona