I've been asked to make a test for independence of residuals for a logistic regression, but I am not sure that I could use a Durbin Watson test for this (using a PROC REG). I've seen that using PROC GENMOD could be an option, however, I've never used PROC GENMOD.