PROC CATMOD fits models to data that are represented by a contingency table by using weighted least squares and a variety of link functions. Although PROC CATMOD also provides maximum likelihood estimation for logistic regression, PROC LOGISTIC is more efficient for most analyses.