Table IV.

Model characteristics.

Model	Properties	SAS procedurea	Interpretation
Logistic Regression	Requires independent observations	PROC GENMOD
Log-Binomial Regression	Requires independent observations	PROC GENMOD
Mixed Effects Models	Accounts for correlated observations Can yield unbiased estimates with unbalanced cluster sizes given adjustment for covariates predicting imbalance	PROC GLIMMIX	Conditional
Generalized Estimating Equation Models	Accounts for correlated observations Does not require distributional assumptions Assumes imbalance in data follows MCAR structure	PROC GENMOD	Marginal
Cluster-Weighted Generalized  Estimating Equation Models	Accounts for correlated observations Does not require distributional assumptions Accounts for non-ignorable cluster size	PROC GENMOD	Marginal

MCAR, Missing completely at random.

^aSAS statistical procedure (PROC) for generalized linear models (GENMOD) or generalized linear mixed models (GLIMMIX).