PROC GENMOD DATA=mydata; MODEL y = x / DIST=BINOMIAL; OUTPUT OUT=preds PRED=pred_prob; RUN; | Mistake | Consequence | Fix | |---------|-------------|-----| | Forgetting CLASS for categorical vars | Treated as continuous | List all categorical vars in CLASS | | Using TYPE=AR(1) with unbalanced times | Invalid correlation | Use TYPE=UN or EXCH with CORRW | | Ignoring overdispersion | Inflated significance | Try DIST=NEGBIN or SCALE=DEVIANCE | | Not exponentiating coefficients for ratios | Misinterpretation | Use EXP option in ESTIMATE or LSMEANS | 8. Comparison: GENMOD vs. Other SAS Procedures for GLMs | Feature | GENMOD | GLIMMIX | LOGISTIC | REG | |---------|--------|---------|----------|-----| | Normal errors | Yes | Yes | No | Yes | | Binomial | Yes | Yes | Yes | No | | Poisson/NegBin | Yes | Yes | No | No | | Random effects | No (GEE only) | Yes | No | No | | GEE for repeated measures | Yes | No | No | No | | LSMEANS on link scale | Yes | Yes | Limited | No |
1. What is GENMOD? PROC GENMOD fits generalized linear models (GLMs) . It extends ordinary linear regression to response variables that have error distributions other than normal (e.g., binomial, Poisson, gamma). It also handles repeated measures via GEE (Generalized Estimating Equations). genmod