/* Fit a linear logistic model with a clinic-specific effect to a data set from a multi-center clinical trial. (3) Using a marginal approack (i.e. GEE) */ options ls=80 center pagesize=60 number label; data clin; input clinic trt $ fav unfav ; nij=fav+unfav; datalines; 1 drug 11 25 1 cntl 10 27 2 drug 16 4 2 cntl 22 10 3 drug 14 5 3 cntl 7 12 4 drug 2 14 4 cntl 1 16 5 drug 6 11 5 cntl 0 12 6 drug 1 10 6 cntl 0 10 7 drug 1 4 7 cntl 1 8 8 drug 4 2 8 cntl 6 1 ; proc genmod; class clinic trt; model fav/nij = trt / dist=bin link=logit; repeated subject=clinic / type=ind; title 'GEE: using a working independence model'; proc genmod; class clinic trt; model fav/nij = trt / dist=bin link=logit; repeated subject=clinic / type=cs; title 'GEE: using a working CS model'; proc genmod; class clinic trt; model fav/nij = trt / dist=bin link=logit; repeated subject=clinic / type=un; title 'GEE: using a working unstructured model'; /* default maxit=50 */ proc genmod; class clinic trt; model fav/nij = trt / dist=bin link=logit; repeated subject=clinic / type=cs maxit=500; title 'GEE: using MORE ITERATIONS and a working CS model'; run; ****************** GEE: using a working independence model 1 Criterion DF Value Value/DF Deviance 14 90.9602 6.4972 Scaled Deviance 14 90.9602 6.4972 Pearson Chi-Square 14 81.5126 5.8223 Scaled Pearson X2 14 81.5126 5.8223 Log Likelihood -179.1170 Analysis Of Initial Parameter Estimates Standard Wald 95% Chi- Parameter DF Estimate Error Confidence Limits Square Pr > ChiSq Intercept 1 -0.3102 0.1775 -0.6581 0.0378 3.05 0.0806 trt cntl 1 -0.4040 0.2514 -0.8968 0.0887 2.58 0.1080 trt drug 0 0.0000 0.0000 0.0000 0.0000 . . Scale 0 1.0000 0.0000 1.0000 1.0000 Analysis Of GEE Parameter Estimates Empirical Standard Error Estimates Standard 95% Confidence Parameter Estimate Error Limits Z Pr > |Z| Intercept -0.3102 0.3875 -1.0696 0.4493 -0.80 0.4235 trt cntl -0.4040 0.2462 -0.8866 0.0785 -1.64 0.1008 trt drug 0.0000 0.0000 0.0000 0.0000 . . GEE: using a working CS model 3 WARNING: Iteration limit exceeded. Analysis Of GEE Parameter Estimates Empirical Standard Error Estimates Standard 95% Confidence Parameter Estimate Error Limits Z Pr > |Z| Intercept 1.9203 1.7200 -1.4508 5.2914 1.12 0.2642 trt cntl -0.5823 0.6913 -1.9371 0.7725 -0.84 0.3996 trt drug 0.0000 0.0000 0.0000 0.0000 . . GEE: using a working unstructured model 5 WARNING: Iteration limit exceeded. Analysis Of GEE Parameter Estimates Empirical Standard Error Estimates Standard 95% Confidence Parameter Estimate Error Limits Z Pr > |Z| Intercept 1.9203 1.7200 -1.4508 5.2914 1.12 0.2642 trt cntl -0.5823 0.6913 -1.9371 0.7725 -0.84 0.3996 trt drug 0.0000 0.0000 0.0000 0.0000 . . GEE: using MORE ITERATIONS and a working CS model WARNING: Iteration limit exceeded. Analysis Of GEE Parameter Estimates Empirical Standard Error Estimates Standard 95% Confidence Parameter Estimate Error Limits Z Pr > |Z| Intercept 1.9203 1.7200 -1.4508 5.2914 1.12 0.2642 trt cntl -0.5823 0.6913 -1.9371 0.7725 -0.84 0.3996 trt drug 0.0000 0.0000 0.0000 0.0000 . .