The SAS System 15:38 Sunday, July 27, 2008 1 Obs clinic trt fav unfav n y 1 1 drug 11 25 36 -0.79633 2 1 cntl 10 27 37 -0.96281 3 2 drug 16 4 20 1.29928 4 2 cntl 22 10 32 0.76214 5 3 drug 14 5 19 0.96940 6 3 cntl 7 12 19 -0.51083 7 4 drug 2 14 16 -1.75786 8 4 cntl 1 16 17 -2.39790 9 5 drug 6 11 17 -0.57054 10 5 cntl 0 12 12 -3.21888 11 6 drug 1 10 11 -1.94591 12 6 cntl 0 10 10 -3.04452 13 7 drug 1 4 5 -1.09861 14 7 cntl 1 8 9 -1.73460 15 8 drug 4 2 6 0.58779 16 8 cntl 6 1 7 1.46634 When PROC GENMOD is used with trt as discrete 2 15:38 Sunday, July 27, 2008 The GENMOD Procedure Model Information Data Set WORK.CLINICS Distribution Binomial Link Function Logit Response Variable (Events) fav Response Variable (Trials) n Number of Observations Read 16 Number of Observations Used 16 Number of Events 102 Number of Trials 273 Class Level Information Class Levels Values clinic 8 1 2 3 4 5 6 7 8 trt 2 cntl drug Criteria For Assessing Goodness Of Fit Criterion DF Value Value/DF Deviance 7 9.7463 1.3923 Scaled Deviance 7 9.7463 1.3923 Pearson Chi-Square 7 8.0256 1.1465 Scaled Pearson X2 7 8.0256 1.1465 Log Likelihood -138.5100 Algorithm converged. Analysis Of Parameter Estimates Standard Wald 95% Confidence Chi- Parameter DF Estimate Error Limits Square Intercept 1 1.6628 0.6951 0.3005 3.0252 5.72 clinic 1 1 -2.2079 0.7195 -3.6181 -0.7977 9.42 clinic 2 1 -0.1525 0.7381 -1.5992 1.2942 0.04 Analysis Of Parameter Estimates Parameter Pr > ChiSq Intercept 0.0167 clinic 1 0.0022 clinic 2 0.8363 When PROC GENMOD is used with trt as discrete 3 15:38 Sunday, July 27, 2008 The GENMOD Procedure Analysis Of Parameter Estimates Standard Wald 95% Confidence Chi- Parameter DF Estimate Error Limits Square clinic 3 1 -1.0550 0.7457 -2.5166 0.4066 2.00 clinic 4 1 -3.6264 0.9071 -5.4043 -1.8484 15.98 clinic 5 1 -2.7278 0.8184 -4.3318 -1.1238 11.11 clinic 6 1 -4.3548 1.2293 -6.7642 -1.9455 12.55 clinic 7 1 -3.0056 1.0200 -5.0047 -1.0065 8.68 clinic 8 0 0.0000 0.0000 0.0000 0.0000 . trt cntl 1 -0.7769 0.3067 -1.3780 -0.1758 6.42 trt drug 0 0.0000 0.0000 0.0000 0.0000 . Scale 0 1.0000 0.0000 1.0000 1.0000 Analysis Of Parameter Estimates Parameter Pr > ChiSq clinic 3 0.1571 clinic 4 <.0001 clinic 5 0.0009 clinic 6 0.0004 clinic 7 0.0032 clinic 8 . trt cntl 0.0113 trt drug . Scale NOTE: The scale parameter was held fixed. Observation Statistics Observation Resraw Reschi Resdev StResdev StReschi Reslik 1 -2.212118 -0.764929 -0.775296 -1.458717 -1.439212 -1.444748 2 2.2121182 0.8921091 0.864869 1.3952661 1.4392118 1.4224873 3 -0.382232 -0.222042 -0.219137 -0.291474 -0.295337 -0.293159 4 0.3822318 0.1443281 0.1448001 0.2963025 0.2953368 0.2955677 5 1.698627 0.8156524 0.8348975 1.3035073 1.2734604 1.2858718 6 -1.698627 -0.782171 -0.788376 -1.283562 -1.27346 -1.277281 7 0.0306628 0.0233331 0.0232814 0.0400292 0.0401181 0.040088 8 -0.030663 -0.031163 -0.031309 -0.040307 -0.040118 -0.040232 When PROC GENMOD is used with trt as discrete 4 15:38 Sunday, July 27, 2008 The GENMOD Procedure Observation Statistics Observation Resraw Reschi Resdev StResdev StReschi Reslik 9 1.6419134 0.9120537 0.880921 1.6766908 1.7359468 1.7197939 10 -1.641914 -1.379196 -1.879186 -2.365267 -1.735947 -2.154686 11 0.3020916 0.373657 0.3522753 0.6393332 0.6781382 0.6665956 12 -0.302092 -0.558124 -0.783261 -0.951688 -0.678138 -0.87285 13 -0.035248 -0.038904 -0.03907 -0.055713 -0.055476 -0.055593 14 0.0352482 0.0379796 0.0377793 0.0551837 0.0554764 0.0553394 15 -1.043697 -1.164067 -1.047163 -1.338984 -1.488466 -1.398939 16 1.0436971 0.8676335 0.9293394 1.5943252 1.4884658 1.5252588 Using PROC GENMOD with hospital and trt treated as continuous 5 15:38 Sunday, July 27, 2008 Obs clinic cli1 cli2 cli3 cli4 cli5 cli6 cli7 cli8 trt treat fav unfav 1 1 1 0 0 0 0 0 0 0 drug 1 11 25 2 1 1 0 0 0 0 0 0 0 cntl 0 10 27 3 2 0 1 0 0 0 0 0 0 drug 1 16 4 4 2 0 1 0 0 0 0 0 0 cntl 0 22 10 5 3 0 0 1 0 0 0 0 0 drug 1 14 5 6 3 0 0 1 0 0 0 0 0 cntl 0 7 12 7 4 0 0 0 1 0 0 0 0 drug 1 2 14 8 4 0 0 0 1 0 0 0 0 cntl 0 1 16 9 5 0 0 0 0 1 0 0 0 drug 1 6 11 10 5 0 0 0 0 1 0 0 0 cntl 0 0 12 11 6 0 0 0 0 0 1 0 0 drug 1 1 10 12 6 0 0 0 0 0 1 0 0 cntl 0 0 10 13 7 0 0 0 0 0 0 1 0 drug 1 1 4 14 7 0 0 0 0 0 0 1 0 cntl 0 1 8 15 8 0 0 0 0 0 0 0 1 drug 1 4 2 16 8 0 0 0 0 0 0 0 1 cntl 0 6 1 Using PROC GENMOD with hospital and trt treated as continuous 6 15:38 Sunday, July 27, 2008 The GENMOD Procedure Model Information Data Set WORK.DETAIL Distribution Binomial Link Function Logit Response Variable (Events) fav Response Variable (Trials) n Number of Observations Read 16 Number of Observations Used 16 Number of Events 102 Number of Trials 273 Criteria For Assessing Goodness Of Fit Criterion DF Value Value/DF Deviance 7 9.7463 1.3923 Scaled Deviance 7 9.7463 1.3923 Pearson Chi-Square 7 8.0256 1.1465 Scaled Pearson X2 7 8.0256 1.1465 Log Likelihood -138.5100 Algorithm converged. Analysis Of Parameter Estimates Standard Wald 95% Chi- Parameter DF Estimate Error Confidence Limits Square Pr > ChiSq Intercept 0 0.0000 0.0000 0.0000 0.0000 . . cli1 1 -1.3220 0.3165 -1.9423 -0.7018 17.45 <.0001 cli2 1 0.7334 0.3294 0.0877 1.3791 4.96 0.0260 cli3 1 -0.1691 0.3634 -0.8813 0.5431 0.22 0.6417 cli4 1 -2.7405 0.6400 -3.9949 -1.4860 18.33 <.0001 cli5 1 -1.8419 0.5101 -2.8417 -0.8421 13.04 0.0003 cli6 1 -3.4689 1.0494 -5.5257 -1.4121 10.93 0.0009 cli7 1 -2.1197 0.7854 -3.6591 -0.5804 7.28 0.0070 cli8 1 0.8859 0.6755 -0.4380 2.2098 1.72 0.1897 treat 1 0.7769 0.3067 0.1758 1.3780 6.42 0.0113 Scale 0 1.0000 0.0000 1.0000 1.0000 NOTE: The scale parameter was held fixed. Lagrange Multiplier Statistics Parameter Chi-Square Pr > ChiSq Intercept . . Using PROC GENMOD with hospital and trt treated as continuous 7 15:38 Sunday, July 27, 2008 The GENMOD Procedure Observation Statistics Observation Resraw Reschi Resdev StResdev StReschi Reslik 1 -2.212118 -0.764929 -0.775296 -1.458717 -1.439212 -1.444748 2 2.2121182 0.8921091 0.864869 1.3952661 1.4392118 1.4224873 3 -0.382232 -0.222042 -0.219137 -0.291474 -0.295337 -0.293159 4 0.3822318 0.1443281 0.1448001 0.2963025 0.2953368 0.2955677 5 1.698627 0.8156524 0.8348975 1.3035073 1.2734604 1.2858718 6 -1.698627 -0.782171 -0.788376 -1.283562 -1.27346 -1.277281 7 0.0306628 0.0233331 0.0232814 0.0400292 0.0401181 0.040088 8 -0.030663 -0.031163 -0.031309 -0.040307 -0.040118 -0.040232 9 1.6419134 0.9120537 0.880921 1.6766908 1.7359468 1.7197939 10 -1.641914 -1.379196 -1.879186 -2.365267 -1.735947 -2.154686 11 0.3020916 0.373657 0.3522753 0.6393332 0.6781382 0.6665956 12 -0.302092 -0.558124 -0.783261 -0.951688 -0.678138 -0.87285 13 -0.035248 -0.038904 -0.03907 -0.055713 -0.055476 -0.055593 14 0.0352482 0.0379796 0.0377793 0.0551837 0.0554764 0.0553394 15 -1.043697 -1.164067 -1.047163 -1.338984 -1.488466 -1.398939 16 1.0436971 0.8676335 0.9293394 1.5943252 1.4884658 1.5252588 PROC MIXED with logit transformation 15:38 Sunday, July 27, 2008 8 The Mixed Procedure Model Information Data Set WORK.CLINICS Dependent Variable y Covariance Structure Variance Components Estimation Method REML Residual Variance Method Profile Fixed Effects SE Method Model-Based Degrees of Freedom Method Satterthwaite Class Level Information Class Levels Values clinic 8 1 2 3 4 5 6 7 8 trt 2 cntl drug Dimensions Covariance Parameters 2 Columns in X 3 Columns in Z 8 Subjects 1 Max Obs Per Subject 16 Number of Observations Number of Observations Read 16 Number of Observations Used 16 Number of Observations Not Used 0 Iteration History Iteration Evaluations -2 Res Log Like Criterion 0 1 55.16822239 1 1 48.96217261 0.00000000 Convergence criteria met. Covariance Parameter Estimates Cov Parm Estimate clinic 1.7161 Residual 0.5220 PROC MIXED with logit transformation 15:38 Sunday, July 27, 2008 9 The Mixed Procedure Fit Statistics -2 Res Log Likelihood 49.0 AIC (smaller is better) 53.0 AICC (smaller is better) 54.1 BIC (smaller is better) 53.1 Type 3 Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F trt 1 7 4.79 0.0647