S10 File. SAS output Glimmix outcomes above 75 years of age (quadratic analyses) The SAS System 16:24 Monday, October 19, 2015 1 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable OUT_smoking Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 101689 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 482451.94631 0.98499881 . 1 0 0 543259.71938 0.48221803 . 2 0 0 562485.79049 0.09082568 . 3 0 0 564440.34092 0.00193679 . 4 0 0 564459.92406 0.00000051 . 5 0 0 564459.92619 0.00000000 . The SAS System 16:24 Monday, October 19, 2015 2 The GLIMMIX Procedure Convergence criterion (PCONV=1.11022E-8) satisfied. Fit Statistics -2 Res Log Pseudo-Likelihood 564459.9 Generalized Chi-Square 100960.8 Gener. Chi-Square / DF 0.99 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 0.9929 0.004404 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 2.6820 0.3023 32186 8.87 <.0001 Age -0.07459 0.002968 69496 -25.14 <.0001 Gender 0 2.1559 0.2340 32186 9.21 <.0001 Gender 1 0 . . . . Year 0.06484 0.03371 69496 1.92 0.0545 YearSQ -0.00160 0.001468 69496 -1.09 0.2750 Year*Gender 0 -0.1790 0.04218 69496 -4.24 <.0001 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.004406 0.001868 69496 2.36 0.0183 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 69496 631.78 <.0001 Gender 1 32186 84.87 <.0001 Year 1 69496 1.37 0.2423 YearSQ 1 69496 0.41 0.5201 Year*Gender 1 69496 18.01 <.0001 YearSQ*Gender 1 69496 5.56 0.0183 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 -2.1717 0.01678 32186 -129.41 <.0001 0.05 -2.2046 -2.1388 1 -2.7895 0.01724 32186 -161.82 <.0001 0.05 -2.8233 -2.7557 The SAS System 16:24 Monday, October 19, 2015 3 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable OUT_microalbuminuria Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 81083 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 370287.89225 2.00000000 . 1 0 0 364393.16552 0.00846855 . 2 0 0 364506.63388 0.00000671 . 3 0 0 364506.68418 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 4 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 364506.7 Generalized Chi-Square 80969.31 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 0.9987 0.004960 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept -5.4567 0.1727 28471 -31.60 <.0001 Age 0.06726 0.001788 52605 37.62 <.0001 Gender 0 0.6622 0.1615 28471 4.10 <.0001 Gender 1 0 . . . . Year -0.1843 0.01793 52605 -10.28 <.0001 YearSQ 0.006612 0.000812 52605 8.14 <.0001 Year*Gender 0 0.02379 0.02890 52605 0.82 0.4104 Year*Gender 1 0 . . . . YearSQ*Gender 0 -0.00126 0.001278 52605 -0.99 0.3236 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 52605 1415.29 <.0001 Gender 1 28471 16.82 <.0001 Year 1 52605 142.41 <.0001 YearSQ 1 52605 87.59 <.0001 Year*Gender 1 52605 0.68 0.4104 YearSQ*Gender 1 52605 0.97 0.3236 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 -0.5249 0.01174 28471 -44.72 <.0001 0.05 -0.5479 -0.5019 1 -1.2727 0.01107 28471 -114.94 <.0001 0.05 -1.2944 -1.2510 The SAS System 16:24 Monday, October 19, 2015 5 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable OUT_macroalbuminuria Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 61347 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 15 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 277136.93279 0.89257476 . 1 0 0 334323.56688 0.50477681 . 2 0 0 363641.43872 0.37697800 . 3 0 0 370763.57231 0.06326179 . 4 0 0 371114.35117 0.00030690 . 5 0 0 371115.26009 0.00000000 . The SAS System 16:24 Monday, October 19, 2015 6 The GLIMMIX Procedure Convergence criterion (PCONV=1.11022E-8) satisfied. Fit Statistics -2 Res Log Pseudo-Likelihood 371115.3 Generalized Chi-Square 60631.52 Gener. Chi-Square / DF 0.99 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 0.9884 0.005644 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept -9.0267 0.3872 24401 -23.31 <.0001 Age 0.08461 0.003943 36939 21.46 <.0001 Gender 0 1.4397 0.2994 24401 4.81 <.0001 Gender 1 0 . . . . Year -0.1138 0.04211 36939 -2.70 0.0069 YearSQ 0.001140 0.001972 36939 0.58 0.5630 Year*Gender 0 -0.05629 0.05658 36939 -0.99 0.3197 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.002373 0.002624 36939 0.90 0.3659 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 36939 460.37 <.0001 Gender 1 24401 23.12 <.0001 Year 1 36939 25.17 <.0001 YearSQ 1 36939 3.14 0.0763 Year*Gender 1 36939 0.99 0.3197 YearSQ*Gender 1 36939 0.82 0.3659 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 -2.3983 0.02451 24401 -97.86 <.0001 0.05 -2.4464 -2.3503 1 -3.5249 0.03032 24401 -116.26 <.0001 0.05 -3.5843 -3.4654 The SAS System 16:24 Monday, October 19, 2015 7 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable PRO_AlbKreat Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 107758 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 506461.9685 2.00000000 . 1 0 0 508096.40818 2.00000000 . 2 0 0 508488.58512 0.00871922 . 3 0 0 508489.33986 0.00000006 . 4 0 0 508489.33986 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 8 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 508489.3 Generalized Chi-Square 108106.2 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.0033 0.004322 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 4.7276 0.1451 33985 32.58 <.0001 Age -0.06598 0.001611 73766 -40.95 <.0001 Gender 0 0.05150 0.1148 33985 0.45 0.6537 Gender 1 0 . . . . Year 0.1348 0.01281 73766 10.53 <.0001 YearSQ 0.000997 0.000612 73766 1.63 0.1032 Year*Gender 0 0.01103 0.02223 73766 0.50 0.6196 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.000044 0.001053 73766 0.04 0.9663 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 73766 1677.26 <.0001 Gender 1 33985 0.20 0.6537 Year 1 73766 159.45 <.0001 YearSQ 1 73766 3.75 0.0529 Year*Gender 1 73766 0.25 0.6196 YearSQ*Gender 1 73766 0.00 0.9663 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 1.4839 0.01318 33985 112.58 <.0001 0.05 1.4580 1.5097 1 1.2840 0.009904 33985 129.64 <.0001 0.05 1.2646 1.3034 The SAS System 16:24 Monday, October 19, 2015 9 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable OUT_HbA1c Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 102681 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 461087.69567 0.14194998 . 1 0 0 450513.44541 0.00793610 . 2 0 0 450777.80305 0.00001159 . 3 0 0 450777.98849 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 10 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 450778.0 Generalized Chi-Square 102658.0 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 0.9998 0.004413 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 0.7891 0.1377 32865 5.73 <.0001 Age -0.01673 0.001478 69809 -11.32 <.0001 Gender 0 -0.04646 0.1216 32865 -0.38 0.7023 Gender 1 0 . . . . Year 0.2483 0.01314 69809 18.90 <.0001 YearSQ -0.01049 0.000605 69809 -17.34 <.0001 Year*Gender 0 -0.00729 0.02240 69809 -0.33 0.7448 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.000843 0.001015 69809 0.83 0.4063 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 69809 128.11 <.0001 Gender 1 32865 0.15 0.7023 Year 1 69809 477.29 <.0001 YearSQ 1 69809 393.82 <.0001 Year*Gender 1 69809 0.11 0.7448 YearSQ*Gender 1 69809 0.69 0.4063 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 0.8107 0.01093 32865 74.15 <.0001 0.05 0.7893 0.8322 1 0.8041 0.008703 32865 92.40 <.0001 0.05 0.7871 0.8212 The SAS System 16:24 Monday, October 19, 2015 11 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable PRO_HbA1c Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 107758 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 478847.86826 2.00000000 . 1 0 0 588036.63406 2.00000000 . 2 0 0 651349.68492 0.51672521 . 3 0 0 671476.6513 0.07992179 . 4 0 0 673067.99536 0.00082873 . 5 0 0 673076.78602 0.00000005 . 6 0 0 673076.78629 0.00000000 . The SAS System 16:24 Monday, October 19, 2015 12 The GLIMMIX Procedure Convergence criterion (PCONV=1.11022E-8) satisfied. Fit Statistics -2 Res Log Pseudo-Likelihood 673076.8 Generalized Chi-Square 113505.9 Gener. Chi-Square / DF 1.05 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.0534 0.004538 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 4.0813 0.2710 33985 15.06 <.0001 Age -0.03548 0.003163 73766 -11.21 <.0001 Gender 0 -0.01907 0.1525 33985 -0.12 0.9005 Gender 1 0 . . . . Year 0.04336 0.01960 73766 2.21 0.0269 YearSQ 0.008717 0.001057 73766 8.25 <.0001 Year*Gender 0 0.006330 0.03358 73766 0.19 0.8505 Year*Gender 1 0 . . . . YearSQ*Gender 0 -0.00047 0.001772 73766 -0.26 0.7929 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 73766 125.76 <.0001 Gender 1 33985 0.02 0.9005 Year 1 73766 7.68 0.0056 YearSQ 1 73766 91.62 <.0001 Year*Gender 1 73766 0.04 0.8505 YearSQ*Gender 1 73766 0.07 0.7929 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 3.2464 0.02874 33985 112.95 <.0001 0.05 3.1901 3.3028 1 3.2648 0.02341 33985 139.48 <.0001 0.05 3.2189 3.3106 The SAS System 16:24 Monday, October 19, 2015 13 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable OUT_SBP Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 104505 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 448392.66276 1.25365633 . 1 0 0 445753.19083 0.03413458 . 2 0 0 445786.57262 0.00002027 . 3 0 0 445786.58349 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 14 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 445786.6 Generalized Chi-Square 104523.2 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.0002 0.004376 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept -1.9172 0.1402 33266 -13.68 <.0001 Age 0.000328 0.001389 71232 0.24 0.8135 Gender 0 0.6214 0.1360 33266 4.57 <.0001 Gender 1 0 . . . . Year 0.1682 0.01508 71232 11.15 <.0001 YearSQ -0.00388 0.000663 71232 -5.85 <.0001 Year*Gender 0 -0.04746 0.02412 71232 -1.97 0.0491 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.001513 0.001056 71232 1.43 0.1519 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 71232 0.06 0.8135 Gender 1 33266 20.89 <.0001 Year 1 71232 143.46 <.0001 YearSQ 1 71232 34.94 <.0001 Year*Gender 1 71232 3.87 0.0491 YearSQ*Gender 1 71232 2.05 0.1519 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 -0.1311 0.01006 33266 -13.04 <.0001 0.05 -0.1509 -0.1114 1 -0.4061 0.008173 33266 -49.69 <.0001 0.05 -0.4221 -0.3901 The SAS System 16:24 Monday, October 19, 2015 15 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable PRO_SBP Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 107758 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 451056.15191 2.00000000 . 1 0 0 588471.12357 2.00000000 . 2 0 0 672982.90932 1.66729228 . 3 0 0 700574.78984 0.74664572 . 4 0 0 703239.06719 0.02878890 . 5 0 0 703264.8095 0.00000631 . 6 0 0 703264.81241 0.00000000 . The SAS System 16:24 Monday, October 19, 2015 16 The GLIMMIX Procedure Convergence criterion (PCONV=1.11022E-8) satisfied. Fit Statistics -2 Res Log Pseudo-Likelihood 703264.8 Generalized Chi-Square 107817.0 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.0006 0.004311 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 4.1596 0.3172 33985 13.11 <.0001 Age -0.03881 0.003742 73766 -10.37 <.0001 Gender 0 -0.05117 0.1590 33985 -0.32 0.7476 Gender 1 0 . . . . Year 0.2967 0.02136 73766 13.89 <.0001 YearSQ -0.00703 0.001158 73766 -6.07 <.0001 Year*Gender 0 0.001407 0.03677 73766 0.04 0.9695 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.000584 0.001966 73766 0.30 0.7665 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 73766 107.58 <.0001 Gender 1 33985 0.10 0.7476 Year 1 73766 261.76 <.0001 YearSQ 1 73766 46.94 <.0001 Year*Gender 1 73766 0.00 0.9695 YearSQ*Gender 1 73766 0.09 0.7665 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 3.6654 0.03263 33985 112.34 <.0001 0.05 3.6015 3.7294 1 3.5983 0.02525 33985 142.49 <.0001 0.05 3.5488 3.6478 The SAS System 16:24 Monday, October 19, 2015 17 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable Out_Obesity Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 94583 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 424675.00409 0.15253203 . 1 0 0 418179.82888 0.00648610 . 2 0 0 418305.83407 0.00000245 . 3 0 0 418305.90099 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 18 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 418305.9 Generalized Chi-Square 94578.71 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.0000 0.004599 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 4.1979 0.1617 30373 25.97 <.0001 Age -0.06051 0.001696 64203 -35.69 <.0001 Gender 0 -0.6386 0.1626 30373 -3.93 <.0001 Gender 1 0 . . . . Year 0.02112 0.01593 64203 1.33 0.1848 YearSQ -0.00060 0.000705 64203 -0.85 0.3936 Year*Gender 0 -0.04403 0.02880 64203 -1.53 0.1262 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.002623 0.001258 64203 2.09 0.0370 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 64203 1273.59 <.0001 Gender 1 30373 15.43 <.0001 Year 1 64203 0.00 0.9504 YearSQ 1 64203 1.27 0.2591 Year*Gender 1 64203 2.34 0.1262 YearSQ*Gender 1 64203 4.35 0.0370 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 -1.2588 0.01247 30373 -100.94 <.0001 0.05 -1.2832 -1.2343 1 -0.5116 0.008751 30373 -58.46 <.0001 0.05 -0.5287 -0.4944 The SAS System 16:24 Monday, October 19, 2015 19 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable Out_Overweight Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 94583 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 403051.76139 0.25381924 . 1 0 0 401418.85529 0.00934850 . 2 0 0 401429.68763 0.00000090 . 3 0 0 401429.68818 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 20 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 401429.7 Generalized Chi-Square 94581.83 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.0001 0.004599 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept -0.8361 0.1470 30373 -5.69 <.0001 Age 0.005876 0.001491 64203 3.94 <.0001 Gender 0 -0.03633 0.1415 30373 -0.26 0.7974 Gender 1 0 . . . . Year 0.01270 0.01542 64203 0.82 0.4102 YearSQ -0.00090 0.000684 64203 -1.32 0.1860 Year*Gender 0 0.07381 0.02519 64203 2.93 0.0034 Year*Gender 1 0 . . . . YearSQ*Gender 0 -0.00303 0.001104 64203 -2.75 0.0060 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 64203 15.53 <.0001 Gender 1 30373 0.07 0.7974 Year 1 64203 15.51 <.0001 YearSQ 1 64203 19.20 <.0001 Year*Gender 1 64203 8.59 0.0034 YearSQ*Gender 1 64203 7.54 0.0060 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 0.02970 0.01035 30373 2.87 0.0041 0.05 0.009406 0.04999 1 -0.3572 0.008544 30373 -41.81 <.0001 0.05 -0.3739 -0.3405 The SAS System 16:24 Monday, October 19, 2015 21 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable OUT_BMI Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 94583 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 441876.81001 0.54990853 . 1 0 0 433368.93606 0.04038802 . 2 0 0 433558.22357 0.00004228 . 3 0 0 433558.32931 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 22 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 433558.3 Generalized Chi-Square 94600.97 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.0003 0.004600 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept -5.8259 0.1704 30373 -34.19 <.0001 Age 0.05855 0.001697 64203 34.50 <.0001 Gender 0 0.6629 0.1575 30373 4.21 <.0001 Gender 1 0 . . . . Year -0.04676 0.01833 64203 -2.55 0.0107 YearSQ 0.002156 0.000814 64203 2.65 0.0081 Year*Gender 0 -0.03555 0.02828 64203 -1.26 0.2088 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.000685 0.001250 64203 0.55 0.5838 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 64203 1190.28 <.0001 Gender 1 30373 17.71 <.0001 Year 1 64203 20.83 <.0001 YearSQ 1 64203 15.98 <.0001 Year*Gender 1 64203 1.58 0.2088 YearSQ*Gender 1 64203 0.30 0.5838 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 -1.0108 0.01179 30373 -85.75 <.0001 0.05 -1.0339 -0.9877 1 -1.3328 0.01042 30373 -127.90 <.0001 0.05 -1.3532 -1.3123 The SAS System 16:24 Monday, October 19, 2015 23 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable PRO_BMI Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 107758 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 511532.68145 0.90559637 . 1 0 0 555284.19432 0.44691118 . 2 0 0 567955.86417 0.04087435 . 3 0 0 568732.90515 0.00026558 . 4 0 0 568736.54527 0.00000001 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 24 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 568736.5 Generalized Chi-Square 107666.0 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 0.9992 0.004305 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 7.1759 0.1721 33985 41.70 <.0001 Age -0.09277 0.001970 73766 -47.10 <.0001 Gender 0 0.3296 0.1202 33985 2.74 0.0061 Gender 1 0 . . . . Year 0.2107 0.01395 73766 15.11 <.0001 YearSQ -0.00175 0.000696 73766 -2.52 0.0117 Year*Gender 0 -0.07688 0.02475 73766 -3.11 0.0019 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.005319 0.001234 73766 4.31 <.0001 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 73766 2218.52 <.0001 Gender 1 33985 7.52 0.0061 Year 1 73766 193.79 <.0001 YearSQ 1 73766 2.16 0.1421 Year*Gender 1 73766 9.65 0.0019 YearSQ*Gender 1 73766 18.59 <.0001 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 2.3251 0.01829 33985 127.10 <.0001 0.05 2.2892 2.3609 1 2.0617 0.01305 33985 158.04 <.0001 0.05 2.0362 2.0873 The SAS System 16:24 Monday, October 19, 2015 25 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable OUT_CholHDL Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 98458 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 440983.13672 2.00000000 . 1 0 0 431295.33909 0.14939801 . 2 0 0 431531.74412 0.00017841 . 3 0 0 431531.90885 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 26 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 431531.9 Generalized Chi-Square 98473.23 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.0002 0.004508 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 0.8653 0.1464 32023 5.91 <.0001 Age -0.01829 0.001526 66428 -11.99 <.0001 Gender 0 -0.3492 0.1392 32023 -2.51 0.0121 Gender 1 0 . . . . Year 0.2120 0.01495 66428 14.18 <.0001 YearSQ -0.00727 0.000675 66428 -10.76 <.0001 Year*Gender 0 0.009805 0.02509 66428 0.39 0.6959 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.000086 0.001113 66428 0.08 0.9383 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 66428 143.68 <.0001 Gender 1 32023 6.29 0.0121 Year 1 66428 298.95 <.0001 YearSQ 1 66428 168.47 <.0001 Year*Gender 1 66428 0.15 0.6959 YearSQ*Gender 1 66428 0.01 0.9383 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 0.6519 0.01082 32023 60.26 <.0001 0.05 0.6307 0.6731 1 0.8587 0.009020 32023 95.21 <.0001 0.05 0.8411 0.8764 The SAS System 16:24 Monday, October 19, 2015 27 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable PRO_CholHDL Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 107758 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 502478.29564 2.00000000 . 1 0 0 574627.44375 2.00000000 . 2 0 0 604856.31452 1.96738599 . 3 0 0 608810.45331 2.00000000 . 4 0 0 608870.46649 0.00046728 . 5 0 0 608870.48338 0.00000001 . The SAS System 16:24 Monday, October 19, 2015 28 The GLIMMIX Procedure Convergence criterion (PCONV=1.11022E-8) satisfied. Fit Statistics -2 Res Log Pseudo-Likelihood 608870.5 Generalized Chi-Square 109129.0 Gener. Chi-Square / DF 1.01 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.0128 0.004363 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 6.1469 0.1990 33985 30.90 <.0001 Age -0.07878 0.002316 73766 -34.02 <.0001 Gender 0 0.1296 0.1233 33985 1.05 0.2932 Gender 1 0 . . . . Year 0.2189 0.01519 73766 14.41 <.0001 YearSQ 9.507E-7 0.000791 73766 0.00 0.9990 Year*Gender 0 -0.01517 0.02649 73766 -0.57 0.5670 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.000961 0.001365 73766 0.70 0.4811 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 73766 1157.40 <.0001 Gender 1 33985 1.11 0.2932 Year 1 73766 254.20 <.0001 YearSQ 1 73766 0.50 0.4802 Year*Gender 1 73766 0.33 0.5670 YearSQ*Gender 1 73766 0.50 0.4811 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 2.6710 0.02148 33985 124.36 <.0001 0.05 2.6289 2.7131 1 2.5696 0.01649 33985 155.86 <.0001 0.05 2.5373 2.6019 The SAS System 16:24 Monday, October 19, 2015 29 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable OUT_EyeExamination Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 73028 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 14 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 351943.12415 0.71443105 . 1 0 0 386133.75424 0.24616666 . 2 0 0 391790.84408 0.00904237 . 3 0 0 391941.05662 0.00000719 . 4 0 0 391941.18065 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 30 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 391941.2 Generalized Chi-Square 72993.92 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 0.9996 0.005232 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept -3.2005 0.2568 26519 -12.46 <.0001 Age 0.03177 0.002810 46502 11.31 <.0001 Gender 0 -0.5484 0.2619 26519 -2.09 0.0363 Gender 1 0 . . . . Year -0.2358 0.02398 46502 -9.83 <.0001 YearSQ 0.008348 0.001128 46502 7.40 <.0001 Year*Gender 0 0.04614 0.04677 46502 0.99 0.3240 Year*Gender 1 0 . . . . YearSQ*Gender 0 -0.00153 0.002106 46502 -0.72 0.4689 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 46502 127.86 <.0001 Gender 1 26519 4.38 0.0363 Year 1 46502 82.68 <.0001 YearSQ 1 46502 51.87 <.0001 Year*Gender 1 46502 0.97 0.3240 YearSQ*Gender 1 46502 0.52 0.4689 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 -2.4674 0.02198 26519 -112.28 <.0001 0.05 -2.5105 -2.4244 1 -2.2566 0.01647 26519 -137.03 <.0001 0.05 -2.2888 -2.2243 The SAS System 16:24 Monday, October 19, 2015 31 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable PRO_EyeExamination Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 107758 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 499722.01665 2.00000000 . 1 0 0 523409.97349 1.04383572 . 2 0 0 526475.17349 0.01505264 . 3 0 0 526516.4807 0.00000465 . 4 0 0 526516.49062 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 32 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 526516.5 Generalized Chi-Square 121334.6 Gener. Chi-Square / DF 1.13 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.1261 0.004851 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 4.3830 0.1577 33985 27.80 <.0001 Age -0.03951 0.001733 73766 -22.80 <.0001 Gender 0 -0.1642 0.1239 33985 -1.33 0.1850 Gender 1 0 . . . . Year -0.4460 0.01404 73766 -31.77 <.0001 YearSQ 0.03331 0.000687 73766 48.52 <.0001 Year*Gender 0 -0.01611 0.02417 73766 -0.67 0.5050 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.002622 0.001170 73766 2.24 0.0250 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 73766 519.74 <.0001 Gender 1 33985 1.76 0.1850 Year 1 73766 1411.24 <.0001 YearSQ 1 73766 3496.87 <.0001 Year*Gender 1 73766 0.44 0.5050 YearSQ*Gender 1 73766 5.02 0.0250 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 1.3075 0.01468 33985 89.10 <.0001 0.05 1.2788 1.3363 1 1.2266 0.01118 33985 109.72 <.0001 0.05 1.2046 1.2485 The SAS System 16:24 Monday, October 19, 2015 33 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable OUT_FeetExamination Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 56603 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 257902.58377 2.00000000 . 1 0 0 252572.17607 0.05603609 . 2 0 0 252704.81315 0.00001935 . 3 0 0 252704.91342 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 34 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 252704.9 Generalized Chi-Square 56518.24 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 0.9986 0.005936 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept -5.8471 0.1868 24515 -31.30 <.0001 Age 0.07187 0.002031 32081 35.38 <.0001 Gender 0 0.5053 0.1532 24515 3.30 0.0010 Gender 1 0 . . . . Year -0.1975 0.01931 32081 -10.23 <.0001 YearSQ 0.008681 0.000926 32081 9.37 <.0001 Year*Gender 0 -0.01560 0.03140 32081 -0.50 0.6193 Year*Gender 1 0 . . . . YearSQ*Gender 0 -0.00032 0.001491 32081 -0.21 0.8308 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 32081 1251.55 <.0001 Gender 1 24515 10.88 0.0010 Year 1 32081 171.00 <.0001 YearSQ 1 32081 130.75 <.0001 Year*Gender 1 32081 0.25 0.6193 YearSQ*Gender 1 32081 0.05 0.8308 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 -0.8039 0.01470 24515 -54.67 <.0001 0.05 -0.8327 -0.7750 1 -1.0311 0.01252 24515 -82.38 <.0001 0.05 -1.0556 -1.0066 The SAS System 16:24 Monday, October 19, 2015 35 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable PRO_FeetExamination Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 107758 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 518602.57105 1.62792982 . 1 0 0 535912.92848 1.02841787 . 2 0 0 538011.58241 0.01400117 . 3 0 0 538031.1299 0.00000188 . 4 0 0 538031.13208 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 36 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 538031.1 Generalized Chi-Square 108014.3 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.0024 0.004319 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 2.9748 0.1638 33985 18.16 <.0001 Age -0.03621 0.001853 73766 -19.54 <.0001 Gender 0 0.1301 0.1176 33985 1.11 0.2685 Gender 1 0 . . . . Year 0.1381 0.01345 73766 10.27 <.0001 YearSQ -0.00013 0.000662 73766 -0.19 0.8480 Year*Gender 0 -0.00911 0.02340 73766 -0.39 0.6970 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.000503 0.001133 73766 0.44 0.6574 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 73766 381.75 <.0001 Gender 1 33985 1.22 0.2685 Year 1 73766 130.38 <.0001 YearSQ 1 73766 0.05 0.8262 Year*Gender 1 73766 0.15 0.6970 YearSQ*Gender 1 73766 0.20 0.6574 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 1.8882 0.01492 33985 126.52 <.0001 0.05 1.8590 1.9175 1 1.7879 0.01153 33985 155.01 <.0001 0.05 1.7653 1.8105 The SAS System 16:24 Monday, October 19, 2015 37 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable OUT_Insulin Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 107758 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 518997.40874 2.00000000 . 1 0 0 517549.76893 1.05908742 . 2 0 0 517660.6172 0.00055241 . 3 0 0 517660.66149 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 38 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 517660.7 Generalized Chi-Square 107751.2 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.0000 0.004308 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept -3.1194 0.1594 33985 -19.57 <.0001 Age 0.02442 0.001727 73766 14.14 <.0001 Gender 0 -0.1907 0.1393 33985 -1.37 0.1712 Gender 1 0 . . . . Year -0.07143 0.01446 73766 -4.94 <.0001 YearSQ 0.003145 0.000674 73766 4.66 <.0001 Year*Gender 0 -0.00103 0.02593 73766 -0.04 0.9684 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.000058 0.001187 73766 0.05 0.9608 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 73766 199.88 <.0001 Gender 1 33985 1.87 0.1712 Year 1 73766 30.78 <.0001 YearSQ 1 73766 28.59 <.0001 Year*Gender 1 73766 0.00 0.9684 YearSQ*Gender 1 73766 0.00 0.9608 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 -1.7094 0.01368 33985 -124.98 <.0001 0.05 -1.7362 -1.6826 1 -1.5156 0.01020 33985 -148.63 <.0001 0.05 -1.5356 -1.4956 The SAS System 16:24 Monday, October 19, 2015 39 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable OUT_Diet Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 107758 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 516183.88831 0.45698737 . 1 0 0 509409.14927 0.01272093 . 2 0 0 509543.43432 0.00002120 . 3 0 0 509543.48166 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 40 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 509543.5 Generalized Chi-Square 107773.0 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.0002 0.004309 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept -1.6600 0.1655 33985 -10.03 <.0001 Age -0.00448 0.001726 73766 -2.59 0.0095 Gender 0 -0.1632 0.1586 33985 -1.03 0.3036 Gender 1 0 . . . . Year 0.09239 0.01637 73766 5.64 <.0001 YearSQ -0.00364 0.000740 73766 -4.92 <.0001 Year*Gender 0 0.04112 0.02854 73766 1.44 0.1496 Year*Gender 1 0 . . . . YearSQ*Gender 0 -0.00224 0.001266 73766 -1.77 0.0768 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 73766 6.72 0.0095 Gender 1 33985 1.06 0.3036 Year 1 73766 62.66 <.0001 YearSQ 1 73766 56.48 <.0001 Year*Gender 1 73766 2.08 0.1496 YearSQ*Gender 1 73766 3.13 0.0768 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 -1.4923 0.01272 33985 -117.29 <.0001 0.05 -1.5173 -1.4674 1 -1.4686 0.01002 33985 -146.53 <.0001 0.05 -1.4882 -1.4489 The SAS System 16:24 Monday, October 19, 2015 41 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable OUT_DMoralMed Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 107758 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 466690.67449 0.19074372 . 1 0 0 461489.10774 0.00534375 . 2 0 0 461554.19562 0.00000176 . 3 0 0 461554.20895 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 42 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 461554.2 Generalized Chi-Square 107761.8 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.0001 0.004309 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 1.4599 0.1276 33985 11.44 <.0001 Age -0.01363 0.001368 73766 -9.97 <.0001 Gender 0 0.1379 0.1125 33985 1.23 0.2202 Gender 1 0 . . . . Year 0.03386 0.01194 73766 2.84 0.0046 YearSQ -0.00212 0.000551 73766 -3.85 0.0001 Year*Gender 0 -0.01661 0.02069 73766 -0.80 0.4221 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.001171 0.000937 73766 1.25 0.2112 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 73766 99.32 <.0001 Gender 1 33985 1.50 0.2202 Year 1 73766 6.10 0.0135 YearSQ 1 73766 10.73 0.0011 Year*Gender 1 73766 0.64 0.4221 YearSQ*Gender 1 73766 1.56 0.2112 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 0.5527 0.01022 33985 54.09 <.0001 0.05 0.5327 0.5728 1 0.4253 0.008000 33985 53.16 <.0001 0.05 0.4096 0.4410 The SAS System 16:24 Monday, October 19, 2015 43 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable OUT_AntihypertensiveMed Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 107758 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 514809.3648 2.00000000 . 1 0 0 508680.98358 0.57226458 . 2 0 0 508787.96373 0.00030574 . 3 0 0 508788.01218 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 44 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 508788.0 Generalized Chi-Square 107859.1 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.0010 0.004313 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 0.3865 0.1535 33985 2.52 0.0118 Age -0.00240 0.001710 73766 -1.40 0.1608 Gender 0 -0.3528 0.1145 33985 -3.08 0.0021 Gender 1 0 . . . . Year 0.2052 0.01319 73766 15.55 <.0001 YearSQ -0.00746 0.000633 73766 -11.78 <.0001 Year*Gender 0 0.002416 0.02202 73766 0.11 0.9126 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.000767 0.001034 73766 0.74 0.4583 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 73766 1.97 0.1608 Gender 1 33985 9.49 0.0021 Year 1 73766 351.66 <.0001 YearSQ 1 73766 187.21 <.0001 Year*Gender 1 73766 0.01 0.9126 YearSQ*Gender 1 73766 0.55 0.4583 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 1.3376 0.01217 33985 109.88 <.0001 0.05 1.3137 1.3614 1 1.5278 0.01027 33985 148.81 <.0001 0.05 1.5076 1.5479 The SAS System 16:24 Monday, October 19, 2015 45 The GLIMMIX Procedure Model Information Data Set SYS.ZODIAC_45_OVER_75 Response Variable OUT_LipidLoweringMed Response Distribution Binomial Link Function Logit Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Residual PL Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 107758 Dimensions R-side Cov. Parameters 1 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique None Parameters 0 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Subiterations Function Change Gradient 0 0 0 477636.75258 0.69288663 . 1 0 0 475049.08962 0.06884842 . 2 0 0 475887.17843 0.00196092 . 3 0 0 475907.13051 0.00000230 . 4 0 0 475907.14484 0.00000000 . Convergence criterion (PCONV=1.11022E-8) satisfied. The SAS System 16:24 Monday, October 19, 2015 46 The GLIMMIX Procedure Fit Statistics -2 Res Log Pseudo-Likelihood 475907.1 Generalized Chi-Square 108108.1 Gener. Chi-Square / DF 1.00 Covariance Parameter Estimates Standard Cov Parm Estimate Error Residual (VC) 1.0033 0.004323 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 4.9255 0.1638 33985 30.08 <.0001 Age -0.1104 0.001490 73766 -74.08 <.0001 Gender 0 -0.4246 0.2019 33985 -2.10 0.0355 Gender 1 0 . . . . Year 0.5238 0.01988 73766 26.35 <.0001 YearSQ -0.01429 0.000844 73766 -16.93 <.0001 Year*Gender 0 0.09360 0.03467 73766 2.70 0.0069 Year*Gender 1 0 . . . . YearSQ*Gender 0 -0.00354 0.001457 73766 -2.43 0.0150 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 73766 5488.42 <.0001 Gender 1 33985 4.42 0.0355 Year 1 73766 1083.14 <.0001 YearSQ 1 73766 486.00 <.0001 Year*Gender 1 73766 7.29 0.0069 YearSQ*Gender 1 73766 5.91 0.0150 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 0.3884 0.01109 33985 35.03 <.0001 0.05 0.3667 0.4102 1 0.2281 0.008573 33985 26.60 <.0001 0.05 0.2113 0.2449 The GLIMMIX Procedure Model Information Data Set WORK.ZODIAC_45_OVER_75 Response Variable HbA1c Response Distribution Gaussian Link Function Identity Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Restricted Maximum Likelihood Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 102681 Dimensions R-side Cov. Parameters 2 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique Dual Quasi-Newton Parameters in Optimization 1 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Evaluations Function Change Gradient 0 0 4 721554.15019 . 3084.375 1 0 4 721117.61407 436.53611933 407.8937 2 0 2 721105.17203 12.44203475 127.4992 3 0 2 721103.91364 1.25838932 8.415752 4 0 2 721103.90829 0.00535807 0.160473 5 0 2 721103.90828 0.00000198 0.000197 The SAS System 13:41 Wednesday, October 14, 2015 31 The GLIMMIX Procedure Convergence criterion (GCONV=1E-8) satisfied. Fit Statistics -2 Res Log Likelihood 721103.9 AIC (smaller is better) 721107.9 AICC (smaller is better) 721107.9 BIC (smaller is better) 721124.8 CAIC (smaller is better) 721126.8 HQIC (smaller is better) 721113.3 Generalized Chi-Square 4145952 Gener. Chi-Square / DF 40.38 Covariance Parameter Estimates Standard Cov Parm Subject Estimate Error CS BSISID 55.0403 0.5868 Residual 40.3798 0.2171 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 48.5051 0.7430 32865 65.28 <.0001 Age 0.1064 0.008913 69809 11.93 <.0001 Gender 0 0.4965 0.5829 32865 0.85 0.3944 Gender 1 0 . . . . Year -1.3120 0.05668 69809 -23.15 <.0001 YearSQ 0.05978 0.002470 69809 24.20 <.0001 Year*Gender 0 -0.02693 0.09918 69809 -0.27 0.7860 Year*Gender 1 0 . . . . YearSQ*Gender 0 -0.00120 0.004247 69809 -0.28 0.7783 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 69809 142.44 <.0001 Gender 1 32865 0.73 0.3944 Year 1 69809 713.28 <.0001 YearSQ 1 69809 773.96 <.0001 Year*Gender 1 69809 0.07 0.7860 YearSQ*Gender 1 69809 0.08 0.7783 The SAS System 13:41 Wednesday, October 14, 2015 32 The GLIMMIX Procedure Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 50.5769 0.07569 32865 668.24 <.0001 0.05 50.4285 50.7252 1 50.6356 0.06119 32865 827.57 <.0001 0.05 50.5156 50.7555 The SAS System 13:41 Wednesday, October 14, 2015 33 The GLIMMIX Procedure Model Information Data Set WORK.ZODIAC_45_OVER_75 Response Variable SBP Response Distribution Gaussian Link Function Identity Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Restricted Maximum Likelihood Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 104505 Dimensions R-side Cov. Parameters 2 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique Dual Quasi-Newton Parameters in Optimization 1 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Evaluations Function Change Gradient 0 0 4 883584.92302 . 921.878 1 0 4 883561.36378 23.55924166 18.22894 2 0 2 883561.35351 0.01027160 1.478299 3 0 2 883561.35344 0.00006784 0.002624 Convergence criterion (GCONV=1E-8) satisfied. The SAS System 13:41 Wednesday, October 14, 2015 34 The GLIMMIX Procedure Fit Statistics -2 Res Log Likelihood 883561.4 AIC (smaller is better) 883565.4 AICC (smaller is better) 883565.4 BIC (smaller is better) 883582.2 CAIC (smaller is better) 883584.2 HQIC (smaller is better) 883570.7 Generalized Chi-Square 20030322 Gener. Chi-Square / DF 191.68 Covariance Parameter Estimates Standard Cov Parm Subject Estimate Error CS BSISID 150.06 1.8198 Residual 191.68 1.0086 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 164.72 1.3936 33266 118.20 <.0001 Age -0.09057 0.01620 71232 -5.59 <.0001 Gender 0 -4.4275 1.1450 33266 -3.87 0.0001 Gender 1 0 . . . . Year -1.4488 0.1128 71232 -12.85 <.0001 YearSQ 0.02369 0.004975 71232 4.76 <.0001 Year*Gender 0 0.1227 0.1978 71232 0.62 0.5351 Year*Gender 1 0 . . . . YearSQ*Gender 0 -0.00171 0.008559 71232 -0.20 0.8417 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 71232 31.25 <.0001 Gender 1 33266 14.95 0.0001 Year 1 71232 196.80 <.0001 YearSQ 1 71232 28.39 <.0001 Year*Gender 1 71232 0.38 0.5351 YearSQ*Gender 1 71232 0.04 0.8417 Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 139.78 0.1342 33266 1041.32 <.0001 0.05 139.52 140.04 1 142.93 0.1082 33266 1320.53 <.0001 0.05 142.72 143.14 The SAS System 13:41 Wednesday, October 14, 2015 35 The GLIMMIX Procedure Model Information Data Set WORK.ZODIAC_45_OVER_75 Response Variable BMI Response Distribution Gaussian Link Function Identity Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Restricted Maximum Likelihood Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 94583 Dimensions R-side Cov. Parameters 2 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique Dual Quasi-Newton Parameters in Optimization 1 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Evaluations Function Change Gradient 0 0 4 437291.54688 . 1462.644 1 0 2 435884.26721 1407.2796678 556.1981 2 0 2 435546.09044 338.17677348 217.1126 3 0 2 435470.00233 76.08811634 53.50699 4 0 2 435464.35955 5.64277119 7.06985 5 0 2 435464.25467 0.10488179 0.272971 6 0 3 435464.25451 0.00015767 0.00032 The SAS System 13:41 Wednesday, October 14, 2015 36 The GLIMMIX Procedure Convergence criterion (GCONV=1E-8) satisfied. Fit Statistics -2 Res Log Likelihood 435464.3 AIC (smaller is better) 435468.3 AICC (smaller is better) 435468.3 BIC (smaller is better) 435485.1 CAIC (smaller is better) 435487.1 HQIC (smaller is better) 435473.6 Generalized Chi-Square 201691.3 Gener. Chi-Square / DF 2.13 Covariance Parameter Estimates Standard Cov Parm Subject Estimate Error CS BSISID 18.8263 0.1617 Residual 2.1326 0.01191 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 40.7331 0.3356 30373 121.38 <.0001 Age -0.1560 0.004557 64203 -34.24 <.0001 Gender 0 -1.5036 0.1931 30373 -7.79 <.0001 Gender 1 0 . . . . Year 0.09307 0.01822 64203 5.11 <.0001 YearSQ -0.00271 0.000748 64203 -3.62 0.0003 Year*Gender 0 -0.05159 0.03084 64203 -1.67 0.0944 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.003520 0.001271 64203 2.77 0.0056 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 64203 1172.24 <.0001 Gender 1 30373 60.66 <.0001 Year 1 64203 18.20 <.0001 YearSQ 1 64203 2.23 0.1356 Year*Gender 1 64203 2.80 0.0944 YearSQ*Gender 1 64203 7.68 0.0056 The SAS System 13:41 Wednesday, October 14, 2015 37 The GLIMMIX Procedure Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 27.3342 0.04044 30373 675.90 <.0001 0.05 27.2549 27.4135 1 28.8860 0.03338 30373 865.31 <.0001 0.05 28.8205 28.9514 The SAS System 13:41 Wednesday, October 14, 2015 38 The GLIMMIX Procedure Model Information Data Set WORK.ZODIAC_45_OVER_75 Response Variable CholHDL Response Distribution Gaussian Link Function Identity Variance Function Default Variance Matrix Blocked By BSISID Estimation Technique Restricted Maximum Likelihood Degrees of Freedom Method Between-Within Class Level Information Class Levels Values BSISID 33987 not printed Gender 2 0 1 Number of Observations Read 107758 Number of Observations Used 98458 Dimensions R-side Cov. Parameters 2 Columns in X 10 Columns in Z per Subject 0 Subjects (Blocks in V) 33987 Max Obs per Subject 16 Optimization Information Optimization Technique Dual Quasi-Newton Parameters in Optimization 1 Lower Boundaries 0 Upper Boundaries 0 Fixed Effects Profiled Residual Variance Profiled Starting From Data Iteration History Objective Max Iteration Restarts Evaluations Function Change Gradient 0 0 4 255102.96665 . 2184.384 1 0 3 254637.30819 465.65845746 596.2106 2 0 2 254563.59725 73.71094345 213.0364 3 0 2 254554.60439 8.99286394 37.48435 4 0 2 254554.34008 0.26430811 1.914218 5 0 2 254554.33938 0.00069933 0.016184 The SAS System 13:41 Wednesday, October 14, 2015 39 The GLIMMIX Procedure Convergence criterion (GCONV=1E-8) satisfied. Fit Statistics -2 Res Log Likelihood 254554.3 AIC (smaller is better) 254558.3 AICC (smaller is better) 254558.3 BIC (smaller is better) 254575.2 CAIC (smaller is better) 254577.2 HQIC (smaller is better) 254563.7 Generalized Chi-Square 40471.18 Gener. Chi-Square / DF 0.41 Covariance Parameter Estimates Standard Cov Parm Subject Estimate Error CS BSISID 0.9841 0.009388 Residual 0.4111 0.002256 Solutions for Fixed Effects Standard Effect Gender Estimate Error DF t Value Pr > |t| Intercept 3.9442 0.09186 32023 42.94 <.0001 Age 0.008380 0.001134 66428 7.39 <.0001 Gender 0 0.2834 0.07178 32023 3.95 <.0001 Gender 1 0 . . . . Year -0.1506 0.006899 66428 -21.83 <.0001 YearSQ 0.005233 0.000292 66428 17.92 <.0001 Year*Gender 0 -0.01252 0.01191 66428 -1.05 0.2932 Year*Gender 1 0 . . . . YearSQ*Gender 0 0.000026 0.000498 66428 0.05 0.9577 YearSQ*Gender 1 0 . . . . Type III Tests of Fixed Effects Num Den Effect DF DF F Value Pr > F Age 1 66428 54.63 <.0001 Gender 1 32023 15.59 <.0001 Year 1 66428 689.28 <.0001 YearSQ 1 66428 442.73 <.0001 Year*Gender 1 66428 1.10 0.2932 YearSQ*Gender 1 66428 0.00 0.9577 The SAS System 13:41 Wednesday, October 14, 2015 40 The GLIMMIX Procedure Gender Least Squares Means Standard Gender Estimate Error DF t Value Pr > |t| Alpha Lower Upper 0 3.7146 0.009677 32023 383.85 <.0001 0.05 3.6956 3.7336 1 3.5890 0.007881 32023 455.41 <.0001 0.05 3.5735 3.6044