Full Generalized Gamma Model 1 The NLMIXED Procedure Specifications Data Set WORK.HIV Dependent Variable exit Distribution for Dependent Variable General Optimization Technique Dual Quasi-Newton Integration Method None Full Generalized Gamma Model 2 ----------------------------------- period=1 ----------------------------------- The NLMIXED Procedure Dimensions Observations Used 633 Observations Not Used 0 Total Observations 633 Parameters 3 Parameters beta sigma lambda NegLogLike 0 1 0.5 830.99793 Iteration History Iter Calls NegLogLike Diff MaxGrad Slope 1 8 807.103491 23.89444 54.07252 -1224.24 2 17 804.660361 2.44313 69.96223 -26.2953 3 25 790.265847 14.39451 78.42907 -70.3027 4 29 786.2438 4.022046 45.9616 -20.7043 5 33 783.945846 2.297955 10.11967 -4.49136 6 39 783.867826 0.07802 0.93938 -0.34102 7 44 783.858552 0.009274 0.456952 -0.01072 8 49 783.8584 0.000152 0.029676 -0.00032 9 54 783.858399 4.681E-7 0.000104 -9.58E-7 NOTE: GCONV convergence criterion satisfied. Fit Statistics -2 Log Likelihood 1567.7 AIC (smaller is better) 1573.7 AICC (smaller is better) 1573.8 BIC (smaller is better) 1587.1 Parameter Estimates Standard Parameter Estimate Error DF t Value Pr > |t| Alpha Lower beta 0.6737 0.07156 633 9.41 <.0001 0.05 0.5332 sigma 0.7731 0.06785 633 11.40 <.0001 0.05 0.6399 lambda 1.3689 0.1889 633 7.25 <.0001 0.05 0.9980 Full Generalized Gamma Model 3 ----------------------------------- period=1 ----------------------------------- The NLMIXED Procedure Parameter Estimates Parameter Upper Gradient beta 0.8142 0.00002 sigma 0.9064 0.000104 lambda 1.7398 0.000052 Covariance Matrix of Parameter Estimates Row Parameter beta sigma lambda 1 beta 0.005120 -0.00337 0.01093 2 sigma -0.00337 0.004603 -0.01123 3 lambda 0.01093 -0.01123 0.03567 Additional Estimates Standard Label Estimate Error DF t Value Pr > |t| Alpha Lower logsigma -0.2573 0.08776 633 -2.93 0.0035 0.05 -0.4296 Additional Estimates Label Upper logsigma -0.08497 Full Generalized Gamma Model 4 ----------------------------------- period=2 ----------------------------------- The NLMIXED Procedure Dimensions Observations Used 660 Observations Not Used 0 Total Observations 660 Parameters 3 Parameters beta sigma lambda NegLogLike 0 1 0.5 884.728546 Iteration History Iter Calls NegLogLike Diff MaxGrad Slope 1 8 816.421039 68.30751 64.94807 -1871.72 2 13 813.638065 2.782974 44.77012 -49.4942 3 21 803.90722 9.730844 46.30055 -19.1269 4 25 802.173452 1.733768 3.894104 -3.7189 5 29 802.162219 0.011233 1.383005 -0.02849 6 34 802.161194 0.001025 0.433458 -0.0028 7 39 802.161075 0.000119 0.027252 -0.00025 8 44 802.161074 5.995E-7 0.001298 -1.23E-6 NOTE: GCONV convergence criterion satisfied. Fit Statistics -2 Log Likelihood 1604.3 AIC (smaller is better) 1610.3 AICC (smaller is better) 1610.4 BIC (smaller is better) 1623.8 Parameter Estimates Standard Parameter Estimate Error DF t Value Pr > |t| Alpha Lower beta 0.6493 0.06032 660 10.76 <.0001 0.05 0.5308 sigma 0.8475 0.04848 660 17.48 <.0001 0.05 0.7523 lambda 0.8510 0.1267 660 6.71 <.0001 0.05 0.6021 Full Generalized Gamma Model 5 ----------------------------------- period=2 ----------------------------------- The NLMIXED Procedure Parameter Estimates Parameter Upper Gradient beta 0.7677 0.001298 sigma 0.9427 0.000894 lambda 1.0998 -0.00058 Covariance Matrix of Parameter Estimates Row Parameter beta sigma lambda 1 beta 0.003638 -0.00161 0.005795 2 sigma -0.00161 0.002350 -0.00462 3 lambda 0.005795 -0.00462 0.01606 Additional Estimates Standard Label Estimate Error DF t Value Pr > |t| Alpha Lower logsigma -0.1655 0.05721 660 -2.89 0.0039 0.05 -0.2778 Additional Estimates Label Upper logsigma -0.05318 Full Generalized Gamma Model 6 ----------------------------------- period=3 ----------------------------------- The NLMIXED Procedure Dimensions Observations Used 472 Observations Not Used 0 Total Observations 472 Parameters 3 Parameters beta sigma lambda NegLogLike 0 1 0.5 739.44748 Iteration History Iter Calls NegLogLike Diff MaxGrad Slope 1 8 458.458576 280.9889 138.454 -6058.12 2 12 408.903546 49.55503 97.23428 -84.0469 3 20 367.754817 41.14873 34.58549 -43.8861 4 24 359.29848 8.456337 4.30789 -15.1586 5 32 358.807227 0.491253 3.159006 -0.5739 6 37 358.760505 0.046722 0.43217 -0.06862 7 42 358.759354 0.00115 0.399859 -0.00167 8 50 358.746019 0.013335 1.181404 -0.0005 9 63 358.578363 0.167656 1.396707 -0.0252 10 68 358.568624 0.009739 2.280446 -0.07127 11 76 358.528866 0.039758 0.755968 -0.07365 12 81 358.510217 0.018648 0.167112 -0.01279 13 86 358.509853 0.000365 0.007637 -0.00069 14 91 358.509843 0.00001 0.001262 -0.00002 15 96 358.509843 2.377E-8 0.000023 -4.05E-8 NOTE: GCONV convergence criterion satisfied. Fit Statistics -2 Log Likelihood 717.0 AIC (smaller is better) 723.0 AICC (smaller is better) 723.1 BIC (smaller is better) 735.5 Full Generalized Gamma Model 7 ----------------------------------- period=3 ----------------------------------- The NLMIXED Procedure Parameter Estimates Standard Parameter Estimate Error DF t Value Pr > |t| Alpha Lower beta 2.4456 0.2470 472 9.90 <.0001 0.05 1.9602 sigma 2.0082 0.8030 472 2.50 0.0127 0.05 0.4303 lambda 0.5491 0.6969 472 0.79 0.4312 0.05 -0.8204 Parameter Estimates Parameter Upper Gradient beta 2.9311 -0.00001 sigma 3.5862 0.000014 lambda 1.9186 0.000023 Covariance Matrix of Parameter Estimates Row Parameter beta sigma lambda 1 beta 0.06103 -0.05384 0.07823 2 sigma -0.05384 0.6448 -0.5383 3 lambda 0.07823 -0.5383 0.4857 Additional Estimates Standard Label Estimate Error DF t Value Pr > |t| Alpha Lower logsigma 0.6973 0.3999 472 1.74 0.0819 0.05 -0.08848 Additional Estimates Label Upper logsigma 1.4830 Full Generalized Gamma Model 8 ----------------------------------- period=4 ----------------------------------- The NLMIXED Procedure Dimensions Observations Used 549 Observations Not Used 0 Total Observations 549 Parameters 3 Parameters beta sigma lambda NegLogLike 0 1 0.5 1358.15559 Iteration History Iter Calls NegLogLike Diff MaxGrad Slope 1 8 318.970036 1039.186 9.553147 -76593.5 2 16 313.769939 5.200097 4.121508 -0.74909 3 20 312.716315 1.053624 5.962889 -2.01102 4 28 311.616697 1.099619 7.439396 -0.15092 5 32 309.971921 1.644776 5.643953 -2.43471 6 36 309.390109 0.581812 3.457354 -3.13786 7 40 308.647536 0.742572 3.961785 -1.14968 8 45 308.548963 0.098573 4.15116 -0.047 9 53 308.228095 0.320868 7.5631 -0.17132 10 61 307.20566 1.022435 3.595716 -1.72162 11 70 306.524597 0.681063 0.709292 -0.89367 12 75 306.408844 0.115753 1.034213 -0.11544 13 79 306.330383 0.07846 1.431717 -0.13097 14 83 306.267515 0.062868 1.016851 -0.18614 15 88 306.248869 0.018645 0.219733 -0.04215 16 93 306.242319 0.00655 0.238985 -0.01615 17 98 306.238128 0.004191 0.409411 -0.00509 18 103 306.236545 0.001583 0.022717 -0.00295 19 108 306.236481 0.000065 0.003198 -0.00011 20 113 306.236479 1.103E-6 0.000505 -2.12E-6 NOTE: GCONV convergence criterion satisfied. Fit Statistics -2 Log Likelihood 612.5 AIC (smaller is better) 618.5 AICC (smaller is better) 618.5 Full Generalized Gamma Model 9 ----------------------------------- period=4 ----------------------------------- The NLMIXED Procedure Fit Statistics BIC (smaller is better) 631.4 Parameter Estimates Standard Parameter Estimate Error DF t Value Pr > |t| Alpha Lower beta 3.0710 0.4827 549 6.36 <.0001 0.05 2.1228 sigma 1.3764 1.0357 549 1.33 0.1844 0.05 -0.6581 lambda 0.8447 1.8791 549 0.45 0.6532 0.05 -2.8464 Parameter Estimates Parameter Upper Gradient beta 4.0192 0.000505 sigma 3.4109 -0.00051 lambda 4.5357 -0.00039 Covariance Matrix of Parameter Estimates Row Parameter beta sigma lambda 1 beta 0.2330 -0.4267 0.8500 2 sigma -0.4267 1.0728 -1.8897 3 lambda 0.8500 -1.8897 3.5309 Additional Estimates Standard Label Estimate Error DF t Value Pr > |t| Alpha Lower logsigma 0.3195 0.7525 549 0.42 0.6713 0.05 -1.1586 Additional Estimates Label Upper logsigma 1.7976