. * do-file for lecture 4a of VHM 802/812, Winter 2016 . version 14 /* works also for version 13 */ . set more off . cd "h:\vhm\vhm802\data_stata" h:\vhm\vhm802\data_stata . . use nocardia.dta, clear . tabulate casecont dclox, chi2 row exact lrchi2 +----------------+ | Key | |----------------| | frequency | | row percentage | +----------------+ | Cloxacillin used on Case - | farm Control | no yes | Total -----------+----------------------+---------- no | 35 19 | 54 | 64.81 35.19 | 100.00 -----------+----------------------+---------- yes | 46 8 | 54 | 85.19 14.81 | 100.00 -----------+----------------------+---------- Total | 81 27 | 108 | 75.00 25.00 | 100.00 Pearson chi2(1) = 5.9753 Pr = 0.015 likelihood-ratio chi2(1) = 6.1130 Pr = 0.013 Fisher's exact = 0.025 1-sided Fisher's exact = 0.013 . cc casecont dclox Proportion | Exposed Unexposed | Total Exposed -----------------+------------------------+------------------------ Cases | 8 46 | 54 0.1481 Controls | 19 35 | 54 0.3519 -----------------+------------------------+------------------------ Total | 27 81 | 108 0.2500 | | | Point estimate | [95% Conf. Interval] |------------------------+------------------------ Odds ratio | .3203661 | .1091298 .8852277 (exact) Prev. frac. ex. | .6796339 | .1147723 .8908702 (exact) Prev. frac. pop | .2391304 | +------------------------------------------------- chi2(1) = 5.98 Pr>chi2 = 0.0145 . * logistic regression for case-control design . logit casecont dclox Iteration 0: log likelihood = -74.859896 Iteration 1: log likelihood = -71.810582 Iteration 2: log likelihood = -71.803418 Iteration 3: log likelihood = -71.803418 Logistic regression Number of obs = 108 LR chi2(1) = 6.11 Prob > chi2 = 0.0134 Log likelihood = -71.803418 Pseudo R2 = 0.0408 ------------------------------------------------------------------------------ casecont | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- dclox | -1.138291 .4774329 -2.38 0.017 -2.074042 -.2025395 _cons | .2732933 .2243002 1.22 0.223 -.1663269 .7129136 ------------------------------------------------------------------------------ . logit casecont dclox, or /* displays ORs */ Iteration 0: log likelihood = -74.859896 Iteration 1: log likelihood = -71.810582 Iteration 2: log likelihood = -71.803418 Iteration 3: log likelihood = -71.803418 Logistic regression Number of obs = 108 LR chi2(1) = 6.11 Prob > chi2 = 0.0134 Log likelihood = -71.803418 Pseudo R2 = 0.0408 ------------------------------------------------------------------------------ casecont | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- dclox | .3203661 .1529533 -2.38 0.017 .1256768 .8166542 _cons | 1.314286 .2947945 1.22 0.223 .8467694 2.039926 ------------------------------------------------------------------------------ . logistic casecont dclox /* same using logistic command */ Logistic regression Number of obs = 108 LR chi2(1) = 6.11 Prob > chi2 = 0.0134 Log likelihood = -71.803418 Pseudo R2 = 0.0408 ------------------------------------------------------------------------------ casecont | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- dclox | .3203661 .1529533 -2.38 0.017 .1256768 .8166542 _cons | 1.314286 .2947945 1.22 0.223 .8467694 2.039926 ------------------------------------------------------------------------------ . logit dclox casecont /* to demonstrate same log(OR) */ Iteration 0: log likelihood = -60.732196 Iteration 1: log likelihood = -57.735076 Iteration 2: log likelihood = -57.675769 Iteration 3: log likelihood = -57.675718 Iteration 4: log likelihood = -57.675718 Logistic regression Number of obs = 108 LR chi2(1) = 6.11 Prob > chi2 = 0.0134 Log likelihood = -57.675718 Pseudo R2 = 0.0503 ------------------------------------------------------------------------------ dclox | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- casecont | -1.138291 .4774329 -2.38 0.017 -2.074042 -.2025396 _cons | -.6109091 .2849614 -2.14 0.032 -1.169423 -.052395 ------------------------------------------------------------------------------ . . * logistic regression for categorical predictor . tabulate casecont dbarn, chi2 row exact lrchi2 +----------------+ | Key | |----------------| | frequency | | row percentage | +----------------+ Enumerating sample-space combinations: stage 3: enumerations = 1 stage 2: enumerations = 3 stage 1: enumerations = 0 Case - | dry cow housing Control | freestal tiestall other | Total -----------+---------------------------------+---------- no | 13 38 3 | 54 | 24.07 70.37 5.56 | 100.00 -----------+---------------------------------+---------- yes | 22 29 3 | 54 | 40.74 53.70 5.56 | 100.00 -----------+---------------------------------+---------- Total | 35 67 6 | 108 | 32.41 62.04 5.56 | 100.00 Pearson chi2(2) = 3.5232 Pr = 0.172 likelihood-ratio chi2(2) = 3.5531 Pr = 0.169 Fisher's exact = 0.157 . logit casecont i.dbarn Iteration 0: log likelihood = -74.859896 Iteration 1: log likelihood = -73.083934 Iteration 2: log likelihood = -73.083342 Iteration 3: log likelihood = -73.083342 Logistic regression Number of obs = 108 LR chi2(2) = 3.55 Prob > chi2 = 0.1692 Log likelihood = -73.083342 Pseudo R2 = 0.0237 ------------------------------------------------------------------------------ casecont | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- dbarn | tiestall | -.7963834 .4279909 -1.86 0.063 -1.63523 .0424632 other | -.5260931 .8882816 -0.59 0.554 -2.267093 1.214907 | _cons | .5260931 .3498251 1.50 0.133 -.1595516 1.211738 ------------------------------------------------------------------------------ . logistic casecont i.dbarn Logistic regression Number of obs = 108 LR chi2(2) = 3.55 Prob > chi2 = 0.1692 Log likelihood = -73.083342 Pseudo R2 = 0.0237 ------------------------------------------------------------------------------ casecont | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- dbarn | tiestall | .4509569 .1930054 -1.86 0.063 .1949075 1.043378 other | .5909091 .5248937 -0.59 0.554 .1036129 3.36998 | _cons | 1.692308 .5920118 1.50 0.133 .852526 3.359317 ------------------------------------------------------------------------------ . testparm i.dbarn /* multiple Wald test for dbarn */ ( 1) [casecont]2.dbarn = 0 ( 2) [casecont]3.dbarn = 0 chi2( 2) = 3.46 Prob > chi2 = 0.1771 . . * multiple logistic regression . logit casecont dneo dclox dcpct Iteration 0: log likelihood = -74.859896 Iteration 1: log likelihood = -54.156164 Iteration 2: log likelihood = -53.993934 Iteration 3: log likelihood = -53.993656 Iteration 4: log likelihood = -53.993656 Logistic regression Number of obs = 108 LR chi2(3) = 41.73 Prob > chi2 = 0.0000 Log likelihood = -53.993656 Pseudo R2 = 0.2787 ------------------------------------------------------------------------------ casecont | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- dneo | 2.212564 .5780423 3.83 0.000 1.079622 3.345506 dclox | -1.412516 .5572076 -2.53 0.011 -2.504623 -.3204093 dcpct | .0226682 .0071871 3.15 0.002 .0085817 .0367547 _cons | -2.984272 .7722467 -3.86 0.000 -4.497848 -1.470697 ------------------------------------------------------------------------------ . logistic casecont dneo dclox dcpct /* shows ORs */ Logistic regression Number of obs = 108 LR chi2(3) = 41.73 Prob > chi2 = 0.0000 Log likelihood = -53.993656 Pseudo R2 = 0.2787 ------------------------------------------------------------------------------ casecont | Odds Ratio Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- dneo | 9.139118 5.282797 3.83 0.000 2.943566 28.37493 dclox | .2435298 .1356966 -2.53 0.011 .0817064 .7258519 dcpct | 1.022927 .0073519 3.15 0.002 1.008619 1.037439 _cons | .0505763 .0390574 -3.86 0.000 .0111329 .2297654 ------------------------------------------------------------------------------ . test dneo /* Wald test as chi^2-test, same P as z-test */ ( 1) [casecont]dneo = 0 chi2( 1) = 14.65 Prob > chi2 = 0.0001 . * profile likelihood CI for dneo, using logprof add-on command . logprof dneo Maximum Profile Log-Likelihood = -53.993656 ------------------------------------------------------------------------------- casecont | Coef. [95% Conf. Interval] ---------------+--------------------------------------------------------------- dneo | 2.2125639 1.1441697 3.4453488 ------------------------------------------------------------------------------- . . * display of iterations of ML estimation . logit casecont dneo dclox dcpct, trace ------------------------------------------------------------------------------------------------ Iteration 0: Parameter vector: casecont: casecont: casecont: casecont: dneo dclox dcpct _cons r1 0 0 0 0 log likelihood = -74.859896 ------------------------------------------------------------------------------------------------ Iteration 1: Parameter vector: casecont: casecont: casecont: casecont: dneo dclox dcpct _cons r1 2.235522 -1.60015 .0228446 -2.857956 log likelihood = -54.156164 ------------------------------------------------------------------------------------------------ Iteration 2: Parameter vector: casecont: casecont: casecont: casecont: dneo dclox dcpct _cons r1 2.204562 -1.406852 .0225647 -2.974098 log likelihood = -53.993934 ------------------------------------------------------------------------------------------------ Iteration 3: Parameter vector: casecont: casecont: casecont: casecont: dneo dclox dcpct _cons r1 2.212543 -1.4125 .022668 -2.984243 log likelihood = -53.993656 ------------------------------------------------------------------------------------------------ Iteration 4: Parameter vector: casecont: casecont: casecont: casecont: dneo dclox dcpct _cons r1 2.212564 -1.412516 .0226682 -2.984272 log likelihood = -53.993656 ------------------------------------------------------------------------------------------------ Logistic regression Number of obs = 108 LR chi2(3) = 41.73 Prob > chi2 = 0.0000 Log likelihood = -53.993656 Pseudo R2 = 0.2787 ------------------------------------------------------------------------------ casecont | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- dneo | 2.212564 .5780423 3.83 0.000 1.079622 3.345506 dclox | -1.412516 .5572076 -2.53 0.011 -2.504623 -.3204093 dcpct | .0226682 .0071871 3.15 0.002 .0085817 .0367547 _cons | -2.984272 .7722467 -3.86 0.000 -4.497848 -1.470697 ------------------------------------------------------------------------------ . . * likelihood-ratio tests . logit casecont dneo dclox dcpct /* need to rerun using logit */ Iteration 0: log likelihood = -74.859896 Iteration 1: log likelihood = -54.156164 Iteration 2: log likelihood = -53.993934 Iteration 3: log likelihood = -53.993656 Iteration 4: log likelihood = -53.993656 Logistic regression Number of obs = 108 LR chi2(3) = 41.73 Prob > chi2 = 0.0000 Log likelihood = -53.993656 Pseudo R2 = 0.2787 ------------------------------------------------------------------------------ casecont | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- dneo | 2.212564 .5780423 3.83 0.000 1.079622 3.345506 dclox | -1.412516 .5572076 -2.53 0.011 -2.504623 -.3204093 dcpct | .0226682 .0071871 3.15 0.002 .0085817 .0367547 _cons | -2.984272 .7722467 -3.86 0.000 -4.497848 -1.470697 ------------------------------------------------------------------------------ . test dneo dclox /* Wald test for dneo+dclox */ ( 1) [casecont]dneo = 0 ( 2) [casecont]dclox = 0 chi2( 2) = 21.40 Prob > chi2 = 0.0000 . estimates store full . logit casecont dcpct Iteration 0: log likelihood = -74.859896 Iteration 1: log likelihood = -69.088754 Iteration 2: log likelihood = -69.074669 Iteration 3: log likelihood = -69.074669 Logistic regression Number of obs = 108 LR chi2(1) = 11.57 Prob > chi2 = 0.0007 Log likelihood = -69.074669 Pseudo R2 = 0.0773 ------------------------------------------------------------------------------ casecont | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- dcpct | .0189543 .0060179 3.15 0.002 .0071595 .0307492 _cons | -1.462424 .5230518 -2.80 0.005 -2.487586 -.4372608 ------------------------------------------------------------------------------ . lrtest full /* likelihood-ratio test for dneo+dclox */ Likelihood-ratio test LR chi2(2) = 30.16 (Assumption: . nested in full) Prob > chi2 = 0.0000 . lrtest full, stats /* same with all fit statistics */ Likelihood-ratio test LR chi2(2) = 30.16 (Assumption: . nested in full) Prob > chi2 = 0.0000 Akaike's information criterion and Bayesian information criterion ----------------------------------------------------------------------------- Model | Obs ll(null) ll(model) df AIC BIC -------------+--------------------------------------------------------------- . | 108 -74.8599 -69.07467 2 142.1493 147.5136 full | 108 -74.8599 -53.99366 4 115.9873 126.7158 ----------------------------------------------------------------------------- Note: N=Obs used in calculating BIC; see [R] BIC note. . estimates store red . logit casecont Iteration 0: log likelihood = -74.859896 Iteration 1: log likelihood = -74.859896 Logistic regression Number of obs = 108 LR chi2(0) = -0.00 Prob > chi2 = . Log likelihood = -74.859896 Pseudo R2 = -0.0000 ------------------------------------------------------------------------------ casecont | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- _cons | 0 .1924501 0.00 1.000 -.3771952 .3771952 ------------------------------------------------------------------------------ . estimates store null . estimates stats * Akaike's information criterion and Bayesian information criterion ----------------------------------------------------------------------------- Model | Obs ll(null) ll(model) df AIC BIC -------------+--------------------------------------------------------------- full | 108 -74.8599 -53.99366 4 115.9873 126.7158 red | 108 -74.8599 -69.07467 2 142.1493 147.5136 null | 108 -74.8599 -74.8599 1 151.7198 154.4019 ----------------------------------------------------------------------------- Note: N=Obs used in calculating BIC; see [R] BIC note. . estimates table * ----------------------------------------------------- Variable | full red null -------------+--------------------------------------- dneo | 2.2125639 dclox | -1.412516 dcpct | .02266825 .01895434 _cons | -2.9842724 -1.4624236 0 ----------------------------------------------------- . . use mice.dta, clear . logit dead dose Iteration 0: log likelihood = -75.669723 Iteration 1: log likelihood = -64.201632 Iteration 2: log likelihood = -63.944944 Iteration 3: log likelihood = -63.944713 Iteration 4: log likelihood = -63.944713 Logistic regression Number of obs = 120 LR chi2(1) = 23.45 Prob > chi2 = 0.0000 Log likelihood = -63.944713 Pseudo R2 = 0.1549 ------------------------------------------------------------------------------ dead | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- dose | 14.6369 3.332533 4.39 0.000 8.10526 21.16855 _cons | -3.569738 .7053432 -5.06 0.000 -4.952185 -2.187291 ------------------------------------------------------------------------------ . estimates store red . * does not work: logit dead i.dose . * because dose has non-integer values . egen Dose=group(dose) /* create dose groups (in new variable Dose) */ . logit dead i.Dose /* dose as categorical predictor */ note: 4.Dose != 0 predicts failure perfectly 4.Dose dropped and 10 obs not used Iteration 0: log likelihood = -71.523665 Iteration 1: log likelihood = -59.113208 Iteration 2: log likelihood = -58.82041 Iteration 3: log likelihood = -58.817856 Iteration 4: log likelihood = -58.817856 Logistic regression Number of obs = 110 LR chi2(10) = 25.41 Prob > chi2 = 0.0046 Log likelihood = -58.817856 Pseudo R2 = 0.1776 ------------------------------------------------------------------------------ dead | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- Dose | 2 | .81093 1.317616 0.62 0.538 -1.771549 3.393409 3 | -9.56e-17 1.490712 -0.00 1.000 -2.921742 2.921742 4 | 0 (empty) 5 | -9.56e-17 1.490712 -0.00 1.000 -2.921742 2.921742 6 | .81093 1.317616 0.62 0.538 -1.771549 3.393409 7 | 1.791759 1.236033 1.45 0.147 -.6308209 4.214339 8 | 2.60269 1.236033 2.11 0.035 .1801093 5.02527 9 | 1.791759 1.236033 1.45 0.147 -.6308209 4.214339 10 | 2.197224 1.229273 1.79 0.074 -.2121055 4.606554 11 | 2.197224 1.229273 1.79 0.074 -.2121055 4.606554 12 | 3.583519 1.317616 2.72 0.007 1.00104 6.165998 | _cons | -2.197224 1.054092 -2.08 0.037 -4.263208 -.1312411 ------------------------------------------------------------------------------ . logit dead i.Dose, asis /* avoids exclusion of dose=0.1413 */ Iteration 0: log likelihood = -75.669723 Iteration 1: log likelihood = -60.073875 Iteration 2: log likelihood = -59.076968 Iteration 3: log likelihood = -58.874743 Iteration 4: log likelihood = -58.831027 Iteration 5: log likelihood = -58.820395 Iteration 6: log likelihood = -58.81828 Iteration 7: log likelihood = -58.817905 Iteration 8: log likelihood = -58.817866 Iteration 9: log likelihood = -58.817858 Iteration 10: log likelihood = -58.817856 Logistic regression Number of obs = 120 LR chi2(11) = 33.70 Prob > chi2 = 0.0004 Log likelihood = -58.817856 Pseudo R2 = 0.2227 ------------------------------------------------------------------------------ dead | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- Dose | 2 | .8110233 1.31765 0.62 0.538 -1.771523 3.39357 3 | -7.83e-16 1.490771 -0.00 1.000 -2.921858 2.921858 4 | -14.73911 1505.568 -0.01 0.992 -2965.599 2936.12 5 | 2.90e-16 1.490771 0.00 1.000 -2.921858 2.921858 6 | .8110233 1.31765 0.62 0.538 -1.771523 3.39357 7 | 1.791833 1.23607 1.45 0.147 -.6308193 4.214485 8 | 2.602659 1.236064 2.11 0.035 .1800167 5.0253 9 | 1.791833 1.23607 1.45 0.147 -.6308193 4.214485 10 | 2.197249 1.229309 1.79 0.074 -.2121518 4.606649 11 | 2.197249 1.229309 1.79 0.074 -.2121518 4.606649 12 | 3.583471 1.317628 2.72 0.007 1.000967 6.165976 | _cons | -2.197324 1.054134 -2.08 0.037 -4.26339 -.1312584 ------------------------------------------------------------------------------ . logit dead i.Dose, asis trace /* showing convergence process */ ------------------------------------------------------------------------------------------------ Iteration 0: Parameter vector: c1 c2 c3 c4 c5 c6 c7 c8 > c9 c10 r1 0 0 0 0 0 0 0 0 > 0 0 c11 c12 r1 0 -.7308875 log likelihood = -75.669723 ------------------------------------------------------------------------------------------------ Iteration 1: Parameter vector: c1 c2 c3 c4 c5 c6 c7 c8 > c9 c10 r1 .5128205 1.14e-14 -.5128205 1.14e-14 .5128205 1.538462 2.564103 1.538462 2.0 > 51282 2.051282 c11 c12 r1 3.589744 -1.884734 log likelihood = -60.073875 ------------------------------------------------------------------------------------------------ Iteration 2: Parameter vector: c1 c2 c3 c4 c5 c6 c7 c8 > c9 c10 r1 .8756869 -2.01e-15 -1.630307 -2.01e-15 .8756869 1.840062 2.554934 1.840062 2.2 > 03774 2.203774 c11 c12 r1 3.484351 -2.26729 log likelihood = -59.076968 ------------------------------------------------------------------------------------------------ Iteration 3: Parameter vector: c1 c2 c3 c4 c5 c6 c7 c8 > c9 c10 r1 .7838782 2.26e-15 -3.132326 2.26e-15 .7838782 1.770934 2.616282 1.770934 2.1 > 92047 2.192047 c11 c12 r1 3.607197 -2.16817 log likelihood = -58.874743 ------------------------------------------------------------------------------------------------ Iteration 4: Parameter vector: c1 c2 c3 c4 c5 c6 c7 c8 > c9 c10 r1 .8339775 -4.60e-15 -4.962829 -4.60e-15 .8339775 1.809369 2.589805 1.809369 2.2 > 01125 2.201125 c11 c12 r1 3.560565 -2.222022 log likelihood = -58.831027 ------------------------------------------------------------------------------------------------ Iteration 5: Parameter vector: c1 c2 c3 c4 c5 c6 c7 c8 > c9 c10 r1 .8077708 -4.61e-17 -6.116858 -1.71e-16 .8077708 1.789285 2.603871 1.789285 2.1 > 96459 2.196459 c11 c12 r1 3.585359 -2.193845 log likelihood = -58.820395 ------------------------------------------------------------------------------------------------ Iteration 6: Parameter vector: c1 c2 c3 c4 c5 c6 c7 c8 > c9 c10 r1 .8168411 -1.36e-15 -8.982863 -1.49e-15 .8168411 1.796387 2.600456 1.796387 2.1 > 98647 2.198647 c11 c12 r1 3.580031 -2.203547 log likelihood = -58.81828 ------------------------------------------------------------------------------------------------ Iteration 7: Parameter vector: c1 c2 c3 c4 c5 c6 c7 c8 > c9 c10 r1 .8101734 -3.32e-16 -10.11501 -4.42e-16 .8101734 1.791163 2.602943 1.791163 2.1 > 97029 2.197029 c11 c12 r1 3.583905 -2.196416 log likelihood = -58.817905 ------------------------------------------------------------------------------------------------ Iteration 8: Parameter vector: c1 c2 c3 c4 c5 c6 c7 c8 > c9 c10 r1 .811592 -1.16e-15 -11.9885 -1.27e-15 .811592 1.792281 2.602468 1.792281 2.1 > 97395 2.197395 c11 c12 r1 3.58318 -2.197931 log likelihood = -58.817866 ------------------------------------------------------------------------------------------------ Iteration 9: Parameter vector: c1 c2 c3 c4 c5 c6 c7 c8 > c9 c10 r1 .8106818 7.71e-16 -13.36448 6.59e-16 .8106818 1.791564 2.602773 1.791564 2. > 19716 2.19716 c11 c12 r1 3.583645 -2.196959 log likelihood = -58.817858 ------------------------------------------------------------------------------------------------ Iteration 10: Parameter vector: c1 c2 c3 c4 c5 c6 c7 c8 > c9 c10 r1 .8110233 -7.83e-16 -14.73911 2.90e-16 .8110233 1.791833 2.602659 1.791833 2.1 > 97249 2.197249 c11 c12 r1 3.583471 -2.197324 log likelihood = -58.817856 ------------------------------------------------------------------------------------------------ Logistic regression Number of obs = 120 LR chi2(11) = 33.70 Prob > chi2 = 0.0004 Log likelihood = -58.817856 Pseudo R2 = 0.2227 ------------------------------------------------------------------------------ dead | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------- Dose | 2 | .8110233 1.31765 0.62 0.538 -1.771523 3.39357 3 | -7.83e-16 1.490771 -0.00 1.000 -2.921858 2.921858 4 | -14.73911 1505.568 -0.01 0.992 -2965.599 2936.12 5 | 2.90e-16 1.490771 0.00 1.000 -2.921858 2.921858 6 | .8110233 1.31765 0.62 0.538 -1.771523 3.39357 7 | 1.791833 1.23607 1.45 0.147 -.6308193 4.214485 8 | 2.602659 1.236064 2.11 0.035 .1800167 5.0253 9 | 1.791833 1.23607 1.45 0.147 -.6308193 4.214485 10 | 2.197249 1.229309 1.79 0.074 -.2121518 4.606649 11 | 2.197249 1.229309 1.79 0.074 -.2121518 4.606649 12 | 3.583471 1.317628 2.72 0.007 1.000967 6.165976 | _cons | -2.197324 1.054134 -2.08 0.037 -4.26339 -.1312584 ------------------------------------------------------------------------------ . testparm i.Dose /* note: Wald test for dose fails here */ ( 1) [dead]2.Dose = 0 ( 2) [dead]3.Dose = 0 ( 3) [dead]4.Dose = 0 ( 4) [dead]5.Dose = 0 ( 5) [dead]6.Dose = 0 ( 6) [dead]7.Dose = 0 ( 7) [dead]8.Dose = 0 ( 8) [dead]9.Dose = 0 ( 9) [dead]10.Dose = 0 (10) [dead]11.Dose = 0 (11) [dead]12.Dose = 0 chi2( 11) = 19.22 Prob > chi2 = 0.0573 . lrtest red, stats /* likelihood-ratio test ok */ Likelihood-ratio test LR chi2(10) = 10.25 (Assumption: red nested in .) Prob > chi2 = 0.4185 Akaike's information criterion and Bayesian information criterion ----------------------------------------------------------------------------- Model | Obs ll(null) ll(model) df AIC BIC -------------+--------------------------------------------------------------- red | 120 -75.66972 -63.94471 2 131.8894 137.4644 . | 120 -75.66972 -58.81786 12 141.6357 175.0856 ----------------------------------------------------------------------------- Note: N=Obs used in calculating BIC; see [R] BIC note.