* use afb.dta generate event = 1 - censored global time afb global event event global xlist edu sibs white immig lowinc16 city16 describe $time $event $xlist summarize $time $event $xlist * Set data as survival time stset $time, failure($event) stdescribe stsum * List of hazard and survival functions - reproducing Table 8.2 * in W&B textbook ltable afb, hazard survival * Nonparametric estimation * Graph of survival function (Kaplan-Meier survival curve) sts graph, survival * Graph of Kaplan-Meier Failure Function sts graph, failure * Graph of smoothed hazard function sts graph, hazard * Graph of cumulative hazard ratio (Nelson-Aalen cumulative hazard curve) sts graph, cumhaz * Kaplan-Meier survival curves for the white groups sts graph, by(white) * Test for equality of survival functions between white groups * using logrank and Wilcoxon tests sts test white, logrank sts test white, wilcoxon * Kaplan-Meier survival curves for the immig groups sts graph, by(immig) * Test for equality of survival functions between immigrant groups * using logrank and Wilcoxon tests sts test immig, logrank sts test immig, wilcoxon * Kaplan-Meier survival curves for the loinc16 groups sts graph, by(lowinc16) * Test for equality of survival functions between loinc16 groups * using logrank and Wilcoxon tests sts test lowinc16, logrank sts test lowinc16, wilcoxon * Kaplan-Meier survival curves for the city16 groups sts graph, by(city16) * Test for equality of survival functions between city16 groups * using logrank and Wilcoxon tests sts test city16, logrank sts test city16, wilcoxon * Parametric models (accelerared failure time (ACF) parametrization) * Here duplicating the results in Table 8.6 in W&B textbook * Exponential regression coefficients and fitted hazard curve streg $xlist, dist(exponential) time nolog stcurve, hazard * Weibull regression coefficients and fitted hazard curve streg $xlist, dist(weibull) time nolog stcurve, hazard * Lognormal regression coefficients and fitted hazard curve streg $xlist, dist(lognormal) nolog stcurve, hazard * Lognormal regression (Gamma Fraility) coefficients and fitted hazard curve * Note: In order to run this streg command you must have run the previous * lognormal streg command streg $xlist, frailty(gamma) time nolog stcurve, hazard * Cox proportional hazard model coefficients and hazard rates * Notice the nonparametrically fitted hazard curve is possibly a better * match of the smoothed hazard function that was fitted above by the * the nonparametric "sts graph, hazard" command. stcox $xlist, nohr stcurve, hazard stcox $xlist * Test of Cox Proportional Hazards Assumption in Cox Regession estat phtest, log detail * The PH assumption of the Cox Regression is not supported by the data