window push version 11
window push *!!!!!!!!!!!!!!! Not debugged as of Nov 2009. Check function of margins command
window push *!!!!!!!!!!!!!!  Note that margins replaces the adjust command
window push log using adjust.log, replace
window push * adjust.log
window push *
window push * Evaluate the adjust postestimation command after hazard regression
window push * and logistic regression models
window push *
window push * The adjust command tabulates the linear predictor, or the exponentiated linear 
window push * predictor evaluated at one or more catigorical variables.  It can be used to 
window push * tabulate hazard ratios with their confidence intervals over one or two categorical
window push * variables, either with or without interaction terms.  An example follows.
window push *
window push use "C:\WDDtext\2.20.Framingham.dta", clear
window push gen hypertension = dbp > 90
window push stset followup, failure(chdfate)
window push gen men = sex==1
window push stcox men##hypertension, nohr
window push lincom  1.men+ 1.hypertens +1.men#1.hypertens
window push *
window push * The identical linear predictors given above can be calculated by the adjust
window push * command as follows.
window push *
window push margins men##hypertension
window push *
window push *  We now calculate the related hazard ratios from the same model as above
window push *
window push stcox men##hypertension
window push lincom  1.men+ 1.hypertens + 1.men#1.hypertens, hr
window push *
window push * Now repeat these calculations using the adjust command
window push *
window push margins men##hypertension // exp ci label(hazard ratio)
window push *
window push * Unfortunately this approach does not work in general with logistic 
window push * regression.  This is because in logistic regression the linear predictor 
window push * rarely equals the log of an interesting odds ratio.  For example
window push *
window push use "C:\WDDtext\5.5.EsophagealCa.dta", clear
window push gen smoke = tobacco >2
window push logit cancer heavy##smoke [freq=patients]
window push lincom   _cons + 1.heavy
window push lincom   _cons + 1.smoke
window push lincom   _cons +1.heavy+ 1.smoke+ 1.heavy#1.smoke
window push *
window push * The preceding results are replicated by the following adjust command.
window push * Alas, they are not interesting log odds ratios.
window push *
window push margins heavy##smoke 
window push log close