* do-file for home assignment 2 of vhm 811, Fall semester 2008
* Henrik Stryhn

version 10
use "a2.dta", clear

* crude associations
cs dis work [fw=n], or
cs dis gender [fw=n], or
tabulate dis age [fw=n], chi2 expected
* measures against baseline category
csi 22 4 6255 5751, or
csi 48 4 6628 5751, or
csi 95 4 5803 5751, or
csi 160 4 5234 5751, or
* measures at a cutpoint
gen age40=age>=2
gen age50=age>=3
gen age60=age>=4
gen age70=age>=5
cs dis age40 [fw=n], or
cs dis age50 [fw=n], or
cs dis age60 [fw=n], or
cs dis age70 [fw=n], or

* Mantel-Haenszel estimate for gender
cs dis work [fw=n], by(gender) 
cs dis work [fw=n], by(gender) or

* Mantel-Haenszel estimate for age
cs dis work [fw=n], by(age)
cs dis work [fw=n], by(age) or

* association of age with outcome among exposure-negative
tabulate dis age if work==1 [fw=n], chi2 expected exact
cs dis age40 if work==1 [fw=n], or
cs dis age50 if work==1 [fw=n], or
cs dis age60 if work==1 [fw=n], or
cs dis age70 if work==1 [fw=n], or

* association with exposure among controls
tabulate work age if dis==0 [fw=n], chi2 expected

* combined age and gender classification
gen agegender=10*age+gender
* can also use the egen group command:
* egen agegender=group(age gender)
cs dis work [fw=n], by(agegender)
cs dis work [fw=n], by(agegender) or

* epidemiological analysis without work 
* confounding effect of gender on age-disease relation:
cs dis age40 [fw=n], by(gender)
cs dis age50 [fw=n], by(gender)
cs dis age60 [fw=n], by(gender)
cs dis age70 [fw=n], by(gender)
cs dis age40 [fw=n], or by(gender)
cs dis age50 [fw=n], or by(gender)
cs dis age60 [fw=n], or by(gender)
cs dis age70 [fw=n], or by(gender)
* association between gender and disease controlled for age
cs dis gender [fw=n], by(age)
cs dis gender [fw=n], by(age) or
* association between gender and age among non-diseased
tabulate gender age if dis==0 [fw=n], chi2 expected