*****pgm: descriptive_selective.do****

use "C:\NHANES\Data\analysis_data.dta", clear



***create variable codes*****
generate age=1 if ridageyr >= 20 & ridageyr <=39
replace age=2 if ridageyr >= 40 & ridageyr <=59
replace age=3 if ridageyr >= 60 & ridageyr <.

gen race=1 if ridreth1==3
replace race=2 if ridreth1==4
replace race=3 if ridreth1==1
replace race=4 if ridreth1==2 | ridreth1==5


****format variables******
label define agefmt 1 "20-39"
label define agefmt 2 "40-59", add
label define agefmt 3 "60+", add

label define racefmt 1 "NH White", 
label define racefmt 2 "NH Black", add
label define racfmt  3 "Mex American", add
label define racefmt 4 "Other race/ethn", add

label define sexfmt 1 "male"
label define sexfmt 2 "female", add

label values age agefmt
label values riagendr sexfmt
label values race racefmt

log using "c:\NHANES\log\descriptive_selective.log", replace

sort riagendr age

by riagendr: tabstat lbxtc [w=wtmec4yr], by(age) stat(mean p25 median p75), if (ridageyr >=20 & ridageyr <.) & ridstatr==2

log close