/*
cd "C:\Users\chkim\OneDrive - The University of Kansas\Lectures\Soc910_AdvStat\examples"

use soc910_1_data1.dta, clear

* sample limitation

keep if relate==1 | relate==2 
keep if age>=25 & age<=54
gen married=marst==1
drop if (age_sp<25 | age_sp>54) & married==1

sort serial
by serial: egen nadult = count(serial)
keep if nadult==1 | nadult==2

keep if age>=35 & age<=44

save, replace
*/

cd "C:\Users\chkim\OneDrive - The University of Kansas\Lectures\Soc910_AdvStat\examples"

use soc910_1_data1.dta, clear

* Recode
numlabel, add

gen eqincome  = ftotinc/sqrt(famsize)
gen lneqincome= ln(eqincome)
gen lnpincome = ln(inctot) if incearn>0 
gen lnspincome= ln(inctot_sp)

gen female=sex==2
gen single=marst==6
gen married = marst==1

gen     races=5
replace races=1 if raced==100 & hispan==0
replace races=2 if raced==200 & hispan==0 
replace races=4 if raced>=400 & raced<=679 & hispan==0


* education
gen     edu=1
replace edu=2 if educd==63 | educd==64
replace edu=3 if educd>=65 & educd<=100
replace edu=4 if educd==101
replace edu=5 if educd==114
replace edu=6 if educd==115
replace edu=7 if educd==116

replace edu=5 if edu>=5

gen age2=age*age
gen immig=yrimmig!=0



keep if races==1 | races==2
gen black=races==2

tabstat lnpincome, by(black)

reg lnpincome black
reg lnpincome black female i.edu age married

tab edu, gen(edu)

oaxaca lnpincome female edu2 edu3 edu4 edu5 age married, by(black) w(1)
oaxaca lnpincome female edu2 edu3 edu4 edu5 age married, by(black) w(1) detail
oaxaca lnpincome female edu1 edu2 edu3 edu5 age married, by(black) w(1) detail
oaxaca lnpincome female edu2 edu3 edu4 edu5 age married, by(black) w(1) detail(edu:edu*)
oaxaca lnpincome female edu1 edu2 edu3 edu5 age married, by(black) w(1) detail(edu:edu*)