set more off
clear
set seed 123
set obs 10000

gen id = _n

/* parameters
S1 = runiform() < 0.1
S2 = runiform() < 0.5
*/

** Generate variables ***

gen S2 = runiform() < 0.8

gen U = runiform()<0.3 if S2==1
replace U = runiform()<0.6 if S2==0
*logistic S2 U

gen S1 = runiform() < 1.0
tab S1

gen X = uniform() < invlogit(-2 + 0.73*U) if S1==0
replace X = uniform() < invlogit(-2 + 0.23*U) if S1==1
tab X

gen Y = 2000 - 300*X + 450*U + 3 * invnorm(uniform()) if S1==0 
replace Y = 2000 +197*X - 50*U + 3 * invnorm(uniform()) if S1==1
regress Y X U

*logistic S1 Y X


/**** Run regressions ****

regress Y X U

regress Y X 

regress Y X if S1==1

regress Y X if S2==1

****************************/

*bootstrap, rep(1000) size(500): regress Y X U 

*bootstrap, rep(1000) size(500): regress Y X 

*bootstrap, rep(1000) size(500): regress Y X if S1==1

bootstrap, rep(1000) size(500): regress Y X if S2==1


set more off
clear
set seed 123
set obs 10000

gen id = _n

/* parameters
S1 = runiform() < 0.1
S2 = runiform() < 0.5
*/

** Generate variables ***

gen S2 = runiform() < 0.8

gen U = runiform()<0.3 if S2==1
replace U = runiform()<0.6 if S2==0
*logistic S2 U

gen S1 = runiform() < 1.0
tab S1

gen X = uniform() < invlogit(-2 + 0.73*U) if S1==0
replace X = uniform() < invlogit(-2 + 0.23*U) if S1==1
tab X

gen Y = 1800 + 450*U + 3 * invnorm(uniform()) if S1==0 
replace Y = 1800 - 50*U + 3 * invnorm(uniform()) if S1==1
regress Y X U

bootstrap, rep(1000) size(500): regress Y X if S2==1