insheet using "C:\Users\ahorenstein\Desktop\IR Directors Council\STATA Example\CS_G2_Subgroups_Copy.csv"

* estimate logistic regression and pscore

logit cs usip_participation with27mohsgrad sp10 sp11 sp12 fa10 fa11 need_rem_r need_rem_w need_rem_m1 need_rem_m2 score_r score_w score_m1 score_m2 score_r_impute score_w_impute score_m1_impute 
score_m2_impute bcc csi qbcc kbcc bmcc hostos amerindian white black hispanic female non_english_dummy missinglanguage_dummy non_us_dummy missingcountry_dummy log_age gedrecode 
b_bronx b_brooklyn b_manhattan b_queens b_si medianhouseholdincome     

predict propen

** This is where I modify the propensity score in order to force match on the sg_ft_rm variable

gen propen2=sg_ft_rm*10+propen


* set seed

set seed 1000

* sort randomly prior to doing a 1:1 match, in case there are multiple control cases with the same propensity score

generate x=uniform()
sort x

* 1:1 matching without replacement, with caliper of .2 

psmatch2 cs, pscore(propen2) caliper(0.2) noreplacement descending
sort _id
g match=oirastudentid[_n1]
g treat=cs_student_id if _nn==1
drop if treat==""