線形制約の仮説検定の計算方法
シャープな回帰の不連続デザイン
Stata 3.13 (シャープな回帰の不連続デザイン)
import delimited C:\srdd.csv
generate xtilde = x-40
generate dxtilde = d*xtilde
regress y xtilde d dxtilde
predict yhat, xb
twoway (scatter y x, msize(vsmall)) (scatter yhat x, msize(vsmall))
VIFの解釈
差の差の分析
Stata 5.10 (差の差の分析 クロスセクションデータ)
import delimited C:\housing.csv
regress y dt d2 dtd2 x1-x6
Stata 5.11 (差の差の分析 パネルデータ)
import delimited C:\scrap.csv
xtset id year, yearly
xtreg y d2 dtd2, fe
regress d.y dt
R 5.10 (差の差の分析 クロスセクションデータ)
R 5.11 (差の差の分析 パネルデータ)
因果効果の分析
Stata 6.17 (因果効果の分析 )
use http://www.stata-press.com/data/r16/cattaneo2.dta
generate magesq = mage*mage
teffects ipw (bweight) (mbsmoke mmarried mage magesq fbaby medu)
teffects overlap
teffects ipw (bweight) (mbsmoke mmarried mage magesq fbaby medu, probit)
teffects ipw (bweight) (mbsmoke mmarried mage magesq fbaby medu), atet
teffects aipw (bweight prenatal1 mmarried mage fbaby) (mbsmoke mmarried mage magesq fbaby medu)
teffects aipw (bweight prenatal1 mmarried mage fbaby) (mbsmoke mmarried mage magesq fbaby medu), pomeans aequations
teffects psmatch (bweight) (mbsmoke mmarried mage magesq fbaby medu)
コックス回帰モデル
Stata 7.15 (コックス回帰モデル )
import delimited C:\strike.csv
stset y, failure(status==1) scale(1)
stcox x, nohr
stcurve, survival
stcurve, hazard
局所平均処置効果、ファジーな回帰の不連続デザイン
Stata 8.12 (局所平均処置効果 )
import delimited C:\late.csv
ivregress 2sls y (d = z)
Stata 8.14 (ファジーな回帰の不連続デザイン )
import delimited C:\frdd.csv
generate xtilde = x-5
generate dxtilde = d*xtilde
generate z = 0
replace z = 1 if x >= 5
generate zxtilde = z*xtilde
twoway (scatter y x if d==1, msize(small) msymbol(triangle)) (scatter y x if d==0, msize(small))
twoway (scatter y x if d==1, msize(small) msymbol(triangle)) (scatter y x if d==0, msize(small)) if x >= 3 & x <=7
regress y xtilde d dxtilde