Stata da Difference-in-Differences (DiD) tahlili

DiD — interventsiyaning sababiy ta'sirini aniqlash uchun ishlatiladi.

DiD nima?

Ikkita guruh (davolash va nazorat) va ikkita vaqt (oldin va keyin) solishtirish:

DiD = (Davolash_keyin - Davolash_oldin) - (Nazorat_keyin - Nazorat_oldin)

Asosiy DiD

* O'zgaruvchilar:
* treat = 1 davolash guruhi, 0 nazorat
* post  = 1 interventsiyadan keyin, 0 oldin
* y     = natija o'zgaruvchisi

generate treat_post = treat * post

regress y treat post treat_post

Koeffitsient — DiD ta'sir bahosi.

Panel ma'lumotlar bilan DiD

xtset id t

regress y treat##post, vce(cluster id)

* yoki
regress y i.treat##i.post i.id i.t, vce(cluster id)

Event Study (dinamik DiD)

* Har bir davr uchun ta'sir
regress y i.treat##i.t, vce(cluster id)

* Koeffitsientlar grafigi
coefplot, ///
    keep(1.treat#*.t) ///
    vertical ///
    yline(0) ///
    title("Event Study")

Parallel Trends tekshirish

* Interventsiyadan oldingi davrlar uchun
* Koeffitsientlar 0 ga yaqin bo'lishi kerak
regress y i.treat##i.t if t <= interventsiya_davri, ///
    vce(cluster id)

Sintaktik nazorat (Synthetic Control)

ssc install synth
synth y yosh ta_lim daromad, ///
    trunit(5) trperiod(2015) ///
    figure