[mohamed khalifa]

i have read alot about PSM but iam confused about the stata codes. i want to investigate the impact of female on board of directors on financial reporting quality. i my OLS i control for certain firm characteristics plus industry and year.. so what are PSM codes to do this? if i want to use nearest neighbor method with maximum diff .01

[Nermine Medhioub]

Hello @mohamed khalifa, I used the PSM technique in my previous research article and I employed the following commands:
1- pscore treatment variable independent variables ,pscore (mypscore) blockid(myblock) comsup logit detail
2-qui psmatch2 treatment, outcome(dependent variable) pscore(mypscore) caliper(0.005) neighbor(1) common noreplacement
3-pstest independent variables , treated (treatment variable) both
4-reg dependent variable post*treat treat post independent variables year fixed effect industry fixed effect.
NB: The treatment variable is a binary variable you create to divide the sample into a treatment group (where treatment=1) and a control group (where treatment =0)
the second command aims to match each treatment group with the control group and you will notice that some variables were added ("mypscore myblock comsup _pscore _treated _support _weight _cod _id _n1 _nn _pdif).
after that, you should match manually each treatment group with the control group based on the _n1.
here is the research paper providing more details about the PSM
https://www.researchgate.net/profile...nce-matter.pdf
Kind regards