Dear All,

I would like to ask about matching. If I have a panel data and I want to run Difference in Difference using PSM and my dependent variable is a dummy variable (having access or not), is it correct if I take a difference between post-year minus pre-year program variable, let say this variable called Rediff_access (1 or 0) as my dependent variable?

My second question is how to apply the geographical fixed effects in the model? Below is my PSM model employing nearest neighbour matching.

dprobit treatment $xlistvariable

predict double score if e(sample),pr

psmatch2 treated, pscore(score) outcome(rediff_access) common

I would be really grateful if you could help me inform how to properly specify the DID using PSM with FEs.

Thank you.

Best,
Daim