Dear Statalist,

I am running a Difference in Difference analysis for a policy effect on employment and part time status.

Treated: Alaska
Time: 1982
Control group: the remaining US states
Covariates: Age, industry, education, ethnicity, marital status

In order to build a better control group, I wanted to use propensity scores. Here is what I did:
Code:
probit treated teen twnt yadult oadult older indcat1 indcat2 indcat3 indcat4 indcat5 female lesshs highsc coll white afram nativ asian mix married divorce single employ partt, vce(r)
predict pscore
psmatch2 treated, outcome(employ partt) pscore(pscore) common
I the used the psmatch2 estimated _pscore in the difference in difference, as follows:
Code:
diff employ, t(treated) p(time) cov(agecat indcat female educ race marst) ps(_pscore)
I attach the results as image.

Now, I am confused about two things:

1. How should I interpret the results as follows:
- Diff (T-C) Before: before the policy took action, Alaska experienced an employment rate of 0.031 percentage points lower when compared to the control group
- Diff (T-C) After: after the policy was implemented Alaska experienced an employment rate of 0.022 percentage points lower when compared to the control group
- Diff in Diff: one could estimate the effect of the policy in a 0.010 percentage points increase in the employment rate in Alaska, all other things considered.?

2. The psmatch2 commands creates the _support variable. I should run the diff in diff estimation only if _support==1, right?

Thank you so much in advance for any help, it would be greatly appreciated.
IMArray