Hi everyone,

I want to use the Diff command in Stata to run a propensity score matching diff-in-diff regression. I want to assess the impact of an employment guarantee programme on households' self-perception. However in my case the outcome variable is a categorical variable with 3 levels. The variable records answers to the following question: “Compared to 7 years ago, would you say your household is economically doing the same, better or worse today?”. Answers with "worse" are coded with 1, those with "same" are coded with 2, and those with "better" with 3.

I tried to run the command and I obtained the following results:

DIFFERENCE-IN-DIFFERENCES ESTIMATION RESULTS
Number of observations in the DIFF-IN-DIFF: 51187
Before After
Control: 23239 22488 45727
Treated: 2343 3117 5460
25582 25605

Outcome var. BETTE~F S. Err. t P>t

Before
Control 2.083
Treated 2.028
Diff (T-C) -0.055 0.015 -3.65 0.000***
After
Control 2.033
Treated 2.022
Diff (T-C) -0.011 0.012 0.89 0.374

Diff-in-Diff 0.044 0.019 2.34 0.019**

R-square: 0.07
* Means and Standard Errors are estimated by linear regression
**Robust Std. Errors
**Inference: *** p<0.01; ** p<0.05; * p<0.1



My doubt is on how should I interpret the result of this regression?



p.s.: the command I run is the following:

diff BETTER_OFF, t(TREATED) p(POST) cov(TR1_dummy2-TR1_dummy3 HHLITERATE AN1 NFBN1 ME14A CI4) addcov(ID14_dummy2-ID14_dummy11 GROUPS6_dummy2-GROUPS6_dummy6 NPERSONS FM1) kernel id(HHBASE) support robust