PSM Interpretation

[Josh Spoerere]

Hi all,

I have conducted a propensity score analysis using caliper matching, but am unsure as to how I can interpret the output:

psmatch2 treatment, outcome(Y) pscore(logit1) neighbor(1) caliper(.001) common There are observations with identical propensity score values.
The sort order of the data could affect your results.

Sample | Treated Controls

Unmatched | .746302251 .567882472

ATT | .738078902 .58490566

Then it says common support off match and on match as untreated 0 and treated 50.
Hopefully someone can clarify where to look and what this means. Thanks!

[Gayatri Krishna]

Hello!
Could you please share how you have interpreted this ATT 0.738?

[GUIRA Asmo]

ATT ATU ATE, afer Endogenous Swicth Regression

I use the ESR modèle to estime the effet of subsidies on Food security. After I need the ATT ATU and ATE with the standard error and so on. When I predict only there are not these parameters. I ask a command can help I to have these parameters. I need also tests to verify the robustess of my ATT ATE ATU.

I have another question about the propensity score matching (PSM). Indeed, my variable of treatment is binary (1=subsidy and 0= otherwise). My outcome variable est also binary (1= food security and 0 otherwise). It is possible to use PSM in this situation. If yes, wath command.

Please I need some responses about these questions and thanks you.

[Jeff Wooldridge]

You say you're using PSM and also an endogenous switching regression model. For the latter you need an instrumental variable. Do you have that?