teffects for binary and multinomial outcome variables

[Jun Xu]

Dear Statalisters,

I am trying to estimate the treatment effects of a binary treatment for a binary outcome variable. First off, what does the coefficient column tells us? Is it a coefficient from running a binary logit of the binary outcome variable on the balanced treatment indicator, or is it some probability?

Treatment-effects estimation Number of obs = 17,795
Estimator : propensity-score matching Matches: requested = 1
Outcome model : matching min = 1
Treatment model: logit max = 103
------------------------------------------------------------------------------
| AI robust
usualpl | Coefficient std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
ATE |
notcovR |
(1 vs 0) | -.2735564 .0157245 -17.40 0.000 -.3043759 -.2427369
------------------------------------------------------------------------------

A follow-up question is if I want to estimate if this treatment effect varies across different groups (kind of heterogeneous treatment effects), say race, is there any way to do it in Stata? What if I have multiple treatments for a multinomial outcome variable?

Jun Xu, PhD
Professor
Data Science Program
Department of Sociology
Ball State University
Muncie, IN 47306
Web: sites.google.com/site/socjunxu

[Jun Xu]

Looks like I posted my question without doing enough homework. Below are two threads of posts that I found on the forum that might've provided answers,

https://www.statalist.org/forums/for...binary-outcome

https://www.statalist.org/forums/for...binary-outcome

[Jun Xu]

Dear all,

This a kind of follow-up of what I asked about previously. Where can I find the technical details about how, for example, teffects psmatch calculates ATE/ATET for binary outcomes? Is there any documentation out there about the interpretation of results. Thanks!