I'm using Propensity Score Matching method to estimate a treatment effect (psmatch2 used). On top of that, I want to assess the robustness in the treatment effect to potential unobserved factors. Hence I use rbounds (Roesnbaum bounds analysis) following the estimation results from psmatch2. My question is: if I'm interested in knowing not just the main effect, but also heterogeneous effect across groups of individuals, how would the implementations be done with STATA?
So, let me start with plain model (only looking at main treatment effect). Below are the codes as an example:
Code:
psmatch2 treatment gender age X, out(wage) gen delta = wage12- _wage12 if _treatment==1 & _support==1 rbounds delta, gamma(1 (0.1) 2)
Now, if I want to know how the treatment effect varies across gender, how the PSM and Rosenbuam bounds analysis would be? That is, if we write a regression model:
Code:
wage ~ a*treatment + b*treatment*female + c*age + d*X + e
0 Response to psmath2 and rbounds: estimating 'interaction effect' using Propensity Score Matching, and the sensitivity analysis
Post a Comment