I am doing a propensity score matching analysis. When it came to checking the common support between my treatment and control group, I created the following graph using the code below. I am unsure whether the fact that my curves follow the same trend is a good thing (i.e. my treatment and control group have been well matched) or that the similarities in their properties in fact indicates that the effect of being in a treatment or control group is arbitrary.
Graph pscore.gph
* Check the common support area with two separate kernel density estimates
kdensity pshighcon if highevent==1, ti("Treated")
kdensity pshighcon if highevent==0, ti("Control")
graph twoway (kdensity pshighcon if highevent==1)(kdensity pshighcon if highevent==0), ///
ytitle("Density") xtitle("Propensity Score") legend(label(1 "Treated") label(2 "Control"))
Graph pscore.gph
* Check the common support area with two separate kernel density estimates
kdensity pshighcon if highevent==1, ti("Treated")
kdensity pshighcon if highevent==0, ti("Control")
graph twoway (kdensity pshighcon if highevent==1)(kdensity pshighcon if highevent==0), ///
ytitle("Density") xtitle("Propensity Score") legend(label(1 "Treated") label(2 "Control"))
