Propensity Score Matching: how I can estimate with groups?

[Lucas Arp]

Hi all


I am trying to estimate the impact of an educational strategy on the academic performance of students. I have identified my controls and treatments: students who were part of the program and students who were not. I'm using the psmatch2 command and I do not know how I could specify that student performance be compared by groups. In other words, that students are matched by the independent variables that I added and in turn that they select them according to the variable "classroom" to which each one belongs.

I know that the command "by" it's not allowed.

What command would be used to make this distinction? Or what code other than psmatch2 allows me to do it?

Thank u

[David Radwin]

Do you mean that within each classroom, there are some treated and some control students? If so, please see the section titled "Matching within strata" in the the psmatch2 (Leuven and Sianesi, available from SSC) help file.