Hi all,
I am writing to ask for some guidance. I have found great resources on propensity score matching. The challenge I'm facing however, is that I have clustered data (students nested within schools). I have read as much as I can about different options to do the matching (for example, within-cluster matching, across-cluster matching, or a two state propensity score matching). Admittedly, I'm not an expert on this topic. But I have decided that the best option is to use a two-stage propensity score matching. To do so, two authors (Rickles and Seltzer, 2015) provide a step by step description of the process one should/could employ. While the steps they describe are clear, I am having trouble figuring out what the syntax in stata should be. Does anyone have any examples of having used a two-stage matching strategy? I can provide more specifics, for example, the steps they list. But I'm not sure if that would be helpful. I'll list the first step, in the first phase.
In the design phase, the first step is for the researcher to use a two-stage matching strategy to construct treatment and control groups within each site (in my case schools); those treatment units without an acceptable within-match are then matched to control units in a different site with similar key site-level characteristics.
In my case, I have multiple school-level variables I could match on (for example, % of student who qualify for free/reduced price lunch, % of students categorized as English learners, racial/ethnic composition of the school, etc.). Using these variables, I could group schools into quintiles. But I'm not quite sure how I would go about doing this first step in stata.
This is my first time posting, so please let me know if you need additional information to help me with this puzzle.
Diana
I am writing to ask for some guidance. I have found great resources on propensity score matching. The challenge I'm facing however, is that I have clustered data (students nested within schools). I have read as much as I can about different options to do the matching (for example, within-cluster matching, across-cluster matching, or a two state propensity score matching). Admittedly, I'm not an expert on this topic. But I have decided that the best option is to use a two-stage propensity score matching. To do so, two authors (Rickles and Seltzer, 2015) provide a step by step description of the process one should/could employ. While the steps they describe are clear, I am having trouble figuring out what the syntax in stata should be. Does anyone have any examples of having used a two-stage matching strategy? I can provide more specifics, for example, the steps they list. But I'm not sure if that would be helpful. I'll list the first step, in the first phase.
In the design phase, the first step is for the researcher to use a two-stage matching strategy to construct treatment and control groups within each site (in my case schools); those treatment units without an acceptable within-match are then matched to control units in a different site with similar key site-level characteristics.
In my case, I have multiple school-level variables I could match on (for example, % of student who qualify for free/reduced price lunch, % of students categorized as English learners, racial/ethnic composition of the school, etc.). Using these variables, I could group schools into quintiles. But I'm not quite sure how I would go about doing this first step in stata.
This is my first time posting, so please let me know if you need additional information to help me with this puzzle.
Diana
