2

This is my first time posting here, so please forgive me if I am not following some form of etiquette.

My question is in regard to the following evaluation design:

We are examining the impact of policy X. It was implemented at the same time for all units j. We have the ability to collect measurements of A for all participants (i) in a given unit j in both the pre- and post-implementation periods. We do not have a non-treated post-implementation group for comparison, so the following was suggested:

We can match post-participants to the pre-participants on relevant person level factors. Second, to account for historical changes in j-level factors, we can adjust for changes in said factors at level 2 of a hierarchical model.

Overall, our goal would be to estimate an adjusted avearage change in A at the j-level or even estimate what the predicted change would have been when we assume no change in j-level factors.

What are the downsides of this approach?.

ReliableResearch
  • 416
  • 1
    Hi and welcome to the site. Your question is fine etiquette wise, but as a general comment more specific questions are likely to get better answers. For example, you ask for downsides in general - do you have any specific concerns about the method? – Corvus Feb 25 '13 at 22:29
  • Nothing too specific comes to mind. The goal is to establish a reasonable counterfactual in the absence of a better quasi-experimental technique. How would you rate our approach to establishing a counterfactual? – ReliableResearch Feb 25 '13 at 23:39

0 Answers0