Dear Statalisters,

I am currently writing my thesis on the topic of Multidimensional Empowerment. For the analysis, I have access to the following files: baseline data of the treatment group (measured in 2018), data of the control group (measured in 2020) and end line data of the treatment group (measured in 2020). For the analysis, I want to use PSMatching, for which I first used the cross-sectional data of control (2020) and treatment (2020), however, as some of the demographic questions were not asked AGAIN in the end line I only have 5-6 covariates to match on. It did give me a reasonable match but, of course, I am not happy with this, as it will leave me with a huge bias... and therefore I came up with the following idea: could I merge the baseline data of the treatment group (measured in 2018) with the cross sectional data that I have to match the households on their baseline characteristics, which are measured for the control group in 2020 and for the treatment group in 2018?

Linda