Dear all,

I would like to do a mediation analysis with panel data. This is the first time I conduct a mediation analysis. So I'm sorry for this basic question. My independent variable is binary, my mediator is a count variable and the dependent variable is continuous. I read that there are two possibilities in this case to run a mediation analysis:

1) Run two separate regressions (first: IV -> M, second: M ->DV). For the first regression I would use the -xtbnreg- command as my moderator is a count variable. For the second regression I would use -xtreg-. For both regressions I would use the -fe- option (result of Hausman-test). There is a mediating effect if both regressions are significant.

2) Instead of two separate regressions, there is the possibility to run a mediation analysis via SEM. I read that the command -gsem- is suitable for panel data. However, I am using the Stata 12 version which does not include this command.

Do you have any recommendations for me on how to conduct a mediation with panel data using Stata 12 in my case?