Dear Stata community,
I'm performing an IPD meta-analysis on data that has been derived from about 80 studies, some of which had exclusion criteria for the outcome variable of interest (y) above and below a certain level - meaning I need to use truncated regression as the standard OLS method would result in biased estimates.
However, I also need to take into account random effects due to the clustering of similar individuals within certain studies / study sites.
I can see that an "xttobit" command exists for performing regression analysis on censored data - has anybody written an "xttruncreg" command, or know how I can adjust for random effects using truncreg?
Thank you very much!!
Elke
I'm performing an IPD meta-analysis on data that has been derived from about 80 studies, some of which had exclusion criteria for the outcome variable of interest (y) above and below a certain level - meaning I need to use truncated regression as the standard OLS method would result in biased estimates.
However, I also need to take into account random effects due to the clustering of similar individuals within certain studies / study sites.
I can see that an "xttobit" command exists for performing regression analysis on censored data - has anybody written an "xttruncreg" command, or know how I can adjust for random effects using truncreg?
Thank you very much!!
Elke
Comment