We are trying to work out a regression on poverty analysis using underweight as a proxy, i.e. higher underweight, poorer (percent children below normal weight). Using a 5 year panel data by municipality (143 municipalities). In addition, we made both sides of equation in natural log format for interpretation purposes (i.e. as elasticities). In this case, we have a "censored regression" and therefore we would use xttobit. Censored because you would not have negative values for underweight, nor zero underweight. Wanted to ask for recommendations on what we could use in stata to test for omitted variables and heteroskedasticity? I understand, the LR test, and p-value of chi2 is given by xttobit to indicate significance of model. No R2 given the censored regression that is solved by maximum likelihood (between 0 and 1 values). Any recommendations will be appreciated.