I'm trying to estimate a Multilevel Tobit Regression using Stata. Since fixed effects are not appropriate within Tobit, I tried to run a Tobit regression with random intercepts. In this case the error occurs: numerical overflow. initial values not feasible.
I already tried to downsize the metric variables and also the sample size (down to 2,819) but still the error occurs.
Can anyone help?
Duvan Gallo Gil
Remember that the method to calculate multilevel models consists first of knowing the null models, this is to identify the levels and sub levels, once you have identified the levels and sub levels you add the fixed part that is to say the one that is not going to have a variance. Maybe the levels you are including are not really levels.
0 votes
0 thanks
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Sampling Design
- Sample Size
More Janka Goldan's questions See All
Similar questions and discussions