www.cmm.bristol.ac.uk/forum

Dear Mr. Leckie.

I want to estimate a generalized linear mixed model with gamma distributed error terms and random effects using runmlwin. The runmwlin-command only allows me to specify a normal, binomial, multinomial, poisson or nbinomial distribution. Is this a limitation of MlwiN or do I miss the point somewhere in my modelling?

Thanks for your help!

Hi Andreas,

Yes you are correct, you can't fit these models in MLwiN and this is why there is no discussion of them in the runmlwin help file.

You could try fitting these models using the winbugs command to call winbugs from within Stata.

Best wishes

George

I will do that. Thank you for your answer!

Best,

Andreas