www.cmm.bristol.ac.uk/forum

Hi Guys,

I am running multinomial single level models using MCMC and wanted to know what the minimum acceptable ESS is for MCMC convergence. I seem to remember this being discussed in the workshop and cant remember they said either 50 or 500 I am not sure, anyone know?

If not, then also can you tell me if its ok to use IGLS for a SL multinomial model? Do I have to use MCMC or does it not matter?

Also I wanted to know whether it is ok to do X2 tests on fixed effects levels in Multinomial models, I know its ok in binomial models but not sure if there is any reason why I shouldnt do this in Multinomial.

Thanks

Emma

In a CMM course that I took, the number that I remember was closer to 50 than 500.