Thanks for the reply Sean.
I will look into the possibility of using an empirical distribution in Mathematica to model a categorical distribution. They not the same thing by definition, but it may be a better soution than using the mulitnomial.
Note: A categorical distribution is a generalization of a bernoulii distribution. Where are bernoulli distribution models two categores the categorical distribution models "k" categories. My intestest is tied to doing Bayesian statistics.