Hello everyone, I meet such a problem when applying Stan for model parameter estimation. In my model, I want to set up a person parameter, which can be used to classify people. For example, the people who belong to category 1 used the formula (1) to calculate the response probability, the people who belong to category 2 used the formula (2) to calculate the response probability, and the people who belong to category 3 used the formula (3) to calculate the response probability. In the end, I want to estimate the person parameter, that is, I can know these persons belong to which category.

2 Likes

Hi,

that could be a reasonable approach. I am not sure where you are stuck or which part do you need help with… Could you clarify a bit? Also note that mixture models can have identifiability issues. Identifying Bayesian Mixture Models is a nice intro in both how to implement those models in Rstan and also what are the identifiability problems and how to resolve them.

Best of luck with your model!

2 Likes