Bob, would it be possible to use one different random number in each different iteration?
I am trying to implement a two-step estimator. Estimates from the first step become data for the second stage, but should not be influenced by the second step. The second step needs to account for the uncertainty in the estimates from the first step.
What I am doing so far is estimate the first step and store the chains. Then pass the mean and sd to the second step and use them to generate a parameter, but it seems the draws are still influenced by the likelihood of the second step.