I look at some unusual data where the median was higher than the mode, and show how to model it in Stan as a mixture of two negative binomial …