Logistic Regression with severe divergences

[alj]

Your hunch was right! Sampling quickly and efficiently, thank you so much. That is a great trick! Is there general advice on when to prefer the p versus logit_p parameterisation of Bernoulli?

Thanks again!