Increasing the target_accept to 0.95 resulted in 2 divergences and the diagnosis remains the same. I’ll try a more informative prior to see if that changes things. So as long as the divergences are gone and the model reasonably mimics the data, these coefficient values should be fine?
1 Like