Optimizing Stan Code to estimate variable transformation parameters

Short on time so just quick notes:

  • Remove terms from your formula to find the smallest model that diverges and largest model that converges. You have a big formula, so it is quite possible you just don’t have enough data to inform all the parameters.
  • Try not modelling correlations in the varying intercepts (i.e. use (1||group) instead of 1|group )
  • Check out our new Divergent transitions - a primer