Specify the RE distribution (follow-up)


#1

Just wanted to report that I used this (undocumented) feature to specify a student distribution for a random effect, and it worked well, producing a modest, but nonetheless meaningful (> SE) improvement in estimation via LOO, as compared to the standard multivariate normal distribution for a group.

More background here: https://github.com/paul-buerkner/brms/issues/231

A nice feature addition.

Please also provide the following information in addition to your question:

  • Operating System: Windows 10
  • brms Version: 2.3

#2

Great to hear! Thanks for testing.