Response's standard error when using skew_normal family

Interfaces brms
1

Hi, I want to incorporate measurement errors into the response variable y_i that is left-skewed. Considering the simple example yi | se(sei) ~ 1 when family = skew_normal , should sei be the standard error/deviation of yi or that of log(yi) ? Thanks!

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

  • Operating System: Ubuntu 18.04.3 LTS
  • brms Version: brms_2.10.0 in R version 3.6.1

p/s: apologies for posting it wrongly on the GitHub issue page.

1 Like
2

Hi,
sorry for taking long to respond, yes it is the SE of yi. If in doubt, you can use the make_stancode function to generate actual Stan code from your model. The resulting code is not always 100% comprehensible, but big parts of ti are and it lets you check exactly these kind of things.

Hope that helps!