Rethinking multi-level models

Operating System: Mac OS
Interface Version: Rstudio
Compiler/Toolkit: rethinking package - Richard McElreath
Hi there. Trying once again to start a topic for those using the rethinking package. I have a multi-level model that won’t run, even with non-centered parameterisation. All code available at https://github.com/lottybrand22/GH_prestigeQualtrics in the Round3Summary.R file.
Might it be because some participants provide only one observation, whereas others provide 2, 3, and some up to 30-40? I thought implementing a varying effect/clustering on participant would be able to handle this? Perhaps not?

“Won’t run” is pretty broad - too vague. Do you mean will not compile? Will not converge? Samples slowly? Exits in an error?

More details would be helpful.

1 Like

That is a great use-case for hierarchical modeling. I’d like to second @mespe’s comment - what is the output you’re seeing and what are you expecting to see?