I am running various bayesian path models in r ('blavaan' package). I understand that LOO is a more robust fit index when working with a smaller samples size (and weakly informative priors). I am using LOO-CV (PSIS) method to compare path models; however, I have additional questions.

  1. How do I interpret LOO-CV values to determine which model has the better fit?
  2. Should multiple fit indices be present in bayesian SEM results?
  3. Is there an alternative r package that runs 'loo-cv' analyses? The "loo" package will not load.

Thank you.


1 Answer 1

  1. LOO-CV works like an information criterion (lowest value wins): for comparison, not an index for evaluating fit in an absolute sense.
  2. "present"? Multiple are available.
  3. blavaan provides the DIC, WAIC, and LOO-IC (approximation to LOO-CV) via the loo package. If you have trouble with installing it, that is a separate issue to post about (e.g., on StackOverflow, or the Stan forum). You can read through all their introductory information on the blavaan web site.
  • $\begingroup$ Thank you very much! This was very helpful. (also - apologies presented, "present" was a typo) $\endgroup$
    – kirstenq
    Jun 30, 2022 at 16:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.