Hello,
I was looking for a reference explaining how the variance components of a mixed effect model are computed with the "unbounded variance components" option unthicked.
Thanks!
PY
Hi @PYS There doesn’t appear to be much detail in JMP help. However, as I understand it, it works like SAS proc mixed. REML is the method (bounded or unbounded). However, if bounded and the VC estimate via REML is less than zero, then it sets the VC to zero…and adjusts tests for the fixed effects accordingly. And that adjustment for the fixed effects may or may not be a good thing, but that is a separate issue.
Here is SAS proc mixed documentation.
https://documentation.sas.com/doc/en/pgmsascdc/9.4_3.4/statug/statug_mixed_details01.htm#statug.mixed.mixedmmtgandr
Thank you for the information!