Skip to content

zero random effect sizes with binomial lmer

3 messages · Daniel Ezra Johnson, Gregor Gorjanc

#
Daniel Ezra Johnson <johnson4 <at> babel.ling.upenn.edu> writes:
...
...

BLUPs are essentially shrinkage estimates, where shrinkage is determined with
magnitude of variance. Lower variance more shrinkage towards the mean - zero 
in this case. So this is not a bug.

Gregor
#
Daniel Ezra Johnson <johnson4 <at> babel.ling.upenn.edu> writes:
...
...

You might try with mcmcsamp to get a better view of posterior distributions of
your parameters. It might be the case that MLE for item variance is near 0, while
its posterior distribution covers also values that are higher than 0.

Gregor