Hi all,
Is it possible to run a mixed model with random effects, and use the estimate only variance component option and request 95% confidence intervals estimates for the variance components, similar to the output you receive when you select unbounded variance components option and request 95% confidence intervals from the output?
Thanks,
Mick