Width of Confidence Intervals for Variance Estimates in Contrast to Point Estimates

We are conducting a variance decomposition using a hierarchical linear random effects Bayesian model to investigate the variance in a DV that is affected by three nested layers. We estimate credible (confidence) intervals (from the posterior distribution) for the variance explained by each of the layers. We observe that they are larger than “usual”, when comparing them to the credible intervals of the point estimates. We are wondering if this is to be expected given that the variance is a quadratic term.

It seems you could have asked your two other questions (Zero inflated and right skewed dependent variable – is the Tweedie distribution a good solution? and Control for right skewed DV leading to well fitting posterior) in the same post, and tell more about your model, inference diagnostics, and model checking diagnostics used. It is likely that no-one has commented as there is not enough information about your model and data.

1 Like