Which posterior moments of the commensurability parameter exist

The target marginal likelihood approaches a positive constant as tau goes to infinity, so it does not repair a divergent positive moment of the prior: the posterior mean and standard deviation of tau exist exactly when the prior's do. Where they do not, both inference paths report Inf rather than a finite, run-dependent truncation - the quadrature because its outer nodes are clipped, Stan because its draws are a sample.

Usage

commensurate_tau_moments_exist(heterogeneity_prior_family, heterogeneity_prior)

Arguments

heterogeneity_prior_family

Name of the prior family.

heterogeneity_prior

Prior parameters.

Value

A logical vector with elements mean and sd.