Hierarchical or "multilevel" regression models typically parameterize the mean response conditional on unobserved latent variables or "random" effects and then make simple assumptions regarding their distribution. The ...