Table 2.

Bayesian hierarchical model specification.

Parameter	Proposal
${\hat{\rho }}_{\mathrm{ml}}$	N (ρml, V(ρml, Nml, kml))
logit (ρml)	N (μm, ${\sigma }_w^2$ /wml)
μm	N (μ, ${\sigma }_b^2$ /wm)
μ	N (0, 10 000)
σw	U [0, 5]
σb	U [0, 5]

N_ml and k_ml are the cluster size and number of clusters; vague priors N (0, 10 000) were assigned to mean µ for each model throughout; U [0, 5] is a uniform distribution.