An estimation procedure using Bayesian inference. Suppose that the random variables corresponding to n sample means are denoted by Y¯₁, Y¯₂,…, Y¯_n. The belief is that each sample is obtained from a population having the same variance, σ², but different means, μ₁, μ₂,…, μ_n, with the means themselves being regarded as observations from a distribution with mean μ and variance ω². The parameters μ and ω² are called hyperparameters and are assigned prior distributions referred to as hyperpriors.