Markov chain sampling schemes generate dependent observations {Θ i, 0 ≤ i ≤ n} from a full joint posterior distribution π(θ∣data). Frequently, only certain marginals of this full posterior density are ...