Introduction to Lindley and Smith (1972) Bayes Estimates for the Linear Model

In 1972 there were no MCMC methods readily available to the statistical community in general and to the Bayesian statisticians in particular. Realistic formulations of models for practical situations were dismissed if solutions were neither available in closed form nor amenable to numerical calculation. Thus much effort was devoted to either approximating solutions to these practically desirable models or formulating other models that were both realistic and solvable. It was also the case the Bayesian modeling did not find general acceptance unless it could be shown that such models had desirable frequentist properties. Thus, when viewed in this context, this contribution by Lindley and Smith (hereafter LS) must surely qualify for the accolade, “breakthrough.”