Probability matching priors: Higher order asymptotics (Q1420353)

This monograph presents the first state of the art presentation on probability matching priors and related questions in book form. Since the exposition of the material is not meant to be encyclopedic, the reader can't expect to find any detail on the subject. Recent developments concerning probability matching priors addressing problems in estimation and prediction are covered comprehensively. Not all of the proofs are given in full detail, but the key steps leading to the main arguments are well outlined. Beside being useful for researchers in the field, the monograph might also be used as a basis for a course in Bayesian asymptotics. Matching priors for posterior quantiles of first and second order are characterized via appropriate partial differential equations. The outline is then extended to single matching priors for multiple parameters and parametric functions. Further, priors ensuring approximate frequentist validity of highest posterior density regions are considered. By inversion of commonly used statistics, posterior credible regions can be deduced which allows to narrow down the choice for matching priors. Finally, priors ensuring approximate frequentist validity of Bayesian predictive regions are characterized. The book is divided into 6 Chapters. Chapter 1: Intoduction and the Shrinkage Argument; Chapter 2: Matching Priors for Posterior Quantiles; Chapter 3: Matching Priors for Distribution Functions; Chapter 4: Matching Priors for Highest Posterior Density Regions; Chapter 5: Matching Priors for Other Credible Regions; Chapter 6: Matching Priors for Prediction.