Is Bayes Posterior just Quick and Dirty Confidence?

Abstract

Bayes [Philos. Trans. R. Soc. Lond. 53 (1763) 370--418; 54 296--325] introduced the observed likelihood function to statistical inference and provided a weight function to calibrate the parameter; he also introduced a confidence distribution on the parameter space but did not provide present justifications. Of course the names likelihood and confidence did not appear until much later: Fisher [Philos. Trans. R. Soc. Lond. Ser. A Math. Phys. Eng. Sci. 222 (1922) 309--368] for likelihood and Neyman [Philos. Trans. R. Soc. Lond. Ser. A Math. Phys. Eng. Sci. 237 (1937) 333--380] for confidence. Lindley [J. Roy. Statist. Soc. Ser. B 20 (1958) 102--107] showed that the Bayes and the confidence results were different when the model was not location. This paper examines the occurrence of true statements from the Bayes approach and from the confidence approach, and shows that the proportion of true statements in the Bayes case depends critically on the presence of linearity in the model; and with departure from this linearity the Bayes approach can be a poor approximation and be seriously misleading. Bayesian integration of weighted likelihood thus provides a first-order linear approximation to confidence, but without linearity can give substantially incorrect results.Comment: Published in at http://dx.doi.org/10.1214/11-STS352 the Statistical Science (http://www.imstat.org/sts/) by the Institute of Mathematical Statistics (http://www.imstat.org

Valid Prior-Free Probabilistic Inference and Its Applications in Medical Statistics

“…Since (1.1) implies that θ = x − Z, the distribution on Z defines a distribution on θ when x is fixed at its observed value. That distribution is θ ∼ N (x, 1) conditional on x, which is the same as the Bayesian posterior of θ obtained by putting the flat prior on θ; see and Fraser (2011). When used for inference, this posterior distribution has nice frequency properties for certain assertions or hypotheses on the unknown quantity θ.…”

“…In any case, when a prior is taken for everything, the inference problem is reduced to an exercise of usual probability calculus. Following, e.g., Fraser (2011), for conceptual clarity we simply refer to such models as probability models. This chapter is concerned with statistical inference when there is no known prior for some unknown quantity.…”

Bayesian Inference of GNSS Failures

“…It is assumed that the physical process's true failure rate exists and may be estimated using Bayesian analysis as opposed to a classical approach (Neyman, 1937). While a confidence distribution may be constructed by extending the classical notion of the confidence interval, such an approach has not been extensively formalised (Fraser, 2011). The advantages of a Bayesian approach are that it allows a more natural means to express the posterior distribution for the estimated random parameter.…”

Vector Exponential Models and Second Order Inference

“…The approximations also typically use two intermediate measures of departure, the signed likelihood root r and the maximum likelihood departure q in the canonical  scaling: with k constant to that order. These approximations from Daniels (1954) and Barndorff-Nielsen(1991) provide exceptional access to statistical inference, both theoretical and practical; for some recent discussion see Fraser (2011).…”