Limiting behavior of the Jeffreys Power-Expected-Posterior Bayes Factor\n in Gaussian Linear Models

2013/07/09 by Dimitris Fouskakis, Fouskakis, Dimitris, Ioannis Ntzoufras +1
Computer Science · Decision Sciences · Mathematics · #Advanced Statistical Methods and Models #Bayesian Modeling and Causal Inference #Computation (stat.CO) #FOS: Computer and information sciences #Multi-Criteria Decision Making

paper · pdf · doi:10.48550/arxiv.1307.2435

openalex publication_date 2013/07/09 · openalex created_date 2022/10/01 · openalex updated_date 2026/07/28

Abstract

Expected-posterior priors (EPP) have been proved to be extremely useful for\ntesting hypothesis on the regression coefficients of normal linear models. One\nof the advantages of using EPPs is that impropriety of baseline priors causes\nno indeterminacy. However, in regression problems, they based on one or more\n\training samples, that could influence the resulting posterior\ndistribution. The power-expected-posterior priors are minimally-informative\npriors that diminishing the effect of training samples on the EPP approach, by\ncombining ideas from the power-prior and unit-information-prior methodologies.\nIn this paper we show the consistency of the Bayes factors when using the\npower-expected-posterior priors, with the independence Jeffreys (or reference)\nprior as a baseline, for normal linear models under very mild conditions on the\ndesign matrix.\n

Related

• Power-Expected-Posterior Priors as Mixtures of g-Priors
• On posterior propriety for the Student-t linear regression model under Jeffreys priors
• Power-Expected-Posterior Priors for Generalized Linear Models
• Laplace Power-expected-posterior priors for generalized linear models with applications to logistic regression
• Consistency of Bayes factors for linear models