Skip to main content
Journal cover image

On the prevalence of information inconsistency in normal linear models

Publication ,  Journal Article
Mulder, J; Berger, JO; Peña, V; Bayarri, MJ
Published in: Test
March 1, 2021

Informally, ‘information inconsistency’ is the property that has been observed in some Bayesian hypothesis testing and model selection scenarios whereby the Bayesian conclusion does not become definitive when the data seem to become definitive. An example is that, when performing a t test using standard conjugate priors, the Bayes factor of the alternative hypothesis to the null hypothesis remains bounded as the t statistic grows to infinity. The goal of this paper is to thoroughly investigate information inconsistency in various Bayesian testing problems. We consider precise hypothesis tests, one-sided hypothesis tests, and multiple hypothesis tests under normal linear models with dependent observations. Standard priors are considered, such as conjugate and semi-conjugate priors, as well as variations of Zellner’s g prior (e.g., fixed g priors, mixtures of g priors, and adaptive (data-based) g priors). It is shown that information inconsistency is a widespread problem using standard priors while certain theoretically recommended priors, including scale mixtures of conjugate priors and adaptive priors, are information consistent.

Duke Scholars

Published In

Test

DOI

EISSN

1863-8260

ISSN

1133-0686

Publication Date

March 1, 2021

Volume

30

Issue

1

Start / End Page

103 / 132

Related Subject Headings

  • Statistics & Probability
  • 4905 Statistics
  • 0104 Statistics
 

Citation

APA
Chicago
ICMJE
MLA
NLM
Mulder, J., Berger, J. O., Peña, V., & Bayarri, M. J. (2021). On the prevalence of information inconsistency in normal linear models. Test, 30(1), 103–132. https://doi.org/10.1007/s11749-020-00704-4
Mulder, J., J. O. Berger, V. Peña, and M. J. Bayarri. “On the prevalence of information inconsistency in normal linear models.” Test 30, no. 1 (March 1, 2021): 103–32. https://doi.org/10.1007/s11749-020-00704-4.
Mulder J, Berger JO, Peña V, Bayarri MJ. On the prevalence of information inconsistency in normal linear models. Test. 2021 Mar 1;30(1):103–32.
Mulder, J., et al. “On the prevalence of information inconsistency in normal linear models.” Test, vol. 30, no. 1, Mar. 2021, pp. 103–32. Scopus, doi:10.1007/s11749-020-00704-4.
Mulder J, Berger JO, Peña V, Bayarri MJ. On the prevalence of information inconsistency in normal linear models. Test. 2021 Mar 1;30(1):103–132.
Journal cover image

Published In

Test

DOI

EISSN

1863-8260

ISSN

1133-0686

Publication Date

March 1, 2021

Volume

30

Issue

1

Start / End Page

103 / 132

Related Subject Headings

  • Statistics & Probability
  • 4905 Statistics
  • 0104 Statistics