Abstract

The power-expected-posterior prior is used in this paper for comparing nested linear models. The asymptotic behaviour of the method is investigated for different values of the power parameter of the prior. Focus is given on the consistency of the Bayes factor of comparing the full model MpMp versus a generic submodel MℓMℓ. In each case, we allow the true generating model to be either MpMp or MℓMℓ and we keep the dimension of MℓMℓ fixed, while the dimension of MpMp can be either fixed or (grow as) O(n)O(n), with ndenoting the sample size.

References

1. Berger, J., & Pericchi, L. (2004). Training samples in objective Bayesian model selection. Annals of Statistics, 32, 841–869. doi: 10.1214/009053604000000229 [Crossref], [Web of Science ®], [Google Scholar]
2. Casella, G., Girón, F. J., Martínez, M. L., & Moreno, E. (2009). Consistency of Bayesian procedures for variable selection. Annals of Statistics, 37, 1207–1228. doi: 10.1214/08-AOS606 [Crossref], [Web of Science ®], [Google Scholar]
3. Fouskakis, D., & Ntzoufras, I. (2016a). Limiting behaviour of the Jeffreys power-expected-posterior Bayes factor in Gaussian linear models. Brazilian Journal of Probability and Statistics, 30, 299–320. doi: 10.1214/15-BJPS281 [Crossref], [Web of Science ®], [Google Scholar]
4. Fouskakis, D., & Ntzoufras, I. (2016b). Power-conditional-expected priors: Using g-priors with random imaginary data for variable selection power-conditional-expected priors. Journal of Computational and Graphical Statistics, 25, 647–664. doi: 10.1080/10618600.2015.1036996 [Taylor & Francis Online], [Web of Science ®], [Google Scholar]
5. Fouskakis, D., Ntzoufras, I., & Draper, D. (2015). Power-expected-posterior priors for variable selection in Gaussian linear models. Bayesian Analysis, 10, 75–107. doi: 10.1214/14-BA887 [Crossref], [Web of Science ®], [Google Scholar]
6. Girón, F. J., Moreno, E., & Casella, G. (2010). Consistency of objective Bayes factors as the model dimension grows. Annals of Statistics, 38, 1937–1952. doi: 10.1214/09-AOS754 [Crossref], [Web of Science ®], [Google Scholar]
7. Ibrahim, J. G., & Chen, M. H.. (2000). Power prior distributions for regression models. Statistical Science, 15, 46–60. doi: 10.1214/ss/1009212673 [Crossref], [Web of Science ®], [Google Scholar]
8. Innocent, J. K. (2016). Bayes factors consistency for nested linear models with increasing dimensions (Unpublished doctoral dissertation). University of Puerto Rico. [Crossref], [Google Scholar]
9. Kass, R. E., & Wasserman, L. (1995). A reference Bayesian test for nested hypotheses and its relationship to the Schwarz criterion. Journal of the American Statistical Association, 90, 928–934. doi: 10.1080/01621459.1995.10476592 [Taylor & Francis Online], [Web of Science ®], [Google Scholar]
10. Pérez, J. M., & Berger, J. O. (2002). Expected-posterior prior distributions for model selection. Biometrika, 89, 491–512. doi: 10.1093/biomet/89.3.491 [Crossref], [Web of Science ®], [Google Scholar]

To cite this article: D. Fouskakis, J. K. Innocent & L. Pericchi (2020) Power-expected-posterior prior Bayes factor consistency for nested linear models with increasing dimensions, Statistical Theory and Related Fields, 4:2, 162-171, DOI: 10.1080/24754269.2020.1719355 To link to this article: https://doi.org/10.1080/24754269.2020.1719355