A note on the solution of the intrinsic equations
- Cano Sánchez, Juan Antonio
- Kessler, Mathieu
- Moreno Bas, Elías
Verlag: Jaén : Universidad de Jaén, 2001
ISBN: 84-8439-080-2
Datum der Publikation: 2001
Kongress: Congreso Nacional de Estadística e Investigación Operativa (26. 2001. Úbeda)
Art: Konferenz-Beitrag
Zusammenfassung
Defaults priors are topically used in Bayesian inference when there is no subjective information on the unknown parameters. In model selection and prediction the conventional default priors are not suitable because they are improper. However, they can be converted in appropiate priors via the solution of the functional intrinsic equations introduced in Berger and Pericchi (1996) and partially studied in Moreno, Bertolino and Racugno (1998). While the solution to these equations is known for nested models, to establishing conditions to characterize the solution for non-nested models is still an open problem. In this paper we push forward this problem and give illustrations of the obtained results