Restricted type II maximum likelihood priors on regression coefficients

In Bayesian hypothesis testing and model selection, prior distributions must be chosen carefully. For example, setting arbitrarily large prior scales for location parameters, which is common practice in estimation problems, can lead to undesirable behavior in testing (see Lindley’s paradox; Lindley...

Descripción completa

Detalles Bibliográficos
Autores: Peña Pizarro, Víctor|||0000-0002-3801-5203, Berger, James O.
Tipo de recurso: artículo
Fecha de publicación:2020
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/433737
Acceso en línea:https://hdl.handle.net/2117/433737
https://dx.doi.org/10.1214/19-ba1188
Access Level:acceso abierto
Palabra clave:Mathematical statistics
Estadística matemàtica
Classificació AMS::62 Statistics::62F Parametric inference
Àrees temàtiques de la UPC::Matemàtiques i estadística::Estadística matemàtica::Inferència estadística
Descripción
Sumario:In Bayesian hypothesis testing and model selection, prior distributions must be chosen carefully. For example, setting arbitrarily large prior scales for location parameters, which is common practice in estimation problems, can lead to undesirable behavior in testing (see Lindley’s paradox; Lindley (1957)). We study the properties of some restricted type II maximum likelihood (type II ML) priors on regression coefficients. In type II ML, hyperparameters are “estimated” by maximizing the marginal likelihood of a model. In this article, we define priors by estimating their variances or covariance matrices, adding restrictions which ensure that the resulting priors are at least as vague as conventional proper priors for model uncertainty. We find that these type II ML priors typically yield results that are close to answers obtained with the Bayesian Information Criterion (BIC; Schwarz (1978)).