Deviance information criteria for missing data models. (English) Zbl 1331.62329

Celeux, G.; Forbes, F.; Robert, C. P.; Titterington, D. M.

Bayesian Anal. 1, No. 4, 651-674 (2006).

Summary: The deviance information criterion (DIC) introduced by D. J. Spiegelhalter et al. [J. R. Stat. Soc., Ser. B, Stat. Methodol. 64, No. 4, 583–639 (2002; Zbl 1067.62010)] for model assessment and model comparison is directly inspired by linear and generalised linear models, but it is open to different possible variations in the setting of missing data models, depending in particular on whether or not the missing variables are treated as parameters. In this paper, we reassess the criterion for such models and compare different DIC constructions, testing the behaviour of these various extensions in the cases of mixtures of distributions and random effect models.

MSC:

62J05	Linear regression; mixed models
62J12	Generalized linear models (logistic models)
62C10	Bayesian problems; characterization of Bayes procedures
62F15	Bayesian inference
62B10	Statistical aspects of information-theoretic topics

Keywords:

completion; deviance; DIC; EM algorithm; MAP; model comparison; mixture model; random effect model

Software:

WinBUGS