Biometrika Advance Access originally published online on January 22, 2009
Biometrika 2009 96(1):163-173; doi:10.1093/biomet/asn060
Articles |
Wilcoxon-type generalized Bayesian information criterion
School of Statistics, University of Minnesota, 313 Ford Hall, 224 Church Street S.E., Minneapolis, Minnesota 55455, U.S.A. lan{at}stat.umn.edu
Received for publication 1 June 2007. Revision received 1 June 2008.
We develop a generalized Bayesian information criterion for regression model selection. The new criterion relaxes the usually strong distributional assumption associated with Schwarz's BIC by adopting a Wilcoxon-type dispersion function and appropriately adjusting the penalty term. We establish that the Wilcoxon-type generalized BIC preserves the consistency of Schwarz's BIC without the need to assume a parametric likelihood. We also show that it outperforms Schwarz's BIC with heavier-tailed data in the sense that asymptotically it can yield substantially smaller L2 risk. On the other hand, when the data are normally distributed, both criteria have similar L2 risk. The new criterion function is convex and can be conveniently computed using existing statistical software. Our proposal provides a flexible yet highly efficient alternative to Schwarz's BIC; at the same time, it broadens the scope of Wilcoxon inference, which has played a fundamental role in classical nonparametric analysis.
Key Words: BIC • Bayesian information criterion • Consistency of model selection • Heavier-tailed distribution • L2 risk • Rank • Wilcoxon inference
References
-
Broman K. W., Speed T. P. A model selection approach for the identification of quantitative trait loci in experimental crosses. J. R. Statist. Soc. (2002) B 64:641–56.[CrossRef]
Burnham K. P., Anderson D. R. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach (2002) 2nd ed. New York: Springer.
Fan J., Li R. Variable selection via nonconcave penalized likelihood and its oracle properties. J. Am. Statist. Assoc. (2001) 96:1348–60.[CrossRef][Web of Science]
Fan J., Peng H. On non-concave penalized likelihood with diverging number of parameters. Ann. Statist. (2004) 32:928–61.[CrossRef]
Hettmansperger T. P., McKean J. W. A geometric interpretation of inferences based on ranks in the linear model. J. Am. Statist. Assoc. (1983) 78:885–93.[CrossRef][Web of Science]
Hettmansperger T. P., McKean J. W. Robust Nonparametric Statistical Methods (1998) London: Arnold.
Jaeckel L. A. Estimating regression coefficients by minimizing the dispersion of residuals. Ann. Math. Statist. (1972) 43:1449–58.[CrossRef]
Jure
ková J. Nonparametric estimate of regression coefficients. Ann. Math. Statist. (1971) 42:1328–38.[CrossRef]
Kass R. E., Wasserman L. A reference Bayesian test for nested hypotheses and its relationship to the Schwarz criterion. J. Am. Statist. Assoc. (1995) 90:928–34.[CrossRef][Web of Science]
Konishi S., Ando T., Imoto S. Bayesian information criteria and smoothing parameter selection in radial basis function networks. Biometrika (2004) 91:27–43.
Lazar N. A. Bayesian empirical likelihood. Biometrika (2003) 90:319–26.
McKean J. W., Hettmansperger T. P. Tests of hypotheses of the general linear models based on ranks. Commun. Statist. (1976) A 5:693–709.
McKean J. W., Hettmansperger T. P. A robust analysis of the general linear model based on one-step R-estimates. Biometrika (1978) 65:571–9.
McKean J. W., Schrader R. M. The geometry of robust procedures in linear models. J. R. Statist. Soc. (1980) B 42:366–71.
Nishii R. Asymptotic properties of criteria for selection of variables in multiple regression. Ann. Statist. (1984) 12:758–65.[CrossRef]
Pettitt A. N. Inference for the linear model using a likelihood based on ranks. J. R. Statist. Soc. (1982) B 44:234–43.
Raftery A. E. Approximate Bayes factors and accounting for model uncertainty in generalized linear models. Biometrika (1996) 83:251–66.
Schwarz G. Estimating the dimension of a model. Ann. Statist. (1978) 6:461–4.[CrossRef]
Shi P., Tsai C.-L. A note on the unification of the Akaike information criterion. J. R. Statist. Soc. (1998) B 60:551–8.[CrossRef]
Siegmund D. Model selection in irregular problems: Applications to mapping quantitative trait loci. Biometrika (2004) 91:785–800.
Sievers G. L., Abebe A. Rank estimation of regression coefficients using iterated reweighted least squares. J. Statist. Comp. Simul. (2004) 74:821–31.[CrossRef]
Terpstra J., McKean J. Rank-based analysis of linear models using R. J. Statist. Soft. (2005) 14:1–26.
Tibshirani R. Regression shrinkage and selection via the Lasso. J. R. Statist. Soc. (1996) B 58:267–88.
Tierney L., Kadane J. B. Accurate approximations for posterior moments and marginal densities. J. Am. Statist. Assoc. (1986) 81:82–6.[CrossRef][Web of Science]
Volinsky C. T., Raftery A. E. Bayesian information criterion for censored survival models. Biometrics (2000) 56:256–62.[CrossRef][Web of Science][Medline]
Weisberg S. Applied Linear Regression (2005) 3rd ed. Hoboken, NJ: John Wiley.
White H. Consequences and detection of misspecified nonlinear regression models. J. Am. Statist. Assoc. (1981) 76:419–33.[CrossRef][Web of Science]
Zhan X., Hettmansperger T. P. Bayesian R-estimates in two-sample location models. Comp. Statist. Data Anal. (2007) 51:5077–89.[CrossRef]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||