A note on bootstrap model selection criterion. (English) Zbl 0843.62050
Summary: We show that a bootstrap model selection criterion constructed by directly plugging-in a consistent estimator in place of the unknown parameter has a downward bias of amount roughly equivalent to the number of parameters in the approximating model. An example is provided to illustrate the points discussed.
MSC:
| 62G09 | Nonparametric statistical resampling methods |
Keywords:
iterated bootstrap; plug-in estimator; Kullback-Leibler discrepancy; parametric bootstrap; bootstrap model selection criterion; biasReferences:
| [1] | Akaike, H., Information theory and an extension of the maximum likelihood principle, (Petrov, B. N.; Csáki, F., Proc. 2nd Internat. Symp. on Information Theory (1973), Akadémia Kiadó: Akadémia Kiadó Budapest), 267-281 · Zbl 0283.62006 |
| [2] | Beran, R.; Ducharme, G., Asymptotic Theory for Bootstrap Methods in Statistics (1991), Centre de Recherches Mathématiques, Univ. of Montreal · Zbl 0733.62050 |
| [3] | Efron, B., Bootstrap methods: Another look at the jackknife, Ann. Statist., 7, 1-26 (1979) · Zbl 0406.62024 |
| [4] | Linhart, H.; Zucchini, W., Model Selection (1986), Wiley: Wiley New York · Zbl 0665.62003 |
| [5] | Sakamoto, Y.; Ishiguro, M.; Kitagawa, G., Akaike Information Criterion Statistics (1986), KTK Scientific Publishers: KTK Scientific Publishers Tokyo · Zbl 0608.62006 |
| [6] | Schall, R.; Zucchini, W., Model selection and the estimation of odds ratios in the presence of extraneous factors, Statist. Med., 9, 1131-1141 (1990) |
| [7] | Shao, J., Linear model selection by cross-validation, J. Amer. Statist. Assoc., 88, 486-494 (1993) · Zbl 0773.62051 |
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.
