Regression Model Selection—A Residual Likelihood Approach
- 1 May 2002
- journal article
- Published by Oxford University Press (OUP) in Journal of the Royal Statistical Society Series B: Statistical Methodology
- Vol. 64 (2) , 237-252
- https://doi.org/10.1111/1467-9868.00335
Abstract
Summary. We obtain the residual information criterion RIC, a selection criterion based on the residual log-likelihood, for regression models including classical regression models, Box–Cox transformation models, weighted regression models and regression models with autoregressive moving average errors. We show that RIC is a consistent criterion, and that simulation studies for each of the four models indicate that RIC provides better model order choices than the Akaike information criterion, corrected Akaike information criterion, final prediction error, Cp and Radj2, except when the sample size is small and the signal-to-noise ratio is weak. In this case, none of the criteria performs well. Monte Carlo results also show that RIC is superior to the consistent Bayesian information criterion BIC when the signal-to-noise ratio is not weak, and it is comparable with BIC when the signal-to-noise ratio is weak and the sample size is large.Funding Information
- National Institutes of Health (DA-01-0433)
- National Science Foundation (DMS-95-10511)
This publication has 24 references indexed in Scilit:
- Bias Reduction of Autoregressive Estimates in Time Series Regression Model through Restricted Maximum LikelihoodJournal of the American Statistical Association, 2000
- Smoothing Parameter Selection in Nonparametric Regression Using an Improved Akaike Information CriterionJournal of the Royal Statistical Society Series B: Statistical Methodology, 1998
- Model Selection and InferencePublished by Springer Nature ,1998
- Miscellanea. Linear regression after spline transformationBiometrika, 1997
- Sensitivity Analysis in Linear RegressionWiley Series in Probability and Statistics, 1988
- Transformation and Weighting in RegressionPublished by Springer Nature ,1988
- An Analysis of Transformations RevisitedJournal of the American Statistical Association, 1981
- Restricted Maximum Likelihood (REML) Estimation of Variance Components in the Mixed ModelTechnometrics, 1976
- Bayesian inference for variance components using only error contrastsBiometrika, 1974
- Statistical predictor identificationAnnals of the Institute of Statistical Mathematics, 1970