Abstract
The asymptotic error probability of Linhart's model selection test isevaluated, and compared with the nominal significance level. We examine thecase where the expected discrepancies of the candidate models from the truemodel are asymptotically equal. The local alternatives method is employed inthe limiting operation of the asymptotic evaluation. Although the errorprobability under the null hypothesis is actually shown to be equal to orless than the level for most situations, intolerable violations of the errorcontrol are observed for nested models: It is often erroneously concludedthat the smaller model is significantly better than the larger model. Toprevent this violation, a modification of Linhart's test statistic isproposed. The effectiveness of the proposed test is confirmed throughtheoretical analysis and numerical simulations.
Similar content being viewed by others
References
Cox, D. R. (1962). Further results on tests of separate families of hypotheses, J. Roy. Statist. Soc. Ser. B, 24, 406–424.
Hotelling, H. (1936). Relations between two sets of variables, Biometrika, 28, 321–377.
Linhart, H. (1988). A test whether two AIC's differ significantly, South African Statist. J., 22, 153–161.
Shimodaira, H. (1995). A Study on Statistical Model Selection—Construction of the Confidence Set of Models, Ph.D. dissertation, Department of mathematical engineering and information physics, University of Tokyo (in Japanese).
Shimodaira, H. (1996). An application of multiple comparison techniques to model selection, Ann. Inst. Statist. Math. (to appear).
Steiger, J. H., Shapiro, A. and Browne, M. W. (1985). On the multivariate asymptotic distribution of sequential chi-square statistics, Psychometrika, 50, 253–264.
Author information
Authors and Affiliations
About this article
Cite this article
Shimodaira, H. Assessing the Error Probability of the Model Selection Test. Annals of the Institute of Statistical Mathematics 49, 395–410 (1997). https://doi.org/10.1023/A:1003140609666
Issue Date:
DOI: https://doi.org/10.1023/A:1003140609666