On Effect of Nonnormality on Size of Test for Dimensionality in Discriminant Analysis

  • Changha Hwang (Assistant Professor, Dept. of Statistics, Catholic University of Taegu-Hyosung, Kyungbuk, 713-702, Korea)
  • Published : 1996.12.01


In discriminant analysis the procedures commonly used to estimate the dimensionality involve testing a sequence of dimensionality hypotheses. There is a problem with the size of the test since dimensionality hypotheses are tested sequentially and thus they are actually conditional tests. The focus of this paper is to investigate in asymptotic sense what happens to the sequential testing procedure if the assumption of normality does not hold.



  1. J. Statist. Appl. Prob. v.2 Moments of a random vector and its quadratic forms Gang, L.
  2. Reprinted in Statistical Inference in Elliptically Contoured and Related Distribution Gang, L.;Fang, K.(ed.);Anderson, T.W.(ed.)
  3. Ph. D. Thesis, Univ. of Michigan Model Selection Methods in Discriminant Analysis Hwang, C.
  4. Communication in Statistics:Theory ana Methods v.23 On estimating the dimensionality in discriminant analysis Hwang, C.
  5. Journal of the Korean Statistical Society v.23 Characterization of the asymptotic distributions of certain eigenvalues in a general setting Hwang, C.
  6. The Korean Communications in Statistics v.2 A note on the asymptotic distributions of dimensionality Estimators in discriminant analysis Hwang, C.
  7. Journal of Statistical Theory and Methods v.6 Size of test for dimensionality in discriminant analysis Hwang, C.
  8. Technical Report No. 93-9 The effects of nonormality on tests for dimensionality in canonical correlation and MANOVA models Seo, T.;Kanda, T.;Fujikoshi, Y.
  9. Modern Multivariate Statistical Analysis: A Graduate Course and Handbook Siotani, M.;Hayakawa, T.;Fujikoshi, Y.