OUTLIERS IN MULTIVARIATE REGRESSION-MODELS

Authors
SRIVASTAVA MS VONROSEN D
Citation
Ms. Srivastava et D. Vonrosen, OUTLIERS IN MULTIVARIATE REGRESSION-MODELS, Journal of Multivariate Analysis, 65(2), 1998, pp. 195-208
Citations number
31
Categorie Soggetti
Statistic & Probability","Statistic & Probability
Journal title
Journal of Multivariate Analysis → ACNP
ISSN journal
0047259X
Volume
65
Issue
2
Year of publication
1998
Pages
195 - 208
Database
ISI
SICI code
0047-259X(1998)65:2<195:OIMR>2.0.ZU;2-2
Abstract
Likelihood ratio tests for detecting a single outlier in multivariate linear models are considered, where an observation is called an outlie r if there has been a shift in the mean. The test statistics are the m aximum of n nonindependent statistics, where it is the number of obser vations. Relevant distributions to use upper and lower Bonferroni's in equalities are given. (C) 1998 Academic Press.