Relatório de pesquisa 25/08
On Estimation and Local Influence Analysis for Measurement Errors Models under Heavy-tailed Distributions, V. H. Lachos, T. Angolini, and C. A. Abanto-Valle, submitted Nov. 28.
Abstract
Scale mixtures of normal distribution is a class of
symmetric thick--tailed distributions that includes the normal one as a
special case. In this paper we consider local influence analysis for
measurement error model (MEM) when the random error and the unobserved
value of the covariates follows jointly a scale mixtures of normal
distribution, providing an appealing robust alternative to the usual
Gaussian process in measurement error models. As the observed data
log-likelihood associated with this model is intractable, Cook's
well--known approach may be hard to be applied to obtain measures of
local influence. Instead we develop local influence measures following
the approach of Zhu and Lee (2001), which is based on the use of an
EM-type algorithm. Three specific perturbation schemes are discussed.
Results obtained from a real data set are reported, illustrating the
usefulness of the proposed methodology.
Mathematics Subject Classifications
(2000):
Keywords: EM algorithm; Scale mixtures of normal distribution; Mahalanobis distance; Measurement error models.
Copy
of the file:
rp25-08.pdf
(PDF)
rp25-08.pdf.gz (gzipped PDF)
November 28, 2008
Volta ao
indíce de Relatórios de Pesquisa