• JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
 
  Bookmark and Share
 
 
Doctoral Thesis
DOI
https://doi.org/10.11606/T.45.1999.tde-20210729-023307
Document
Author
Full name
Viviana Giampaoli
E-mail
Institute/School/College
Knowledge Area
Date of Defense
Published
São Paulo, 1999
Supervisor
Title in Portuguese
Inferência estatística para modelos lineares com restrições nos parâmetros em condições regulares e não regulares
Keywords in Portuguese
Inferência Estatística
Abstract in Portuguese
Estudamos estatísticas adequadas para testar hipóteses sobre os parâmetros de modelos lineares com restrições no espaço paramétrico. Propomos um teste alternativo para dois casos em que as restrições sobre o espaço paramétrico são impostas peloobjetivo da investigação em condições regulares, (i.e. quando o parâmetro está no interior do espaço paramétrico). Concluímos, através de estudos de simulação, que o teste alternativo é, em geral, mais poderoso que o teste usual. Uma abordagembayesiana utilizando o fator de Bayes para a análise deste tipo de hipótese também foi considerada. Nesse contexto, mostramos que a metodologia proposta por Irony e Pereira (Resenhas IME-USP, (1995) 27-46) para o cálculo do fator de Bayes é maisflexível que proposta por Carlin e Chib (Journal of the Royal Statistical Society B, 57, (1995) 473-484) quanto à escolha das distribuições a priori. Também consideramos testes de hipóteses sob condições não regulares (quando a hipótese coloca oparâmetro na fronteira do espaço paramétrico) para as componentes de variância de um modelo de efeitos aleatórios. Aplicamos os resultados dados por Vu e Zhou (The Annals of Statistics, 25, (1997) 897-916) sobre a estatística da razão deverossimilhanças, identificando casos específicos em que as condições de regularidade são válidas
Title in English
not available
Abstract in English
Suitable statistics for testing of hypothesis on the parameters of linear models with restrictions on the parametric space were studied. An alternative test for two cases in which the restrictions are imposed by the research objective inregularconditions (i.e. when the parameter is in the interior of the parametric space) was proposed. Using simulation studies, we concluded that, in general, that the alternative test is more powerful that the usual test. A Bayesian approach fortheanalyses of this type of hypothesis using the Bayes's factor was also considered. In this context, we showed that the methodology proposed by Irony and Pereira (Resenhas IME-USP, (1995) 27-46) for calculation of Bayes's factor is moreflexiblethat proposed by Carlin and Chib (Journal of the Royal Statistical Society B, 57, (1995) 473-484) with respect to choice of the prior distribution. Tests of hypothesis in nostandard conditions (when the hypothesis sets the parameter intheboundary of the parametric space) for the components of variance in a random model were also proposed. We applied the results of Vu and Zhou (The Annals of Statistics, 25, (1997) 897-916) about the likelihood ratio test to specific casesinwhich the conditions of regularity are valid
 
WARNING - Viewing this document is conditioned on your acceptance of the following terms of use:
This document is only for private use for research and teaching activities. Reproduction for commercial use is forbidden. This rights cover the whole data about this document as well as its contents. Any uses or copies of this document in whole or in part must include the author's name.
GiampaoliViviana.pdf (21.42 Mbytes)
Publishing Date
2021-07-29
 
WARNING: Learn what derived works are clicking here.
All rights of the thesis/dissertation are from the authors
CeTI-SC/STI
Digital Library of Theses and Dissertations of USP. Copyright © 2001-2024. All rights reserved.