Banca de QUALIFICAÇÃO: KAIO BRENO PEREIRA ALVES

Uma banca de QUALIFICAÇÃO de MESTRADO foi cadastrada pelo programa.
STUDENT : KAIO BRENO PEREIRA ALVES
DATE: 27/11/2020
TIME: 10:00
LOCAL: ambiente virtual
TITLE:

Nonnested Hypothesis Testing Inference in GJS Regression Models


KEY WORDS:

Bootstrap, Nonnested Hypothesis, GJS regression, J test, MJ test


PAGES: 46
BIG AREA: Ciências Exatas e da Terra
AREA: Probabilidade e Estatística
SUMMARY:

In several areas of knowledge, regression as a means for data modeling can lead to more than one model with similar adjustments, but with different specifications. When the model specified in the null hypothesis can be obtained through the model represented in the alternative hypothesis, imposing some restrictions on the parameters, these are said to be unmatched. With regard to the GJS regression models proposed by Lemonte and Bazán (2016), whose support is the interval (0.1), these models may not be fitted, as for example, when the models differ in the regressors. In this case, the methods proposed in order to assess which of the models are correctly specified are suitable only for linear regression models, not being applied in this case. Thus, the present work aims to show an adaptation in the J test presented by Davidson and MacKinnon (1981) - which is the most used to evaluate, in regression, unmatched hypotheses - and in the MJ test, modification of the J test. some scenarios, where the models differ in the connection functions and in the regressors, considering small samples since in large samples the asymptotic approximations are valid. The analysis took into account the zero rejection rates of the tests, comparing their results with their bootstrap versions. According to the results, when we increased the sample size, the rejection rates of the tests tended to the desired values and the size distortions were reduced considerably with the use of a bootstrap scheme.


BANKING MEMBERS:
Presidente - 2310429 - ARTUR JOSE LEMONTE
Interno - 2612836 - FRANCISCO MOISES CANDIDO DE MEDEIROS
Interna - 2312009 - MARIANA CORREIA DE ARAUJO
Notícia cadastrada em: 05/11/2020 10:38
SIGAA | Superintendência de Tecnologia da Informação - (84) 3342 2210 | Copyright © 2006-2024 - UFRN - sigaa13-producao.info.ufrn.br.sigaa13-producao