Robust testing in semiparametric partly linear regression models

BIANCO, ANA; BOENTE, GRACIELA; MARTINEZ, ELENA

Amberes, Belgica

Congreso; International Conference on Robust Statistics (ICORS 2003); 2003

In some situations, we are interested in finding out the impact of a covariable on the response variable y under a partial linear model. That is, we would like to make inference on the slope parameter or on some of its components. In this talk, we will focus on the problem of testing on the partly linear model the following parametric hypothesis i) H0 : beta= beta_o, ii) H01 : beta_1 = beta_{1,o}, beta_2 unspecified and we will propose a class of tests based on the three step robust procedure described above. The asymptotic behavior of the test statistics, under the null hypothesis and contiguous alternatives, is obtained. Through a Monte Carlo study we illustrate the performance of the proposed tests. and we will propose a class of tests based on the three step robust procedure described above. The asymptotic behavior of the test statistics, under the null hypothesis and contiguous alternatives, is obtained. Through a Monte Carlo study we illustrate the performance of the proposed tests. ii) H01 : beta_1 = beta_{1,o}, beta_2 unspecified and we will propose a class of tests based on the three step robust procedure described above. The asymptotic behavior of the test statistics, under the null hypothesis and contiguous alternatives, is obtained. Through a Monte Carlo study we illustrate the performance of the proposed tests. and we will propose a class of tests based on the three step robust procedure described above. The asymptotic behavior of the test statistics, under the null hypothesis and contiguous alternatives, is obtained. Through a Monte Carlo study we illustrate the performance of the proposed tests. i) H0 : beta= beta_o, ii) H01 : beta_1 = beta_{1,o}, beta_2 unspecified and we will propose a class of tests based on the three step robust procedure described above. The asymptotic behavior of the test statistics, under the null hypothesis and contiguous alternatives, is obtained. Through a Monte Carlo study we illustrate the performance of the proposed tests. and we will propose a class of tests based on the three step robust procedure described above. The asymptotic behavior of the test statistics, under the null hypothesis and contiguous alternatives, is obtained. Through a Monte Carlo study we illustrate the performance of the proposed tests. ii) H01 : beta_1 = beta_{1,o}, beta_2 unspecified and we will propose a class of tests based on the three step robust procedure described above. The asymptotic behavior of the test statistics, under the null hypothesis and contiguous alternatives, is obtained. Through a Monte Carlo study we illustrate the performance of the proposed tests. and we will propose a class of tests based on the three step robust procedure described above. The asymptotic behavior of the test statistics, under the null hypothesis and contiguous alternatives, is obtained. Through a Monte Carlo study we illustrate the performance of the proposed tests. y under a partial linear model. That is, we would like to make inference on the slope parameter or on some of its components. In this talk, we will focus on the problem of testing on the partly linear model the following parametric hypothesis i) H0 : beta= beta_o, ii) H01 : beta_1 = beta_{1,o}, beta_2 unspecified and we will propose a class of tests based on the three step robust procedure described above. The asymptotic behavior of the test statistics, under the null hypothesis and contiguous alternatives, is obtained. Through a Monte Carlo study we illustrate the performance of the proposed tests. and we will propose a class of tests based on the three step robust procedure described above. The asymptotic behavior of the test statistics, under the null hypothesis and contiguous alternatives, is obtained. Through a Monte Carlo study we illustrate the performance of the proposed tests. ii) H01 : beta_1 = beta_{1,o}, beta_2 unspecified and we will propose a class of tests based on the three step robust procedure described above. The asymptotic behavior of the test statistics, under the null hypothesis and contiguous alternatives, is obtained. Through a Monte Carlo study we illustrate the performance of the proposed tests. and we will propose a class of tests based on the three step robust procedure described above. The asymptotic behavior of the test statistics, under the null hypothesis and contiguous alternatives, is obtained. Through a Monte Carlo study we illustrate the performance of the proposed tests. H0 : beta= beta_o, ii) H01 : beta_1 = beta_{1,o}, beta_2 unspecified and we will propose a class of tests based on the three step robust procedure described above. The asymptotic behavior of the test statistics, under the null hypothesis and contiguous alternatives, is obtained. Through a Monte Carlo study we illustrate the performance of the proposed tests. and we will propose a class of tests based on the three step robust procedure described above. The asymptotic behavior of the test statistics, under the null hypothesis and contiguous alternatives, is obtained. Through a Monte Carlo study we illustrate the performance of the proposed tests. H01 : beta_1 = beta_{1,o}, beta_2 unspecified and we will propose a class of tests based on the three step robust procedure described above. The asymptotic behavior of the test statistics, under the null hypothesis and contiguous alternatives, is obtained. Through a Monte Carlo study we illustrate the performance of the proposed tests.