M-tests for multivariate regression model
Document Type
Article
Publication Date
1-1-2011
Abstract
The M-estimation method is used to define the unrestricted test, restricted test and pre-test test (PTT) for testing the intercept vector of a multivariate simple regression model when it is a priori suspected that the slope vector has some specified values. The asymptotic distribution of the test statistics and the asymptotic power functions of the proposed M-tests are derived. Performances of the M-tests are compared both analytically and graphically. The analytical results as well as an illustrative simulation study to compare the size and power of the M-tests reveal a reasonable dominance of the PTT over the other two tests. © American Statistical Association and Taylor & Francis 2011.
Keywords
pre-test test, non-sample prior information, asymptotic distribution, asymptotic power, M-estimation, bivariate non-central chi-square distribution
Divisions
MathematicalSciences
Publication Title
Journal of Nonparametric Statistics
Volume
23
Issue
1
Publisher
Taylor and Francis