Academic Journals Database
Disseminating quality controlled scientific knowledge

Wald-Type Statistics in Heteroscedastic Linear Models

ADD TO MY LIST
 
Author(s): Esra Akdeniz | Michael G. Akritas

Journal: Sel├žuk Journal of Applied Mathematics
ISSN 1302-7980

Volume: 7;
Issue: 2;
Start page: 91;
Date: 2006;
Original page

Keywords: ANOVA (Analysis of Variance) design | Heteroscedasticity | Linear Models | Wald Statistic.

ABSTRACT
In this paper we consider the heteroscedastic linear models and consider Wald-type test statistics for the hypothesis of no factor effects and no interactions. Designs with independent data are considered. The statistics based on the original observations appear to be new. They test the usual parametric null hypotheses and provide useful alternatives to the classical F-test especially in the presence of variance heterogeneity. The statistics based on ranks have been proposed in Akritas and Arnold (1994), Akritas, Arnold and Brunner (1997) and in Akritas and Brunner (1997) for the purpose of testing certain nonparametric hypotheses. Simulation programs in MATLAB are developed and the results are presented.
RPA Switzerland

RPA Switzerland

Robotic process automation

    

Tango Rapperswil
Tango Rapperswil