A comparison of testing procedures for least absolute deviation regression

by Moawad El-Fallah Abd El -Sallam.

Abstract: In this paper, we compare three procedures for assessing the significance of coefficients in least absolute deviation (LAD) multiple regression. The three test procedures studied are the Wald, likelihood ratio (LR), and Lagrange multiplier (LM) tests. Empirical levels are compared based on Monte Carlo simulation. This paper extends previous studies that concentrated only on simple regression. Further, the performance of overall goodness of fit tests for multiple regression is also examined. The results of the study suggest that the LR and LM tests are preferable to the Wald test for this situation.

Key Words: regression; least absolute deviations; Lagrange multiplier test; Likelihood ratio test; Wald test; simulation

Moawad El-Fallah Abd El -Sallam, moawadothman@hotmail.com

Editor: Sapra, Sunil K., ssapra@exchange.calstatela.edu

READING THE ARTICLE: You can read the article in portable document (.pdf) format (201903 bytes.)

NOTE: The content of this article is the intellectual property of the authors, who retains all rights to future publication.

This page has been accessed 2190 times since MAY 18, 2008.

Return to the InterStat Home Page.