A note on hypothesis testing in LAV multiple regression: A small sample comparison

被引：2
|
作者：
Dielman, TE
Rose, EL
机构：
[1] TEXAS CHRISTIAN UNIV,MJ NEELEY SCH BUSINESS,FT WORTH,TX 76129
[2] UNIV SO CALIF,SCH BUSINESS ADM,LOS ANGELES,CA 90089
来源：
COMPUTATIONAL STATISTICS & DATA ANALYSIS | 1997年 / 23卷 / 03期
关键词：
L(1)-regression; least absolute value regression; robust regression; least absolute deviations; Lagrange multiplier test; likelihood ratio test; Wald test;
D O I：
10.1016/S0167-9473(97)81021-3
中图分类号：
TP39 [计算机的应用];
学科分类号：
081203 ; 0835 ;
摘要：
We compare three approaches for assessing the significance of coefficients in least absolute value (LAV) multiple regression. The three test procedures studied are the Wald, likelihood ratio (LR), and Lagrange multiplier (LM) tests. Empirical levels of significance and power for the three test procedures 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, and a more extensive examination of the power of the tests is performed. The results of the study suggest that the LR and LM tests are preferable to the Wald test for this situation.
引用
收藏
页码：381 / 388
页数：8
相关论文
共 29 条
123