|
Activity Number:
|
110
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 30, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Nonparametric Statistics
|
| Abstract - #310155 |
|
Title:
|
Robust and Powerful Two-Sample Semiparametric Tests
|
|
Author(s):
|
Zhong Guan*+ and Hongyu Zhao
|
|
Companies:
|
Indiana University South Bend and Yale University
|
|
Address:
|
Dept of Mathematical Sciences, South Bend, IN, 46634,
|
|
Keywords:
|
Empirical likelihood ; Two-sample test ; Semiparametric test ; Behrens-Fisher problem ; Bootstrap ; Power
|
|
Abstract:
|
A semiparametric model is constructed for testing the equality of parameters of two populations, such as means or medians. The semiparametric likelihood ratio based tests are studied and compared with some parametric and nonparametric ones. The numerical comparison shows that the semiparametric approach is more robust than the parametric tests and more powerful than the nonparametric tests in the sense that it has similar powers to some optimum parametric tests for some distributions, it is more powerful than two-sample $t$ and Wilcoxon tests when the parameters are not location parameters. The validity of the bootstrap procedure for estimating $p$-values is also proved.
|
