|
Activity Number:
|
419
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 9, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Nonparametric Statistics
|
| Abstract - #307589 |
|
Title:
|
Empirical Likelihood Method for Heteroscedastic Linear Model
|
|
Author(s):
|
Hua Zhu*+ and Mi-Ok Kim and Mai Zhou
|
|
Companies:
|
University of Kentucky and University of Kentucky and University of Kentucky
|
|
Address:
|
4070 Victoria Way, Apt. 84, Lexington, KY, 40515,
|
|
Keywords:
|
Wilks theorem ; empirical likelihood ; variance modeling ; right censored data
|
|
Abstract:
|
For a heteroscedastic regression model, variance modeling has been considered to improve the efficiency of mean parameter estimates or to estimate variance if it is interesting in its own right. Among others, Owen (2001) proposed the empirical likelihood (EL) method. We propose a new estimating equation and consider the EL method for the conditional means. We show that properly defined EL ratio statistic has an asymptotic chi-square distribution under null hypothesis. We also examine the possibility to extend the EL method to variance modeling in censored case with the proposed estimating equation.
|
