|
Activity Number:
|
159
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Monday, August 3, 2009 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Nonparametric Statistics
|
| Abstract - #303628 |
|
Title:
|
Quantile Regression: New Research Directions
|
|
Author(s):
|
Mi-Ok Kim*+
|
|
Companies:
|
Cincinnati Children's Hospital Medical Center
|
|
Address:
|
MLC 5041, Cincinnati, OH, 41042,
|
|
Keywords:
|
Censored Quantile Regression ; Empirical Likelihood ; Nonproportional Hazard
|
|
Abstract:
|
Censored quantile regression (QR) was proposed by Portnoy (2003) as an alternative modeling framework for the analysis of time-to-event data where standard approaches are deemed inadequate because their assumptions do not hold and/or unduly constrain covariate effects, thereby precluding many interesting forms of heterogeneity. Although interesting heterogeneous covariate effects can be examined via their interval estimates, a more rigorous and formal inference is desired. We propose empirical likelihood analysis of censored quantile regression where the simultaneous inference of coefficient estimates of more than one quantiles is readily accommodated. The proposed method will be illustrated with a real life example of the Cancer and Leukemia Group B (CALGB) 9082 study.
|
