Adapting for Heteroscedasticity in Regression Models.
Abstract
This document investigates the limiting behavior of a class of one-step M-estimators in heteroscedastic regression models. The mean function is assumed to be known up to parameters, but the variance function is considered an unknown function of a dimensional vector. The variance function is to be estimated nonparametrically by a function of the absolute residuals from the current fit to the mean. Under a variety of conditions when the estimates adapt for scale, i.e., the regression parameter is estimated just as well as if the scale function was known. Connections with the theory of optimal semiparametric estimation are made. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1986
- Accession Number
- ADA174584
Entities
People
- David Ruppert
- Leonard A. Stefanski
- Raymond J. Carroll
Organizations
- University of North Carolina at Chapel Hill