Logo image
Nonparametric and Semiparametric Regression for Independent Data
Book chapter

Nonparametric and Semiparametric Regression for Independent Data

Hua Liang
The Work of Raymond J. Carroll, pp 293-370
13 May 2014

Abstract

Independent Data Least Squares Semiparametric Regression Unknown Parameter Vector Usual Confidence Interval
Consider the linear model yi=xiTβ+σi𝜀i,i=1,⋯,n,\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document} $$\displaystyle{y_{i} = \mathbf{x}_{i}^{T}\beta +\sigma _{ i}\varepsilon _{i},i = 1,\cdots \,,n,}$$ \end{document} where β is an unknown parameter vector and the {𝜀i}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document} $$\{\varepsilon _{i}\}$$ \end{document} are i.i.d. errors. It is well known that ordinary least squares (LS) estimators are unbiased and consistent, but are not efficient when errors are heteroscedastic, and the usual standard error estimators of LS estimators are biased. Hence the usual confidence intervals and test statistics are biased and may lead to incorrect conclusions.

Metrics

13 Record Views

Details

Logo image