Estimation in covariate-adjusted regression

Damla Şentürk, Danh V. Nguyen

Research output: Contribution to journalArticle

11 Citations (Scopus)

Abstract

The method of covariate-adjusted regression was recently proposed for situations where both predictors and response in a regression model are not directly observed, but are observed after being contaminated by unknown functions of a common observable confounder in a multiplicative fashion. One example is data collected for a study on diabetes, where the variables of interest, systolic and diastolic blood pressures and glycosolated hemoglobin levels are known to be influenced by an observable confounder, body mass index. An estimation procedure based on equidistant binning (EB), currently available, gives consistent estimators for the regression coefficients adjusted for the confounder. In this paper, we propose two new estimation procedures based on nearest-neighbor binning (NB) and local polynomial modeling (LP). Even though, the three methods perform similarly in terms of their bias, it is shown through simulation studies that NB has smaller variance compared to EB, and LP yields substantially lower variance relative to the two binning methods for small to moderate sample sizes. The consistency and convergence rates of the proposed estimators of LP, with the smallest MSE, are also established. We illustrate the proposed method of LP with the above-mentioned diabetes data, where the goal is to uncover the regression relation between the response, glycosolated hemoglobin levels, and the predictors, systolic and diastolic blood pressures, adjusted for body mass index.

Original languageEnglish (US)
Pages (from-to)3294-3310
Number of pages17
JournalComputational Statistics and Data Analysis
Volume50
Issue number11
DOIs
StatePublished - Jul 20 2006

Fingerprint

Binning
Local Polynomial
Covariates
Regression
Polynomials
Hemoglobin
Blood pressure
Medical problems
Diabetes
Equidistant
Blood Pressure
Modeling
Predictors
Consistent Estimator
Regression Coefficient
Nearest Neighbor
Multiplicative
Regression Model
Sample Size
Rate of Convergence

Keywords

  • Equidistant binning
  • Local polynomial regression
  • Multiplicative effects
  • Nearest-neighbor binning
  • Smoothing
  • Varying-coefficient models

ASJC Scopus subject areas

  • Computational Theory and Mathematics
  • Statistics, Probability and Uncertainty
  • Electrical and Electronic Engineering
  • Computational Mathematics
  • Numerical Analysis
  • Statistics and Probability

Cite this

Estimation in covariate-adjusted regression. / Şentürk, Damla; Nguyen, Danh V.

In: Computational Statistics and Data Analysis, Vol. 50, No. 11, 20.07.2006, p. 3294-3310.

Research output: Contribution to journalArticle

Şentürk, Damla ; Nguyen, Danh V. / Estimation in covariate-adjusted regression. In: Computational Statistics and Data Analysis. 2006 ; Vol. 50, No. 11. pp. 3294-3310.
@article{9251040ee8d64d9c9bd975a0ae1656d7,
title = "Estimation in covariate-adjusted regression",
abstract = "The method of covariate-adjusted regression was recently proposed for situations where both predictors and response in a regression model are not directly observed, but are observed after being contaminated by unknown functions of a common observable confounder in a multiplicative fashion. One example is data collected for a study on diabetes, where the variables of interest, systolic and diastolic blood pressures and glycosolated hemoglobin levels are known to be influenced by an observable confounder, body mass index. An estimation procedure based on equidistant binning (EB), currently available, gives consistent estimators for the regression coefficients adjusted for the confounder. In this paper, we propose two new estimation procedures based on nearest-neighbor binning (NB) and local polynomial modeling (LP). Even though, the three methods perform similarly in terms of their bias, it is shown through simulation studies that NB has smaller variance compared to EB, and LP yields substantially lower variance relative to the two binning methods for small to moderate sample sizes. The consistency and convergence rates of the proposed estimators of LP, with the smallest MSE, are also established. We illustrate the proposed method of LP with the above-mentioned diabetes data, where the goal is to uncover the regression relation between the response, glycosolated hemoglobin levels, and the predictors, systolic and diastolic blood pressures, adjusted for body mass index.",
keywords = "Equidistant binning, Local polynomial regression, Multiplicative effects, Nearest-neighbor binning, Smoothing, Varying-coefficient models",
author = "Damla Şent{\"u}rk and Nguyen, {Danh V.}",
year = "2006",
month = "7",
day = "20",
doi = "10.1016/j.csda.2005.06.001",
language = "English (US)",
volume = "50",
pages = "3294--3310",
journal = "Computational Statistics and Data Analysis",
issn = "0167-9473",
publisher = "Elsevier",
number = "11",

}

TY - JOUR

T1 - Estimation in covariate-adjusted regression

AU - Şentürk, Damla

AU - Nguyen, Danh V.

PY - 2006/7/20

Y1 - 2006/7/20

N2 - The method of covariate-adjusted regression was recently proposed for situations where both predictors and response in a regression model are not directly observed, but are observed after being contaminated by unknown functions of a common observable confounder in a multiplicative fashion. One example is data collected for a study on diabetes, where the variables of interest, systolic and diastolic blood pressures and glycosolated hemoglobin levels are known to be influenced by an observable confounder, body mass index. An estimation procedure based on equidistant binning (EB), currently available, gives consistent estimators for the regression coefficients adjusted for the confounder. In this paper, we propose two new estimation procedures based on nearest-neighbor binning (NB) and local polynomial modeling (LP). Even though, the three methods perform similarly in terms of their bias, it is shown through simulation studies that NB has smaller variance compared to EB, and LP yields substantially lower variance relative to the two binning methods for small to moderate sample sizes. The consistency and convergence rates of the proposed estimators of LP, with the smallest MSE, are also established. We illustrate the proposed method of LP with the above-mentioned diabetes data, where the goal is to uncover the regression relation between the response, glycosolated hemoglobin levels, and the predictors, systolic and diastolic blood pressures, adjusted for body mass index.

AB - The method of covariate-adjusted regression was recently proposed for situations where both predictors and response in a regression model are not directly observed, but are observed after being contaminated by unknown functions of a common observable confounder in a multiplicative fashion. One example is data collected for a study on diabetes, where the variables of interest, systolic and diastolic blood pressures and glycosolated hemoglobin levels are known to be influenced by an observable confounder, body mass index. An estimation procedure based on equidistant binning (EB), currently available, gives consistent estimators for the regression coefficients adjusted for the confounder. In this paper, we propose two new estimation procedures based on nearest-neighbor binning (NB) and local polynomial modeling (LP). Even though, the three methods perform similarly in terms of their bias, it is shown through simulation studies that NB has smaller variance compared to EB, and LP yields substantially lower variance relative to the two binning methods for small to moderate sample sizes. The consistency and convergence rates of the proposed estimators of LP, with the smallest MSE, are also established. We illustrate the proposed method of LP with the above-mentioned diabetes data, where the goal is to uncover the regression relation between the response, glycosolated hemoglobin levels, and the predictors, systolic and diastolic blood pressures, adjusted for body mass index.

KW - Equidistant binning

KW - Local polynomial regression

KW - Multiplicative effects

KW - Nearest-neighbor binning

KW - Smoothing

KW - Varying-coefficient models

UR - http://www.scopus.com/inward/record.url?scp=33646583548&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33646583548&partnerID=8YFLogxK

U2 - 10.1016/j.csda.2005.06.001

DO - 10.1016/j.csda.2005.06.001

M3 - Article

VL - 50

SP - 3294

EP - 3310

JO - Computational Statistics and Data Analysis

JF - Computational Statistics and Data Analysis

SN - 0167-9473

IS - 11

ER -