Two-step Estimation for Longitudinal Data When the Working Correlation Matrix is a Linear Combination of Some Known Matrices

被引：0

作者：

Yu-ling LI ^{[1
,2
]}

Wei GAO ^{[1
]}

Man-Lai TANG ^{[3
]}

Shu-rong ZHENG ^{[1
]}

机构：

[1] Key Laboratory for Applied Statistics of MOE,School of Mathematics and Statistics,Northeast Normal University

[2] School of Applied Mathematics,Beijing Normal University

[3] Department of Mathematics and Statistics,Hang Seng Management College

来源：

Acta Mathematicae Applicatae Sinica | 2019年 / 35卷 / 02期

基金：

中国国家自然科学基金;

关键词：

generalized estimating equations; longitudinal data; quadratic inference functions; quasi-likelihood; two-step estimation;

D O I：

暂无

中图分类号：

O151.21 [矩阵论];

学科分类号：

070104 ;

摘要：

The generalized estimating equations(GEE) approach is perhaps one of the most widely used methods for longitudinal data analysis. While the GEE method guarantees the consistency of its estimators under working correlation structure misspecification, the corresponding efficiency can be severely affected. In this paper, we propose a new two-step estimation method in which the correlation matrix is assumed to be a linear combination of some known working matrices. Asymptotic properties of the new estimators are developed.Simulation studies are conducted to examine the performance of the proposed estimators. We illustrate the methodology with an epileptic data set.

引用

页码：264 / 273

页数：10

共 6 条

[1] Informative Estimation and Selection of Correlation Structure for Longitudinal Data
Zhou, Jianhui
Qu, Annie
[J]. JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 2012, 107 (498) : 701 - 710
[2] Semiparametric Mean–Covariance Regression Analysis for Longitudinal Data [J] . Chenlei Leng,Weiping Zhang,Jianxin Pan.&nbsp&nbspJournal of the American Statistical Association . 2010 (489)
[3] Semiparametric Estimation of Covariance Matrixes for Longitudinal Data
Fan, Jianqing
Wu, Yichao
[J]. JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION, 2008, 103 (484) : 1520 - 1533
[4] Working correlation structure misspecification, estimation and covariate design: Implications for generalised estimating equations performance [J] . You&hyphen,Gan Wang,Vincent Carey.&nbsp&nbspBiometrika . 2003
[5] Improving generalised estimating equations using quadratic inference functions [J] . Annie Qu.&nbsp&nbspBiometrika . 2000
[6] QUASI-LIKELIHOOD FUNCTIONS
MCCULLAGH, P
[J]. ANNALS OF STATISTICS, 1983, 11 (01) : 59 - 67

← 1 →