Robust estimation in linear regression models for longitudinal data with covariate measurement errors and outliers (Q1795590)

In this paper, a robust estimating equation method to estimate the linear regression model for longitudinal data with covariate measurement errors and outliers is proposed. First, the independence between replicate measurements is used to reduce the biases caused by measurement errors. Then centralization of the observed covariate matrix is used to correct the biases caused by outliers. The method does not require specifying the distributions of the true covariates, response and measurement errors. Standard generalized estimating equations algorithms are used to implement it. Asymptotic normality of the proposed estimator under regularity conditions is established in the paper, which also includes simulation studies showing that the proposed method performs better in handling measurement errors and outliers than other existing methods. An application of the proposed method to a real data set from the Lifestyle Education for Activity and Nutrition (LEAN) study is also presented.