A study of bias in the naive estimator in longitudinal linear mixed-effects models with measurement error and misclassification in covariates

摘要

This research presents a generalized least square approach to estimate the pa-udrameters in a longitudinal linear mixed-effects model. In this model, we considerudmeasurement error and misclassification in the covariates. Moreover, a classical mea-udsurement error for continuous covariates, and misclassification for discrete covariatesudup to three categories, is considered. Through simulation studies, we observe theudimpact of each parameter of the model on the bias of the naive estimation, when theudother parameters stay unchanged.