|
主讲人 |
Jun Shao |
简介 |
<p><span style="font-size: small;"><span style="font-family: Arial;">Estimation with longitudinal data having nonignorable dropout is considered when the joint distribution of the study variable and covariate is nonparametric and the dropout propensity follows a parametric model. To deal with the identifiability problem caused by nonignorable dropout, we use an instrument which is a covariate unrelated to the dropout propensity. We apply the generalized method of moments to estimate the parameters in the dropout propensity model based on some estimating equations constructed using the instrument. Population means and other parameters in the nonparametric distribution of the study variable can be estimated based on inverse propensity weighting with weights constructed after addressing the nonignorable dropout. To improve efficiency, we derive a model-assisted regression estimator making use of extra information provided by the covariates and previously observed study variables in the longitudinal setting. The model-assisted regression estimator is protected from model misspecification, and is shown to be asymptotically normal and more efficient than the estimator based on inverse propensity weighting when the working models are correct or nearly correct and some other conditions are satisfied. The finite-sample performance of the estimators is studied through simulation, and an application to the HIV-CD4 data set is also presented as illustration.</span></span></p> |
