figshare
Browse
- No file added yet -

Robust statistical inference for longitudinal data with nonignorable dropouts

Download (316.07 kB)
journal contribution
posted on 2022-08-12, 06:00 authored by Yujing Shao, Wei Ma, Lei Wang

In this paper, we propose robust statistical inference and variable selection method for generalized linear models that accommodate the outliers, nonignorable dropouts and within-subject correlations. The purpose of our study is threefold. First, we construct the robust and bias-corrected generalized estimating equations (GEEs) by combining the Mallows-type weights, Huber's score function and inverse probability weighting approaches to against the influence of outliers and account for nonignorable dropouts. Subsequently, the generalized method of moments is utilized to estimate the parameters in the nonignorable dropout propensity based on sufficient instrumental estimating equations. Second, in order to incorporate the within-subject correlations under an informative working correlation structure, we borrow the idea of quadratic inference function and hybrid-GEE to obtain the improved empirical likelihood procedures. The asymptotic properties of the proposed estimators and their confidence regions are derived. Third, the robust variable selection and algorithm are investigated. We evaluate the performance of proposed estimators through simulation and illustrate our method in an application to HIV-CD4 data.

Funding

This work was supported by the National Natural Science Foundation of China [Grant Nos. 11871287, 11771144, 11801359], Fundamental Research Funds for the Central Universities [Grant No. ZB22000102] and the Key Laboratory for Medical Data Analysis and Statistical Research of Tianjin.

History