Unified approach for regression models with nonmonotone missing at random data

Weighted generalized estimating equations and unified estimation for longitudinal data with nonmonotone missing data patterns

“…However, it is inefficient for nonmonotone missing data patterns because it can only use the information from the subjects with R X it = 1, which include the complete cases and the subjects with fully observed covariates. In this section we will introduce a new method, called unified approach, 18,19 which can use the observations with partially observed covariates to improve the estimation efficiency of the WGEE estimator.…”

“…A unified approach 18 to improve the efficiency of the WGEE estimator includes the following four steps.…”

“…Recently, a unified approach was developed to improve the estimation efficiency of the (weighted) complete case estimates for generalized linear models with MAR data. 18 In the unified approach, one first defines a set of parametric working models to deal with nonmonotone missing data patterns, then estimates each working model using standard complete data methods. Finally, a statistic is defined to combine the results of the complete case estimates from the working models and use it to improve the estimation efficiency of the complete case estimates in the regression model of interest.…”

“…The unified approach produces a consistent estimator which can be as efficient as other alternative estimators. 18,19 Furthermore, it is computationally simple and can be directly implemented in standard software.…”

Unified estimation for Cox regression model with nonmonotone missing at random covariates

“…

This article investigates a unified estimator for Cox regression model (Cox, 1972) when covariate data are missing at random (Rubin, 1976). It extends the idea of using parametric working models (Zhao and Liu, 2021) to extract the partial information contained in the incomplete observations. The working models are flexible and convenient to deal with nonmonotone missing data patterns.It can also incorporate auxiliary variables into the analysis to reduce estimation bias and improve efficiency.

…”