A unified framework of multiply robust estimation approaches for handling incomplete data

Area of Science:

• Statistics
• Data Science

Background:

• Missing data are prevalent in practical applications.
• Existing methods like inverse probability weighting and imputation rely on specific model assumptions for validity.

Purpose of the Study:

• To propose a novel, general framework for multiply robust estimation procedures.
• To address limitations of current missing data handling techniques by combining multiple models.

Main Methods:

• Development of a general framework integrating multiple nonresponse and imputation models.
• Application to estimate smooth and non-smooth parameters, including population means, quantiles, and distribution functions.
• Establishment of asymptotic theoretical results for the proposed methods.

Main Results:

• The proposed framework offers multiply robust estimation for various statistical parameters.
• Simulation studies and real data application demonstrate good performance.
• The methods show improvements in terms of bias and efficiency compared to existing approaches.

Conclusions:

• The novel framework provides a flexible and robust approach to handling missing data.
• The methods are applicable to a wide range of statistical estimation problems.
• The findings suggest practical utility in improving the accuracy of analyses with incomplete datasets.