On varieties of doubly robust estimators under missingness not at random with a shadow variable

Area of Science:

• Statistics
• Biostatistics
• Epidemiology

Background:

• Missing data is a common challenge in statistical analysis.
• Missing not at random (MNAR) data poses significant estimation difficulties.
• Existing methods may not fully address MNAR scenarios.

Purpose of the Study:

• To develop novel statistical estimators for outcome means with MNAR data.
• To leverage a fully observed shadow variable for improved identification.
• To provide methods with distinct properties and validation techniques.

Main Methods:

• Proposed two alternative semiparametric estimators.
• Utilized a shadow variable, associated with the outcome but independent of missingness.
• Employed goodness-of-fit tests to assess working model validity.

Main Results:

• The new estimators extend methods for data missing at random.
• These estimators exhibit different, potentially advantageous, properties.
• Goodness-of-fit tests provide straightforward model correctness assessment.

Conclusions:

• The proposed methods offer viable alternatives for handling MNAR data.
• The use of shadow variables is effective for identifying outcome means.
• Validation through goodness-of-fit tests enhances the reliability of the estimators.