Partial F-tests with multiply imputed data in the linear regression framework via coefficient of determination

Area of Science:

• Statistics
• Biostatistics
• Data Analysis

Background:

• F-tests for regression coefficients are standard in applied research.
• Existing methods for hypothesis testing with multiply imputed data are computationally intensive.
• Simultaneous hypothesis testing with multiply imputed data presents computational challenges.

Purpose of the Study:

• To propose a computationally simple method for F-tests with multiply imputed data.
• To enable global, local, and partial F-tests using a scalar measure.
• To address the computational burden of existing methods for simultaneous hypothesis testing.

Main Methods:

• Developed a novel method utilizing the coefficient of determination.
• Applied the method to perform global, local, and partial F-tests.
• Evaluated the method's performance using simulated data.

Main Results:

• The proposed method offers a computationally efficient alternative for F-tests.
• The coefficient of determination-based approach simplifies hypothesis testing with multiply imputed data.
• The method was successfully applied to real-world suicide prevention data.

Conclusions:

• The new method provides a practical and efficient approach for F-tests in multiple imputation settings.
• This simplifies complex statistical analyses in various research fields.
• The approach is validated through simulation and real-data application.