An extended sequential goodness-of-fit multiple testing method for discrete data

Area of Science:

• Biostatistics
• Statistical methodology
• High-dimensional data analysis

Background:

• The sequential goodness-of-fit (SGoF) method is a recent approach for multiple testing, particularly in high-dimensional settings.
• Traditional SGoF can be overly conservative when dealing with discrete test statistics.
• Existing methods for controlling familywise error rate (FWER) and false discovery rate (FDR) may lack power in high-dimensional scenarios.

Purpose of the Study:

• To introduce a modified SGoF procedure that accounts for the discreteness of test statistics.
• To develop a more powerful multiple testing method for discrete data compared to the original SGoF.
• To achieve false discovery rate (FDR) levels closer to the nominal level while maintaining weak control.

Main Methods:

• Development of an alternative SGoF-type procedure incorporating discrete test statistic adjustments.
• Simulation studies to evaluate the performance and power of the new method.
• Application of the proposed method to a real-world pharmacovigilance dataset.

Main Results:

• The proposed SGoF-type method demonstrates improved power compared to the original SGoF.
• The new procedure achieves false discovery rate (FDR) levels closer to the desired nominal level.
• The method shows effective control of the familywise error rate (FWER) and false discovery rate (FDR) in simulations.

Conclusions:

• The modified SGoF procedure offers a more powerful and accurate alternative for multiple testing with discrete data.
• This method provides a valuable tool for analyzing high-dimensional data where test statistics are discrete.
• The approach is applicable to real-world problems, such as in pharmacovigilance studies.