Multiple testing via relative belief ratios

TL;DR

This paper introduces a method for multiple testing using relative belief ratios, providing a measure of evidence that is advantageous for assessing sparsity without restrictive priors.

Contribution

The paper proposes a novel multiple testing approach based on relative belief ratios, offering a flexible and evidence-based method for assessing sparsity in large-scale inference.

Findings

Effective in large-scale inference problems

Does not require restrictive prior assumptions

Provides a measure of statistical evidence

Abstract

Some large scale inference problems are considered based on using the relative belief ratio as a measure of statistical evidence. This approach is applied to the multiple testing problem. A particular application of this is concerned with assessing sparsity. The approach taken to sparsity has several advantages as it is based on a measure of evidence and does not require that the prior be restricted in any way.