False Discovery Variance Reduction in Large Scale Simultaneous Hypothesis Tests

TL;DR

This paper introduces a Bayesian approach to reduce false discovery variance in large-scale multiple hypothesis testing by modeling data dependence as a stationary ergodic process, improving stability despite unknown correlations.

Contribution

It develops a novel Bayesian method that incorporates local data dependence to significantly reduce false discovery variance in large-scale tests.

Findings

Variance of false positive proportion is substantially reduced.

Method remains effective under unknown short-range dependence.

Simulation results validate the approach's robustness.

Abstract

Statistical dependence between hypotheses poses a significant challenge to the stability of large scale multiple hypotheses testing. Ignoring it often results in an unacceptably large spread in the false positive proportion even though the average value is acceptable [21, 39, 40, 49]. However, the statistical dependence structure of data is often unknown. Using a generic signalprocessing model, Bayesian multiple testing, and simulations, we demonstrate that the variance of the false positive proportion can be substantially reduced even under unknown short range dependence. We do this by modeling the data generating process as a stationary ergodic binary signal process embedded in noisy observations. We derive conditional probabilities needed for the Bayesian multiple testing by incorporating nearby observations into a second order Taylor series approximation. Simulations under general conditions are carried out to assess the validity and the variance reduction of the approach. Along the way, we address the problem of sampling a random Markov matrix with specified stationary distribution and lower bounds on the top absolute eigenvalues, which is of interest in its own right.