Open Internet by MindsNet

Failure of Benjamini–Hochberg Procedure in Correlated Gaussian Tests

The standard Benjamini-Hochberg procedure can fail to control the false discovery rate for two-sided tests when the underlying test statistics follow a correlated multivariate Gaussian distribution. This limitation poses a significant challenge in statistical analysis, particularly in fields where test statistics are often correlated. The failure to control the false discovery rate can lead to incorrect conclusions and decisions. This challenge highlights the need for more robust procedures that can handle correlated test statistics.

Mathematics & logic, Mathematics, Statistics & Probability