FDR Upper Bound Results in Common-Factor Models

lihua_lei_stat · x · 2026-07-17

The author continues to explain their results in a thread: for a common class of common-factor Gaussian models, they proved a matching-order upper bound, inspired by a counterexample from @EdgarDobriban.

The core conclusion is that in such models, the scale of worst-case FDR inflation is accurately characterized by a lower bound. For a smaller nominal level q, the worst-case FDR is at least

Therefore, as q approaches 0, FDR/q grows unbounded, demonstrating that no constant upper bound universally applicable across nominal levels, number of hypotheses, means, and correlation matrices exists.

Related event: Lihua Lei sharpens worst-case FDR bounds in common-factor Gaussian models(5 posts)→

Original post →

More from Research

Research channel →