Backend
A Local False Discovery Rate Framework for Compound p-values
arXiv stat.TH0 views
arXiv:2611.24555v1 Announce Type: cross Abstract: Compound p-values relax individual p-value validity to an average condition across the null hypotheses. This flexibility, however, need not protect particular rejections, especially when paired with an average error criterion such as the false discovery rate (FDR). In this work, we introduce "density-compound p-values," a concept that connects compound p-values and the frequentist local FDR paradigm. Under independence, we establish boundary FDR (bFDR) guarantees for the support-line (SL) procedure, with nominal-level control for weighted and compound-e-weighted p-values. For general density-compound p-values, our bounds depend on a tractable null-distribution parameter, and we show that without further restrictions on this parameter, the bFDR can substantially exceed its nominal level. Our theory extends to discrete compound p-values constructed by pooling permutation tests. Applications illustrate the method in low-resolution testing problems, where individually valid permutation p-values are too coarse to yield discoveries after multiplicity correction.
Source:https://arxiv.org/abs/2611.24555PreviousAsymptotic efficiency comparison of explicit Dirichlet shape estimatorsNextFaithful Generative Modeling for Censored Time-to-Event Data: From Distributional Fidelity to Statistical Inference
Related posts
Splunk's Empty Cluster Secret Turns a Standalone Default into an Authentication Bypass
0dev.to
Pragmatic zero-trust: what to ship in the first 90 days
0dev.to
The EKS cost playbook: 7 levers we pull on every audit
0dev.to
Building a hearing test in the browser with the Web Audio API (and why we refuse to call it an audiogram)
0dev.to
How we score French speaking with AI: an honest look at limits
0dev.to
What a Four-Letter Word Ladder Taught Me About Planning Under Constraints
0dev.to