References & Citations
Mathematics > Statistics Theory
Title: Post hoc inference via joint family-wise error rate control
(Submitted on 7 Mar 2017 (v1), last revised 8 Jan 2018 (this version, v3))
Abstract: We introduce a general methodology for post hoc inference in a large-scale multiple testing framework. The approach is called "user-agnostic" in the sense that the statistical guarantee on the number of correct rejections holds for any set of candidate items selected by the user (after having seen the data). This task is investigated by defining a suitable criterion, named the joint-family-wise-error rate (JER for short). We propose several procedures for controlling the JER, with a special focus on incorporating dependencies while adapting to the unknown quantity of signal (via a step-down approach). We show that our proposed setting incorporates as particular cases a version of the higher criticism as well as the closed testing based approach of Goeman and Solari (2011). Our theoretical statements are supported by numerical experiments.
Submission history
From: Pierre Neuvial [view email][v1] Tue, 7 Mar 2017 10:12:22 GMT (214kb,D)
[v2] Mon, 13 Mar 2017 15:44:35 GMT (215kb,D)
[v3] Mon, 8 Jan 2018 16:27:48 GMT (204kb,D)
Link back to: arXiv, form interface, contact.