MMCTest—A Safe Algorithm for Implementing Multiple Monte Carlo Tests. (1st April 2014)

• Record Type: Journal Article
• Authors: Gandy, Axel; Hahn, Georg
• Is Part Of: Scandinavian journal of statistics. Volume 41:Number 4(2014:Dec.)
• Page Start: 1083
• Page End: 1101
• Publication Date: 2014-04-01
• Subjects: Statistics -- Periodicals; 310
• DOI: 10.1111/sjos.12085
• Languages: English
• ISSNs: 0303-6898

Abstract

<abstract abstract-type="main" id="sjos12085-abs-0001"> <title>ABSTRACT</title> <p id="sjos12085-para-0002">Consider testing multiple hypotheses using tests that can only be evaluated by simulation, such as permutation tests or bootstrap tests. This article introduces <monospace>MMCTest</monospace>, a sequential algorithm that gives, with arbitrarily high probability, the same classification as a specific multiple testing procedure applied to ideal <italic>p</italic>‐values. The method can be used with a class of multiple testing procedures that include the Benjamini and Hochberg false discovery rate procedure and the Bonferroni correction controlling the familywise error rate. One of the key features of the algorithm is that it stops sampling for all the hypotheses that can already be decided as being rejected or non‐rejected. <monospace>MMCTest</monospace> can be interrupted at any stage and then returns three sets of hypotheses: the rejected, the non‐rejected and the undecided hypotheses. A simulation study motivated by actual biological data shows that <monospace>MMCTest</monospace> is usable in practice and that, despite the additional guarantee, it can be computationally more efficient than other methods.</p> </abstract>