Abstract
Statistical practice often requires combining evidence from independent sources. A popular approach is to combine p-values. Motivated by the observation that p-values under the alternative are stochastically smaller than p-values under the null, we develop new combination rules that explicitly account for this ordering. The new combination rules are broadly applicable, optimal against some alternatives, and highly efficient against optimal procedures.
Original language | English |
---|---|
Pages (from-to) | 2433-2444 |
Number of pages | 12 |
Journal | Computational Statistics and Data Analysis |
Volume | 55 |
Issue number | 7 |
DOIs | |
State | Published - 1 Jul 2011 |
Bibliographical note
Funding Information:This research was supported by the Israeli Science Foundation Grants No. 875/09 and 1049/06 .
Keywords
- Combining functions
- Optimality
- Restricted inference
- Stochastic order relations
- Testing uniformity
- p-values
ASJC Scopus subject areas
- Statistics and Probability
- Computational Mathematics
- Computational Theory and Mathematics
- Applied Mathematics