Reference : Impact of dependency on the distribution of p-value
Scientific congresses and symposiums : Unpublished conference/Abstract
Physical, chemical, mathematical & earth Sciences : Mathematics
Impact of dependency on the distribution of p-value
Ernst, Marie mailto [Université de Liège > Département de mathématique > Statistique mathématique >]
Swan, Yvik mailto [Université de Liège > Département de mathématique > Probabilités et statistique mathématique >]
Joint Statistical Meetings - JSM 2017
29 juillet 2017 au 3 août 2017
[en] p-value ; Stein's method ; dependent random variables ; risk measure ; Binomial approximation ; runs
[en] We study the impact of dependence assumptions on the distribution of p-values and quantiles for repeated testing on dependent data. This leads us to considering the general problem of the quality of a binomial approximation to the distribution of a sum of dependent indicator variables. Whenever possible we use classical and adhoc versions of Stein’s method to provide tight bounds on classical probability distances. In many cases, however, the relevant expressions are intractable and we resort to empirical analysis by extensive simulations. We apply our findings to a realistic real-life scenario.
IAP Project StUDyS (2012-17)

File(s) associated to this reference

Fulltext file(s):

Open access
JSM2017MErnst.pdfPresentationAuthor preprint822.96 kBView/Open

Bookmark and Share SFX Query

All documents in ORBi are protected by a user license.