The Wasserstein Impact Measure (WIM): A practical tool for quantifying prior impact in Bayesian statistics

Reference : The Wasserstein Impact Measure (WIM): A practical tool for quantifying prior impact i...


	Document type :	Scientific journals : Article
	Discipline(s) :	Physical, chemical, mathematical & earth Sciences : Mathematics Engineering, computing & technology : Multidisciplinary, general & others
	To cite this reference:	http://hdl.handle.net/10993/51742

Title :	The Wasserstein Impact Measure (WIM): A practical tool for quantifying prior impact in Bayesian statistics
Language :	English
Author, co-author :	Ley, Christophe [University of Luxembourg > Faculty of Science, Technology and Medicine (FSTM) > Department of Mathematics (DMATH) >]
	Ghaderinezhad, Fatemeh [Ghent University > Department of Applied Mathematics, Computer Science and Statistics]
	Serrien, Ben [Vrije Universiteit Brussel - VUB > Experimental Anatomy Research Group]
Publication date :	Oct-2022
Journal title :	Computational Statistics and Data Analysis
Volume :	174
Peer reviewed :	Yes
Keywords :	[en] Effective sample size ; Neutrality ; Prior distribution ; Vallender formula ; Wasserstein distance
Abstract :	[en] The prior distribution is a crucial building block in Bayesian analysis, and its choice will impact the subsequent inference. It is therefore important to have a convenient way to quantify this impact, as such a measure of prior impact will help to choose between two or more priors in a given situation. To this end a new approach, the Wasserstein Impact Measure (WIM), is introduced. In three simulated scenarios, the WIM is compared to two competitor prior impact measures from the literature, and its versatility is illustrated via two real datasets.
Permalink :	http://hdl.handle.net/10993/51742

File(s) associated to this reference

Fulltext file(s):

	File	Commentary	Version	Size	Access
Open access	1-s2.0-S0167947321001869-main.pdf		Publisher postprint	1.33 MB	View/Open

All documents in ORBi^lu are protected by a user license.