Gamma minimax robustness of bayes rules

Roger L. Berger

Research output: Contribution to journalArticle

1 Scopus citations

Abstract

Bayes rules are shown to be robust in multiple decision problems in the sense that they retain an optimality property, T-minimaxity, when the original distributions are replaced by families of E-contaminated versions of themselves and the prior is replaced by a family of priors on these e-contaminations. Thus, the Bayes rule is robust against inaccurately specified parameter spaces and, hence, inaccurately specified priors. Bounds are obtained on the amount of contamination which can be present with the Bayes rule remaining T-minimax.

Original languageEnglish (US)
Pages (from-to)543-560
Number of pages18
JournalCommunications in Statistics - Theory and Methods
Volume8
Issue number6
DOIs
StatePublished - Jan 1 1979

Keywords

  • E-contamination
  • least favorable prior

ASJC Scopus subject areas

  • Statistics and Probability

Fingerprint Dive into the research topics of 'Gamma minimax robustness of bayes rules'. Together they form a unique fingerprint.

  • Cite this