SUFFICIENT CONDITIONS FOR ASSOCIATION OF BINARY VARIABLES,

Abstract

In connection with studies of reliability of coherent binary systems, it has been suggested that a set of binary random variables be called associated if all pairs of binary nondecreasing functions of these variables have nonnegative covariance. In this note it is shown that there exists a unique smallest set A of pairs of binary nondecreasing functions on the variables such that nonnegative covariance for the pairs in A is necessary and sufficient condition for association of the variables. This set A is characterized and it is shown that its size increases rapidly with the number of variables. It is also shown that for four or more variables there exist pairs in A corresponding to coherent systems with no nontrivial modules, thus suggesting it is impossible to find an equivalent definition of association of binary random variables which makes material use of partitioning of the variables. (Author)

Document Details

Document Type
Technical Report
Publication Date
Dec 01, 1966
Accession Number
AD0648147

Entities

People

  • D. W. Walkup

Organizations

  • Boeing

Tags

DTIC Thesaurus Topics

  • Covariance
  • Data Science
  • Information Science
  • Materials
  • Mathematics
  • Random Variables
  • Reliability

Fields of Study

  • Mathematics

Readers

  • Statistical inference.