ABSTRACT

ACT In this article we review positive ant1 negative depcridencc structuiw for multivariate distributiorls. We prcscnt conditions under which product-type i~lequalities hold fbr the clmlulative distritn1tion function anti the survival function of a scqnence of random varial)lcs. When thcse conditions are not satisfied, we survey the Bonferroni-type inequalities that are often ~ ~ s c d . Applications to sequeritial inference, simultaneous inkrerice, testing for randorr~ness and outlier detection are discussed. Ni~merical results are presented to evaluate tlic perfornmlce of the ineqllalities discussed in this article.