A comparison theorem on moment inequalities between negatively associated and independent random variables. (English) Zbl 0971.60015

Two sequences of random variables, negatively associated (NA) and independent, are considered. It is shown that the expectation of any convex function of the partial sum (or the maximum partial sum) of NA random variables can be bounded by those of independent random variables. Such a comparison result is useful to obtain limit theorems, especially strong laws of large numbers, functional central limit theorems, Berry-Esseen bounds and laws of the iterated logarithm for NA sequences of random variables.


60E15 Inequalities; stochastic orderings
60F15 Strong limit theorems
62N05 Reliability and life testing
60F17 Functional limit theorems; invariance principles
Full Text: DOI