Second-order stochastic comparisons of order statistics

Abstract: We study the problem of comparing ageing patterns of the lifetime of k-out-of-n systems. Mathematically, this reduces to being able to decide about a stochastic ordering relationship between different order statistics. We discuss such relationships with respect to second-order stochastic dominance, obtaining characterizations through the verification of relative convexity with respect to a suitably chosen reference distribution function. We introduce a hierarchy of such reference functions leading to classes, each expressing different and increasing knowledge precision about the distribution of the component lifetimes. Such classes are wide enough to include popular families of distributions, such as, for example, the increasing failure rate distributions. We derive sufficient dominance conditions depending on the identification of the class which includes the component lifetimes. Concerning the conditions, as expected, relying on a larger class of distributions, meaning that we have less precise information about the components behavior, leads to the need of stronger assumptions. We discuss the applicability of this method and characterize a test for the relative convexity, as this notion plays a central role in the proposed approach.