The Utility of Survivability
Nozer D. Singpurwalla, (The George Washington University), email@example.com
The motivation for this research was a problem that the author confronted during some consulting work done, several years ago, for the Marine Corps. The general issue is, how does the DoD or other such agencies specify reliability requirements?. Numbers like .95 and .99 seem to be routine. Is there a justification for this choice? If so, the agency's utility should come into play. The problem therefore is an elicitation of utilities. But this is a general topic and should be of interest to statisticians as well as engineers doing control theory, and information integration, and decision makers in general. Statisticians and engineers use squared error and absolute error as a matter of routine. The matter of utility was of interest to statisticians like Savage, Chernoff, Rubin (Herman, that is), and Mosteller, and economists like Friedman, Arrow and Stigler, to name a few. Of late the matter seems to have become dormant. This talk is about peeking a t the old world through a new window. Modern decision theorists and economists write a lot about utilities. But the use of statistical ideas and methods seems to not have been exploited. We argue how this can be done.