Within the First World, observed differences across air carriers in passenger death risk are almost never statistically significant. But given the rarity of fatal crashes, even a lopsided split of crashes across airlines is unlikely to achieve significance. For greater perspective about relative airline safety performance, we supplemented an analysis of overall mortality data with a two-part statistical procedure. The first part--which was inspired by the way the Oakland Athletics Major League baseball team evaluates individual baseball players--allows for a sharp test of a key aspect of an equal-safety hypothesis, related to the frequency with which individual airlines suffer life-threatening events. The second test considers a further aspect of safety, namely the ability to recover from emergencies given that they have occurred. When applied to U.S. domestic jet services over 1983-2002, the test procedures indicate that long-established airlines have been about equally (and hugely) successful in protecting their passengers from accidental death and that established airlines are not demonstrably safer than the "new-entrant" jet carriers formed after U.S. airline deregulation.
Available at: http://works.bepress.com/david_czerwinski/3/