Deriving Ratios in Aviation Safety Analysis when Data Unavailability Precludes Probability Normalization

Abstract

A runway incursion operational error data is used to illustrate the problem of stating rates and probabilities using inappropriate denominators. The author suggests meaningful probability ratios can be derived by using a simple contingent probability manipulation known as Bayes’ rule. The author argues that due to the way data are recorded and published, data required for proper normalization may not be available. Yet in the absence of such normalizing data, he shows that it is possible to determine certain desired features of the data, such as ratios of types of violations, given joint frequency of salient factors.