ESTIMATING THE LIKELIHOOD OF INCIDENTS

"Any theory based on experience is necessarily statistical; that is to say, it formulates an ideal average which abolishes all exceptions at either end of the scale and replaces them by an abstract mean. This mean is quite valid, though it need not necessarily occur in reality. " (Dr Carl Gustav Jung)

In Chapter 3, the two-dimensional model discussed has two basic parameters; the severity of incident consequences and the likelihood of occurrence. In order to estimate the risk, we need both parameters. In Chapters 5 to 7, we covered the first of these parameters, namely consequence analysis. In this Chapter, we shall discuss the second parameter, namely estimation of likelihood.
The likelihood of an incident is a probability, and therefore is not deterministic. We cannot say when an incident may occur. We can only say what the probability of occurrence is. If this probability is high, it requires us to do something about it; if it is low, we may decide to live with it, taking every care that it remains low through a Safety Management System and administrative controls. Therefore, the likelihood estimate is one of the major sources of uncertainty in risk analysis. Consequently, it also poses significant challenges to the analyst.