ABSTRACT
Published data on oil spillage incidents are analysed using a variety of statistical techniques. Such data represents considerable past experience of such incidents in a form suitable for the application of Bayesian and time series analysis methods.
The data includes various measures of the severity of the spillage (eg quantity of oil) and of the response (eg cost). Questions about changes in these aspects during the last 20 years are considered. For example, is there evidence that we are “getting better” at dealing with, or at preventing, such incidents? Connected with this, we could ask ‘What are the predictions for the future levels of incidence and costs?’
The techniques and style of analysis may find application in other incident data, for example nuclear and chemical, for which published figures may not be so readily available.
