A BAYESIAN AUDITING MODEL

A Bayesian procedure is demonstrated that binds together multiple audit tests in a multiple-stage decision analysis. The analysis uses information provided by one test in the design of subsequent tests even when subsequent tests are directed at a different set of system characteristics. The analysis is applied to compliance and substantive tests undertaken to ascertain the fairness of a particular account balance. Although the analysis is not operational as presented, it provides a logical framework that clarifies the complex conceptual structure of the problem and enables the identification of impediments to its implementation.     

Continuous forecasting methods

The concept of a continuously operating system of expert opinion, a “collective brain” involving the combination of Delphi and network chart procedures, is discussed. The computer-aided process described by the author uses four types of information: the forecast of external actions of the environment, the values of control parameters as function of time, the current values of internal parameters, and estimates obtained from the principal experts.