David J. Laws and Anthony O'Hagan
University of Sheffield
Publication details: The Statistician 51, 431-450, 2002.
This paper provides a Bayesian analysis of a practical problem in auditing that deals with modelling and methodological difficulties which arise in other applica tions. The hierarchical model we propose has the flexibility to deal with statified populations in which some strata are not sampled and in which strata are not ex changeable. Inference from our model is by Monte Carlo simulation and, to facilitate this, we propose two general purpose rejection sampling algorithms. An example is given in the context of an audit carried out by the National Audit Office.
Keywords: Auditing; Bayesian inference; Cauchy distribution; Dirichlet process; Hierarchical modelling; Nonparametric; Rare errors; Simulation.