Knowledge of the occurrence and detection of errors in accounting populations is of great importance to auditors in assessing risks, evaluating the efficacy of statistical sampling methods, and planning effective and efficient audit procedures to address risks. A significant body of research exists that examines these issues. Prior studies have focused primarily on auditordetected errors. A basic assumption of these studies is that detected errors are an accurate reflection of all significant errors present. That is, there are not a substantial number of undetected errors, or that undetected errors share the same characteristics (e.g., error direction) as detected errors. However, little evidence exists regarding the accuracy of this assumption. Further, there has been little consideration of factors that may affect differences between detected and actual errors and the implications of these differences on research conclusions.
This paper presents a model of the variables involved in the error generation and error detection processes. Variables that have been explored in prior research are discussed along with those requiring further investigation. Finally, the paper identifies confounding variables to be controlled in future studies and makes suggestions for improving extant error study methods.
Available at: http://works.bepress.com/paul_caster/19/