Abstract ABSTRACT: The auditing profession, in its use of existing statistical estimators and statistical procedures, has been assuming implicitly that the effect of nonsampling error on these estimators and procedures is negligible. This study investigates the potential effect of one type of nonsampling error, errors of measurement on the part of the auditor, on the behavior of alternative estimators and statistical sampling procedures in a Monte Carlo simulation study. The results indicate that the potential existence of independently occurring measurement error can significantly affect the auditor's statistical results and that the actual warranted level of confidence obtained by a statistical procedure may often be far below the desired level of confidence.
Anderson et al. (Tue,) studied this question.