Abstract ABSTRACT: A new method of statistical analysis for DUS is presented in this paper. This method is based on a three-moment representation of the sampling distribution for the mean error. The performance of the resulting upper confidence bound is examined in 236 separate simulation studies covering the error rates and tainting patterns observed in actual accounts receivable and inventory accounts. The bound is easily calculated and reliable. It would result in the correct acceptance of more accounts receivable populations than the multinomial bound when the materiality limit is less than four percent. It does a significantly better job of dealing with inventory populations with both overstatement and understatement errors than the currently available methods.
Dworin et al. (Sun,) studied this question.