Dollar-Unit Sampling: A Comparison of the Quasi-Bayesian and Moment Bounds.
Test results are presented on the relative performance of two new auditing procedures for dollar-unit sampling--the quasi-Bayesian method developed by McCray [1984] and a slightly modified version of the moment method developed by Dworin and Grimlund [1984]. Extensive simulation studies of 96 representative accounts receivable and inventory populations were conducted. Both methods were found to be reliable over the tested range, and both provided tighter confidence bounds than those provided by the Stringer bound. This is the first study that has been made of the reliability of the quasi-Bayesian bound and of the comparative performance of the quasi-Bayesian and the moment methods. The moment method provided a tighter confidence bound at low error rates. For populations with high error rates, for which the true mean error was either very nearly zero or very large, the quasi-Bayesian method provided a tighter bound. Both bounds provided comparable results for the remaining intermediate cases. While either can be successfully used in these cases, the moment method appears to be easier to implement on currently available personal computers.