The Auditor's Sampling Objectives: Four or Two?: A Reply
We do not have any quarrel with the statistical analysis in Kinney's note. In our first paper on this subject, we acknowledged that an auditor may sample high-valued items because of their normal association with high variance.' Similarly, sampling from a high error rate category may be explained as contributing to representative sampling because of the higher degree of contribution in reducing the error of estimate. Finally, random sampling is consistent with the auditor wishing to make statistically valid statements about the population from which a sample is chosen. Our papers,2 however, attempted to point out that there may be nonstatistical reasons which also explain the above set of auditors' behavior; i.e., auditors may sample high-valued items to cover as much of the total dollar value in the population as possible (protective sampling); they may sample high error rate categories to correct as many items as possible; and they may use random sampling to create a greater degree of uncertainty as to the scope of future audits to deter potential errors in the future (preventive sampling).