Statistical tests for verification sampling plans

摘要

Some of the IAEA's verification activities are based on verifying operator's declarations. These activities usually rely on a sampling plan that is designed to achieve a specified probability to detect falsification of operator records. Currently, some of the IAEA's sampling plans assume item-by-item tests in which the difference between the reported and the measured value of each item selected for verification is compared to a threshold. If a difference exceeds this threshold, then an "alarm" occurs, and the item is investigated further. In this paper two additional sampling plans are analyzed: First, instead of the item-by-item tests, only the overall difference statistic of the verified items is used as test statistic. Second, in addition to the item-by-item tests, the overall difference statistic of the verified items is applied as test statistic. This paper investigates analytically and by using simulation the relation between the three sampling plans.