Year
1969
Abstract
Who needs statistics? A favorite story among statisticians is that too often many people use statistics as a drunk uses a lamppost. He is depending on the lamppost for support rathe- than enlightenment. When we speak of establishing a safeguards measurement system, the intent is to establish a good working procedure whereby each measurement taken will shed some new, necessary light in answering the question, \"How well am I accounting for the special nuclear material in my possession?\" A well-planned measurement system, then, should not onfly use statistics to support the data obtained, but also to provide a more illuilnated view of the facility's safeguards picture. Since it is necessary to make measurements to establish a material balance, a good safeguards program should contain a strong statistical program for determining and evaluating the uncertainties associated with these measurements and their effect on the resulting material balance. The statistical parameter which characterizes these measurement uncertainties is the \"limits of error\" which is defined as \"the boundaries within which the true (best) value of the parameter being measured lies with a probability of 95%.\" It has recently and appropriately been christened \"the elusive limit of error.\"