, , Alexander L. Enders
Often, experimenters wish to confirm that a test artifact meets some predefined, fixed performance criterion or claim. For example: does the newly formulated pharmaceutical reduce the disease rate by 10 %; will the composite overwrapped pressure vessel fail prior to the completion of the 15-year space mission; or, can the radiation detection system detect the specified radiological source with at least 80 % probability? The answers to these questions are a simple yes or no, but because of the inherent uncertainty in the measurements used in the assessment, there is a risk of answering the question incorrectly. This chapter provides guidance on developing an experimental sample size and acceptance criterion to determine whether a test artifact satisfies a predefined and fixed performance criterion when the response variable observed is binary, such as a success or failure to detect. We provide a sample size and rejection criterion table to define a fixed sample test that will satisfy a variety of performance thresholds and levels of acceptable risk. We conclude with a general discussion of sequential sampling tests and provide important considerations and contrasts to their fixed sample counterparts.
Technical Note (NIST TN) - 2045
hypothesis testing, sample size, sequential test, binary response, radiation detection