The article discusses the use of deterioration tests to enhance the efficiency of superiority and non-inferiority metrics in decision-making. It outlines how making assumptions about the rejection regions of inferiority tests—so they do not overlap with those of superiority tests—can lead to improved performance in type I and type II error rates. The focus is on a layered approach that includes both success and deterioration tests, specifically addressing how these tests interplay in the framework of statistical hypotheses and decision rules.
We describe how deterioration tests affect the type I and type II error rates for superiority and non-inferiority tests and under what conditions the rates are not affected.
It is possible to improve the efficiency by making assumptions that makes the inferiority test rejection region not overlap with the rejection region of the superiority test.
The composite hypotheses for success and deterioration metrics tell us about the shipping decision based on both passing and failing metrics.
We highlight how new notations can specify impacts on testing when comparing success metrics against additional deterioration tests.
Collection
[
|
...
]