| Popis: |
Only the reproducibility part of uncertainty is under the control of a production facility; making it in their favor to monitor, maintain, and improve production reproducibility. However, it is disturbingly common to find production facilities, and sometimes research organizations, making qualification decisions based on the statistically weak metric of maximum-minimum (range). Although range is easily computed and comprehended, it is not a good statistical measure. We show that reproducibility can be calculated easily with sound statistics that can be used by production facilities. Using a statistically strong metric like standard deviation provides more insight into the quality of data and potential contributors to the reproducibility than the often-used range metric (maximum and minimum). Range uses two points of many collected and does not give an accurate value for tester reproducibility. |
