How important is all this rigor – what is the “worst case scenario” for mis-stating equivalence?
2002-08-26 “QDE” is not the sole conveyor of rigor – for any pair of participating labs it is a good one-parameter description of equivalence. It is much better than the binary pass/fail decision, but it is only a one-parameter summary of the two-parameter description of bilateral equivalence that is given in the MRA degrees-of-equivalence table. The deficiencies of the traditional null hypothesis approach (basing a lab’s pass/fail decision on whether its departure from the KCRV is less than or more than twice the combined uncertainty) can be calculated from the two-parameter degrees-of-equivalence, and summarized as the confidence in the Calibration Measurement Capabilities in the MRA Appendix C. This calculation can be done by the QDE Toolkit Version 2.05 macro “tk_mraQDC_TableBuilder”, and can give the specifics of the “worst case scenario” for any particular comparison. For the traditional null hypothesis approach, the global “worst case scenario” would be for two labs, one offset
