Untitled

Meaning and models in key comparisons, with measures of operability and interoperability

“…Some authors provide more details. Chunovkina [3] contrasts a key comparison with two other activities in metrology; those of certifying a reference material and certifying a method of measurement. Her analysis highlights an important distinction to be made between the kind of interlaboratory comparison that is envisaged in the MRA and the kind of multi-laboratory study that has the determination of a shared true value as the primary goal.…”

“…' Elster and Link [10] note that as a result of a key comparison 'it is judged whether a laboratory ... is measuring as accurately as it claims to do.' Similarly, Chunovkina [3] identifies one major purpose of a key comparison as being the provision of a check on the uncertainties in the results, and relates this to the task of determining reasons for systematic deviations between laboratories [9], while Pavese [11] considers the basic purpose to be the detection of possible inconsistencies in measurement results. Also, Chunovkina [9] states 'The quantitative expression of equivalence, as a matter of fact, is a new way of ... [expressing] ... closeness (reproducibility) of the measurement results obtained from participating laboratories.…”

“…Indeed, because the purposes of comparisons mentioned in section 1.1 include those of (i) comparing standards realized at different institutes, (ii) checking claims of accuracy, (iii) detecting possible inconsistencies and (iv) determining reasons for systematic deviations, it does not seem appropriate to let a measurement result contribute to its own point of reference. The influence of the measurement result on the point of reference would undoubtedly be to draw the two closer together, which would promote overestimation of the measurement ability of the NMI 3 .…”

“…There would seem to be several reasons why this approach should be considered, including the following. 3 It might be thought that accounting for the correlation between a measurement result and the KCRV compensates for this effect. However, this is only the case if measurement ability relates exclusively to the idea of 'being correct within our uncertainty' via a normalized error such as the figure…”

“…Several papers have presented methods of analysis (see list of references in [1]), while others contain more discussion about the purpose of key comparisons and the meaning of their outputs, e.g. [2][3][4]. The latter kind seems valuable because without a clear idea of the meaning of outputs of the key comparison there cannot be agreement on a proper means of analysis.…”