Scribd
Upload
15 views

BF02503651

The document examines the relationship between the concepts of "error" and "uncertainty" of measurement results. It argues that the concepts of "standard uncertainty", "combined standard uncertainty", and "expanded uncertainty" from the Guide to the Expression of Uncertainty in Measurement are not applicable to non-metric scales of quantities and properties, such as order and designation scales. The document recommends using the general concept of "uncertainty" in the broad sense and expressing measurement result uncertainty for non-metric scales through scale-specific methods rather than those recommended in the Guide.

Uploaded by

ChuckPeter Martinorris
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
15 views

BF02503651

The document examines the relationship between the concepts of "error" and "uncertainty" of measurement results. It argues that the concepts of "standard uncertainty", "combined standard uncertainty", and "expanded uncertainty" from the Guide to the Expression of Uncertainty in Measurement are not applicable to non-metric scales of quantities and properties, such as order and designation scales. The document recommends using the general concept of "uncertainty" in the broad sense and expressing measurement result uncertainty for non-metric scales through scale-specific methods rather than those recommended in the Guide.

Uploaded by

ChuckPeter Martinorris
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Measurement Techniques. Vol. 43. No. 5.

2000

UNCERTAINTY OF MEASUREMENT RESULTS


IN VARIOUS SCALES

L. N. Bryanskii, A~ S. Doinikov, UDC 62-533:389


and B. N. Krupin

The relationship between the conc~7~ts of "'error" cazd "uncertainty" of measurement results is examined. The
concepts o f "standard uncertabTt3.;'" "'combbred standard uncertainty," card "'e.rpanded uncertabz~ '" are shown
to be bTapplicable b7 nomnetrh" ~'cales of quantit&s and properties, h~ which the general concept of
"'uncertainty'" b7 the broad ~'ense is recommended.

Use of the Guide to the Expressbin o[" U, certab~, in Measurement [l] is causing some difficulties among domestic
metrologists who are accustomed to employing the concept of "'error of a measurement result." This is aggravated by the fact
that this concept is also used in the entire nommtive documentation. At the same time, it has been established by an interna-
tional a~eement on the reciprocal recognition of national measurement standards and calibration certificates issued by nation-
al metrolo~cal institutes [2], in which the State Standards Institutes of Russia participate, that in key comparisons of national
standards the d e ~ e e of their equivalence "'is expressed quantitatively in values of deviations from the reference value of the unit
obtained in a key comparison and by the uncertainty in those deviations" [3]. As a result of an analysis of the situation, the
desire for "an exact definition of measurement uncertainty in the broad sense" has been expressed [4], so that the discussion of
this topic is interesting and acceptable decisions can be made.
We propose that the question of the relationship between the concepts of "'measurement-result error'" (MILE) and mea-
surement-result uncertainty" (MRU) be resoh'ed in the following most general and economical manner. First of all, we distin-
guish the general dictionary and metrological meanings of the words "error" and "'uncertainty.'" In the general sense, the word
"uncertainty" does not need to be defined. For metrology, we propose that MRU in the broad sense [1] be considered a more
general concept than MRE. This allows both terms to be combined harmoniously and prevents misunderstandings when exist-
ing normative documentation is used.
Then we turn our attention to the fact that the Guide Ill does not indicate any constraints on the use of the recom-
mended method for assessment and expression of the uncertainty of a measurement result. Moreover, it is stated that "... the
method must be applicable to all types of measurements and all types of input data used in measurements.'" Strictly speaking,
the Guide does not meet this requirement, since the methods described in it are applicable to far from all types of measurements.
The developers of the Guide failed to note that not only quantitative properties (quantities) are measured, but also qualitative
properties, to which designation scales correspond in metrology [5, 6]. In addition, there are quantities that do not possess the
property of proportionality, such as non-Archimedean quantities [7], on which a very extensive class of order scales is based in
metrology [5, 6, 8]. Also of importance is the fact that the concept of measurement unit is not in principle applicable to order
and designation scales.
It has been noted repeatedly [5, 6] that the traditional procedure of defining a measurement result as an arithmetic mean
of the results of individual observations, error calculation, and the concept of error itself are not applicable to measurements in
order and designation scales. Since the statistics of the arithmetic mean and standard deviation are not adequate for order and
designation scales [9, 10] (the median and range are applicable here), the Guide's cornerstone concept of "'standard uncertain-
ty'" is unsuitable for describing the uncertainty of a-measurement result in these scales. The concept of the "extended uncer-
tainty" of measurement results in the broadest sense is a convenient way out of this situation.

Translated from Izmeritel'naya Tekhnika, No. 5, pp. 29-30, May, 2000.

398 0543-1972/0014305-0398525.00 9 Kluwer Academic/Plenum Publishers


Uncertainty is present in any measurement, in any scale, and it is desirable to evaluate it somehow. These circum-
stances are responsible for the proposals that the concept of "measurement-result uncertainty" be employed in the new metro-
logical paradigm [5] for the correct description of the accuracy of measurement results in any scale. According to MI 2365-96
[6], the MRU should be understood as the region of the measurement scale in which an evaluation of a measured (quantitative
or qualitative) property is assumed to be found.
The meaning of this definition is similar to that of "measurement uncertainty" in [ 1], as the spread of values that can
justifiably be attributed to the measured quantity. This understanding of MRU is similar in meaning to that of "measurement
uncertainty" in [11] but is more-general than the definition of MRU in [l, 4] or "measurement error" according to [12].
On the strength of what has been said, the characteristics "standard uncertainty," "combined standard uncertainty," and
"extended uncertainty" and algorithms recommended in the Guide do not have the required generality. Just as "measurement
error;' these concepts are, in essence, directly applicable only for measurements in one-dimensional metric scales. Therefore,
the formulas for quantitative description of uncertainty in [ 1] cannot be used, for example, to evaluate the results of comparison
of national standards that reproduce the hardness scales of metals or color-coordinate and chroma scales, to compare the results
of measurements of the light sensitivity of photo~aphic materials, octane numbers, acid numbers, etc. [8]. It is also not possi-
ble to use directly the formulas of [ 1] to calculate the standard uncertainty of measurement results in logarithmic scales, such as
in the comparison of national standards of the unit of sound pressure.
The known methods for evaluating measurement error and uncertainty also cannot be used directly in studies of sur-
face shape or location or direction in a coordinate system [ 13]. Moreover, "soft measurements" and smart instruments have
already appeared on the metrology horizon, which are based on fuzzy logic, neural networks, and genetic algorithms (see [ 14],
for example). The results of soft measurements in many publications are also characterized by uncertainty in the broad sense.
Advances in the theory and calculus of fuzzy boundaries and the theory of granulation of fuzzy data, in particular, are used for
this [14]. A special concept -"uncertainty of identification" - has been introduced in scientific circles in the identification of
substances to assess the probability of correct or erroneous identification [ 15].
Since the differences in numerical estimates are insignificant, MRU according to the Guide [1] and the usual MRE
according to the normative documents in effect can be used in metric scales. Determination of the boundaries (zones) of MRU
for nonmetric scales has not yet been formalized; they are found for each specific scale [10]. This understanding of the term
allows it to be used to describe the accuracy of measurement results in all possible metric and nonmetric scales in use today [5].
Thus, the concept of MRU as well as those of "standard uncertainty;' "'combined standard uncertainty," and "extended
uncertainty" are not applicable to measurement scales in nonmetric order and designation scales; MRU in nonmetric scales
should be expressed by specific methods stipulated in specific normative documents that differ from those recommended in the
Guide. The fairly wide use of nonmetric scales and the presence of state standards and primary precision instruments that imple-
ment those scales make it important to find generalized methods for expression of uncertainty in measurements.

REFERENCES

1. Guide to the Expression of Uncertain~ h7 Measurement, 1st ed., ISO, Switzerland (1993).
2. Mutual Recognition of National Measurement Standards and of Calibration and Measurement Certificates Issued by
National Metrology h~stitutes, Comit~ international des poids et mesures, Paris, 14 October (1999).
3. T. J. Quine, Izmer. Tekh., No. 5, 67 (1998).
4. Yu. V. Tarbeev, V. A. Slaev, andA. G. Chunovkina, Izmer. Tekh.. No. 1, 69 (1997).
5. L. N. Bryanskii, A. S. Doinikov, and B. N. Krupin, Izmer. Tekh., No. 8. 15 (1998).
6. MI 2365-96, Measurement Scales, Basic Positions, Terms and Definitions [in Russian].
7. A. N. Kolmogorov, "Quantity." Mathematical Encyclopedia, Vol. 1, Sov. l~ntsiklopedia, Moscow (1977), p. 651.
8. L. N. Bryanskii, A. S. Doinikov, and B. N. Krupin, Izmm: Tekh., No. 9, 3 (1999).
9. I. Pfanzagl', Measurement Theory [Russian translation], Mir, Moscow (19761.
10. L. N. Bryanskii, A. S. Doinikov, and B. N. Krupin, Zakonodat. Prikl. Metrolog., No. 6. 30 (1996).
11. E E Stakhov, Introduction to the Algorithmic Theory o f Measurements [in Russian], Sovetskoe Radio, Moscow (1977).
12. GOST 16236-70, Metrology, Terms and Definitions [in Russian].

399
13. E E. El'yasberg, Measurement Data: How Much Is Required? How Is It Processed? [in Russian], Nauka, Moscow
(1983).
14. L. Reznik, in: Abstracts of Proceedings of lnten~ational Conference on Soft CalcLdations and Measurements
[in Russian], Vol. 1, St. Petersburg (1999). p. 21.
15. B. L. MiFman and L. A. Konopel'ko, in: Abstracts of Proceedings of AlI-Russian Conference "Metrological Problems
of the Development and lntrodltction of Measllrement Methods'" [in Russian], St. Petersburg (1999), p. 80.

400

You might also like