You are on page 1of 3

technical circular

1
National Association of Testing Authorities, Australia

measurement
uncertainty in
chemical testing
March 2002
National Association of Testing Authorities, Aus-
tralia www.nata.asn.au
A copy of this circular can be found at www.nata.asn.au in the Publications
section under Policy and Technical Circulars. Published by the National
Association of Testing Authorities,
Implementation Date and Applicability Australia (NATA).
This technical policy is applicable to all laboratories accredited in the field of Chemical
Copies may be obtained from the
Testing and will apply at assessments carried out from 30 December 2001.
website or by contacting NATA
Corporate Communications on
Introduction Ph: (02) 9736 8222 or
Competent laboratories know the quality of their test methods and are able to estimate Fx: (02) 9743 5311
the uncertainty of the results obtained from these methods. Copyright National Association of
Testing Authorities, Australia 2002.
With the application of ISO/IEC 17025 at NATA assessments, laboratories must now satisfy
the requirement of this standard for the determination of measurement uncertainty (MU).
This is not really a new requirement. In the past, NATA has asked laboratories to be able to Contents
define their position when reporting results found at the specification limit. That is, is the Implementation Date and
result within specification or not? The only way this can be properly determined is if the Applicability
laboratory knows the uncertainty of the measurement. Introduction
ISO/IEC 17025 (clause 5.4.6) now requires that MU be determined for all testing undertaken Estimation of Measurement
within the laboratory, and when the laboratory performs its own in-house calibrations. Uncertainty
The latter must be of a comparable standard to similar calibrations performed by external Policy
accredited calibration laboratories. References
The knowledge of the measurement uncertainty of testing results is of particular Worked Examples
importance to laboratories and their clients who need to compare test results with defined Queries about this Policy
limits or requirements eg specifications, maximum residue limits.

Clients should be able to draw optimum advantage from a laboratory’s services. Clients Distribution
are interested in: Via website
■ How reliable the results are and if they can be complemented by a statement about All Chemical Testing Laboratories
their uncertainty;
All Chemical Testing Assessors
■ Knowing the certainty with which a conformity statement about the product can
be made; Relevant NATA Staff

■ Whether the test reports are factually correct, useful and comprehensive for the
client’s purposes. NATA is Australia’s national laboratory
accreditation system. Established in
It should be remembered that measurement uncertainty is but one feature of a test 1947, it is the largest association of
method. Laboratories will also consider the measurement specification, method validation laboratories in the world and serves as
(including reproducibility, repeatability, linearity, robustness, and selectivity) and the model for many overseas accredi-
traceability. tation systems. NATA accredits labora-
tories in many different fields of test-
Estimation of Measurement Uncertainty ing covering areas such as engineer-
ISO/IEC 17025 recognises the principles of “fitness for purpose” and this approach can ing, pathology, construction, applied
physics, chemistry, biology, forensic,
be used in the determination and application of measurement uncertainty of a result.
veterinary, non-destructive testing
It is recognised that the degree of thoroughness required for the determination of
and metrology. NATA also offers train-
measurement uncertainty of a particular method/task will impact on a laboratory’s time ing courses in laboratory manage-
and resources. It will be expected that the rigour of the evaluation of measurement ment and auditing.
uncertainty would be commensurate with the purpose of the testing.
technical circular
measurement uncertainty in
chemical testing

The degree of rigour with which MU is determined involved in the taking of the sample, then it does
will depend on a number of factors associated with not have to estimate the measurement uncertainty
the method and the client’s requirements, eg: of this process.
■ The risk associated with decisions that will be
Relying on the number of significant figures quoted
based on the measurements, where known;
for a result is not a valid way to estimate the un-
■ The cost-effectiveness of the method; certainty.
■ The need to comply with ‘good practice’ in the
Note 2 to clause 5.4.6.2 of ISO/IEC 17025 allows
particular sector;
for the use of a ‘well recognized test method’ that
■ Whether the method is fully quantitative or specifies limits to the major sources of uncertainty
only semi-quantitative; and specifies the form of reporting. The laboratory
■ In the case of compliance decisions, that those is considered to have satisfied the MU requirements
decisions are clear. by following the verified method and reporting in
accordance with the method.
Measurement uncertainty can be established and
reported by the following steps: In the absence of appropriate data, professional
judgement may be used but must be supported
1. The method specification is clearly described by objective evidence or clear previous expertise. It
and significant sources of uncertainty (> 20% of is not recommended that professional judgement
the total) are identified. This may be displayed be used to establish the uncertainties of major
on a ‘cause and effect’ diagram. components.
2. Where possible, sources of uncertainty are
grouped together to be covered by single deter- While laboratories may not currently have expertise
minations of uncertainty as may be available in the determination of MU, a professional labora-
from, for example, method validation studies. tory should have staff with the background and
training to develop this expertise. Laboratories
3. Each uncertainty component (or grouped com- may also need to use this expertise to estimate
ponent) is estimated by reference to available uncertainty by professional judgement when no
data, literature or professional judgement. formal evaluation is available.
4. The measurement uncertainty is established by
combining the uncertainties estimated in 3. As is usual NATA policy, it is not intended that we
tell laboratories how to estimate MU. However,
5. The measurement uncertainty is reported as in the Chemical Testing area, an acceptable
a combined standard uncertainty (u) or, after method is provided in the EURACHEM/CITAC
multiplying by a suitable coverage factor (k), as Guide - Quantifying Uncertainty in Analytical Meas-
an expanded uncertainty (U) which represents a urement. Other methods will also be considered, if
certain probability range (eg 95% for k=2). it can be demonstrated that they present a valid
approach to the estimation of MU.
The ISO Guide to the Expression of Uncertainty of
Measurement (GUM) provides details of how to This circular describes NATA’s current policy with
build up an uncertainty estimate from the most regard to the estimation of MU. As the transition to
basic effects (temperature, weighings, etc). For ISO/IEC 17025 continues and laboratories begin to
the most rigorous requirements, for example the estimate MU, this policy may need to be reviewed.
establishment of the concentration of a reference Updates to this policy will be found at the NATA
material, the ‘bottom up’ method described in GUM website and will be published in NATA News.
is recommended. However GUM also recognises
single estimates of grouped sources.
Policy
a) Laboratories must review their standard meth-
In estimating the measurement uncertainty, the
ods to determine if they provide sufficient data
laboratory needs to estimate those factors under
to allow MU estimations to be made. NATA will
its control. For example, if the laboratory is not
review progress with regard to this activity at
assessment.
2
Technical Circular # 1 - measurement uncertainty in chemical testing - March 2002
technical circular
measurement uncertainty in
chemical testing

b) MU must be established for all non-standard/in-


house methods.
It is expected that laboratories will have com-
pleted approximately 50% of the required MU
estimations by 31 December 2002.
Progress toward these targets will be reviewed
at assessment.
c) In order to demonstrate progress and facilitate
the assessment process, Chemical Testing
laboratories will be asked to provide, prior to
assessment:
i) their procedure(s) for estimating MU;
ii) example(s) of a completed MU estimation
using the procedure(s) implemented; and
iii) a proposed schedule of how the laboratory is
going to achieve the above time frame.

NOTE 1: If the laboratory has a large number of


in-house methods, consideration will be given to a
schedule that extends the time frame.
NOTE 2: It is acceptable to make use of examples
found in reputable literature. The laboratory must,
however, be able to demonstrate the relevance of
the example to their own operation.
NOTE 3: Standard methods are as defined in ISO/IEC
17025 section 5.4.6.2 Note 2. Any deviation from
the Standard method is considered to be a non-
standard method.

References
■ NATA Policy Circular No 3 NATA Accreditation
Requirements (ISO/IEC 17025) Implementation
and Timeline (Nov 2000)
■ EURACHEM/CITAC Quantifying Uncertainty in
Analytical Measurement (2nd Edition).
This is available on the Internet at www.vtt.fi/
ket/eurachem/quam2.pdf
■ ISO Guide to the Expression of Uncertainty in
Measurement (GUM) 1993

Worked Examples
The NATA website will include worked measurement
uncertainty examples relating to Chemical Testing.
Examples may also be found in the Eurachem/CITAC
Guide referred to above.

Queries About this Policy


Queries should be directed to John Widdowson,
Manager, Chemical Testing on
John.Widdowson@nata.asn.au or 3
(03) 9329 1633 (NATA Melbourne).
Technical Circular # 1 - measurement uncertainty in chemical testing - March 2002

You might also like