ISO/TR 11843-8:2021

Capability of detection — Part 8: Guidance for the implementation of the ISO 11843 series

This document provides guidance for implementing the theories of the ISO 11843 series in various practical situation. As defined in this series, the term minimum detectable value corresponds to the limit of detection or detection limit defined by the IUPAC. The focus of interest is placed on the practical applications of statistics to quantitative analyses.

Capacité de détection — Partie 8: Recommandations pour la mise en œuvre de la série ISO 11843

General Information

• Status: Published
• Publication Date: 21-Nov-2021
• ICS: 03.120.30 - Application of statistical methods
• ICS: 17.020 - Metrology and measurement in general
• Technical Committee: ISO/TC 69/SC 6 - Measurement methods and results
• Drafting Committee: ISO/TC 69/SC 6/WG 5 - Capability of detection
• Current Stage: 6060 - International Standard published
• Start Date: 22-Nov-2021
• Due Date: 26-Jun-2022
• Completion Date: 22-Nov-2021

Standards Content (Sample)

TECHNICAL ISO/TR
REPORT 11843-8
First edition
2021-11
Capability of detection —
Part 8:
Guidance for the implementation of
the ISO 11843 series
Capacité de détection —
Partie 8: Recommandations pour la mise en œuvre de la série ISO
Reference number
© ISO 2021
All rights reserved. Unless otherwise specified, or required in the context of its implementation, no part of this publication may
be reproduced or utilized otherwise in any form or by any means, electronic or mechanical, including photocopying, or posting on
the internet or an intranet, without prior written permission. Permission can be requested from either ISO at the address below
or ISO’s member body in the country of the requester.
ISO copyright office
CP 401 • Ch. de Blandonnet 8
CH-1214 Vernier, Geneva
Phone: [phone]
Email: [e-mail]
Website: www.iso.org
Published in Switzerland
ii
Contents Page
Foreword .iv
0       Introduction .v
1 Scope . 1
2 Normative references . 1
3 Terms, definitions and symbols . 1
3.1 Terms and definitions . 1
3.2 Symbols . 1
4 Historical survey of terms . 3
5 Fundamental concepts of detection limit (minimum detectable value in ISO 11843) .4
5.1 General . 4
5.2 General definition of detection limit . 4
5.3 Detection limit with probability α . 5
5.4 Detection limit with probabilities α and β . 6
6 Pragmatic view of α and β . 9
6.1 Statistical definitions of α and β . 9
6.2 Actual examples of α and β values . 9
7 In-depth explanations and examples of the Parts in the ISO 11843 series .9
7.1 General . 9
7.2 ISO 11843-3 and ISO 11843-4 . 10
7.2.1 General . 10
7.2.2 Number of repeated measurements, J and K . 10
7.2.3 Determination of the minimum detectable value . 11
7.2.4 Confirmation of the minimum detectable value for an obtained
experimental value with the number of repeated measurements, N . 11
7.2.5 Number of repeated measurements, J and K, in ISO 11843-5 and ISO 11843-7 .13
7.3 ISO 11843-6 . 13
7.3.1 Overview of ISO 11843-6 . 13
7.3.2 Features of pulse count measurement . 13
7.4 Example from ISO 11843-7 . 19
Annex A (informative) Standard normal random variable .23
Annex B (informative) Difference between the power of test and
the minimum detectable value .25
Annex C (informative) Calculation example from ISO 11843-4 .27
Annex D (informative) Calculation example from ISO 11843-6:2019, Annex E (Measurement
of hazardous substances by X-ray diffractometer) .28
Annex E (informative) Comparison between the Poisson exact arithmetic and
the approximations .31
Annex F (informative) Association of IUPAC recommended detection limit
with the ISO 11843 series .36
Bibliography .38
iii
Foreword
ISO (the International Organization for Standardization) is a worldwide federation of national standards
bodies (ISO member bodies). The work of preparing International Standards is normally carried out
through ISO technical committees. Each member body interested in a subject for which a technical
committee has been established has the right to be represented on that committee. International
organizations, governmental and non-governmental, in liaison with ISO, also take part in the work.
ISO collaborates closely with the International Electrotechnical Commission (IEC) on all matters of
electrotechnical standardization.
The procedures used to develop this document and those intended for its further maintenance are
described in the ISO/IEC Directives, Part 1. In particular, the different approval criteria needed for the
different types of ISO documents should be noted. This document was drafted in accordance with the
editorial rules of the ISO/IEC Directives, Part 2 (see www.iso.org/directives).
Attention is drawn to the possibility that some of the elements of this document may be the subject of
patent rights. ISO shall not be held responsible for identifying any or all such patent rights. Details of
any patent rights identified during the development of the document will be in the Introduction and/or
on the ISO list of patent declarations received (see www.iso.org/patents).
Any trade name used in this document is information given for the convenience of users and does not
constitute an endorsement.
For an explanation of the voluntary nature of standards, the meaning of ISO specific terms and
expressions related to conformity assessment, as well as information about ISO's adherence to
the World Trade Organization (WTO) principles in the Technical Barriers to Trade (TBT), see
www.iso.org/iso/foreword.html.
This document was prepared by Technical Committee ISO/TC 69, Applications of statistical methods,
Subcommittee SC 6, Measurement methods and results.
A list of all parts in the ISO 11843 series can be found on the ISO website.
Any feedback or questions on this document should be directed to the user’s national standards body. A
complete listing of these bodies can be found at www.iso.org/members.html.
iv
0 Introduction
0.1  General
The purpose of this document is to facilitate the dissemination of the principles and methods of the
ISO 11843 series on a global scale by providing a brief explanation of the background of its development,
the significance of defining detection limits, the historical variation of the term detection limit, the
modern concept of detection limit, and basic ideas of statistics and of each part of this series, intelligible
to analytical chemists, biologists, operators, technicians, and others in various fields.
The series ISO 11843 provides statistical theories and some practical applications in a mathematically
strict way. This guidance is put forth with the goal of guiding laymen in statistics in practicing the
statistics of detection limits, not offering the in-depth knowledge of the relevant mathematics, but
making them aware of some of the challenges of using statistical theory and the reasons for success and
failure in using the formulae included in the series.
0.2  Background
[1]
The concept of detection limit was first described in 1949 ; after that, a number of scientists submitted
[2][3]
papers on the definition of detection limit . Scientists in different countries have used detection
limits with different definitions.
In order to avoid such global confusion, the International Union of Pure and Applied Chemistry
(IUPAC) began considering the introduction of a modern detection limit using a new definition based
on statistics. Representatives of the IUPAC and the International Organization for Standardization
(ISO) met between 1993 and 1997 to begin efforts to develop a harmonized international chemical-
metrological position on detection and quantification capabilities. The IUPAC nomenclature document
was published in 1995 to help establish a uniform and meaningful approach to terminology, notation,
and formulation for performance characteristics of the chemical measurement process, and in 1997
ISO published its standard (ISO 11843) for the international metrological community. IUPAC has
incorporated the 1995 recommendations into its basic nomenclature volume, the Compendium on
Analytical Nomenclature (IUPAC, 1998).
0.3  Parts of ISO 11843
The ISO 11843 series consists of the following published parts:
— ISO 11843-1, Capability of detection — Part 1: Terms and definitions;
— ISO 11843-2, Capability of detection — Part 2: Methodology in the linear calibration case;
— ISO 11843-3, Capability of detection — Part 3: Methodology for determination of the critical value
for the response variable when no calibration data are used;
— ISO 11843-4, Capability of detection — Part 4: Methodology for comparing the minimum detectable
value with a given value;
— ISO 11843-5, Capability of detection — Part 5: Methodology in the linear and non-linear calibration
cases;
— ISO 11843-6, Capability of detection — Part 6: Methodology for the determination of the critical value
and the minimum detectable value in Poisson distributed measurements by normal approximations;
— ISO 11843-7, Capability of detection — Part 7: Methodology based on stochastic properties of
instrumental noise.
v
0.4  Social purposes
0.4.1 Significance of defining the minimum detectable value
The determination of the minimum detectable value is sometimes important in practical work. The
value provides a criterion for deciding when “the signal is certainly not detected”, or when “the signal
is significantly different from the background noise level". For example, it is valuable when measuring
the presence of hazardous substances, the degree of calming of radioactive contamination, and surface
contamination of semiconductor materials, as follows.
— RoHS (Restrictions on Hazardous Substances) sets limits on the use of six hazardous materials
(hexavalent chromium, lead, mercury, cadmium and the flame retardant agents perbromobiphenyl,
PBB, and perbromodiphenyl ether, PBDE) in the manufacturing of electronic components and
related goods sold in the EU.
— Environmental pollution by radioactive materials due to accidents at nuclear power plants is a major
problem. While it takes a considerable amount of time for the contaminated environment to return
to its original state, it is important to monitor the state of contamination during that time.
— The condition of an analyser to be quantified when assessing the limiting performance of an
instrument.
0.4.2  Trouble prevention with stakeholders
To avoid problems with stakeholders, concerning the presence or absence of hazardous substances,
a kind of agreement or rule based on the scientific theory for judging the presence or absence of the
hazardous substance is set up.
a) Health hazard trouble of hazardous substances.
b) Product quality assurance in commerce (non-inclusion of hazardous substances, product
contamination).
0.4.3  Performance evaluation of measuring instruments
The series of ISO 11843 provides conditions for judgment on whether the detection capability of
measuring instruments is adequate.
vi
TECHNICAL REPORT ISO/TR 11843-8:2021(E)
Capability of detection —
Part 8:
Guidance for the implementation of the ISO 11843 series
1 Scope
This document provides guidance for implementing the theories of the ISO 11843 series in various
practical situation. As defined in this series, the term minimum detectable value corresponds to the
limit of detection or detection limit defined by the IUPAC. The focus of interest is placed on the practical
applications of statistics to quantitative analyses.
2 Normative references
The following documents are referred to in the text in such a way that some or all of their content
constitutes requirements of this document. For dated references, only the edition cited applies. For
undated references, the latest edition of the referenced document (including any amendments) applies.
ISO 11843-1, Capability of detection — Part 1: Terms and definitions
3 Terms, definitions and symbols
3.1 Terms and definitions
For the purposes of this document, the terms and definitions given in ISO 11843-1 apply.
ISO and IEC maintain terminology databases for use in standardization at the following addresses:
— ISO Online browsing platform: available at https:// www .iso .org/ obp
— IEC Electropedia: available at https:// www .electropedia .org/
3.2 Symbols
X
state variable or probability density
Y
response variable
J
number of replications of measurements on the reference material representing the value of
the basic state variable (blank sample)
k constant for minimum detectable values and critical values, e.g. x = k × standard deviation
D
K
number of replications of measurements on the actual state (test sample)
N
number of replications of measurements of each reference material in assessment of the capa-
bility of detection
x
value of a state variable
y
value of a response variable
y
critical value of the response variable defined by ISO 11843-1 and ISO 11843-3
C
x
given value, tested to determine whether it is greater than the minimum detectable value
g
x
minimum detectable value of the state variable
D
σ
standard deviation under actual performance conditions for the response in the basic state
b
σ
standard deviation under actual performance conditions for the response in a sample with
g
the state variable equal to x
g
η
expected value under the actual performance conditions for the response in the basic state
b
η
expected value under the actual performance conditions for the response in a sample with
g
the state variable equal to x
g
y
arithmetic mean of the actual measured response in the basic state
b
y
arithmetic mean of the actual measured response in a sample with the state variable equal to
g
x
g
y minimum detectable response value with the state variable equal to x
D d
λ
mean value corresponding to the expected number of events in Poisson distribution
α
probability of an error of the first kind
β
probability of an error of the second kind
1−α
confidence level
1−β
confidence level
s estimate of the standard deviation of responses for the basic state
b
estimate of the standard deviation of responses for a sample with the net state variable equal
s
g
to x
g
z ()1−α -quantile of the standard normal distribution
1−α
NOTE Further information is provided in Annex A.
z 1−β -quantile of the standard normal distribution
()
1−β
NOTE Further information is provided in Annex A.
t ()ν ()1−γ -quantile of the t-distribution with ν degrees of freedom
1−γ
T
lower confidence limit
4 Historical survey of terms
Key
X net state variable or probability density Y response variable
minimum detectable value of the net state variable critical value of the net state variable
x x
D C
minimum detectable value of the response
reference state the response variable
y y
b D
variable
critical value of the response variable α probability of an error of the first kind
y
C
β probability of an error of the second kind 1 calibration function
Figure 1 — Critical value of both the response variable and the net state variable, and
minimum detectable value of both the response variable and the net state variable
[4]
In Figure 1, x is called the limit of identification by Boumans and the limit of guarantee by Kaiser . As
D
shown in Figure 1, when analysing a sample containing an x component, the probability that the value
D
of the response variable (output) becomes smaller than y is as small as β. Indeed, x is the minimum
C D
detectable amount with a very low probability of being missed by analytical methods. Currie named it
[5]
the detection limit .
In order to ensure consistency with ISO standards, IUPAC has defined detection limit as x since 1994,
D
[6]
and minimum detectable quantity is also sometimes used . This interpretation has not necessarily
become widespread among analysts, but it is correct as long as detection limit is defined as the
minimum amount that can be detected.
The term "detection limit," which is very familiar to most chemists, has been abolished in
ISO 11843-1 because some chemists disagree with dividing the definition of "detection limit" into
two interpretations, x and x . They feel that x alone is sufficient for the detection limit. Statistical
C D C
interpretations of detection limits are given in 5.4 and Annex B.
Instead, the term “critical value of the response variable” was assigned to y , “critical value of the net
C
state variable” to x , and “minimum detectable value of the net state variable” to x . In addition, the
C D
term “sensitivity” or “detection sensitivity” has often been used to express the detection capability of
the measuring method.
Detection sensitivity, which is most frequently used on a daily basis, can represent the change rate of
a response variable (equivalent to the slope of a calibration curve) with respect to the change per unit
[7]
of a state variable . In consideration of this situation, neither detection limit nor sensitivity has been
used in this document as a term representing detection capability. The terms used in ISO 11843-1 and
[6]
the terms the IUPAC recommended in 1994 are summarized in Table 1. The association of the IUPAC
recommended detection limit with the ISO 11843 series is also described in Annex F.
Table 1 — Terms used in ISO 11843-1 and IUPAC
ISO 11843-1 IUPAC
y Critical value of the response variable Minimum significant signal (critical level)
C
x Minimum detectable value of the net state variable Minimum detective quantity (detection limit)
D
5 Fundamental concepts of detection limit (minimum detectable value in
ISO 11843)
5.1 General
Widely, and for many years, detection limits (DLs) have been recognized as a figure of merit of vital
importance and utilized in every discipline of analytical chemistry to ensure statistical reliability and
practical suitability of analytical systems. A fundamental quantity underlying DLs is the standard
deviation (SD) of response variables or measurements. ISO 11843-1 provides general definitions of DLs
on the basis of theoretical SDs (population SDs), while the other Parts of the ISO 11843 series are all
devoted to the externalization of DLs with SD estimates (sample SDs), i.e. how to obtain SD estimates
in practice. This clause shares a brief but comprehensive explanation of DLs in terms of population SDs.
Estimation methods of sample SDs are given in detail in Clause 7.
In the ISO 11843 series, the definition of detection limits, referred to there as minimum detectable
values, is founded on probabilities, α and β, of errors of the first and second kind, respectively. However,
a DL with probability α alone has also played an important role in some fields of industry. This clause
clarifies the theoretical backgrounds and similarities and differences of these DL definitions, which are
recommended to be noted in practical applications.
5.2 General definition of detection limit
Detection limits are defined in X- and Y-axes that are spanned by a calibration function, y = f(x). The
X-axis denotes objective quantities of analyses, e.g. concentration or weight, and the Y-axis instrumental
responses or measurements such as absorbance or electric current. However, the definitions in the
different scales seem ostensible, because they come from a traditional understanding that a DL has
necessarily been specified in the X-scale, whereas stochastic uncertainty of measurements is directly
observable in the Y-scale. The ISO 11843 series takes the following approaches to interpreting the
lingua franca expressed in the different dimensions.
Y-axis DL, y , is estimated from an observable distribution of measurements or responses, y. Then, y
D D
-1
is transformed into its corresponding quantity, x , through the calibration function: x = f (y ). In this
D D D
clause, uncertainty of calibration functions is not taken into consideration. Therefore, errors of the final
quantity, x, are totally attributable to those of y.
X-axis DL, x , is straightforwardly evaluated in the X-axis. As such, this treatment requires a
D
distribution of the quantity, x, which is to be transformed from a distribution of observable y through
the mathematical relationship between x and y. An example is given in ISO 11843-5.
As is well-known in probability theory, a function of a random variable is a random variable. For
example, under the simplest calibration function, y = ax where a is a constant, a normal distribution of y
produces a normal distribution of x (= y/a) (called reproducibility in probability theory).
5.3 Detection limit with probability α
Let the normal distributions along the Y-axis of Figure 2 and the X-axis of Figure 3 be population
-1
distributions of observable y and estimable x (= f (y)), respectively, the averages of which are y and x
0 0
-1
(= f (y )) and the SDs of which are σ(y ) and σ(x ). With probability α, detection limits, y and x , are
0 0 0 D D
defined as k times the SDs for blank samples, respectively,
y = y + z σ(y) (1)
D 0 1−α 0
x = x + z σ(x) (2)
D 0 1−α 0
where blank samples mean x = 0 and z is a constant. Lowercase k is often used in Formulae (1)
0 1−α
and (2) in analytical chemistry, but z (and z ) is preferred throughout this document. The DL
1−α 1−β
value, y , is first determined in the Y-axis and then transformed into the final quantity, x , through y
D D
= f(x) (Figure 2), whereas the DL, x , is directly evaluated in the X-axis (Figure 3). The DL definitions
D
of Formulae (1) and (2) correspond to the decision limits or critical values defined in the following
subclause.
-1
Symbol α denotes the probability of observable y or estimable x (= f (y)) exceeding the DL, y or x ,
D D
when blank samples at a concentration of x (= 0) are measured repeatedly under exactly the same
experimental conditions. If z = 3 and the distribution of y or x is normal, α is 0,14 %.
1−α
Key
X net state variable or probability density Y response variable
1 detected 2 not detected
−1
3 y = f(x) X
D xf= ()y
DD
Figure 2 — Definition of detection limit in Y-axis with probability α
Key
X state variable Y probability density
1 detected 2 not detected
Figure 3 — Definition of detection limit in X-axis with probability α
For a sample of an unknown concentration, it can safely be said that with a risk of at most 0,14 %,
— if y < y or x < x , nothing is detected;
D D
— if y ≥ y or x ≥ x , something is detected.
D D
These judgments are confirmed by the assumption that the blank samples have a 99,86 % probability
of measurement, y, (or concentration estimates, x) falling below the DL and have a 0,14 % probability of
being above the DL.
In the above definitions, comments on concentration are too speculative to make. Only distributions for
blank samples are drawn in Figures 2 and 3, whereas distributions for blank samples and samples of the
DL concentration, x , called DL samples, are referred to in Figures 4 and 5 (see the following subclause).
D
If DL samples, instead of blank samples, are measured repeatedly, the probability of observable y (or
estimable x) rising above the DL is equal to that of being below the DL. As long as a decision is made by
comparing y or x with y or x , the probability of detecting a target material in DL samples is 50 %. To
D D
achieve a higher probability of detection, e.g. 95 %, another criterion needs be introduced for judging
detection that is less than y or x .
D D
5.4 Detection limit with probabilities α and β
In this subclause, a detection limit is regarded as a target quantity of detection, but not a criterion for
judging whether or not a target material is detected in an analytical system. The presence of a material
in a sample is judged by a decision limit or critical value.
Figures 4 and 5 illustrate not only the distributions for blank samples shown in Figures 2 and 3, but
also those for DL samples. Probability β appends a new limit referred to as a critical value, y or x (=
C C
-1
f (y )), in the ISO 11843 terminology, which is also known as a decision limit in the fields of analytical
C
chemistry. If blank samples at a concentration of x (= 0) are measured repeatedly, the probability of a
measurement running over the decision limit, y , is α, which is called the probability of an error of the
C
first kind (false positive). The error and decision limit, α and x , in the X-axis can be interpreted in the
C
same manner. As for DL samples of concentration x , the probability of a measurable y or estimable x
D
failing to reach y or x is β, which is called the probability of an error of the second kind (false negative).
C C
Key
X net state variable or probability density Y response variable
1 detected 2 not detected
−1
3 y = f(x) X
D xf= y
()
DD
Figure 4 — Definition of detection limit in Y-axis with α and β
Key
X state variable Y probability density
1 detected 2 not detected
Figure 5 — Definition of detection limit in X-axis with α and β
Let α and β be both 5 %. It can safely be said that with a risk of at most 5 % (see Figures 4 and 5),
— if y < y or x < x , the analyte of the concentration x is not detected;
C C D
— if y ≥ y or x ≥ x , the analyte of the concentration x is detected.
C C D
If an analytical result is slightly more than the decision limit, y (or x ), a material of the DL concentration
C C
can be regarded as being detected under the assumptions that the probability of a response being
attributed to a blank sample is at most 5 % and that the probability of missing the presence of the
analyte is at most 5 %. If a result falls slightly below the decision limit, the probability of the wrong
decision (false negative) is less than 5 %. There is a possibility of detection, even if a measurement is
less than the detection limit (see Figures 4 and 5).
The detection limit is a target quantity for detection, and the decision limit is a criterion for judging
detection. The meaning of the term detection limit in this subclause is different from that of the
previous subclause, 5.3, in which a decision is made at y (or x ), i.e. the detection limit takes the same
D D
role as the decision limit.
The definitions of detection limits, y and x , with α and β are written as
D D
y = y + z σ(y ) + z σ(y) (3)
D 0 1−α 0 1−β D
x = x + z σ(x ) + z σ(x) (4)
D 0 1−α 0 1−β D
where σ(y ) and σ(x ) denote the standard deviations of measurements and their concentration
D D
estimates, respectively, for DL samples of concentration x , and the other symbols are the same as those
D
of Formulae (1) and (2). If the distributions with SDs of σ(y ) and σ(y ) (σ(x ) and σ(x )) are normal,
0 D 0 D
z = 1,645 and z = 1,645, the probabilities, α and β, are 5 %.
1−α 1−β
In general, σ(y ) and σ(y ) (σ(x ) and σ(x )) are not necessarily the same (heteroscedastic), but in many
0 D 0 D
real situations, they may not substantially be dissimilar (homoscedastic). In homoscedastic situations,
if z = z = z , Formulae (3) and (4) take the simple forms
1−α 1−β
y – y = 2zσ(y ) = 2zσ(y) (5)
D 0 0 D
x – x = 2zσ(x ) = 2zσ(x) (6)
D 0 0 D
respectively. If z = 1,645, σ = σ(y ) = σ(y ) and σ = σ(x ) = σ(x ), the DLs are written as
y 0 D x 0 D
y – y = 3,290σ ≈ 3,3σ (7)
D 0 y y
x – x = 3,290σ ≈ 3,3σ (8)
D 0 x x
Neglecting the blank measurement, y , and taking into account x = 0, the following useful expressions
0 0
are obtained:
σ / y = 1 / 3,290 ≈ 30 % (9)
y D
σ / x = 1 / 3,290 ≈ 30 % (10)
x D
where the left sides of Formulae (9) and (10) represent the relative standard deviations (RSD) or
coefficients of variation (CV) of measurements, y, and concentration estimates, x, respectively. The
above formulae are another indicator for detection: the DL samples are characterized by 30 % RSD or
CV.
A sample of an unknown concentration is usually handled for the purpose of DLs. When its measurement,
y, is obtained, consideration of its corresponding concentration is feasible with a calibration function,
but makes no sense from the viewpoints of DLs. The question is which state an analytical system
characterized by a measurement, y, belongs to, basic (blank) or reference (detection limit). Other
concentrations than the blank or DL are outside the purview of the definitions of DLs.
In the above definitions of detection limits (Formulae (1)-(4)), the distributions of measurements
and concentration estimates are assumed to be population distributions. In practice, however, the
population SDs of Formulae (1)-(4) need to be replaced by sample SDs. Parts 2 to 7 of the ISO 11843
series focus on methodologies for estimating sample SDs in real situations.
6 Pragmatic view of α and β
6.1 Statistical definitions of α and β
In the ISO 11843 series, α is defined as the probability of erroneously detecting that a system is not
in the basic state when really it is in that state, and β is the probability of erroneously detecting that
a system is in the basic state when the value of the state variable is equal to the minimum detectable
value.
More simply stated, a type I error corresponding to α is to falsely infer the existence of unacceptable
hazardous substances that are not there, while a type II error corresponding to β is to falsely infer the
absence of unacceptable hazardous substances that are present.
In other words, since α is the probability at which it is judged that an unacceptable hazardous substance
is present, it is necessary to decide what numerical values that prioritize the producer's interests should
be selected on the producer's side; the default value is set to 5 %.
Since β is the probability at which it is judged that an unacceptable hazardous substance is not present
even though one actually is, it is a risk factor considered on the basis of the user's side of goods such as
food and home electric appliances. As with α and β, 5 % is used in the ISO 11843 series as default value;
however, if there is an appropriate reason, the use of other values is permitted, and in the past 10 %
was widely used for α.
6.2 Actual examples of α and β values
The minimum detectable value defined by ISO 11843-4 is actually obtained in the most simplified
conditions shown as αβ==,,JK1 =∞ in Formulae (11) and (12).
Supposing that α is 0,10 and β is 0,05, the corresponding Z and Z values are 1,282 and 1,645,
1−α 1−β
respectively, the minimum detectable value is written as follows:
minimum detectable value = background value + 1,282 σ + 1,645 σ .
α β
Assuming σ = σ , the minimum detectable value = background value + 2,927σ . When the background
α β α
value is 0, it is approximately the minimum detectable value = 3σ .
α
Increasing the probability of α and β lowers the minimum detectable value, and it becomes easier
to detect even a smaller measurement value or a smaller measurement peak. At the same time, the
probability that the target substance is present when it is not present increase. In this way, the user
of the ISO 11843 series can understand the meaning of α and β and then select appropriate numerical
values for them.
7 In-depth explanations and examples of the Parts in the ISO 11843 series
7.1 General
The ISO 11843 series consists of Standards from Part 1 to Part 7. This Clause provides helpful
commentaries on ISO 11843-3, ISO 11843-4, ISO 11843-6 and ISO 11843-7.
7.2 ISO 11843-3 and ISO 11843-4
7.2.1 General
ISO 11843-3 defines how to determine the critical value. The critical value itself is not used alone but is
used in combination as a reference value when determining the minimum detectable value. Therefore,
the exposition of ISO 11843-3 together with ISO 11843-4 is appropriate.
7.2.2 Number of repeated measurements, J and K
When determining the critical value, not only J blank measurements but K actual sample measurements
are carried out in order to define a value calculated closer to the measurement result of the actual
sample. In the actual measurement of the sample, it is defined on the assumption that the measured
value is almost equal to the blank value.
As shown in ISO 11843-3 the critical value is defined by Formula (11).
yy=±z σ + (11)
Cb 1−α b
JK
where
J is the number of measurements on the reference material representing the value of the basic
state variable (blank sample); the basic state is defined as a state in which the target materi-
al to be measured does not exist in the sample;
K is the number of measurements on the actual state (test sample); for measurement of an
actual sample in which x = 0, see ISO 11843-3:2003, Annex A and Annex B.
The number of repeated measurements K is applied to an actual sample (test sample), but it is equivalent
to a blank sample with the concentration of the target component set to 0 (zero). Two types of blank
samples are measured J times and K times independently. The reliability of measurement is improved
by measuring the variation of blank samples for two types.
On the other hand, in ISO 11843-4, when the minimum detectable value is examined, the numbers of
repeated measurements J and K are used. The same variables J and K as those for obtaining the critical
value y are used. Originally, different variables are supposed to be used, and this can mislead the user.
c
However, when the critical value and the minimum detectable value are actually determined, it can be
seen that the definition formula is simplified by unifying the conditions such as J = K. The process is
shown below.
The minimum detectable value described in ISO 11843-4 is defined as the estimated value of the
response variable y by Formula (12) in the measurement of the actual sample
g
η  yz++σσ (12)
gC 1−β bg
JK
where
η is the expected value under actual performance conditions for responses of a sample with
g
the net state variable equal to x ;
g
J is the number of measurements on the reference material representing the value of the basic
state variable (blank sample);
K
is the number of measurements on the actual state (test sample).
By replacing y with yz=+ησ + , Formula (13) is obtained,
c Cb 1−α b
JK
11 11
η −+ησz ++z σσ (13)
gb 11−−αβbb g
JK JK
The formula σσ+ is adopted. Since the left-hand side is expressed as a difference of two kinds
bg
JK
22 2
of mean, the additivity of variance, σσ=+σ , is applied.
bg
For Formula (13), the formula is simplified by finally unifying the conditions. Specifically, by setting
α = β and K = J, even if K is adopted as another variable, the criterion is ultimately simplified and the
measurement is performed as shown in Formula (14). The number of repeated measurements can be
unified by J.
22 2
η −+ησz ()2 σσ+ (14)
gb 1−α bb g
J
7.2.3 Determination of the minimum detectable value
When analogue measurements are used, such as in spectrophotometry, it is possible to approximate
with σσ , so that Formula (14) is modified to Formula (15),
gb
2z
1−α 22
η −+ησ σ (15)
gb ( bg )
J
or modified to Formula (16),
ηη−
2z
gb
1−α
 (16)
J
σσ+
bg
Furthermore, in analogue measurement such as spectrophotometry, since it can be assumed that
σσ==, η 0 depending on the measurement conditions, Formula (16) can be simplified as shown in
bg b
Formula (17),
η ≥2z σ (17)
gb1−α
J
As described in the definitions of the critical value and the minimum detectable value, expressing the
definition formula using the number of repeated measurements, J and K, it is necessary to understand
the theoretical background, but in reality, those formulae are greatly simplified by various assumptions
so that this makes those calculations easier.
When obtaining the minimum detectable value, J = 1 is set in Formula (17) in order to set a state in which
a larger value cannot be taken because overly severe criteria can lead to a false result: that the target
substance is present when it is not. If α is 5 %, Z becomes 1,645, so Formula (17) can be simplified to
1-α
η (the minimum detectable value) = 4,65σ . Alternatively, when 10 % is adopted for α, Z is 1,282, so it
g b 1-α
can be simplified to η (the minimum detectable value) = 3,625σ .
g b
7.2.4 Confirmation of the minimum detectable value for an obtained experimental value with
the number of repeated measurements, N
In practice, the obtained value is often compared with the minimum detectable value. For this purpose,
it is consistent to use a formula representing a confidence interval of the difference in mean. This is
because the definition of the minimum detectable value adopts the form of the difference in mean for
simplification.
The number of measurements N is adopted to derive the confidence limit T for η −η in ISO 11843-4:2003,
0 g b
5.4 (Confirmation of the criterion for sufficient capability of detection). In both the measurement of the
basic state and the measurement of the actual state (test sample), the adoption of the same number of N
as the number of repeated measurements is recommended for simpler calculation.
As for the confidence interval on a difference in mean, y and y , a confidence interval of η −η is
g b gb
shown by Formula (18),
1 111
22 2 2
((yy−−) Z σσ+−≤η η≤ yy−+)+Z σ σσ (18)
gb 12−α/ bg gb gb (1-/α 2) b g
()
N N N N
where
N
is the number of measurements of each reference material in assessment of the capability of
detection;
y is the observed mean response of a sample with the state variable equal to x ;

g g
η
is the expected value under the actual performance conditions for the responses of the basic
b
state;
η
is the expected value under the actual performance conditions for the responses of a sample
g
with the state variable equal to x .
g
To confirm the sufficient capability of detection criterion, a one-sided test is used. With α and β ,
100(1−α ) % of the one-sided lower confidence bound on η −η is shown in Formula (19),
g b
1 1
η −≥ησ(yy−−)Z + σ (19)
gb gb (1-)α bg
N N
The one-sided lower confidence bound on η −η of Formula (19) is compared to the right-hand side of
g b
Formula (13), giving Formula (20),
1 111 11
22 2 2
ηη− ≥−(yy )Z−+σσ ≥+Z ασ+ Z + σ (20)
()
gb gb (1-)ααbg 1- bb1−β g
N NKJ J KK
The 100(1-α) % value of the confidence limit T with respect to η −η can be calculated by Formula (21).
0 g b
z
1−α
()
Ty=− y −+σσ (21)
()
0 gb bg
N
When using the standard deviation of the sample, the t-distribution with the degree of freedom v is
used to calculate the value using Formula (22),
t
()v
1−γ
()
Ty=− y −+ss (22)
()
0 gb bg
N
or Formula (23),
t
()v
1−γ
yy−
() ()
gb
′
T = − (23)
N
ss+
bg
If the left side of the equality satisfies Formulae (14) or (15), it can be concluded that the minimum
detectable response value y is more than or equal to the minimum detectable response value y and
g D
that x is therefore more than or equal to x .
g D
An actual calculation example described in ISO 11843-4:2003, Annex B is shown in Annex C to help
understanding.
7.2.5 Number of repeated measurements, J and K, in ISO 11843-5 and ISO 11843-7
In ISO 11843-5 and ISO 11843-7, the number of repeated measurements such as J and K is not applied,
since these Parts are based on the theory of o
...