Accuracy(trueness and precision) of measurement methods and results - Part 6: Use in practice of accuracy values
Some standard content:
ICS 03. 120.30
National Standard of the People's Republic of China
GB/T 6379.6—2009/ISO5725-6:1994 Accuracy (trueness and precision) of measurement methods and results-Part 6, Use in practice of accuracy values(ISO 5725-6:1994,IDT)
Published on March 13, 2009
Digital Anti-Counterfeiting
General Administration of Quality Supervision, Inspection and Quarantine of the People's Republic of China Administration of Standardization of the People's Republic of China
Implemented on September 1, 2009
GB/T 6379.6—2009/1SO 5725-6:1994 Pre-regulatory
Normative references
Terms and definitions
4 Determination of limits
4.1 Repeatability limit and reproducibility limit
4.2 Comparison based on more than two values
5 Methods for checking the acceptability of test results and determining the final reported results 5.1 General
5.2 Methods for checking the acceptability of test results obtained under repeatability conditions 5.3 Methods for checking the acceptability of test results obtained under reproducibility conditions 6 Methods for checking the stability of test results in the laboratory 6.1 Background
6.2 Methods for checking stability
7 Application of repeatability standard deviation and reproducibility standard deviation in laboratory assessment - 7.I Assessment methods
7.2 Assessment by non-assessed 7.3 Reassessment of accredited laboratories 8 Comparison with alternative measurement methods 8.1 Reasons for considering alternative measurement methods 8.2 Purpose of comparing measurement methods 8.3 Method B: Alternative standard method as a candidate (\standardized test\ not established) 8.4 Validity test -
8.5 Method B as a candidate conventional method Appendix A (Normative Appendix) Symbols and abbreviations used in GB/T 6379 10
GB/T 6379.6—2009/ISO 5725-6:1994 GB/T 6379 "Accuracy (trueness and precision) of measurement methods and results" is divided into the following six parts. Its structure and corresponding international standards are: - Part 1: General principles and definitions (ISO 5725-6:1994) 5725-1:1994, IDT); - Part 2: Basic methods for determining repeatability and reproducibility of standard inference methods (ISO5725-2:1994 IDT); - Part 3: Intermediate measures of precision of standard measurement methods (ISO5725-3:1994, IDT); - Part 4: Basic methods for determining the correctness of standard measurement methods (ISO5725-4:1994, IDT); - Part 5: Alternative methods for determining the precision of standard measurement methods (ISO5725-5:1998, IDT) - Part 6: Practical application of accuracy values (ISO5725-6:1994, IDT). This part is Part 6 of GB/T6379.
This part is equivalent to the international standard ISO5725-6:1994 Accuracy (Trueness and Precision) of measurement methods and results Part 6: Practical application of accuracy values and the technical amendment to the 1994 version of ISO5725-6 issued by ISO on October 15, 2001. The following corrections have been made to the errors of ISO5725-6:1994: Figure 10 has been redrawn based on the calculation results; the calculation error of z in the original Table 6, totaling 16.84, has been changed to 16.74. GB/T6379 Part 1 to Part 6 replace GB/T6379-1986 and GB/T11792-1989 as a whole. The original concept of precision has been expanded in the standard, and the concept of trueness has been added, collectively referred to as accuracy. In addition to the repeatability condition and the reproducibility condition, the intermediate precision condition has been added.
The contents of this part partially replace GB/T6379-1986 and GB/T11792--1989. Appendix A of this part is a normative appendix.
This part was proposed by Guangdong Entry-Exit Inspection and Quarantine Bureau. This part is under the jurisdiction of the National Technical Committee for Standardization of Statistical Method Application. Drafting units of this part: Guangdong Entry-Exit Inspection and Quarantine Bureau, Institute of Mathematics and Systems Science of the Chinese Academy of Sciences, China National Institute of Standardization. Main drafters of this part: Li Chengming, Feng Shiyong, Zhang Kun, Jiang Jian, Zhou Qi, Ding Wenxing, Song Wuyuan, Yu Zhenfan, Li Zhengjun, Xiao Hui, Liu Jianbin, Chen Yuzhong.
This part was first published in 2009.
GB/T6379.6--2009/ISO5725-6:1994 Introduction
0.1GB/T6379 uses two terms, "trueness" and "precision" to describe the accuracy of a measurement method. Trueness refers to the degree of agreement between the (arithmetic) mean of a large number of test results and the true value or accepted reference value: while precision refers to the degree of agreement between test results. 0.2 The reason for considering precision is mainly because it is assumed that the same - or the same material (substance/material) is tested under the same conditions, and generally the same results will not be obtained. This is mainly because random errors are inevitable in each measurement procedure, and those factors that affect the measurement results cannot be completely controlled. In the actual interpretation of the measurement data, This variation must be taken into account in the process. For example, the difference between the test result and the specified value may be within the range of unavoidable random errors. In this case, the true deviation between the test value and the specified value cannot be determined. Similarly, when comparing the test results of two batches of materials, if the difference between them comes from inherent changes in the measurement procedure, the essential difference between the two batches of materials cannot be revealed. 0.3 Parts 1 to 5 of GB/T 6379 discuss the background of the assessment of density (expressed by the standard deviation of repeatability and the standard deviation of reproducibility) and correctness (expressed by the components of bias), and give some methods for assessing precision and correctness using the retrospective test results obtained from a standard measurement method. However, if the assessment results cannot be used in practice: these assessment methods lose their meaning, 0.4 Measurement method The accuracy of the method - and determine it. This part of GB/T6379 will use this knowledge to facilitate business and trade. It is also used to monitor and improve laboratory operations.
1 Scope
GB/T6379.6--2009/ISO5725-6:1994 Accuracy of measurement methods and results (trueness and precision)
Part 6: Practical application of accuracy values
1.1 The purpose of this part of GB/T6379 is to describe various practical situations in which accuracy data can be applied: to give standard methods for calculating repeatability limits, reproducibility limits and other limits, which will be used to check the test results obtained using standard measurement methods; to propose the application of repeatability to the test results; c) describe how to assess the stability of a laboratory's test results over a period of time, thereby providing a "quality control" method for laboratory operations; d) describe how to assess whether a particular laboratory has the ability to correctly use a given standard measurement method; e) describe how to compare alternative measurement methods. 1.2 The measurement methods involved in this part are specifically those that measure continuous quantities and take only one measurement value as the test result each time, although this value may be the result of a calculation of a set of observations. 1.3 This part assumes that estimates of the trueness and precision of the measurement method have been obtained in accordance with the methods of Parts 1 to 5 of GB/T 6379. 1.4 Any additional information on the use scenario should be given at the beginning of each specific use. 2 Normative references The provisions of the following documents become provisions of this part through reference to this part of GB/T 6379. For all referenced documents with dates, all subsequent amendments (excluding errata) or revised versions are not applicable to this part. However, parties that reach an agreement based on this part are encouraged to study whether the latest versions of these documents can be used. For all referenced documents without dates, the latest versions apply to this part.
GB/T3358.1-1993 Statistical terminology Part 1 General statistical terminology GB/T4091-2001 Conventional control charts (ISO8258:1991, IDT) GB/T6379.GB/T6379.1-2004 Accuracy (Trueness and Precision) of Measurement Methods and Results Part 1: General Principles and Definitions (ISO 5725-1.1994, IDT) GB/T6379.2-2004 Accuracy (Trueness and Precision) of Measurement Methods and Results Part 2: Basic Methods for Determining the Repeatability and Reproducibility of Standard Measurement Methods (IS05725-2:1994, IDT) GB/T6379.4-2006 Accuracy (Trueness and Precision) of Measurement Methods and Results Part 4: Basic Methods for Determining the Trueness of Standard Measurement Methods (ISD5725-4:1994, IDT) GB/T6379.5-2006 Accuracy (Trueness and Precision) of Measurement Methods and Results Part 5: Basic Methods for Determining the Trueness of Standard Measurement Methods (ISD5725-4:1994, IDT) Density) Part 5: Alternative methods for determining the precision of standard measurement methods (ISO5725-5: 1998, IDT) GB/T27025 General requirements for the competence of testing and calibration laboratories (GB/T27025-2008, ISO/1EC17025: 2005, IDT
ISO3534-1: 1993 Statistical vocabulary and symbols Part 1: Probability and general statistical terms IS05725-3: 1994 Accuracy (trueness and precision) of measurement methods and results Part 3: Intermediate measures of precision of standard measurement methods
CB/T 6379.6-2009/1S0 5725-6:1994 JSO Guide 33:1989 Use of certified reference materials (standard materials) 1S0 Guide 35:1989 General principles and statistical principles for the determination of values of reference materials (standard materials) 3 Terms and definitions
The terms and definitions given in GB/T3358.1 and GB/T6379.1 are still applicable in this part of CB/T6379. The symbols used in GB/T6379 are given in Appendix A. 4 Determination of limits
4.1 Repeatability limit and reproducibility limit
4.1.1 GB/T6379.2 mainly studies the estimation of various standard deviations of measurements made under repeatability conditions and reproducibility conditions. However, in normal laboratory work, it is often required to check the difference between the observed values of two (or more) test results. For this purpose, it is necessary to determine Some measure like critical difference, not just standard deviation, can be defined. 4.1.2 If an estimator is the sum or difference of n independent estimators, and the standard deviation of each estimator is , the standard deviation of the sum or difference is . The reproducibility limit R and the repeatability limit are both the difference between two test results, so the corresponding standard deviation is V2. In routine statistical work, in order to check the difference between two test results, 10 times of this standard deviation is often used as the critical difference. The value of the critical difference coefficient 10 depends on the probability level corresponding to the critical difference and the distribution of the test results. For the repeatability limit and the reproducibility limit, the probability level is specified as 95%. In all analyses of GB/T 6379, it is assumed that the basic distribution is approximately normal. For a normal distribution, at a probability level of 95%, f=1.96, so fV2=2.77. The goal of this part of GB/T 6379 is to give some simple rules of thumb for use by non-statistical experts, and v2 will be revised to 2.8.
4.1.3 As mentioned above, when the true value of the standard deviation is unknown, the result of the estimated precision gives an estimate of the standard deviation. In statistical practice, the estimate of the standard deviation is denoted by s, not α. According to the procedures given in GB/T 6379,1 and GB/T 6379,2, these estimates are based on a certain number of test results and give us the best possible information about the true value of the standard deviation. Therefore, in other applications in the future, 5 is denoted as the estimated value of the standard deviation based on relatively limited test results, and α is denoted as the value obtained from the complete precision test, and this value is used as the true value of the standard deviation for comparison with other estimates. 4.1. From 4.1.1 to 4.1.3, it is known that when two single test results obtained under repeatability conditions or reproducibility conditions are tested, they should be compared with the repeatability limit r = 2.8g, or the reproducibility limit R2.8g. 4.2 Comparisons based on more than two results
4.2.1 Comparison of two groups of test results in one laboratory In a laboratory, if two groups of measurements are carried out under repeatability conditions, the number of test results in the first group is n1 and its arithmetic mean is n2; the number of test results in the second group is n2 and its arithmetic mean is n2, then the standard deviation of (3:-2) is: (l+l
erlnng
At a probability level of 95%, the critical difference of one is: CD.oe = 2.80-~/2n +2n2
Note 1: If n-nz = 1, the above critical difference is simplified to 7 = 2. 8g,. 4.2.2 Comparison of two groups of results in two tests If under repeatability conditions, the number of test results of the first laboratory is n1, and its arithmetic mean is 1; the number of test results of the second laboratory is n, and its arithmetic mean is 2; then the standard deviation of (a) is: a+?+1g
/20+(1+1)
GB/T 6379.6—2009/IS0 5725-6:1994/2(+0)-2g(1-2-2)
At a probability level of 95%, the critical difference of i31-32 is: /2. 80r)-(2.80,)(1-2m2m)
Note 2: If m1 is nz=1, the above critical difference is simplified to R=2.80g. 4.2.3 Comparison of a laboratory's test results with reference values If a laboratory obtains n test results under repeatability conditions, and its arithmetic mean is, then it should be compared with a certain reference value\. In the case where the biased laboratory component has not been determined, the standard deviation (for example) is: o+
2(0+)-20(1)
/2(0. +)-20(\=1
At a probability of 95%, the critical difference of\ is: CD./(2. 8x)(2. 8g,)
4.2.4 Comparison of test results of multiple laboratories with reference values If there are n (i=1, 2, \, p) test results obtained by a number of laboratories under repeatability conditions, the arithmetic mean of the test results of each laboratory is and, the total average of the test results of all laboratories is: 5=
Comparing the total average with the reference value, the standard deviation is: Foi+
2(+)2+21
/2(a+)—28 (1-)
Therefore, at a probability level of 95%, the critical difference of and is: 4.2.5 Report the comparison results
/(2. 80)(2. 80)(1-
If the absolute difference of the test results exceeds the reasonable limits given in the above clauses, the corresponding absolute difference should be considered suspicious. At this time, all measurements used to calculate this absolute difference should be considered suspicious and require further review. 5 Methods for checking the acceptability of test results and determining the final reported results 5.1 General
5.1.1 The inspection method described in this chapter is only applicable to situations where the measurement method used has been standardized and the repeatability standard deviation and reproducibility standard deviation are known. Therefore, when the range of N test results exceeds the reasonable limits given in Chapter 4, one, two, or all of the N test results are considered abnormal. It is recommended to find the cause of the abnormality from a technical point of view. However, it may be necessary to obtain a certain acceptable value for commercial reasons. At this time, the test results should be processed according to the methods specified in this chapter. 5.1.2 This clause assumes that all test results were obtained under repeatability or reproducibility conditions and that a probability level of 95% is used in the calculations. If the test results were obtained under intermediate precision conditions (see ISO 5725-3:1994), the corresponding intermediate precision value shall be used in place of a.
5.1.3 In some cases, the method described in 5.2 will result in the median of the test results being reported as the final result. These data are best discarded.
5.2 Methods for checking the acceptability of test results obtained under conditions of repeatability NOTE 3 In 5.2.2.1 and 5.2.2.2, the low or high cost of a measurement refers not only to the cost, but also to the complexity of the measurement, the difficulty of execution and the timeliness of the measurement.
5.2.1 Single test result
It is not common to have only one test result in commodity testing. When only one test result is available, it is not possible to immediately perform a statistical test of the acceptability of a particular repeatability measure. If there is any doubt about the validity of the test results, a second test result should be obtained. The following describes the more common cycle with two test results. 5.2.2 Two test results
Both test results should be obtained under repeatability conditions, and the absolute value of the difference between the test results should be compared with the repeatability limit r = 2.8g. 5.2.2.1 Cases with low test costs
If the absolute value of the difference between the two test results is not greater than r, the two test results can be accepted. The final report result is the arithmetic mean of the two test results. If the absolute value of the difference between the two test results is greater than r, the laboratory should take two more test results. At this time, if the range of the four test results (zm-min) is equal to or less than the critical range CR..9s (4) with a probability level of 95% when n = 4, the arithmetic mean of these four test results is taken as the final report result. The critical range is calculated as follows:
CRa(n) - f(n)a,
where f(n) is called the critical range coefficient. Table 1 lists the values of some critical range coefficients for n from 2 to 40 and from 45 to 100.
If the range of the four test results is greater than the critical range of repeatability, the median of the four test results is taken as the final report result. The above process can be represented by the flow chart in Figure 1. 5.2.2.2 Case of high test cost
If the absolute value of the difference between two test results is not greater than r, the two test results can be accepted. The final report result is the arithmetic mean of the two test results. If the absolute value of the difference between the two test results is not greater than r, the laboratory should take another test result. At this time, if the range of the three test results (=-m) is equal to or less than the critical range CR with a probability level of 95% when n = 3. .9§5 (3), the average of the three test results is taken as the final report result. If the range of the three test results is greater than the critical range CR.9s (3), the final report result is determined by one of the following two situations. a) The situation where it is impossible to obtain the fourth test result: the laboratory declares that the median is taken as the final report result. This process can be illustrated by the hidden diagram in Figure 2.
b) It is possible to obtain the fourth test result: The laboratory should take the fourth test result. If the range (-) of the four test results is equal to or less than the critical range CRa,9s (4), the arithmetic mean of the four test results is taken as the final report result. If the range is greater than CR..5 (4), the median of the four test results is taken as the final report result. This error can be represented by the process diagram in Figure 3.
GB/T 6379.6-2009/ISO 5725-6:1994 Table 1 Critical range coefficient (n)
Note: The critical range coefficient F(n) is the 95% quantile of the distribution of (zm-Iein)/a, where Tal and Tal are the maximum and minimum values of the sample size \ drawn from a normal population with standard deviation \. Starting from two results
I --≤r
2 more results
Tmu—Zmi≤CRo.g (4)
a is the final report result
where, is the second smallest test result, is the third smallest test result + sa
is the most complete report result
is the final report result
Figure 1 Check method for acceptability of test results obtained under repeatability conditions (when starting from two test results and the test cost is low: Case 5.2.2.1)5
GB/T 6379.6-2009/ISO 5725-6:1994 Starting from two results
1 more result
mn≤CRo.gs(3)
Z(2) is the final report result
where (2) The second smallest test result is the final report result
is the final report result
Figure 2 Check method for the acceptability of test results obtained under repeatability conditions (starting from two test results and the test cost is high: Case 5.2.2.2a)) Start from two results
zt\cal≤r
then 1 result
then another result
nnSCR(4)
is the final report result
where: (is the second smallest test result, (a) is the third smallest test result +
is the final report result
is the final report result
is the most complete report result
Figure 3 Check method for the acceptability of test results obtained under repeatability conditions (starting from two test results and the test cost is high: Case 5.2.2.2a)) Start from two results
zt\cal≤r
then 1 result
then another result
nnSCR(4)
is the final report result
where: (is the second smallest test result, (a) is the third smallest test result +
is the final report result
is the final report result
is the most complete report result
2b))6
5.2.3 Starting from more than two results
GB/T 6379,6-2009/1SO 5725-6: 1994In practice, it is common for the number of initial results to be greater than 2. Under repeatability conditions, the method for determining the final report result when n>2 is similar to the method when n=2.
Compare the range of the n results (zmx without ain) with the critical range CR.ss(n) calculated according to Table 1: If the range is equal to or less than the critical range, take the arithmetic mean of the n results as the final report result. If the range is greater than the critical range, the final report result is determined by one of the three cases A, B, and C shown in Figures 4 to 6 below. Case A and Case B correspond to the cases with lower and higher test costs, respectively. Case C is optional and is recommended for n≥5 and lower test costs, or n≥4 and higher test costs. For the case of low test cost, the difference between case A and case C is that case A requires n additional measurements, while case C only requires less than n/2 additional measurements. Which case determines the final report result depends on the size of n and the difficulty of each measurement.
For the case of high test cost, the difference between case B and case C is that case C requires additional measurements, while case B does not require additional measurements; when the cost of additional measurements is very high and therefore impossible, case B is the only option. Starting from n results
The range of 1 result
SCRaas(n)
Test n more results
The range of 2n results
CRa,ss(2n)
The arithmetic meanbzxz.net
mean of all 1 result is the final reported result
Yes, the arithmetic mean
mean of all 2n results is the final reported result
The median of all 2n results is the final reported result Figure 4 Method for checking the acceptability of test results obtained under repeatability conditions (when starting from n test results and the test cost is low: Case A) Starting from n results
The range of n results
≤CRo,s (n)
The median of all n results is the final reported result. The arithmetic mean of all n results is the final reported result.
Figure 5 Method for checking the acceptability of test results obtained under repeatability conditions (starting from n test results and when the test cost is high: Case B)7
GB/ 6379.6--2009/1S05725-6.1994Starting from n results
The maximum cost of n results
SCRa.s (n)
next m results
the range of (n+m) results
$CRg(n+m)
the arithmetic mean of all n results
the mean is the most complete report result
the median of all (+m) results is the final report result1) m should be selected as an integer that satisfies the condition n/3≤m≤a/2 Figure 6 Method for checking the acceptability of test results obtained under childlike conditions (starting from n test results and when the test width is high: Case C) 5.2.4 Example of Case B. In complex and time-consuming chemical analyses, expensive situations are often encountered, often requiring 2, 3 or more days to complete an analysis. If technically questionable data or outliers are found in the first analysis, it will be time-consuming and expensive to reanalyze. Therefore, usually 3 or 4 test results are obtained under repetitive conditions at the beginning, and then analyzed according to the procedure of Case B. See Figure 5.
For example, when determining the gold and silver content of ore by fire assay, although there are many methods available, all methods require expensive special equipment, highly skilled operators and considerable time, usually about 2 days. When the ore contains platinum group elements or other co-existing gold dust, it takes more time to complete a complete measurement and analysis process. The following are 4 test results of gold content in fine steel ore obtained under repetitive conditions, and these test results are processed by the method of Case B:
Gold content (g/t): 11.011,010.810.5 Currently there is no international standard for determining the gold and silver content in ore. However, when a,=0.12g/t is given, f(4)=3.6 is obtained from Table 1, and the corresponding critical range is:
CR.95(4)=3.6X0,12=0.438/t Because the range of the above four test results is: 11.0-10.5=0.5g/t, which is greater than the critical range, the final reported result is the median of the four results, namely:
5.2.5 Explanation on sugar density test
11.0±10.8=10.9 g/t
If the results frequently exceed the critical value when using the method described in 5.2.2 or 5.2.3, the precision of the laboratory's measurement method and/or precision test should be investigated.
5.2.6 Final report nesting
If only the final test results are required, the following two points should be explained: 1. The number of test results used to calculate the final report test nesting
Tip: This standard content only shows part of the intercepted content of the complete standard. If you need the complete standard, please go to the top to download the complete standard document for free.