ISO 11843-2:2000 pdf download

ISO 11843-2:2000 pdf download.Capability of detection 一Part 2:
Methodology in the linear calibration case.
If, for a particular measurement process, in consecutive calibrations have been carried out in order to determine the minimum detectable value of the net state variable -1d the in minimum detectable values .vdl can be used to determine a minimum detectable value of the measurement process under the following conditions:
a) the measurement process is not changed;
b) the distribution of the values xd is unimodal and there are no outlying values xd,
c) the experimental design (including the number of reference states, I, and the numbers of replications of procedure, J, K and L) was identical for each of the calibrations.
Under these conditions the median of the values xdl,is recommended as the minimum detectable value of the measurement process; if another summary statistic of the values Xdi is used instead of the median, the statistic used shall be reported.
If any of these conditions are violated, the minimum detectable value of the measurement process is not sufficiently well-defined and the determination of a common value shall not be attempted.
If the same measurement method is applied in p laboratories and for each of them a minimum detectable value of the measurement process within the laboratory were to be determined, then under the same conditions as for the determination of the minimum detectable value of the measurement process, the median of the p minimum detectable values of the laboratories is recommended as the minimum detectable value of the measurement method; if another summary statistic of the minimum detectable values of the laboratories is used instead of the median, the statistic used shall be reported.
7 Reporting and use of results
NOTE Examples of the determination of critical and minimal detectable values are given in annex C.
7.1 Critical values
For decisions regarding the investigation of actual states only the critical value of the net state variable or of the response variable is to be applied. These values derived from a calibration of the measurement process are decision limits to be used to assess the unknown states of systems included in this series. Looking at consecutive calibrations of the same measurement process, the critical values may vary from one calibration to another. However, since each of the critical values is a decision limit belonging to a particular measurement series, it is meaningless to calculate overall critical values across calibrations and logically inappropriate to use these as critical values.
If a value of the net state variable or of the response variable is not greater than the critical value, it can be stated that no difference can be shown between the observed actual state and the basic state. However, due to the possibility of committing an error of the second kind, this value should not be construed as demonstrating that the observed system definitely is in its basic state. Therefore, reporting such a result as “zero” or as “smaller than the minimum detectable value” is not permissible. The value (and its uncertainty) should always be reported; if it does not exceed the critical value, the comment “not detected” should be added.
7.2 Minimum detectable values
The minimum detectable value derived from a particular calibration shows whether the capability of detection of the actual measurement process is sufficient for the intended purpose. If it is not, the number J, K or L may be modified.
A minimum detectable value derived from a set of calibrations following the conditions mentioned in clause 6 may serve for the comparison, the choice or the judgement of different laboratories or methods, respectively.
probability of erroneously accepting the null hypothesis “the state under consideration isnot different from the basic state with respect to the state variable”for each of theobserved actual states in the measurement series for which the net state variable isequal to the minimum detectable value to be determined (probability of the error of thesecond kind).