1	Uncertainty of Measurement
2	Presentation Summary What ISO 17025 requires regarding Uncertainty of Measurement How to measure uncertainty of measurement When to measure uncertainty of measurement What to do about reporting and recording UM
3	ISO 17025 and UM Clause 5.4.6.2 requires that: ... Testing laboratories shall have and shall apply procedures for estimating uncertainty of measurement
4	Quantitative data UM only applies to quantitative data
5	UM and Records Clause 4.13.2.1 requires that: ... The records for each test or calibration shall contain sufficient information to facilitate, if possible, identification of factors affecting the uncertainty ...
6	UM and General Validation Clause 5.4.1 requires that ...(when validating a method the laboratory shall include) ... where appropriate, an estimation of the measurement uncertainty ...
7	UM and Non Standard Methods Clause 5.4.4 requires that... ...For new test and/or calibration methods, procedures should be developed prior to the tests and/or calibrations being performed and should contain at least the following information: ... k) the uncertainty or the procedure for estimating uncertainty.
8	UM and Performance of Methods Clause 5.4.5.2 requires the laboratory to confirm that the methods are fit for the intended use and states in NOTE 2 that the techniques used for the determination of the performance of a method should be one of, or a combination of, the following: ... assessment of the uncertainty of the results based on scientific understanding of the theoretical principles of the method and practical experience.
9	UM and calibrations in the testing laboratory Clause 5.4.6.1 requires that a calibration laboratory, or a testing laboratory performing its own calibrations, shall have and shall apply a procedure to estimate the uncertainty of measurement for all calibrations and types of calibrations. Best advice on this one: DON'T GO THERE
10	Measuring UM Direct measurement Mean and SD from repeatability studies during validation Control charts Estimation Experience and conservative distributions Uncertainty budgets
11	Indirect measurement Clause 5.4.6.2 states that in certain cases the nature of the test method may preclude rigorous, metrologically and statistically valid, calculation of uncertainty of measurement. In these cases the laboratory shall at least attempt to identify all the components of uncertainty and make a reasonable estimation
12	UM estimation and reporting Clause 5.4.6.2 goes on to state that the laboratory: ...shall ensure that the form of reporting of the result does not give a wrong impression of the uncertainty. ...Reasonable estimation shall be based on knowledge of the performance of the method and on the measurement scope and shall make use of, for example, previous experience and validation data.
13	Degree of rigor - Factors Clause 5.4.6.2 ends with NOTES NOTE 1 The degree of rigor needed in an estimation of uncertainty of measurement depends on factors such as: the requirements of the test method; the requirements of the customer; the existence of narrow limits on which decisions on conformity to a specification are based.
14	Degree of rigor – Standard methods NOTE 2 to clause 5.4.6.2 states: ... In those cases where a well-recognized test method specifies limits to the values of the major sources of uncertainty of measurement ... the laboratory is considered to have satisfied this clause by following the test method
15	UM estimations and components Clause 5.4.6.3 requires that when estimating the uncertainty of measurement, all uncertainty components which are of importance in the given situation shall be taken into account using appropriate methods of analysis.
16	UM and reporting Clause 5.10.3.1 requires that test reports shall, where necessary for the interpretation of the test results, include the following: ... c) where applicable, a statement on the estimated uncertainty of measurement; information on uncertainty is needed in test reports when it is relevant to the validity or application of the test results, when a customer's instruction so requires, or when the uncertainty affects compliance to a specification limit;
17	Reports and records ILAC Guide 19 states that the reporting requirements of ISO 17025 clause 10 can be met by ensuring that the case record relating to a specific investigation contains all the relevant information required by ISO/ IEC 17025.
18	The Nitty-Gritty UM definition Direct measurement Estimation based on professional judgement Uncertainty budgets
19	UM Definition A parameter associated with the result of a measurement, that characterizes the dispersion of the values that could reasonably be attributed to the measurand
20	Direct Measurement Replicate measurements on a typical sample Concentration range Operators and time influences Matrix effects Replicate measurements on spiked samples Shewhart control charts Calculate standard deviation assuming a normal distribution of results This is a direct measure of dispersion This is known as a Type A estimate of UM
21	Indirect Measurement Use professional judgement to set a reasonable range about the observed value The expanded uncertainty, U Set the acceptable degree of latitude around the mean The coverage factor, k Define distribution nature and calculate standard deviation This is called a Type B estimate
22	Distribution and estimating UM If we do not know how the results are distributed around the mean and have no reason to assume any bias for example a clustering around the mean, then the measurement dispersion (standard deviation) to be used is: Half width interval /√3 This distribution is sometimes referred to as a rectangular distribution
23	Estimating UM (continued) If we can reasonably assume that there is a clustering around the mean, then the measurement dispersion (standard deviation) to be used is: Half-width interval/√6
24	Type A Example 30 replicate analyses of a blood sample containing approximately 0.09 g/dl of ethanol gave a standard deviation of 0.0018 g/dl This establishes the UM for that test as 0.0018 For k = 99.5% we use three times the standard deviation Therefore 99.5% of values for repeated assays would lie in the range mean +/- 0.0054 g/dl Note that the 0.5% not included is made up of 0.25% above the (mean + 3 SD) and 0.25% below the (mean – 3 SD)
25	Type B example A firearms examiner conducts a distance estimation by measuring the diameter of residue deposition, and concludes that “the gun was fired from 18 inches give or take 6 inches either way” This translates as a mean of 18 inches and a half width interval of 6 inches Measurement dispersion (standard deviation) is 6 divided by the square root of 6 if we assume results are clustered around the mean or 6/√3 if we assume a rectangular distribution
26	Example results Rectangular distribution: Standard Deviation = 6/√3 = 6/1.73 = 3.47 Triangular distribution: Standard Deviation = 6/√6 = 6/2.45 = 2.45
27	Apply k We know the relationship between standard deviation and dispersion: 67% of values lie in the range mean +/- 1SD 95% of values lie in the range mean +/- 2 SD 99.7% of values lie in the range mean +/- 3 SD
28	Decide on desired coverage Say we wish to have 95% coverage and assume a triangular distribution: 95% of derived distance values will lie in the range 18 plus or minus 2 times 2.45 That is, between 13.1 and 22.9 inches Note that this is not the same as saying there is 95% probability that the gun was fired from between 13.1 and 22.9 inches
29	Uncertainty budget Identify all contributing factors Calculate variances Add variances Calculate dispersion from square root of the sum of variances
30	When to report UM You MUST have a record of UM for every quantitative measurement that you will report and that is either required by the customer or where a decision on conformity will be made based on your reported measurements
31	When to measure UM Remember that Clause 5.4.6.2 requires that: ... Testing laboratories shall have and shall apply procedures for estimating uncertainty of measurement All quantitative method validations must include a UM estimate
32	Summary ISO 17025 has several requirements regarding uncertainty of measurement for quantitative testing One of these is that you must have and apply procedures for estimating UM for all your quantitative procedures You do not need to report UM in every case