INTERNATIONAL « Iso STANDARD 5725-2 First edition 1994-12-15, Accuracy (trueness and precision) of measurement methods and results — Part 2: Basic method for the determination of repeatability and reproducibility of a standard measurement method Exactitude (justesse et fidéite) des résultats et méthodes de mesure — Partie 2: Méthode de bese pour la détermination de la répétabilité et de la reproductibilté d'une méthode de mesure normalisée Reference nur ISO 5725251951SO 5725-2:1994(E) Contents Page 1 Scope Bee 2 Normative t6f2r2NG85 neuron al 3. Definitions 2 4 Estimates of the parameters in the basic model mr 5 Requirements for a precision experiment z 5.1 Layout of the experiment a cnnnnnnne 2 5.2 Recruitment of the laboratories 3 5.3 Preperation of the materials 3 6 Personnel involved in a precision experiment. ours 6.1 Panel . eee 6.2 Statistical functions seve B 63 Executive functions 5 64 Supervisors a 5 BB Operators renee Bee Ete 7 Statistical analysis of a precision experiment oneanines 6 7.1. Preliminary considerations z 6 7.2. Tabulation of the results and notation used 7.3 Scrutiny of results for consistency and outliers: 9 7.4 Calculation of the general mean and variances oI 13 75 Establishing a functional tae between precision values and the mean level ma. vee 14 7.6 Statistical analysis as o step-by-step procedure an. 16 7.7 The report to, and the decisions to be taken by, the panel 20 8 Statistical tables AAAAARAAD a © ISO. 1998 [Aight reserved. Unless otherwise specified, no part of his publication may be reproduced ‘ot uized in any form or by any means, eleciron o* mechanical, inchxSing photocopying and ‘ierofim, without permission in writing from the publishes, {ternational Orgeization for Standardizstion Case Postale 86 » CH-1211 Genave 20 * Switeerland Printed in Swtzerond1s ISO 5725-2:1994(E) Annexes A Symbols and abbreviations used in ISO 5725 senses 2B. B Examples of the statistical analysis of precision experiments 27 B.1_ Example 1: Determination of the sulfur contant of coal Several levels with no missing or outlying data) nn. = 27 B.2_ Example 2: Softening point of pitch (Several levels with missing data) 32 B3 Example 3: Thermometric tration of creosote oil (Several levels with outing data) 38 © Bibliogrephy iit1SO 5725-2:1994(E) Iso Foreword ISO the International Organization for Steridardization) 1s @ worldwide federation of national standards bodies (|SO member bodies). The ware of Preparing Intemational Standards is normally carried out through 1S0) technical committees. Each member body interested in a subject. for which @ technical committee has been established has the tight to be Fepresented on that committee. International uryanizations, governmental and nongovernmental, in liaison with ISO, also take part in the work ISO collaborates closely with the International Electrotechnical Commission (FEC) on ali matters of electrotechnical standardization, Draft Intemational Standards aclopted by the technical committees a1 Grrculated to the member bodies for voting, Publication as an international Standard requires approval by at least 75 % of the member bodies casting a vote. {atgrnational Standard 1SO 6726-2 was prepared by Technical Cammittoo SOME 69. Applications of statistical methods, Subcommittee SC 6, Measurement methods and results, [SO 8725 consists of the following parts, under the general tie Accuracy {trueness and precision) of measurement methods and results — Part 1: General principles and definitions — Part 2: Basic method for the determination of repeatability and re- Producibility of a standard measurement method — Fart 3: Intermediate measures of the precision of a standard ‘measurement method — Port 4: Basic methods for the determination of the tueness of a Standard measures (rethiod — Part §: Alternative methods for the determination of the precision of a standard measurement method — Part 6: Use in practice of accuracy values Parts I to 6 of ISO 5725 together cancel and replace ISO 5725:1986, which hes been extended to cover trueness {in addition to precision) and intermediate precision conditions (in addition to repeatability and repro- duclbility conditions) ‘Annex A forms an integral part of this part of ISO 5725. Annexes B and C are for information only.elso ISO 5725-2:1994(E) Introduction 0.1 180 5725 uses two terms “trueness” and “precision” to describe the accuracy of a measurement method. “Trueness” refers to the close- ness of agreement between the arithmetic mean of a large number of test results and the true or accepted reference value. “Precision” refers to the closeness of agreement between test results. 0.2 General consideration of these quantities is given in ISO 5725-1 and 80 is not repeated in this part of ISO 5725. ISO 5725-1 should be read in conjunction with all other parts of !SO 6726, including thie part, because it gives the underlying definitions and general principles. 0.3. This part of ISO 5725 is concernad solely with estimating by means of the repeatability standard deviation and reproducibility standard devi ation. Although other types of experiment (such as the split-level exper- iment) are used in certain circumstences for the estimation of precision, they are not dealt with in this part of ISO §725 but rather are the subject of ISO 57255. Nor does this part of ISO 5725 consider any other meas- ues of precision intermediate between the two principel measures; those are the subject of ISO 57253. 0.4 In certain circumstances, the data obteined from an experiment cartied out to estimate precision are used also to estimate trueness. The estimation of trueness is not considered in this part of ISO 5725; all as- pects of the estimation of trueness are the subject of ISO 5725-4.INTERNATIONAL STANDARD © !SO Accuracy (trueness and precision) of measurement methods and results — Part 2: Basic method for the determination of repeatability and reproducibility of a standard measurement method 1 Scope 4.1. This part of ISO 5725 — amplifies the general principles to be abserved in designing experiments for the numerical esti mation of the precision of measurement methods by means of a collaborative interlaboratory exper- iment: — nrevides a detailed practical description of the basic method for routine use in estimating the Precision of measurement methods; — provides guidance to all personnel concerned with designing, performing or analysing tne results of the tests for estimating precision NOTE 1 Modifications to this basic method for particular urposes are given in ather parts of ISO 5725, Annex B provides practical examples of estimating the precision of measurement methods by exper iment, 1.2 This part of ISO 5725 is concerned exclusively ‘with meesuremont methods which yield measure- Ments on a rantinnous sesla and give 8 inglo value 8s the test resuit, although this single value may be the outcome of a calculation from a set of observa tions. 1.3. It assumes that in the design and performance of the precision experiment, all the principles as laid own in iSO 5725-1 have been observer! The basic ‘method uses the same numbor of test results in each laboratory, with each laboretory analysing the same levels of test sample; i.e. a balanced uniformlevel ‘experiment. The basic method applies to procedures that heve been sleruaiized end are In regular use in a number of laboratories. NOTE 2 Worked examples are given to demonstrate bal ‘anced uniform sets of test results, although in one example @ vatiable nuiniber of replicates per cell wore reported (un bolenced design) and in another some data were missing This is because an experiment designed ta be balanced can tum out to be unbalanced, Stregolers and autlers are also considered 1.4 The siatistical model of clause 5 of |SO 5725-1:1894 is accepted as 2 suitable basis for the interpretation and analysis of the test results, the dietribution of which is approximately normal 1.5 The basic method, as described in this part of 'SO 5726, will (usually) estimate the precision of 2 measiirament method: 8) when it is required to determine the repeatability and reproducibility siandard deviations as defined in ISO 5725-1; 5) when the materials to be used are homogeneous, or when the effects of heterogeneity can be in. cluded in the precision values; and1SO 5725-2:1994(E) ©} when the use of a balanced uniform-evel layout is acceptable. 1.6 The same approach can be used to make a preliminary estimate of precision for measurement methods which have not reached standardization or are not in routine use, 2 Normative references The following standards contain provisions which, through reference in this text, constitute provisions of this part of SO 5725. At the time of publication, the. editions indicated were valid. All standards are subject to revision, and parties to agreements based on this part of ISO 6726 orc encouraged to investigate Ui possibility of applying the most recent editions of the standards indicated below. Members of IEC and ISO maintain registers of currently valid International Standards. 1SO 3634-1:1993, Statistics — Vocabulary and sym- bols — Part 1: Probability and general statisticat terms. 1S0 5728-1:1994, Accuracy (trueness and precision) of measurement methods and results — Part 1: General principles and definitions. 3 Definitions For the purposes of this part of iSO 5725, the defi: nitions givan in ISO 3634.1 and in ISO 6726 1 apply. ‘The symbols used in ISO 5725 are given in annex A. 4 Estimates of the parameters in the basic model 4.1 The procedures given in this part of ISO 6726 are based on the statistical model given in clause 5 of ISO 5725-1:1994 and elaborated upon in subslause. 1.2 of ISO 5725-1:1994. In particular, these pro- ‘cedures are based on equations (2) to (6) of clause 5 vf ISO 8725-1:1994, The model is m+Bre where, for the particular material tested, ‘m is the general mean (expectation); Iso B is the laboratory component of bias under re- peatability conditions; e is the random error occuring in every ‘measurement under repeatability conditions. 42 Equations (2) to (6) of ISO 8725-1:1994, Clause ® are expressed in terms of the true standard deviations of the populations considered. 1n practice, the exact values of these standard deviations are not known, and estimates of precision values must be made from a relatively small sample of all the possiblo laboratories, and within those laboratories from a ‘small sample of all the possible test results. 4.3 _In statistical practice, where the true value of 2 standard deviation, a, is not known and is replaced by an estimate based upon a sample, then the symbol « is replaced by s to denote that it is an estimate. This has to be done in cach of the oquetions (2) to (6) uf ISO §725-1:1994, giving: a2 is the estimate of the betweenaboratory variance; shy_ is the estimate of the within-laboratory vari ance; is the arithmetic mean of sa, and is the esti- mate of the repeatebility variance; this arith- metic mean is taken over all those laboratories taking part in the accuracy ex- periment which remain after outliers have. been excluded; 4k is the estimate of the reproducibility veri- ance: a) 5 Requirements for a precision experiment 5.1 Layout of the experiment 5.1.1 In the layout used in the basic method, sam ples from q batches of materials, representing q dif- ferent levels of the test, aro sent to p laboratories which each obtain exactly n replicate test results Un der repeatability concitions at each of the g levels. This type of experiment is called a balanced uniform. level experiment.0180 5.12. The performance of these measurements shall be organized and instructions issued as follows. a) Any preliminary checking at equioment shall be as specified in the stendard method. b) Each group of n measurements belonging to one level shall be carried out under repeatability con- ditions, Le, within a slut interval of time and by the same operator, and without any intermediate recalibration of the apparatus unless this is an in- tegra! part of performing 2 measurement. ) tis essential that @ group of n tests under re peatability conditions be performed independently as if they were n tests on different materials. As a rule, however, the operator will know that he/she is testing identical material, but the point should be stressed in the instructions that the whole purpose of the experiment is to determine what differences in results can occur in actual testing. If itis feared that, despite this warning, previous results may influence subsequent test Tesults and thus the repeatability variance, it should be considered whether to use n separate samples at each of the q levels, coded in such ‘way that tho operator will not know which are the replicates for @ given level. However, such 2 pro- cedure could cause problems in ensuring that re- peatability conditions will apply between replicates. This would only be possible if the measurements were ot such a nature that all the qn measurements could be performed within 2 short interval of time. 0) It is not essential that all the @ groups of m measurements each be performed strictly within a short interval; different groups of measurements may be carried out on different days. ¢) Meesurements of oll q levels shall be performed by one and the same operator and, in addition, the n measurements at a given level shall be per formed using the same equipment throughout. 41) If in the course of the measurements an operator should become unavailable, another operator may ‘complete the measurements, provided that the change does not occur within 3 group of measurements at one level but only occurs be- ‘tween two of the ¢ groups. Any such change shall be reported with the results. @ A time limit shell be given within which all Invasuretnents shall be completed. Thio may bo necessary to limit the time allowed to elapse be- ISO 5725-2:1994(E) tween the day the samples are received and the day the measurements are performed, h) All samples shall be clearly labelled with the name of the experiment and a sample identification. 8.43 In 5.1.2 and elsewhere in this part of 1SO 5726, reference is made to the operator. For some measurements, thoro may in fact be @ team of operators, each of whom performs some specific part of the procedure. In such a case, the team shall be regarded as "the operator” and any change in the team shall be regarded as providing a different “op- erator” 5.1.4 In commercial practice, the test results may be rounded rather crudely, but in a precision exper- iment tot reeulte shall be roported to at least one ‘more digit than specified in the standard method. If the method does not specify the number of digits, the rounding shall not be coarser than half the repeatabil- ity standard deviation estimate. When precision may depend on the level m, different degrees ot rounding may be needed for different levels. 5.2 Recruitment of the laboratories 5.2.1 The general principles regarding recruitment of the laboratories to participate in an interlaboratory experiment are given in 6.3 of ISO 5725-1:1994. In enlisting the cooperation of the requisite number of laboratories, their responsibilities shall be clearly stated. An example of a suitable enlistment question- naire is given in figure 1. 5.2.2 For the purposos of thie part of ISO 5726, a “laboratory” is considered to be a combination of the operator, the equipment and the test site. One test site (or laboratory in the conventional sense) may thus produce several "laboratories" if it can provide severe! ‘operators each with independent sets of equipment and situations in which to performn the work. 5.3. Preparation of the materials 5.3.1 A discussion of the points that need to be considered when selecting materials for use in a pre~ cision experiment is given in 6.4 of ISO §725-1:1994, 5.3.2 When deciding on the quantities of material to be provided, allowance shall be made for accidental spillage or errors in obtaining some test results which may necessitate using extra material. The amount of material proparod shall be cufficiont to cover tho ox periment and allow an adequate stock in reserve.1SO 5725-2:1994(E) o1so Questionnaire for interlaboratory study Title of measurement method (copy attached) 1. Our laboratory is willing to participate in the precision experiment for this stan- dard measurement method. ves 2. AS a participant, we understand that: NO L_] tick appropriate box) 8) all essential apperatus, chemicals and other requirements specified in the ‘method must be availzble in our laboratory when the programme begins; ©) specified “timing” requirements such as starting date, order of testing specimens and finishing date of the programme must be rigidly met; ©) the method must be strictly adhered to; 4) samples must be handled in accordance with instructions; @] a qualified operator must perform the measurements. Having studied the method and having made a fair appraisal of our capsbilities and facilities, we feel thet we will be adequately prepared for cooperative testing of this method. 3. Comments (Signed) .... (Company or laboratory) Figure 1 — Enlistment questionnai 5.3.3 It should be considered whether it's desirable for some laboratories to obtain some preliminary test results for familiarization with the measurement method before obtaining the official test result and, if so, whether addtional material (not pracision axper iment samples) should be provided for this purpose. 5.3.4 When a material has to be homogenized, this, shall be done in the manner most appropriate for that material. When the material to be tested is not homogeneous, itis important to prepare the samples in the manner specified in the method, preferably starting with one batch of commercial material for each level. In the case of unstable materials, special instructions on storage and treatment shall be spect fied. ei aR RRR. for interlaboratory study 5.3.5 For the samples at each level, m separate con- tainers shall be used for each laboratory if there is any danger of the materials deteriorating once the con- tainer has been opened (e.g. by oxidation, by losing volatile components, of with hygroscopic inateriah, In the case of unstable materials, special instructions on storage and treatment shall be spacified. Precautions may be needed to ensure that samples remain iden tical up to the time the measurements are made. If the material to be measured consists of a mixture of Powders of different relative density or of different grein size, some care is needed because segregation may result from shaking, for exemple during transport. When ‘reaction with the atmoophere may be ux: Pected, the specimens may be sealed into ampoules, either evacuated or filled with an inert gas. For per- ishable materials such as food or blood samples, iti iso may be necessary to send them in a deep-frozen state to the participating laboratories with detailed in- structions for the procedure for thawing, 6 Personnel involved in a preci experiment NOTE3 The methods ot operavon within cifferent lab- oratories are not expected to be identical, Therefore the Contents of this clause are only intended as 2 guide to be modified as appropriate to cater for a particular situation 6.1 Panel 6.1.1 The panel should consist of experts familiar with the measurement method and ite application. 6.1.2. The tasks of the panel are: a) to plan and coordinate the experiment; ) to decide on the number of laboratories, levels and measurements to be made, and the number of significant figures to be required; 0 10 appuint someone for the stetistico! function (see 6.2); ) 0 appoint someone for the executive functions (see 6.3); @) to consider the instructions to be issued to the laboratory supervisors in addition to the standard measurement method; f) x0 decide whether some operators may be ak Towad to carry out a few unotticial measurements in order to regain experience of the method after 2 long interval (such measurements shall never 2e carried out on the official collaborative sam- lee) @) x0 discuss the report of the statistical analysis on completion of the analysis of the test results; 1h) 0 establish final values for the repeatability stan- 32rd doviation and the reproducibility standard Seviation; |) 20 decide if further actions are required to improve, 1 standard for the measurement method or with ard to lahoratories whosa tast rasults have een rejected as outliers. 62 Sta ical functions At east one member of the panel should have ex: Perence in statistical design and analysis of exper- imerts, His(her tasks are: {SO 5725-2:1994(E) a) to contribute hisjher specialized knowledge in de- signing the experiment: b) to analyse the data; ©) to write @ report for submission to the panel fol lowing the instructions contained in 7.7. 6.3 Executive functions 6.3.1 The actual organization of the experiment should be entrusted to a single laboratory. & member of the staff of that laboratory should take full respon- sibility; he/she is called the executive officer and is appointed by the panel. 6.3.2. The tasks of the executive officer are: a) to enlist the cooperation of the requisite number of laboratories and to ensure that supervisors are appointed; b} to orgenize and supervise the preparation of the materials and samples and the dispatch of the samples; for each level, an adequate quantity of material should be set aside as a reserve stock; ©) to draft instructions covering all the points in 6.1.2 2) to hl, and circulate them to the supervisors early enough in advance for them to raise any comments or queries and to ensure that operators, selected are those who would normally carry out such measurements in routine operations; 4) to design suitable forms for the operator to use 2s a working record and for the supervisor to re- port the test results to the requisite number of significant figures (such forms may include the name of the operator, the dates on which sam- ples were received and measured, the equipment used and eny other relevant information); e) to deal with any queries from laboratories regard- ing the performance of the measurements; #} to see that an overall time schedule is maintained; 4} to collect the data forms and present thern to the statistical expert. 6.4 Supervisors 6.4.1 A staff member in each of the participating aboratorias should ba made responsible for argania- ing the actual performance of the measurements, in keeping with instructions received from the executive officer, and for reporting the test results.6.4.2 The tasks of the supervisor are: 2) ensure that the operators selected are those who would normally carry out such meesurerats in routine operations: b) to hand out the samples to the operator(s) in keeping with the instructions of the executive of- ficer {and to provide material for familiarization experiments, if necessary); €) to supervise the execution of the measurements (the supervisor shali not take part in performing the measurements); 4) to ensure that the operators carry out the required number of measurements; fe). to ensure adherence to the set timetable for per- forming the measurements; 4) to collect the test results recorded to the agreed number of decimal places, including eny anom- alies and difficulties experienced, and comments made hy tha oparators, 6.4.3 The supervisor of each laboratory should write a full report which should contain the following infor- mation: a) the test results, entered legibly by their originator ‘on the forms provided, not transcribed or typed {computer or testing machine printout may be ac- coptable as an alternative); b) the original observed values or readings (if any) from which the test results were derived, entered legibly by the operator on the forms provided, not ‘ronscrived of typed; ©) comments by the operators on the standard for the measurement method; d) information about itreguiarities or disturbances that may have occurred during the measure ments, including any change of operator thet may have occurred, together with o statement as to ‘which measurements were performed by which ‘operator, and the reasons for any missing results; €) the datefs) on which the samples were received; #) the date(s) on which each sample was measured; 9) information about the equipment used, if relevant; ‘h)_ any other relevant information. 6.5 Operators 6.5.1. In each laboratory the measurements shall be carried out by one operator selected as being repre- sentative of those likely to perform the measure- ‘ments in normal operations. isc 6.52 Because the object of the experiment is t determine the precision obtainable by the genera population of operators working from the stendarc measurement method, in general the operator: should not be given amplifications to the standard fo the measurement method. However, it should be pointed out to the operators that the purpose of the exercise is to discaver the extent to which results car vary in practice, so that there will be less temptatio: for them to discard or rework results that they feel ar: inconsistent. 65.3 Although normally the operators should re ceive no supplementary emplifications to the standan measurement method, they should be encouraged t ‘comment on the standard and, in particular, to stat whether tho instructions containod in it are sufficient unambiguous and cleer. 6.5.4 The tasks of the operators are: a) 10 perform the measurements according to th standard measurement method; b) to report any anomalias or difficulties exper enced; it is better to report a mistake than to ac just the test results because one or two missin: test results will not spoil the experiment an many indicate a deficiency in the standard; ) to comment on the adequacy of the instruction in the standard; operators should report any oc casions when they sre unable to follow their ir structions as this may also indicate a deficiency | the standard. 7 Statistical analysis of a precision experiment 7.1. Preliminary considerations 7.1.1. The analysis of the data, which should b considered as a statistical problem to ba solved by statistical expert, involves three successive stages: ) critical examination of the data in order to identit and treat outliers or other irregularities and to tes the suitability of the model; b)_ computation of preliminary values of precision an means for each level separately: ©} establishment of final values of precision an means, including the establishment of a relatio: ship between precision and the level m when th analysis indicates that such 2 relationship me exist.0180 7.1.2. The analysis first computes, for each level separately, estimates of _— the repeatability variance 5? _— the betweeneboratory variance sf — the reproducibility variance 32 = 42 + of = the mean m, 7.1.3 The analysis includes a systematic application of statistical tests for outliers, a great variety of which are available from the literature and which could be nised for the purposes of this part of ISO 6725. For practical reasons, only 2 limited number of these tests, as explained in 7.3, have been incorporated. 7.2. Tabulation of the results and notat used 724 Cells Fach combination of a laboratory and a level is called 2 call of the precision experiment. In the ideal case, the resuits of an experiment with p laboratories and q lovels consist of a table with pq cells, each contain- jing replicate test results that can all be used for computing the repeetsbility standord deviation and the reproducibility standard deviation, This ideal situation ig not, however, always attained in practice. Depar- tures occur owing to redundant data, missing data and outliers, 7.22 Redundant data Sometimes 2 laboretory may carry out and report ‘moro than the nm test results officially specified. in that case, the supervisor shall report why this was done and which are the correct test results. if the enswer is that they afe all equally valid, then a random se- lection should be made from those available test re- sults to choose the planned numnber of test results for analysis. 7.2.3 Missing data In other cases, some of the test results may be missing, for example because of loss of,a sample or @ mistake in performing the measurement. The analysis recommended in 7.1 is such that completely ‘empty colle can simply ha ignored. while partly empty cells can be taken into account by the standard com- putational procedure. 7.24 Outliers These are entries among the original test results, or in the tables derived from them, that deviate so much feu dress. 29"E) from the comparable entries in the same table that they are considered irreconcilable with the other data. Experience has taught that outliers cannot always be avoided and they have to be taken into consideration ina similar way to the treatment ot missing oata. 7.25 Outlying laboratories When soveral unexplained abnormal test results occur at different levels within the same laboratory, then that laboratory may be considered to be an outlier, having too high a within-laboratory variance andor too large a systematic error inthe level of its test results. it may then be reasonable to discard some or ail of ‘the data from such an outlying laboratory. This part of ISO 8725 does not provide @ stetistical test by which suspected laboratories may be judged. The primary decision should be the responsibility of the statistical expert, but all rejected laboratories shall be reported to the panel for further action. 7.2.6 Erroneous data Obviously erroneous data should be investigated and corrected or discarded. 7.2.7 Balanced uniform-level test results. The ideal case is p laboratories called i (=1, 2, ph. each testing g levels called j 12g) with n replicates at each level (wach ij combination), giving a total of pqn test results. Be- cause of missing (7.2.3) or deviating (7.2.4) test re- sults, of outlying laboratories (7.2.6) or erroneous data (776), this ideal situation is not always attained. Un- der these conditions the notations given in 7.2.8 to 7.2.10 and the procedures of 7.4 allow for differing numbers of test results. Specimens of recommended forms for the statistical analysis are given in figure 2 For convenience, they will be referred to simply 03 forms A, B and C (of figure 2. 7.28 Original test results See form A of figure 2, where ny is the number of test results in the cell for laboratory i at level j; any one of these test results 2 me ME ya's p, is the number of laboratories reporting at least one test resuilt for level j (after alim= inating any test results designated as outliers or as erroneous).\ ISO 5726-2:1994(E) For A — Rewon tended form for the eoltation of tne original cara Iso fe Level ratory 12 i a-t| ¢ 1 2 4 i oa 2 Form B — Recommended form for the collation of the means Laborat ; sborstory a tf2 i = feila 1 z i i ? Form C — Recommended form for the collation of the measures of spread within cells Level Laboratory rf2 L a-1|¢@ 1 2 i % > Figure 2 — Recommended forms for the collation of results for analysis 7.29 Coll means (form B of figure 2) These ara derived fram form A as follews: % Suna De +2) The cell means should be recorded to one more sig- nificant figure than the test result in form A. 12.10 Measures of cell spread (form C of figure2) These are derived from form A (see 7.2.8) and form B (see 7.2.9) as follows.@ ® In asing these equations, care shall be taken to retain a sufficient number of digits in the calculations; ic. evary intermediate value shall be twice as many digits as in the NCTE 4 Ifa cell 1 contains two test results. the intracell standard devietion is 59 aya f2 © ‘Therefore, for simplicity, absolute differences can be used Instaaa of stancara deviatons i all cols contain two tost results The standard deviation should be expressed to one more significant figure than the results in form A. For values of n; less than 2, @ dash should be inserted inform C. 7.2.41 Corrected or rejected data As some of the data may be corrected or rejected on the basis of the tests mentioned in 7.1.3, 7.3.3 and 7.34, the values of ig, my and p; Used for the final determinations of precision and mean may be differ ent from the values referring to the original test re- suits as recorded in forms A, 8 and C of figure 2. Hence in reporting the final values for precision and ‘ttueness, it shall always be stated what data, if any, have been corrected or discarded, 7.3. Scrutiny of results for consistency and outliers See reference [3]. From data collected on a number of specific levels, 2 Tepeatability and reproducibility standard deviations & &e to be estimated. The presence of individual lab- ISO 5725-2:1994(E) oratories or values that appear to be inconsistent with all other laboratories or values may change the esti- mates, and decisions have to be made with respect to these values. Two approaches aro introduced: a) graphical consistency technique; b) numerical outlier tests, 7.3.1 Graphical consistency technique Two measures called Mandef's A and k statistics are used. It may be noted that, as well as describing the variability of the measurement method, these help in laboratory evaluation. 7.3.1.1 Calculate the betweendaboratory consist- ency statistic, A, for each laboratory by dividing the cell deviation (cell mean minus the grand mean for that level) by the standard deviation among cell means (for thet level): y= 6) a» Bay yo in which, for 5, see 7.2.9, and for j see 7.4.4. Flot the h, values tor each cell in order of laboratory, in groups for each level (end separately grouped for the several levels examined by each laboratory) (see figure B.7). 7.3.1.2. Calculate the withinlaboratory consistency statistic, £, by first calculating the pooled within-cell standard deviation as Fi tor each level, and then calculate =O for each laboratory within each level. Plot the Ay values for each cell In order of laboratory, in groups for each level and separately grouped for the several levels examined by each laboratory) (see figure B.8).‘ ISO 5725-2:1994(E) 7.3.1.3 Examination of the h and & plots may indicate that specific laboratories exhibit patterns of results that are markedly different from the others in the study. This is indicated by consistently high or inw \within-cell variation andor extrame cell means across many levels. If this occurs, the specific laboratory should be contacted to try to ascertain the cause of the discrepant behaviour. On the basis of the findings, Ue statistical expert would, 4). retain the laboratory's data for the moment; b) ask the laboratory to redo the measurement (if feasible); ©) remove the laboratory's data from the study, 7.3.1.4 Various patterns can appear in tho & piots. All laboratories can have both positive and negative A values at different levels of the experiment. Individual laboratories may tend to give either all positive or all negative h values, and the number of faboratories 2) whare Cochran's test has raised Sus picions as to whether the high within-laboratory vari- ation is attributable to only one of the test results in the cell 7.3.3 Cochran's test 7.3.3.1 This part of ISO 5725 assumes that between Jaboretories only small differences exist in the within laboratory variances. Experience, however, shows thet this is not always the case, so that a test has been included here to test the validity of this as- ‘sumption. Several tests could be used for this pur- pose, but Cochran's tost has been chosen. 7332 Given a sot of p standard deviations s, all ‘computed from the same number (n) of replicate re- sults, Cochran’s test statistic, C. is 2 c= -@) Where Spe is the highest standard deviation in the set. a) If the test statistic is less than or equal to its 5 % cities! value, the item tested is accepted as nr. rect. b) If the test statistic is greater than its § % critical value and less then or equal to its 1% critical value, the Item tested Is called @ straggler 2 jicated by a single asterisk ©) If the test statistic is greater than its 1 9% critical value, the item is called a statistical outlier and is indiceted by a double asterisk Critical values for Cochran's test are given in 8.1 {table 4, ISO 5725.2:1994(E) Cochran's test has to be applied to form C of figure 2 at each level separately. 7.3.3.3 Cochran's citerion applies stticly only when all the standard deviations are derived from the same number (n) of test resuits obtained under repeatability conditions. In actual cases, this number may vary owing to missing or discarded deta. This part of 1SO 725 assumes, however, that in a properly or- ganized experiment such variations in the number of test results per cell will be limited and can be ignored, and therefore Cochran's criterion is applied using for nthe number of test results occurring in the majority of cells. 7.3.34 Cochran's criterion tests only the highest value in 9 set of standard deviations and is therefore @ one-sided outlier test. Variance heterogeneity may also, of course, manifest itself in some of the sten- dard deviations being comparatively too low. How- fever, small values of standard deviation may he vary strongly influenced by the degree of rounding of the original data and are for that reason not very reliable. In addition, it seems unreasonable to reject the date from @ laboratory because it has accomplished 2 higher precision in its test results than the other jab- oratories, Hence Cochran's criterion is considered ed- equate. 7.3.3.5 A critical examination of form C of figure2 may sometimes reveal that the standard deviations for a particular laboratory are at all or at most levels lower than those for other laboratories. This may in- dicate that the laboratory works with a lower repost- ability standard deviation than the other laboratories, Which in turn may be caused either by better tech- ique and equipment or by a modified or incorrect application of the standard measurement method. If this occurs it should be reported to the panel, which should then decide whether the point is worthy of more detailed investigation. (An example of this is laboratory 2 in the experiment detailed in B.1.) 7.3.3.6 If the highest standard deviation is classed as an outlier, then the value should be omitted and Cochran's tast repeated on the remaining values. This process can be repeated but it may lead to excessive rejections when, as is sometimes the case, the underiying essumption of normality is not sufficiently well approximated to. The repeated application of Cochran's test is here proposed only as a helpful toot in view of the lack of a statistical test designed tor testing several outliers together. Cochran's test is not designed for this purpose and great caution should be exercised in drawing conclusions. When two or three "1SO 5725-2:1994(E) laboratories give results having high standard devi ations, particularly if this is within only one of the lev- eis, conclusions from Cochran’s test should be examined carefully. On the other hand, if several stagglers andjor statistical outliers are found at dif ferent levels within one laboratory, this may be a strong indication that the laboratory's within- laboratory variance is exceptinnally high, and the whole of the data from that laboratory should ba re- jected. 7.2.4 Grube! toct 7.3.4.1, One outlying observation Givan a sat of data x for i= 1 ». arranged in ‘ascending order, then to determine whether the larg- est obsenation is an outlier using Grubbs’ test, com- pute the Grubb's statistic, G,, G,= (Bs --@ where 2 (10) and pare ay ‘To test the significance of the smallest observation, compute the test statistic G= Fay a) If the test statistic is less than or equal to its 5 % critical value, the item tested is accepted as cor rect. D) If the test statistic is greater than its 5 % critical value and less than or equal to its 1% critical value, the item tested is called a straggler and is indicated by a single asterisk. ©) If the test statistic is greater than its 1 % critical valuo, the item ie called a statistical autliar andl is indicated by a double asterisk. 7.3.4.2 Two outlying observations To test whether the two largest observations may be outliers, compute the Grubbs' test statistic G: Cas phe +e (12) e150, where $= G-9? 3a and 2 Bore D-H)” aa and (15) Alternatively, to test the two smallest observations, computs the Giubbs' Wwsl stalislic G. Gate (16) where 2 fae De a & and = i< i 18 pa 2 ag) Critical values for Grubbs’ test are given in 8.2 (tables). 7.3.4.3 Application of Grubbs’ test ‘When analysing @ procision experiment, Grubbs’ test can be applied to the following 2) The cell averages (form B of figure2) for a given level jin which case 4G and P=7, where js tx0d, Taking the data at one level, apply the Grubbs" test for one outlying observation to cell means as, described in 7.2.4.1. Ifa cell mean is shown to be an outlier by this test, exclude it, and repeat the test at the other extreme cell mean (e.g. if the highest is an outler then look at the lowest with the highest excluded), but do not apply the© 1SG Grubbs’ test for two outlying observations de- scribed in 7.3.4.2. If the Grubbs’ test does not show @ cell mean to be en outlier, then apply the double-Grubbs' test described in 7.3.4.2. b) A single result within a cell, where Cochran's test has shown the cell standard deviation to be sus- pect. 7.4 Calculation of the goneral mean and variances 7.4.1 Method of analysis The method of analysis adopted in this part of 1SO 725 involves carrying out the estimation of m and the precision for each level separately. The results ‘af the computation are expressed in a tabla for aach value of j 7.42 Basic data ‘The basic data needed for the computations are pre- sented in the three tables given in figure 2: — table A containing the original test results; — table B containing the cell means; — table C containing the measures of within-cell spreed. 7.4.3, Non-empty colls As a consequence of the rule stated in 7.3.2.1 d), the number of non-empty cells to be used in the compu tation will, for a specific level, always be the same in tables B and C. An exception might occur if, owing to ‘missing data, a coll in table A contains only a single test result, which will entail an empty cell in table C but not in table B. In thet case it is possible 41. to discard the solitary test result, which will lead to empty cells in both tables B and C, or b) if this is considered an undesirable loss of infor- mation, to insert @ dash in table C. The number of non-empty cells may be different for different levels, hence the index jin pp ISO 5725-2:1994(E) 7.44 Calculation of the general mean fi For level j, the general mean is . Snip yay +(19) Yr Bm 7.45 Calculation of variances ‘Thvee variances are calculated for each level. They are the sepeatability variance, the betweenlaboratory variance and the reproducibility variance. 7.4.5.1 Tho ropoatabilty variance is (oy — 5h 0) 15 Saow'- De] -@2) oF mn and -- (23) These calculations are illustrated in the examples in B.1 and B.3 in annex 8, 3ISU S/zb-2:1994(E) 7.45.3 For the particular case where all my the simpler formulae may be used, giving SED o0- and Sst DO ‘These are illustrated by the example given in 8.2. 7.4.5.4 Where, owing to random effects, a negative value for <2 is obtained from these calculations, the value should be assumed to be zero. 7.48.5 The reproducibiny variance Is. hosed, + (24) 7.46 Dependence of the variances upon m Subsequently, it should be investigated whether the precision depends upon m and, if so, the functional relationship should be determined. 7.5 Establishing a functional relationship 7.5.1 It cannot always be teken for granted thet there exists @ regular functional relationship between precision and m. In particular, where material etarngeneity forms an insanarahie part of the vari ability of the test results, there will be a functional relationship only if, this heterogeneity is @ reguiar funetion of the level m. With solid materials of differ- ant composition and coming from different production processes, @ reguiar functional relationship Is in no way certain, This point should be decided before the following procedure is applied, Alternatively, separate values of precision would have to be established for each material investigated. 7.5.2 The reasoning and computation procedures presented in 7.5.3 to 7.5.9 apply both to repeatability and reproducibility standard deviations, but are pre- sented nere for repeatability only in the interests ot brevity. Only three types of relationship will be con- sidered: 1: 5, =.6m (a straight line through the origin) a+ bm (a straight line with a positive inter- cept) “ 2 18U Il: Ig 5, «6 +d fg m (or s, = Cm‘); d <1 (an expo- ential relationship) It is to be expected that in the majority of cases at least one of these formulae will give a satisfactory fit, If not, the statistical expert carrying out the analysis should seek an altemstive solution. To avoid con- fusion, the constants a, b, ¢, C and d occurring in these equations may be distinguished by subscripts, 4, b,, ... for repeatability and ap, be, ... when consid ering reproducibility, but these have been omitted in this clause again to simplify the notations. Aiso's, has been abbraviated simply to s to allow a suffix for the level j. 7.5.3 In general d > 0 so that relationships | and Ml will lead to s=0 for m=0, which may seem un- eccopleble fron an experimental point of view. How ‘ever, when reporting the precision data, it should be made clear that they apply only within the levels cov- ered by the interlaboratory precision experiment. 7.5.4 For a=0 and d=1, all three relationships are identical, so when a fies near zero andor d lies near unity, two or all three of these reletionships will yield practically equivalent fits, and in such a case relation- ship | should be preferred because it permits the fol- lowing simple statement. “Two test results are considered as suspect when thoy differ by more than (100 1) %6." In statistical terminology, this is a staterient that the coefficient of variation (100 sim) is a constant for all levels, 7.5.5 If in a plot of 5; against fA, or plot of Ia against Ig Ay the set of points are found to lie re: sonably close to a straight line, @ line drawn by hand may provide a satistactory solution; but if for some reason a numerical method of fitting is preferred, the procedure of 7.5.6 is recommended for relationships \ end il, and that of 7.5.8 for relationship Ii 75.6 From a statistical viewpoint, the fiting of a straight line is complicated by the fact that both i and 5 2f8 estimates and thus subject to error. But a8 the Slona h ig ristrally small (af the order of 0,1 or love), then errors in have little influence and the errors in estimating s predominate, 7.8.8.1 A good estimate of the parameters of the regression line requires 2 weighted regression be- cause the standard error of s is proportional to the predicted value of s,©1so The weighting factors have to be proportional to 1(3)", where 3; is the predicted repeatability standard deviation for level j. However § depends on par- ameters that have yet to be caloulatod. A mathematically correct procedure for finding esti- motes corresponding to the weighted least-squares of residuals may be complicated. The following pro- cedure, which has proved 10 be satistactory in prac- tice, is recommended. 7.5.6.2 With weighting factor W, equal to 1/(8,)?, where N=0, 1, 2.... for successive iterations, then the calculated formulae are: n=, n= Dw = Swi ty -2¥s Ts ~ Lin Then for relationship | (s = bm), the value of bis given by Tel For relationship Il (s = a + bm): ++. (25) and. T, Ts—T.Ty pe | DERE 2 [ay 7.5.63. For relationship |, algebraic substitution for the weighting factors W, = 1/(5)* with 5 = bi, leads to the simplified expression: Dia) 7 7) and no iteration is necessary. ISO 5725-2:1994(E) 7.5.6.4 For relationship I, the initial values 4 are the original values of s as obtained by the procedures given in 7.4, These are used to calculate We = MG)? G=1 2 @) and to calculate a, and 6, as in 7.5.6.2. This leads to Sy 0 + byhy The computations are then repeated with Wy = 1)? to produce The same procedure could now be repested once again with weighting factors Wy) = 1/(S3)* derived from these equations, but this will only lead to unim- Portant changes. The step from We, to Wy is effective in eliminating gross errors in the weights, and the equations tor #,, should be considered as the final re- sult 7.5.7 The standard error of Ig.s is independent of s and so an unweighted regression of Ig son Ig A is appropriate 7.5.8 For relationship Ill, the computational formulae are: n= S.leaios and thence TTT. eo RGAE aT (28) aaah ++(29) a7 6ISO 5725-2:1994(E) 7.5.9 Examples of fitting relationships |, and il of 7.8.2 t0 the same set of data are now given in 7.5.9.1 to 7.5.9.3. The data are taken from the case study of 8.3 and have been used here only to illustrate the numerical procedure” They will be further discussed o1so 7.5.9.1 An example of fitting relationship | is given in tablet 7592 An oxample of fitting rolotionchip Il is given in table2 Ui, sare as in 7.5.9.1) in B.3. 75.9.3 An example of fitting relationship Ill is given in table 3. Table 1 — Relationship ks = bm ay 3.94 828 1418 1559 20,1 Ly 0,092 0179 or var 0,383 aff e024 foore {ocose foons foors3 ro) 0098 _goro oar nad 5 sam 0.078 0187 0,268 0.296 0.388 nity Table 2 — Relationship It: s =a + bm 51 = 0,058 + 0,009 0 m 4y 0,083 0.132 0,185 0187 0,240) Wy 6 87 2 6 7 2 = 0,080 + 0,015 6 m by 0082 0.159) 0281 oa ose Way 18 40 16 8 8 a= 0.018 4 m 30 0,093 0.160 0.281 0273 0348 a NOTE — The values of the weiahtina factors are not critical: wn significant figures sutfioe 1) The difference trom s, is negligible. rr—E EO Table 3 — Relationship Il: |g s=c +d lg m 94 +0595 frosts = f4nisz [41.103 a sy -0,747 = 0896 ~0.472 4g s~— 1,508 5 +0724 m oF s= 0.081 m*”” s 0.089 0.188 0.239 0287 ost6 16i i ISO 7.6 Statistical analysis as a step-by-step procedure NOTE 5 Figure’ indicates in a stepwise fashion the pro- cedura given in 7.8 7.6.1 Collect all available test results in one form, form A of figure 2 (see 7.2). It is recommended that this form be arranged into p rows, indexed i=1, 2... p (representing the p laboratories that have contributed data) and q columns, indexed a1, 2, ... 4 (representing the q levels in increasing orden) In a uniformlevel experiment the test results within a cell of form A need not be distinguished and may bbe put in any desired order, 7.6.2 Inspect form.A for any obvious irregularities, investigate and, if necessary, discard any obviously etroneous dete {for example, dota outside tho rango of the measuring instrument or data which are im possible for technical reasons} and report to the panel. Itis sometimes immediately evident that the test re- suits of @ particuiar laboratory or in a particular cell lie at a level inconsistent with the other data. Such obv- ously discordant data shail be discarded immediately, but the fact shall be reported to the panel for further consideration (see 7.7.1) 7.6.3 From form A, comected according to 7.6.2 when needed, compute form 8 containing cell means ‘and form C containing measures of within-cell spread. When @ cell in form A contains only a single test re- ‘sult, one of the options of 7.4.3 should be adopted. 7.6.4 Prepare the Mandel h and & plots as described in 73.1 and examine them for consistency of the data. These plots may indicate the suitability of the deta for further analysis, the presence of any possible outlying voluce or outlying laboratories. However, no definite decisions are taken at this stage, but are de- layed until completion of 7.6.5 to 7.6.9. 7.6.5 Inspect forms 8 and U (see tigurez) level by level for possible stragglers andjor statistical outers [see 7.3.2.1 al]. Apply the statistical tests given in 7.3 to all suspect items, marking the stragglers with 2 single asterisk and the statistical outliers with a dou- ble asterisk. If there are no stragglers or statistical outliers, ignore steps 7.6.6 to 7.6.10 and proceed di rectly with 7.6.11 1dU-D/25-2:1994(E) 7.6.6 Investigate whether there is or may be some technical explanation for the stragglers andjor stat- istical outliers and, if possible, verify such an exple- nation. Correct or discard, as required, those stragglers andjor statistical outiers that have been setistactorly explained, and apply corresponding cor- rections to the forms. if there are no stragglers or statistical outliers left thet have not been explained, ignore etepe 7.6.7 to 7.6.10 and proceed directly with 76.11. NOTE 6 A larga number of stragglers andjor statistical outiars may indicate a pronounced variance inhomogeneity or pronounced differencca betwoon leboraterios and may thereby cast doubt on the suitability of the measurement ‘method. This should be reported to the panel 7.6.7 _{f the distribution of the unexplained stragglers or statistical outers in torm B or C does not suggest ny outlying laboratories (see 7.2.6), ignore step 7.6.8 ‘and proceed directly with 7.6.9. 7.6.8 If the evidence against same suspected ‘outlying laboratories is considered strong enough to justify the rejection of sorne or all the data from those laboratories, then discard the requisite data and report to the panel. ‘The decision to reject some or all data from a partic ular laboratory is the responsibility of the statistical ‘expert carrying out the analysis, but shall be reported to the panel for further consideration (see. 7.7.1). 7.6.9 If any stragglers andjor statistical outliers re- ‘main that have not been explained or attributed to an ‘outlying laboratory, discard the statistical outliers but retain the stragglers. 7.6.10 If in the previous steps any entry in form B has been rejected, then the corresponding entry in form C has to be rejected also, and vice versa. 7.6.11. From the entries that have been retained as correct in forms B and C, compute, by the procedures given in 7.4, for each level separately, the mean level fy and the repeatability and reproducibility standard deviations. 7.6.12. If the experiment only used a single level, or if it has been decided that the reveatability and re- producibilty standard deviations should be given sep- atately for each level (see 7.5.1) and not as functions of the level, ignore steps 7.6.13 to 7.6.18 and proceed directly with 7.6.19, NOTE 7 The following steps 7.6.13 to 7.6.17 are applied to 5, and s separately, but for brevity they are written out only in terms of s, v7BU D/Z>ETYNAIED ©1s0 ‘raw up form A, Ave there any obvious irregularities 7 ‘Compute Terns 6 and, Prepare Mandel’s hand k ptots. Are there Is therea ‘ony straggters/outliers iy technical expianation forthe Discard or correct ase formBor€ 7 Testsere straggtess/outters ? items exptained. ivenin73, istribution of unexptaned se Deo Fectae. Segara Are there ‘any remaining unexptained Stragalere/eutliers? Have any ‘entries in form or C been, Discard corresponding discarded? entry in form Ran C Figure 3 — Flow diagram of the principal steps in the statistical analysis (cnntinued on pago 10) 8e1so 1SO 5725-2:1994(E) Compute, for each evel separately, Ling the procedures given in 7.4. ~repeatablity standard deviation $4: ~reprocuctlity standard deviations. as a single Tevel been used om has P been decided to quote 5, and s» op ‘each evel 7 is srorse ‘pparentty independent ot m7 ye Calculate the values of 5, ond Sh Toppy to all values of m, ‘bialn The near eaationship by applying the computationat. | proceauna givanin 25. Fetationship between 2- or Sy Yee] and m be considered linear 7 an the ‘etanonsmip verweentg 5; OF lyse ond igm be considered, Uinear 7 [ Doran he uncer catationene |] by apotying the computationat procedure given n 7S. a pee ‘ther retationship between # or 5 and m be established ? Establish thet retationship.1S 5725-2:1994(E) 7.8.13 Plot s against i, end judge from this plot whether s depends on m or not. If 5 is considered to depend on m, ignore step 7.6.14 and proceed with 7.6.16. I sis judged to be independent of m, proceed with step 7,6.14. If there should be doubt, it is best 10 work out both cases and let the panel decide. There exists no useful statistical test appropriate for this problem, but the technical experts familiar with the monourement method sliould have sufficient ex- Perience to take a decision. 7.6.14 Use Lzs=5, as the final value of the re- peatability standard deviation. Ignore steps 7.4 15 to 7.6.18 and proceed directly with 7.6.19. 7.6.15 Judge from the plot of 7.6.13 whether the relationship between s and m can be represented by @ straight line and, if so, whiter relationship 1 (s = bm) oF relationship Il (s =a + bm) is appropriate (see 7.5.2). Determine the parameter 6, or the two Parameters a and b, by the procedure of 7.5.6. if the linear relationship is considered satisfactory. ignore step 7.6.16 and proceed directly with 7.6.17. If not, proceed with step 7.6.16. 7.6.16 Plot Ig 5, against Ig A, and judge from this whethor the relationship between Ig s and ig m can reasonably be represented by a straight line. If this is considered satisfactory, fit the relationship ll (ig.s=¢ +4 lg m) using the procedure given in 7.58. 7.8.17 if a satisfactory relation has been established in step 7.6.15 oF 7.6.16, then the final values of s, (or 4x) are the smoothed values obtained from this re- lationship for given values of m. Ignore step 7.6.18 and proceed with 7.6.19. 7.6.18 If no satisfactory relation has been estab- lished in step 7.6.15 or 7.6.16, the statistical expert should decide whether some other relation between 3 and m can be established, or alternatively whether the data are so irregular that the establishment of @ functional relationship is considered to be impossible. 7.8.49 Prepare a report showing the basic data and the results and conclusions from the statistical analy- sis, and present this to the panel. The graphical pres- entations of 7.3.1 may be useful in presenting the consistency or variability of the results. 7.7 The report to, and the decisions to be taken by, the panel 7.74 Report by the statistical expert Having completed the statistical analysis, the statisti. cal expert should write a report to be submitted to the 20 ©1so panel, In this report the following information should be given: a] a full account of the observations received from the operators and/or supervisors concerning the standard for the measurement method; ') 2 full eccount of the laboratories that have been rejected as outlying laboratories in etopo 7.6.2 end 7.68, together with the reasons for their re- jection; ©) a full account of any stragglers andjor statisticet outliers that were discovered, and whothar those were explained and corrected, or discarded; 9) a form of the final results i, 5, and sq and an ac- count of the conclusions reached in steps 7.6.13, 7.6.15 oF 7.6.16. illustrated hy one of the piote recommended in these steps; @) forms A, B and C (figure 2) used in the statistical analysis, possibly as an annex. 7.1.2 Decisions to be taken by the pan« The Panel should then discuss this report and take decisions concerning the following questions. @) Ate the discordant results, stragglers or outliers, if any, due to defects in the description of the standard for the measurement mathad? ©) What action should be taken with respect to re- jected outlying laboratories? ©) Do the results of the outlying laboratories andjor the comments received fruit the operators and Supervisors indicate the need to improve the standard for the measurement method? If so, what are the improvements required? 4) Do the recults of the precision experiment justity the establishment of values of the repeatability standard deviation and reproducibility standard Geviation? If so, what are thase values, in what form should they be published, and what is the region in which the precision data apply? 7.7.3 Full report A report setting out the reasons for the work and how it was organized, including the report by the statisti clan and setting out agreed conclusions, should be Prepared by the executive officer for approval by the Panel. Some graphical presentation of consistency or variability is often useful. The report should be circu- lated to those responsible for authorizing the work and to other interested parties,iso ISO 5725-2:1994(E) 8 Statistical tables 8. 1 Critical values for Cochran's test (see 7.3.3) are given in tabled Table 4 — Critical values for Cochran's test naa n=5 n= 1% [ 5% 1% 5% 1% 5% 1% | 6% 1% | 58% — | 0898 [0975 ] 0979 | 0939 | 0959 | 0908 | ear | 0377 oss | og67 | 0942 | 0871 | 0883 } 0798 | oad | o7as | 0793 | 0.707 2 3 4 | 0968 | 0906 | oge | 0768 | 0781 | 068s | 0721 | 0629 | 0676 | o590 5 | 0928 | 041 | 0788 | 0684 | 0695 | 0598 | 0633 | 0.544 | osse | 0,506 6 7 a oss | 0781 | 0722 | o616 | 0626 | 0532 | ose | 0480 | 0520 | 0,445 ogsa | 0.727 | 0664 | 0,561 | 0562 | 0.480 | 0508 | 0.431 | oes | 0,397 0704 | ogo | osis | ose | o521 | 0.438 | 0463 | 03991 | 0423 | 0,260 a | 0754 | 0638 | 0573 | 0478 | 0481 | 0403 | 0.425 | 0.358 | 0387 | 0.329 1o | o78 | 0602 | 0536 | 0445 | 0247 | 0.373 | 0303 | oa | 0357 | 0.303 11 | 0684 | 0870 | oso | 0417 | 0418 | 0248 | 0386 | 0308 | 0332 | 0.281 12 | 0653 | 0541 | 0475 | 0392 | 9992 | 0926 | 0343 | 0208 | oa10 | 0262 az | nara | osis | o4so | o371 | o369 | o307 | 0322 | o271 | o2m | o2ea 1 | 0599 | 0492 | o4z7 | 0.352 | 0399 | 0.291 | oa | 0255 | o27 | 0,232 16 | 0575 | o471 | 0407 | 0.335 | 0932 | 0.276 | o2a8 | 0242 | 0259 | 0,220 16 | 0553 | 0452 | 038 | 0319 | o3t6 | 0.262 | 0274 | 0230 | oz | 0.208 a7 | osaz | 03a | 0372 | 0.305 | 0301 | 0.250 | 026 | o219 | 0234 | o.198 re | 614 | ois | ose | 0.203 | o2e¢ | 0.240 | 02% | 0,200 | o223 | ogo 19 | 0496 | 0403 | 0343 | 0,281 | 0276 | 0,230 | 0238 | 0200 | o214 | o,1@1 20 | 0.480 | 0289 | 0330 | 0270 | 0265 | 0220 | 0229 | 0.192 | 0.205 | 0174 2 | 0465 | 0377 | 0318 | 0261 | 0265 | o212 | o220 | 0185 | o197 | 0.167 zz | 0450 | 0365 | 0307 | 0.252 | 0286 | 0,204 | o212 | 0,178 | 0,189 | 0,160 23 | 0437 | 0354 | 0297 | 0243 | 0238 | 0,197 | o204 | 0172 | 0,182 | 0185 26 | 0425 | 0343 | 0287 | 0.235 | 0230 | 0191 | 0197 | 0163 | 0176 | 0149 2 | 043 | 0334 | 0278 | 0,228 0.185 | 0190 | 0160 | 0,170 | 144 26 | 0402 | 0325 | 0.270 | 020 0.179 | o1e4 | 0155 | 0164 | 0140 27 | oss1 | oie | 0262 | 0215 0.173 | 0179 | 0150 | 0.159 | 0.136 28 | 0382 | 0208 | ozs | ozo9 ores | or73 | ores | ors | 0137 29 | 0372 | 0200 | 0248 | 0,203 ote | 0168 | 0142 | 0,150 | 0127 30 | 0963 | 0293 | 0.241 | 0,198 tsa | ores | 0198 | 0145 ] 0.124 ay | 0355 | 0286 | 0235 | 0193 0.185 | 0169 | 0.134 | ona | 0.120 32 | 0347 | 0280 | 0229 | 0,188 01st | oss | o131 | orga | 0,117 33 | 0339 | 0273 | 0224 | 0,186 0147 | o1s1 | 0127 | 0134 | one 34 | 0332 | 0267 | 0218 | 0.179 or4s | 0187 | 0124 | ona | ond 35 | 0325 | oz6z | 0213 | 0,175 or4o | ores | 0121 | 0127 | 0708 26 | 0318 | 0286 | 0.208 | 0,172 0.137 } 0140 | 0,118 | 0,128 | 0,106 37 | 0312 | 0251 | 0.204 | 168 0138 | 0137 | o1t6 | 0121 | 0103 38 | 0306 | 0246 | 0.200 | 0.164 131 | 0134 | 0113 | 0119 | oor 38 | 0300 | o242 | o196 | oe 129 | ors | ott | ons | 0099 40 | 0204 | 0237 | o192 | 0,158 0,126 | 0128 | 0108 | 0114 | 0097 '? = number of laboratories at a given level n= number of tost resuits per cell (see 733.3)ISO 5725-2:1994(E) 8.2 Critical values for Grubbs’ test (see 7.3.4) are given in table 5. For the Grubbs’ test for one outlying observation, outliers and stragglers give rise to values which are larger than the tabulated 1% and 5 % critical values respectively. ©1sO For the Grubbs’ test for two outlying observations, outliers and streggless give rise to values which are ‘smaller than the tabulated 1 % and 5 % critical values respectivay 8.3 Indicators for Mandel's h and k statistics (see 7.3.1) ere given in tables 6 and 7. Table 5 — Criticat values for Grubbs' test (One largest or one smallest Two largest or two smallest f ‘Upper 1% ‘Upper 5 % Lower 1% Lower 5%. 3 1.155 1,188 7 = 4 1,496 4,481 0,000 0 0,000 2 5 1,784 1715 0.001 8 0,009 0 6 11973 1,887 006 0,034 0 7 2139 2,020 0.0308 0,070 8 s 2274 228 0,088 3 1101 8 2387 2215 0,085 1 0,149 2 10 2,482 2290 0.1150 0,186 4 " 2584 2385 ogee 02213 2 2,696 2412 01738 0253 7 13 2,699 2,462 0.2016 0.283 6 4 2755 2507 0.2280 o3n2 16 2,806 2549 0.2530 03967 16 2852 2585 uZns 1 usseu 3 7 2,894 2,620 0.2990 082 2 18 2932 2.681 03200 0.4025 19 2,968 2,681 03398 0421 4 20 3,001 2.709 0,358 0.4391 2 3,031 2733 03781 0,485 6 2 3,060 2.788 03927 047i 4 23 3,087 2781 0.408 5 0,485 7 28 3,112 2,802 04234 0,499 4 25 3135 2922 0.437 6 05123 26 3157 2841 0451 0 08245 2 3178 2.959 0,463 8 0,536 0 28 3,199 2876 9.475 9 0547 0 29 3218 2893 0.487 5 0.587 4 30 3.298 2.908 0.498 5 0.867 2 an 3253 2924 0,509 1 0,576 6 32 3270 2998 05192 0,585 6 33 3.286 2962 0528 8 0,584 1 34 3,301 2.965 0.838 1 0,602 3 35 3316 2979 aan @ o8101 36 3,330 2,991 0,555 4 06175 7 3343 3,003 0,563 6 06247 38 3,386 3,014 05714 0,631 6 38 3,369 3,025 0,678 9 0,638 2 40 3381 3,038 0,696 2 0,644 5 Reproduced, with the permission of the American Statistical Association, from reference [4Jin annex C. ‘P= number of laboratories at a given levelTable 6 — Indicaturs for Mandel's / and & ctatietics at the 1% significance level z P h 7 2 3 4 6 é 7 8 3 70 3 as Tum | vee | ass [153 | tao | 146 | 143 | iar | 139 4 aas | nar | az | ver | io | 15s} 1st | 148 | 145 | 1.43 6 qe | 20s | ves | 173 | tes | 159 | nes | isi | 148 | 146 6 | ie | 21a | aso | 277 | tee | te2 | 187 | 183) 150 | 147 7 tse | 220 | u9¢ | wwe | uz | 10s | vse | ase | ner | ie 8 zoe | 225 | 197 | ver | ou | 165 | 1s | 185 | 12 | 149 8 213 | 223 | 199 | vez | 17a | 166 | 160 | 186 | 159 | 1.50 wo | 218 | 232 | 200 | 194 | 174 | ros | ver | 187 | 153 | 1.80 | oa | 234 |} zor | tes | onze | ter | 62 | 3s7 | ie | 151 ta | 2a | 236 | 202 | 19s | 175 | ase | vez | yss | toe | 1s we) oa | te | ee | tS | te | vee | te | dss | va | ie te | 230 | 230 | 20a | tar | 176 | 109 | 163 | 158 | 155 | 1.52 | 232 | 2a | 20s | ter | 176 | rsa | tes | aso | tes | 152 Ye | a3: | 2a | 205 | 193 | r77 | 1s9 | nes | 158.) 155 | 182 3 | 23s | 24a | 208 | vea | 177 | 160 | et | 15 | 185 | toe ye | 23 | 24s | 206 | 198 | uz? | 170 | ea | 188 | 186 | 182 jo | 237 | 24a | 207 | veo | 7a | 170 | 184 | 159 | 188 | 183 2 | 239 | 248 | 207 | 129 | 178 | 170 | vos | 160 | 186) 1.89 RN | X55 | as | aor | va | x78 | 170 | nes) 160 | 158 | 158 % | 2a | 24 | 208 | 190 | 178 | 170 | 165 | 1.60 | 158 | 188 | aa | 247 | 208 | sg | 17a | um | 16s | to | 158 | 153 2a | 2a | 247 | 208 | v0 | 479 | x7 | 165 | ie0 | 186 | 188 3s | ga | 247 | 208 | 190 | 479 | am | 165 | 160 ) 1.56} 1.88 ye | gas | aa | 209 | 190 | x70 | um | ues | neo | 156 | 1.53 | 2a | 2a | 209 | 190 | 179 | a7 | 165 | teo | 186 | 1.88 a | 2a | 249 | 209 | 191 | nya | ui | 165 | 160 | 187 | 183 o | 2a | 249 | 200 | 191 | 179 | um | 165 | 60 | 157 | 1.88 3 | aus | zag | 210 | 19 | ue | um | vos | ner | 157 | 188 pp = number of laboratories at a given level ‘n= ournber of replicates within each laboratory at that level NOTE — Supplied by Dr. J. Mandel and published with his petmission.'SO 5725-2:1994(E) ©1so Table 7 — Indicators for Mandel's i and & statistics at the 5 % significance level & > a » 2 3 4 5 6 7 a 9 70 3 as |) 165 | oass ] 14s [ao | iar | isa | naz | ao | 129 4 waz | 476 | 159 | 180 | 144 | 140 | 137 | 438 | 133 | 131 5 187 | 1ar | 162 | 13 | 146 | raz | 139 | 1396 | 134 | 132 6 iso | a5 | 164 | 154 | tes | ras | tao | naz | 135 | 133 7 um | iar | tes | iss | tes | naa | tar | age | va6 | 134 a 175 | 188 | 167 | 156 | 150 | 145 | 141 | 198 | 136 | 136 9 178 | 190 | 168 | 157 | 150 | 145 | 142 | 139 | 138 | a5 10 190 | 190 | 168 | 457 } v.50 | vas | 142 | ago | aaz | 125 u 182 | ar | 169 | 168 | 181 | a6 | 142 | 139 | saz | 135 12 183 | 192 | 169 | 158 | 151 | 148 | 142 | tao | 137 | 135 13 194 | 192 | 169 | 1588 | 151 | 146 | 143 | 140 | 137 | 135 4 vas | 192 | 130 | 150 aay | aaa | tao | ta7 | tes 18 196 | 193 | 1.70 | 159 ee 16 198 | 193 | 170 | 459 | 152 | aa7 | 143 | nao | 13a | 136 "7 197 | 493 | 170 | 159 | 152 | var | 143 | ago | x38 | 136 18 x98 | 393} 171 | 4s9 | 1s2 | naz | 143 | rao | nae | 136 19 a9} noe | ar | aso | use | az | as | tao | ase | 136 20 189 | 198 | 171 | 489 | s2 | naz | 143 | veo | uae | 136 2 ies | isa | az | neo | ts2 | var | ova | vat | 3a | 36 22 189} 194 | oat | aeo | 182 | a7 fora | oar | ug | 136 2B 190 | 44 | a7 | 160 | 153 | az | tee | tat 138 | 1.36 24 190 | vse | iz noo | ores | aaa | 4s | rar | 138 | 138 25 190 | s9¢ | a7 | reo | a3 | ras | 4s | tar | 132 | 136 26 190 | 196 | 471 | 160 | 183 | tas | ras | tar | tgs | t36 27 ai} aes | a7 to 160 | 163 | nas | vas | ver | rae | 136 28 ust | tea | am | reo | asa | vas | ise | ver | vga | 136 29 naif age | azz | neo } vss | as | tas | tat | 138 | i368 30, aor_| 196 | on72_ | 60 [1s | naa | tas | ova | 138] 136 _P = number of laboratories at @ given level ‘umber of replicates within each laboratory at thet level NOTE — Supplied by Dr. J. Mandel and published with his permission. 24isa Buy. Bay ete. acc Cow © CD, cR, Annex A (normative) ISO 5725-2:1994(E) Symbols and abbreviations used in ISO 5725 Intercept in the relationship s=atbm Factor used to calculate the uncer- tainty of an estimate Slope in the relationship ssatbm Component in a test result repre- senting the deviation of a laboratory from the general average (laboratory ‘component of bias) Component of B representing all factors that do not change in inter- mediate precision conditions Components of B representing fac- tors that vary in intermediate pre- cision conditions Intercept in the relationship Igs=c+digm Test statistics Critical values for statistical tests Critical difference for probability Critical range for probability P Slope in the relationship Igs=etdigm Component in a test result repre- senting the random error occurring in every test result Critical range factor p-quantile of the F-distribution with + and vp degrees of freedom Grubbs’ test statistic Mandel’s betweenlaboratory con: sistency test statistic RM uch Mandel's withinlaboratory consistency statistic test Lower control limit (either action tlmit or warning, limit) General mean of the test property; level Number of factors considered in intermediate precision conditions Number of iterations Number of test resulls obtained in ome labora tory at one level fie. per cell) Number of laboratories participating in the inter- laboratory experiment Probability Number of levels of the test property in the interlaboratory experiment Repeatability limit Reproducibility limit Reference material Estimate of a standard deviation Predicted stancard daviation Total or sum of some expression Number of test objects or groups Upper control limit (either action limit or warning lieit) Weighting factor used in calculating a weighted regression Range of a set of test results Datum used for Grubbs" test Test result 28ISO 5725-2:1994(E) ee ee Arithmetic mean of test results Grand mean of test results Significance level Type Il error probability Ratio of the reproducibility stonderd deviation to the repeatability standard deviation (agla,) Laboratory bias Estimate of 4 Bias of the measurement method Estimate of 6 Detectable difference between two laboratory biases or the biases of two measurement methods True value or accepted reference velue of a test property Number of degrees of freedom Detecteble ratio between the ropoatobilty otan dard deviations of method B and method A ‘True value of a standard deviation Component in 9 toot result roprosenting tho variation due to time since last calibration Detectable ratio between the square roots of the betweerlaboratory mean squares of method 8 and method A 2G) p-quantile of the ;?distrbution with » degrees. 26 of freedom ‘Symbols used as subscr sozage (1), @, @), elso Calibration-ifferent Equipment-different Identifier for a particular laboratory Idontifior for intermediate moacures of precision; in brackets, identification of the type of intermediate situation Identifier for a (iSO 5726-2), Identifier for a group of tests or for a factor (ISO 5725-3) particular level Identifier for @ particular test result in a laboratory é at level j Betweenlaboratory (interlaboratory) Identifier for detectable bias Botween-test-sample Operatordifferent Probability Repeatabil Reproducibility Time-different Withinaboratory (intralaboratory) For toot results, numboring in the order of obtaining them For test results, numbering in the order of increasing magnitude©1so 1SO 5725-2:1994(E) Annex B (informative) Examples of the statistical analysis of precision experiments ‘Table B.1 — Original data: Sulfur content of coal B.1. Example 1: Determination of the sulfur content of coal (Several levels with Laboratory i Level j no missing or outlying data) 1 2 3 4 on | 120 | 168 B.1.1. Background 1 om | 118 | 170 070 | 123 | 198 a) Measurement method on | v2 | 169 Determination of the sulfur content in coal with : per advan iia test results expressed as 2 percentage by mass. oes | 122 | 165 b) Source a 3 oss | 131 | 161 Tomkins, §. S. Industrial and Engineering Chem- oes | 130 | 1.62 istry, (See reference [6] in annex C.) 67 | 123 | 1.68 4 os | 110 | 186 c) Deseription 966 | 1.20 | 1.68 ! Eight laboratories participated in the experiment, prorat tee = carrying out the analysis according to @ standard- 7 eet tttee tt yeo tite ized measurement method described in tho 008 | | tee | gar source cited. Laboratory 1 reported four test re ooo | — | res | 326 sults and laboratory 5 reported four or five; the a or | 1a9 [170 | 327 other laboratories all carried out three measure~ 5 oy | tse | ie | ast ments. ors | 1a7 | 170 | 320 ¢) Graphical presentation on | 120 | 169 | 3.27 7 on | 126 | 1.70 | 324 Mandet's h end k statistics should be plotted, but ogo | 126 | ves | 3.23 because in this example they sliuwed litle of note ovo | use | ve? | 326 they have been omitted in order to allow space for 8 oes | 122 | ve8 | 926 a different example of the graphical presentation oss | 130 | 167 | 3.26 of data. Mandel's plots are fully illustrated and discussed in the example given in B.3. NOTE 8 For the experiment quoted in table B.1, the lab- oratories were not instructed as to how many measure- ‘ments were to be made, only a minimum number. By the | recommended procedures given in this part of SO 5725. for B.1.2. Original data laboratories 1 and § a random selection should be made {rom the values given in order to reduce all cells to exactly These are given, 2s percentage by mass [% (wni], ee test results. However, in order to ilustrste the com in table8.1 in the format of form A of figure? (see butaonal procedures fr variable nares of Test fasits, nd ce ite any specific remark all test resu been retained in this example. The Eectaliledecteestobaiutlldesrasanea teint reader may make random selections to reduce the number Graphical presentations uf these deta are given in fi ‘of test resuits to three in each coll if hefshe wishes to verify ety s ‘© thet such a procedure has relatively litle effect on the val 2 off 5, ad 2ISO 5725-2:1994(E) ©1so B13 Computation of cell means (5,) For level 1, largest value of sis in laboratory &: The cell means are given, as a percentage by mass 0,001 82; test value ~ 0,347 [% Im/m}]. in table R 2 in the format of form B of fig ure 2 (see 7.2.9). For level 2, largest value of - is in laboratory 5: B.1.4 Computation of standard deviations (s) £5? = 0,006 36; test value ~ 0,287 The standard deviations are given, as 2 percentage by mass [%6 (n/n], in table B.3 In the format of form For level 3, argest value of sis in laboratory 5 of figure 2 (see 7.2.10) 0,001 72, twst value = 0,598 B.1.5 Scrutiny for consistency and outliers Cochran's test with n= 3 for p= 8 laboratories gives For level 4, largest value of sis in laboratory 4: critical values of 0,516 for 5 % and 0,816 for 1%, : Es? = 0,004 63; test value = 0,310 Table B.2 — Cell means: Sulfur content of coal Level Laboratory i 1 2 3 4 Jy. my Hi, ry Jy. Li i *y 7 0708 | 4 [1206 [4 [1.688 [4 | 3240 | 4 2 o80 | 3 | 1217 | 3 | 183 | 3 | 3200 | 3 3 ose7 | 3 | 127 | 3 | 1613 | 3 | 337 | 3 a oso | 3 | 1203 | 3 | 1007 | 3 | 3203 | 5 5 690 | 5 | 128 | 4 | 1650 | 5 | 326 | 5 6 0733 | 3 | 1379 | 3 | 1720 | 3 | 3200 | 3 7 0703 | 3 | 120 | 3 | 1690 | 3 | 3247 | 3 8 oe77 | 3 | 1.253 | 3 | 1673 | 3 | 3257 | 3 ‘Table B.3 — Standard deviations: Sulfur content of coal Level Fl Laboratory i 1 2 3 4 ws {[w| % |w] oy [w|i [wy 1 ooo fs | oo2, | 4 | 0010 | 4 | ooz ta] 2 oo10 | 3 | 0006 | 3 3 | 0000 | 3 3 oz | 3 | oo | 3 3 | 0010 | 3 4 ooo | 3 | 002 | 3 3 | 0008 | 3 5 org | 5 | oa | 4 5 | 0038 | 5 . 0006 | 3 | oos | 3 3 | 0020 | 3 7 oo | 3 | oss | 3 3 | 002 | 3 8 02 | 3 | 004 | 3 3 | 0006 | 3 285 iso 2 xx 3 xx ‘ xxx 3 x bax 6 Be n x 8 x xf ae aaa Figure B.1 — Sulfur content of coal, sample 1 1 rox 2{ ei > me . wx 34 bl x 4 x a wx] x uw 2 1 43 ah 1S 0. % Figure B.2 — Sulfur content of coal, sample 2 1S0 5725-2:1994(E) 29ISO 5725-2:1994(E) 30 1 Bex 2 & 3{ Bx { lx st x x shoe “{ x & 7 ee + WPS TT TE TAT a, Figure B.3 — Sulfur content of coal, sample 2 1{ x x af x 3 veo ‘ x xx 5 voor xb ‘ pec , vo at 33a Pad ae aa Figure B.4 — Sulfur content of coal, sample 4 iso2 1s0 This indicates that one cell in level 3 may be regarded as a straggler, and there are no outliers. The streggler is retained in subsequent calculations, Grubbs’ tests were applied to the cell means, giving the values shown in tableB.é, There ate no single stragglers or outliers. At levels 2 and 4, the high re- suits for laboratories 3 and 6 are stragglers according to the doublehigh test; these were retained in the analysis B.1.6 Computation of ii, 5, and sq; ‘The variances defined in 7.4.4 and 7.4.5 are calculated a8 follows, using level 1 as an example. Number of laboratories, p = 8 T, = En ¥,= 18,642 T, 5x, (3)? ~ 12,883 7 T,=2n=27 Ty = En} = 95 Ty =2(q)—1)s2 = 0,004 411 ,000 232 2 ISO 5725-2:1994(E) ge Rh-T | [ Be-v "Be | | aaa, ,000 480 3 0,000 692 5 A= 2 20,890 44 Ty 5, = 0,015 24 Se = 0,026 32 The calculations for levels 2, 3 and 4 may be carried out similarly to give the results shown in table B.5, B.1.7 Dependence of preci On m An examination of the data in table B.§ does not indi- cate any dependence and average velues can be used. B.1.8 Conclusions The precision of the measurement method should be quoted, es 2 percentage by mass, as repeatability standard deviation, s, = 0,022 reproducibility standard deviation, s— = 0,045 Table B.4 — Application of Grubbs’ test to cell means Singla | Single | Double | Double natal tow high tow high Leelee 1 1.24 1,80 0539 0.298 2 0.91 2.09 0.699 0,108 | Grubbs’ test 3 167 158 0378 0,469 | statistics 4 294 2.03 0,679 0,132 Straggiers 2.126 2.126 | 0.1104 | 0.1101 | Grips esitical oie 2278 2274 | 00563 | 0.0563 | valves Table B.S — Computed values of #4, s, and sy for sulfur content of coal Level a % 4 sn 1 8 0,680) 0015 0.026 2 8 4,282 0.029 0.061 3 8 1867 0.017 0.035 5 é 3.250 0.026 o.088 31oU vreo-e9veIE) These values may be applied within a range 0,69 % (mim) 10 3,25 % (mim). They were deter mined from a uniform-evel experiment involving 8 laboratories covering that range of values, in which four stragglers were detected and retamed. B.2 Example 2: Softening point of pitch (Several levels with missing data) B.2.1 Background a) Measurement method The determination of the softening point of pitch by ring and ball, >) Source ‘Standard methods for testing tar and its products; Pitch section; Method Sariai No. PT3 using neutral glycerine (reference [5] in annex C). cl Material Thie was sclected from commercial batches of pitch collected and prepared as specified in the “Semples” chapter of the pitch section of refer- ence [5]. 4d) Description ‘This was the determination of a property involving temperature measurement in degrees Celsius. Sixteen leboratories cooperated. Ik wes intended to measure four specimens at about 87,5 °C, 92,5 °C, 97.5 *C and 102,5 °C to cover the normal ‘commercial range of products, but wrong material was chosen for level 2 with a mean temperature of about 96 °C which was similar to level 3. Lab- oratory 5 applied the method incorrectly at first fon the sample for level 2 (the first one they measured) and there was then insufficient ma- terial remaining for more than one determinotion Laboratory 8 found that they did not have a sam- ple for level 1 (they had two specimens for level 4) e) Gray I presentations 7 Mandel's h and & statistics should be plotted, but again in this example they have been omitted in 32 ©1s0 order to provide for another type of graphical presentation of data. Mandel’s plots are fully illustrated and discussed in the example given in 83. B.2.2 Original data These are presented in table B.6, in degrees Celsivs, in the format of form A of figure 2 (see 7.2.8) Table B.6 — Original data: Softening point of piteh (°C) Laborator Level vita fet sae ; 910 | 970 | v5 | 1080 eas | 972 | 570 | 1040 i 7 | 95 | 972 | 1026 eae | 972 | a70 | toa : wo | o7@ | o42 | 1030 875 | 945 | 958 | 995 i 92 | 968 | e60 | 10258 285 | 975 | 980 | 1035 e9.0 | 97.2 | 982 | 1010 i 9.0 | — | 985 | 1002 i. e5 | 97.8 | 995 | 1022 90.5 | 97.2 | 1032 | 1020 5 9 | 966 | 982 | 1028 v8.2 | 97.5 | 990 | 1022 7 — | 960 | s34 | i026 — | 975 | 974 | 1039 a wr | 955 | 982 | 1028 e34 | 968 | 96.7 | 1020 os 260 | 952 | oe | 98 858 | 950 | 930 | 108 oe | 932 | 36 | 902 saa | 934 | 939 | ove 1 262 | 958 [958 | q017 a74 | 954 | 954 | 1012 a0 | 082 [980 | t045 904 | 995 | 970 | 1056 7 @78 | 970 | 971 | 1052 avs | 95 | 966 | i018 " 13 278 | 950 | 978 | i015 e768 | 952 | 992 | 1009 Hu eae | 950 [972 | 998 50 | 93,2 | 978 | 998 8 NOTE — There are no obvious stragglers or statistical outliers©1s0 B.2.3. Coll means These are given in table B.7, in degrees Celsius, in the forinat uf foun B of figure? ise 7.2.9) A graphical presentation of these data is given in figy weBS, ISO 5725-2:1994(E) B.2.4 Absolute differences within cells In this example there are two test results per cell and the absolute uifference can be used to represent the variability. The absolute differences within cells, in degrees Celsius, are given in tableB.8, in the format of form C of figure2 ise 7.2.10) A graphical presentation of these data is given in fig- ure. Table B.7 — Cell means: Softening point of pitch (°C) Level j Laboratory “ 2 3 4 — —_— 1 90,30 87.10 96,75 104,00 2 80,75 97.88 97,10 103,10 3 87.75 96,15 95,00 101,26 4 88,85 7.15 97,00 108,00, 5 869,50 = 98,35 100,80 6 89,50 97,50 101,35 102,10 7 88,55 87,05 98,60 102,50 8 ~ 96,75 97.80 103,25 9 89,25 96.15 9745 102.40 10 85.90 10 93.80 100.30 u 86,00 93.30 93,75 98,00 2 97,80 95,80 95,60 101,45 13 90,70 98.85 97.50 105.05 uw 87.85 96,25 96,85 103,50 15 87.55 95,10 98,50 101,20 16 36,90 94,10 97,50 99,65, NOTE — The entry for = 2 has been dropped (see 7.4.3) Table B.8 — Absolute differences within cells: Softening point of pitch (°C) Level j Laboratory! ; : : a 1 14 02 05 0.0 2 01 13 02 10 3 05, 33 18 38 4 o7 07 20 10 5 10 = aa ag 6 20 os 37 oz 7 07 og og os 8 = 18 10 13 9 17 13 15 og 10 02 02 18 10 u 32 02 03 04 2 os oa Oa 05 13 08 13 10 wi 4 03 18 Os 34 is 0 oz 14 O68 16 38 18 os 03 33ISO 5725-2:1994(E) B25 Scrutiny for consistency and outliers Application of Cochran's test leads to the values of the test statistic C given in table b.9. The critical values (see 8.1) at the 5% significance level are 0,471 for p = 15 and 0,482 for p = 16 where n= 2. No stragglers are indicated Grubbs’ tests were applied to the cell means. No single of double stragglers or outliers were found. B.2.6 Computation of ii, 5, and 55, ‘These are calculated as in 7.4.4 and 7.4.5. Using level 1 for example, the calculations are as fol lows. To ease the arithmetic, 00,00 has boon sub tracted from all the data, The method for n=2 replicates per cell is used. Number of laboratories, p = 15 eso Number of replicates, » = 2 125,950 0 gages ty a 5= 1.1092 sq = 1,668 7 aa (add 80,00) = 89,296 6 ‘The values for all four levels are given in table B.11 Table B.9 — Values of Cochran's test statistic, C Bal 7 leas 7 —__ 7 canna | omne | oaane | omen NOTE — Number of laboratories is given in parentheses, le B.10 — Application of Grubbs’ test to cell means Single | Single ] Double | Double tow. high low high Type of test 7,69 158 0546 0,862 208 177 0478 0,646 | Grubbs’ test 176 2.27 osae 0,566 | statistics 222 174 0,500 0,672 2,549, 209 | 03367 | 03367 21585, 2.585 | 09603 | 03603 | Grubbs critical values 2,806 2906 | 02630 | 0,250 2,052 2e62_| 02767 | 02767 34‘© 1sO ISO 5725-: 1994(E) Table B.11 — Computed values of f, 5, and sy for softening point of pitch Level j y my PC) Sa Sn 7 18 83.40 7.108) 1.670 2 18 9627 0.925 11897 3 6 97.07 0.933 2010 4 6 101,96 1,004 4915 Be FRE poe evel Frequency 6 30 700 70 Temperature," Figure B.5 — Softening point of pitch: Cell means 351SO 5725-2:1994(E) SETS] Heras ©1so oh A Level 4 ees Frequency Ts) B.2.7 Dependence of precision on m ‘A cursory examination of table B.11 does not reveal any marked dependence, except perhaps in repro-

You might also like

• Her Body and Other Parties: Stories

  

  From Everand

  Her Body and Other Parties: Stories

  Carmen Maria Machado

  Rating: 4 out of 5 stars

  4/5 (822)
• The Sympathizer: A Novel (Pulitzer Prize for Fiction)

  

  From Everand

  The Sympathizer: A Novel (Pulitzer Prize for Fiction)

  Viet Thanh Nguyen

  Rating: 4.5 out of 5 stars

  4.5/5 (122)
• The World Is Flat 3.0: A Brief History of the Twenty-first Century

  

  From Everand

  The World Is Flat 3.0: A Brief History of the Twenty-first Century

  Thomas L. Friedman

  Rating: 3.5 out of 5 stars

  3.5/5 (2259)
• • Magazines
  • Podcasts
  • Sheet music
• A Heartbreaking Work Of Staggering Genius: A Memoir Based on a True Story

  

  From Everand

  A Heartbreaking Work Of Staggering Genius: A Memoir Based on a True Story

  Dave Eggers

  Rating: 3.5 out of 5 stars

  3.5/5 (231)
• A Man Called Ove: A Novel

  

  From Everand

  A Man Called Ove: A Novel

  Fredrik Backman

  Rating: 4.5 out of 5 stars

  4.5/5 (4610)
• Sing, Unburied, Sing: A Novel

  

  From Everand

  Sing, Unburied, Sing: A Novel

  Jesmyn Ward

  Rating: 4 out of 5 stars

  4/5 (1104)
• The Light Between Oceans: A Novel

  

  From Everand

  The Light Between Oceans: A Novel

  M.L. Stedman

  Rating: 4.5 out of 5 stars

  4.5/5 (789)
• Fear: Trump in the White House

  

  From Everand

  Fear: Trump in the White House

  Bob Woodward

  Rating: 3.5 out of 5 stars

  3.5/5 (738)
• Rise of ISIS: A Threat We Can't Ignore

  

  From Everand

  Rise of ISIS: A Threat We Can't Ignore

  Jay Sekulow

  Rating: 3.5 out of 5 stars

  3.5/5 (137)
• Team of Rivals: The Political Genius of Abraham Lincoln

  

  From Everand

  Team of Rivals: The Political Genius of Abraham Lincoln

  Doris Kearns Goodwin

  Rating: 4.5 out of 5 stars

  4.5/5 (234)
• The Woman in Cabin 10

  

  From Everand

  The Woman in Cabin 10

  Ruth Ware

  Rating: 3.5 out of 5 stars

  3.5/5 (2521)
• The Outsider: A Novel

  

  From Everand

  The Outsider: A Novel

  Stephen King

  Rating: 4 out of 5 stars

  4/5 (1850)
• A Tree Grows in Brooklyn

  

  From Everand

  A Tree Grows in Brooklyn

  Betty Smith

  Rating: 4.5 out of 5 stars

  4.5/5 (1929)
• Yes Please

  

  From Everand

  Yes Please

  Amy Poehler

  Rating: 4 out of 5 stars

  4/5 (1898)
• The Yellow House: A Memoir (2019 National Book Award Winner)

  

  From Everand

  The Yellow House: A Memoir (2019 National Book Award Winner)

  Sarah M. Broom

  Rating: 4 out of 5 stars

  4/5 (98)
• The Unwinding: An Inner History of the New America

  

  From Everand

  The Unwinding: An Inner History of the New America

  George Packer

  Rating: 4 out of 5 stars

  4/5 (45)
• John Adams

  

  From Everand

  John Adams

  David McCullough

  Rating: 4.5 out of 5 stars

  4.5/5 (2409)
• The Perks of Being a Wallflower

  

  From Everand

  The Perks of Being a Wallflower

  Stephen Chbosky

  Rating: 4.5 out of 5 stars

  4.5/5 (2104)
• On Fire: The (Burning) Case for a Green New Deal

  

  From Everand

  On Fire: The (Burning) Case for a Green New Deal

  Naomi Klein

  Rating: 4 out of 5 stars

  4/5 (74)
• The Emperor of All Maladies: A Biography of Cancer

  

  From Everand

  The Emperor of All Maladies: A Biography of Cancer

  Siddhartha Mukherjee

  Rating: 4.5 out of 5 stars

  4.5/5 (271)
• Devil in the Grove: Thurgood Marshall, the Groveland Boys, and the Dawn of a New America

  

  From Everand

  Devil in the Grove: Thurgood Marshall, the Groveland Boys, and the Dawn of a New America

  Gilbert King

  Rating: 4.5 out of 5 stars

  4.5/5 (266)
• Brooklyn: A Novel

  

  From Everand

  Brooklyn: A Novel

  Colm Tóibín

  Rating: 3.5 out of 5 stars

  3.5/5 (1947)
• Wolf Hall: A Novel

  

  From Everand

  Wolf Hall: A Novel

  Hilary Mantel

  Rating: 4 out of 5 stars

  4/5 (3811)
• The Constant Gardener: A Novel

  

  From Everand

  The Constant Gardener: A Novel

  John le Carré

  Rating: 3.5 out of 5 stars

  3.5/5 (104)
• Manhattan Beach: A Novel

  

  From Everand

  Manhattan Beach: A Novel

  Jennifer Egan

  Rating: 3.5 out of 5 stars

  3.5/5 (792)
• The Art of Racing in the Rain: A Novel

  

  From Everand

  The Art of Racing in the Rain: A Novel

  Garth Stein

  Rating: 4 out of 5 stars

  4/5 (4203)
• Hidden Figures: The American Dream and the Untold Story of the Black Women Mathematicians Who Helped Win the Space Race

  

  From Everand

  Hidden Figures: The American Dream and the Untold Story of the Black Women Mathematicians Who Helped Win the Space Race

  Margot Lee Shetterly

  Rating: 4 out of 5 stars

  4/5 (897)
• The Little Book of Hygge: Danish Secrets to Happy Living

  

  From Everand

  The Little Book of Hygge: Danish Secrets to Happy Living

  Meik Wiking

  Rating: 3.5 out of 5 stars

  3.5/5 (401)
• Angela's Ashes: A Memoir

  

  From Everand

  Angela's Ashes: A Memoir

  Frank McCourt

  Rating: 4.5 out of 5 stars

  4.5/5 (441)
• APQP Checklist: Supplier: Supplier Code: Part Name: Part Number

  

  Document2 pages

  APQP Checklist: Supplier: Supplier Code: Part Name: Part Number

  RAHUL HERANI

  100% (2)
• Grit: The Power of Passion and Perseverance

  

  From Everand

  Grit: The Power of Passion and Perseverance

  Angela Duckworth

  Rating: 4 out of 5 stars

  4/5 (590)
• The Hard Thing About Hard Things: Building a Business When There Are No Easy Answers

  

  From Everand

  The Hard Thing About Hard Things: Building a Business When There Are No Easy Answers

  Ben Horowitz

  Rating: 4.5 out of 5 stars

  4.5/5 (348)
• The Subtle Art of Not Giving a F*ck: A Counterintuitive Approach to Living a Good Life

  

  From Everand

  The Subtle Art of Not Giving a F*ck: A Counterintuitive Approach to Living a Good Life

  Mark Manson

  Rating: 4 out of 5 stars

  4/5 (5811)
• Bad Feminist: Essays

  

  From Everand

  Bad Feminist: Essays

  Roxane Gay

  Rating: 4 out of 5 stars

  4/5 (1018)
• Elon Musk: Tesla, SpaceX, and the Quest for a Fantastic Future

  

  From Everand

  Elon Musk: Tesla, SpaceX, and the Quest for a Fantastic Future

  Ashlee Vance

  Rating: 4.5 out of 5 stars

  4.5/5 (474)
• Principles: Life and Work

  

  From Everand

  Principles: Life and Work

  Ray Dalio

  Rating: 4 out of 5 stars

  4/5 (608)
• Little Women

  

  From Everand

  Little Women

  Louisa May Alcott

  Rating: 4 out of 5 stars

  4/5 (104)
• Shoe Dog: A Memoir by the Creator of Nike

  

  From Everand

  Shoe Dog: A Memoir by the Creator of Nike

  Phil Knight

  Rating: 4.5 out of 5 stars

  4.5/5 (540)
• The Glass Castle: A Memoir

  

  From Everand

  The Glass Castle: A Memoir

  Jeannette Walls

  Rating: 4.5 out of 5 stars

  4.5/5 (1716)
• Never Split the Difference: Negotiating As If Your Life Depended On It

  

  From Everand

  Never Split the Difference: Negotiating As If Your Life Depended On It

  Chris Voss

  Rating: 4.5 out of 5 stars

  4.5/5 (844)
• Standard ISO TR 22514 4 2007 9658 PDF

  

  Document56 pages

  Standard ISO TR 22514 4 2007 9658 PDF

  RAHUL HERANI

  No ratings yet
• Steve Jobs

  

  From Everand

  Steve Jobs

  Walter Isaacson

  Rating: 4.5 out of 5 stars

  4.5/5 (807)
• The Gifts of Imperfection: Let Go of Who You Think You're Supposed to Be and Embrace Who You Are

  

  From Everand

  The Gifts of Imperfection: Let Go of Who You Think You're Supposed to Be and Embrace Who You Are

  Brene Brown

  Rating: 4 out of 5 stars

  4/5 (1092)
• GUM Guidelineson Mesuarementof Uncertainity

  

  Document16 pages

  GUM Guidelineson Mesuarementof Uncertainity

  RAHUL HERANI

  No ratings yet
• Bsi BS 600 PDF

  

  Document176 pages

  Bsi BS 600 PDF

  RAHUL HERANI

  100% (1)
• Gunavardhan Digest

  

  Document4 pages

  Gunavardhan Digest

  RAHUL HERANI

  No ratings yet