1、Designation: D 2777 08Standard Practice forDetermination of Precision and Bias of Applicable TestMethods of Committee D19 on Water1This standard is issued under the fixed designation D 2777; the number immediately following the designation indicates the year oforiginal adoption or, in the case of re
2、vision, the year of last revision. A number in parentheses indicates the year of last reapproval. Asuperscript epsilon (e) indicates an editorial change since the last revision or reapproval.1. Scope1.1 This practice establishes uniform standards for estimat-ing and expressing the precision and bias
3、 of applicable testmethods for Committee D19 on Water.1.2 Except as specified in 1.3, 1.4, and 1.5, this practicerequires the task group proposing a new test method to carryout a collaborative study from which statements for precision(overall and single-operator standard-deviation estimates) andbias
4、 can be developed. This practice provides general guidanceto task groups in planning and conducting such determinationsof precision and bias.1.3 If a full-scale collaborative study is not technicallyfeasible, due to the nature of the test method or instability ofsamples, the largest feasible scaled-
5、down collaborative studyshall be conducted to provide the best possible limited basis forestimating the overall and single-operator standard deviations.1.3.1 Examples of acceptable scaled-down studies are thelocal-area studies conducted by Subcommittee D19.24 onmicrobiological methods because of inh
6、erent sample instabil-ity. These studies involve six or more completely independentlocal-area analysts who can begin analysis of uniform samplesat an agreed upon time.1.3.2 If uniform samples are not feasible under any circum-stances, a statement of single-operator precision will meet therequirement
7、s of this practice. Whenever possible, this state-ment should be developed from data generated by independentmultiple operators, each doing replicate analyses on indepen-dent samples (of a specific matrix type), which generally fallwithin specified concentration ranges (see 7.2.5.2 (3).1.3.3 This pr
8、actice is not applicable to methodology involv-ing continuous sampling or continuous measurement, or both,of specific constituents and properties.1.3.4 This practice is also not applicable to open-channelflow measurements.1.4 A collaborative study that satisfied the requirements ofthe version of thi
9、s practice in force when the study wasconducted will continue to be considered an adequate basis forthe precision-and-bias statement required in each test method.If the study does not satisfy the current minimum requirementsfor a collaborative study, a statement listing the studysdeficiencies and a
10、reference to this paragraph shall be includedin the precision-and-bias statement as the basis for an exemp-tion from the current requirements.1.5 This paragraph relates to special exemptions not clearlyacceptable under 1.3 or 1.4. With the approval of CommitteeD19 on the recommendation of the Result
11、s Advisor and theTechnical Operations Section of the Executive Subcommitteeof Committee D19, a statement giving a compelling reasonwhy compliance with all or specific points of this practicecannot be achieved will meet both ASTM requirements (1)2and the related requirements of this practice. In addi
12、tion,Committee D19, through a Main Committee ballot, mayapprove publication of a “Preliminary” Standard Method for aperiod not to exceed 5 years. Preliminary Standards mustcontain a minimum of a single-operator precision-and-biasstatement and a Quality Control section based on the singleoperator dat
13、a. Publication of a Preliminary Standard is condi-tional on the approval of a full D 2777 collaborative studydesign for the standard. Precision-and-bias statements autho-rized by this paragraph shall include the date of approval byCommittee D19.1.6 In principle, all test methods are covered by thisp
14、ractice.1.7 In Section 12 this practice shows exemplary precision-and-bias-statement formats for: (1) test methods yielding anumerical measure, (2) test methods yielding a non-numericalreport of success or failure based on criteria specified in theprocedure, and (3) test methods specifying that proc
15、edures inanother ASTM test method are to be used with only insignifi-cant modifications.1.8 All studies, even those exempt from some requirementsunder 1.3 or 1.5, shall receive approval from the ResultsAdvisor before being conducted (see Section 8) and aftercompletion (see Section 13).1This practice
16、 is under the jurisdiction of ASTM Committee D19 on Water andis the direct responsibility of Subcommittee D19.02 on Quality Systems, Specifi-cation, and Statistics.Current edition approved Jan. 15, 2008. Published February 2008. Originallyapproved in 1969. Last previous edition approved in 2006 as D
17、 2777 06.2The boldface numbers in parentheses refer to the list of standards at the end ofthis practice.1Copyright ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959, United States.1.9 This practice satisfies the QC requirements of Practice D5847.1.10 It is the
18、intent of this practice that task groups makeevery effort to retain all the data from their round-robin studies.Values should not be eliminated unless solid evidence exists fortheir exclusion. The Results Advisor should work closely withthe task groups to effect this goal.2. Referenced Documents2.1
19、ASTM Standards:3D 1129 Terminology Relating to WaterD 1141 Practice for the Preparation of Substitute OceanWaterD 1193 Specification for Reagent WaterD 4375 Practice for Basic Statistics in Committee D19 onWaterD 5790 Test Method for Measurement of Purgeable OrganicCompounds in Water by Capillary Co
20、lumn GasChromatography/Mass SpectrometryD 5847 Practice for Writing Quality Control Specificationsfor Standard Test Methods for Water AnalysisD 5905 Practice for the Preparation of Substitute Wastewa-terD 6091 Practice for 99 %/95 % Interlaboratory DetectionEstimate (IDE) for Analytical Methods with
21、 NegligibleCalibration ErrorD 6512 Practice for Interlaboratory Quantitation EstimateE 177 Practice for Use of the Terms Precision and Bias inASTM Test MethodsE 178 Practice for Dealing With Outlying ObservationsE 456 Terminology Relating to Quality and StatisticsE 1169 Practice for Conducting Rugge
22、dness Tests3. Terminology3.1 DefinitionsFor definitions of terms used in this prac-tice, refer to Terminologies D 1129, D 4375 and E 456, andPractice E 177.3.2 Definitions of Terms Specific to This Standard:3.2.1 accuracya measure of the degree of conformity ofa single test result generated by a spe
23、cific procedure to theassumed or accepted true value, and includes both precisionand bias.3.2.2 biasthe persistent positive or negative deviation ofthe average value of a test method from the assumed oraccepted true value.3.2.3 laboratorya single and completely independent ana-lytical system with it
24、s own specific apparatus, source ofreagents, set of internal standard-operating procedures, etc.Different laboratories will differ from each other in all of theseaspects, regardless of how physically or organizationally closethey may be to each other.3.2.4 operatorusually the individual analyst with
25、in eachlaboratory who performs the test method throughout thecollaborative study. However, for complicated test methods,the operator may be a team of individuals, each performing aspecific function throughout the study.3.2.5 precisionthe degree of agreement of repeated mea-surements of the same prop
26、erty, expressed in terms of disper-sion of test results about the arithmetical-mean result obtainedby repetitive testing of a homogeneous sample under specifiedconditions. The precision of a test method is expressedquantitatively as the standard deviation computed from theresults of a series of cont
27、rolled determinations.4. Summary of Practice4.1 After the task group has assured itself that the testmethod has had all preliminary evaluation work completed, thetask group should prepare the test-method write-up in finalform. The plan for collaborative study is developed in accor-dance with this pr
28、actice and submitted along with the test-method write-up to the Results Advisor for concurrence exceptas specified in 1.3, 1.4, and 1.5. Upon receipt of concurrence,the collaborative test is conducted, data analyzed, andprecision-and-bias statements formulated by the task group.Estimates of the lowe
29、r limits of quantitation and detection mayalso be developed. The final precision-and-bias statistics mustbe based on usable data from at least six independent labora-tories. The statements, with backup data including thereported-results summary, the calculations leading up to thestatements, and the
30、test method write-up with precision-and-bias statements included are submitted to the subcommitteevice-chairman, who in turn sends a copy to the ResultsAdvisorfor concurrence before balloting. This procedure assures hav-ing an acceptable copy of the collaborative-study results tosend toASTM for item
31、s on the main-committee ballot. In mostinstances, the collaborative study shall be complete before asubcommittee ballot. If the collaborative study is not complete,the test method may go on the ballot as a provisional testmethod rather than a standard test method. Copies of the testdata, approved ca
32、lculations, and statistical results shall be filedat ASTM Headquarters when the test method is submitted bythe subcommittee chairman as an item for the main-committeeballot.4.1.1 The appendix shows an example of “FormAApproval of Plans for Interlaboratory Testing,” as Fig.X1.1.4.1.2 For examples of
33、data-reporting forms, see AppendixX3, 6.0.4.1.3 In addition, the appendix shows a sample calculationof precision and bias from real collaborative-test data, therelated table of statistics, and the related precision-and-biasstatement.5. Significance and Use5.1 Following this practice should result in
34、 precision-and-bias statements that can be achieved by any laboratory properlyusing the test method studied. These precision-and-bias state-ments provide the basis for generic limits for use in the QualityControl section of the test method. Optionally, the detectionand quantitation values provide es
35、timates of the level at whichmost laboratories should be able to achieve confident detection3For referenced ASTM standards, visit the ASTM website, www.astm.org, orcontact ASTM Customer Service at serviceastm.org. For Annual Book of ASTMStandards volume information, refer to the standards Document S
36、ummary page onthe ASTM website.D2777082and meet the minimum precision (expressed as relative stan-dard deviation) expected.5.2 The method specifies the matrices for which the testmethod is appropriate. The collaborative test corroborates thewrite-up within the limitations of the test design. An exte
37、nsivetest can only use representative matrices so that universalapplicability cannot be implied from the results.5.3 The fundamental assumption of the collaborative studyis that the matrices tested, the concentrations tested, and theparticipating laboratories are a representative and fair evalua-tio
38、n of the scope and applicability of the test method as written.6. Preliminary Studies6.1 Considerable pilot work on a test method must precedethe determination of its precision and bias (2, 3). This pilotwork should explore such variables as preservation require-ments, reaction time, concentration o
39、f reagents, interferences,calibration, and sample size. Potentially significant factorsmust be investigated and controlled in the written test methodin advance of the collaborative test. Also, disregard of suchfactors may introduce so much variation among operators thatresults are misleading or inco
40、nclusive (4) (see 9.3 and 9.4). Aruggedness study conducted in a single laboratory is particu-larly useful for such investigations and should be conducted toprove a test method is ready for interlaboratory testing (seeGuide E 1169 for details).6.2 Only after a proposed test method has been tried,pro
41、ved, and reduced to unequivocal written form should adetermination of its precision and bias be attempted.6.3 If the task group intends to evaluate the method char-acteristics of detection and quantification, Practice D 6091(Interlaboraotry Detection Estimate) or Standard PracticeD 6512 (Interlabora
42、tory Quantitation Estimate), or both,should be evaluated and the recommendations for studydesigns incorporated.6.3.1 To minimize the number of samples required, datawould be gathered in two phases:Phase ISingle-laboratory characterization. In this phase, themethod developer would run a sufficient nu
43、mber of samples ata sufficient number of concentrations to characterize fullyresponse vs. concentration, as well as error vs. concentration.The lowest concentration would be the level of the blank or thelowest concentration that could be measured; the highestconcentration would be at the upper limit
44、 of the analyticalrange.Phase IICollaborative study. Using the results of Phase I, themethod developer would estimate the minimum number andthe magnitude of concentrations necessary to meet the require-ments of the documents of interest.7. Planning the Collaborative Test7.1 Based upon the task group
45、s knowledge of a test methodand the unequivocal write-up, several factors must be consid-ered in planning the collaborative test to assess the precision ofthe test method properly. The testing variables that must beconsidered in planning are discussed below. In the collabora-tive study, it is genera
46、lly not acceptable to control significantsources of variability that cannot be controlled in routine use ofthe test method, because this control leads to false estimates ofthe test-method precision and bias. In addition, the task groupmust determine within the resources available how best toestimate
47、 the precision and bias of the test method.7.2 Testing Variables:7.2.1 It is desirable to develop a test methods precisionstatement that indicates the contribution to overall variation ofselected causes such as laboratory, operator, sample matrix,analyte concentration, and other factors that may or
48、have beenshown to have strong effects on the results. Since any testmethod can be tried in only a limited number of applications,the standard deviation calculated from the results of a study canbe only an estimate of the universe standard deviation. For thisreason, the symbol s (sample standard devi
49、ation) is usedherein. The precision estimates generated from the study datawill usually be the overall standard deviation (sT) at any oneconcentration and the pooled single-operator standard devia-tion (so) for each sample matrix and concentration studied.7.2.2 Laboratories, operators, sample matrices, and analyteconcentrations are the only sources of variability represented inthe precision-and-bias statements resulting from the usualcollaborative study. These sources may not represent theadditional influence that can arise from differences in samplesplitting, field prese
