Designation:D4483−20StandardPracticeforEvaluatingPrecisionforTestMethodStandardsintheRubberandCarbonBlackManufacturingIndustries1ThisstandardisissuedunderthefixeddesignationD4483;thenumberimmediatelyfollowingthedesignationindicatestheyearoforiginaladoptionor,inthecaseofrevision,theyearoflastrevision.Anumberinparenthesesindicatestheyearoflastreapproval.Asuperscriptepsilon(´)indicatesaneditorialchangesincethelastrevisionorreapproval.INTRODUCTIONTheprimaryprecisionstandardforASTMtestmethodstandardsisPracticeE691;agenericstandardthatpresentsthefundamentalstatisticalapproachandcalculationalgorithmsforevaluatingrepeatabilityandreproducibilityprecision.However,certainpartsofPracticeE691arenotcompatiblewithprecisionasevaluatedintherubbermanufacturingandcarbonblackindustriesoverthepastfourdecades.Thusaseparatestandardisrequiredforprecisioninthesetwoindustries.ThispracticeisbeingissuedasamajorrevisionofPracticeD4483,whichhasbeenusedforprecisionevaluationbyCommitteeD11since1985.ThebasicPracticeD4483precisioncalculationalgorithms,thesameasinPracticeE691,areunchanged.ThisnewrevisedPracticeD4483,organizedtoaccommodatetherequirementsoftherubberandcarbonblackmanufacturingindustries,hasthreenewfeaturesthatprovideforamoreformalandstructuredanalysisofinterlaboratorytestprogram(ITP)data.Firstitaddressestheoverridingissueswithprecisionevaluationoverthepastseveraldecades—thefrequentdiscoverythatreproducibilityformanytestmethodsisquitepoor.Experiencehasshownthatfrequentlypoorreproducibilityiscausedbyonlyafewlaboratoriesthatdifferfromtheremainderthatgivegoodagreement.Anewproceduredesignatedasrobustanalysisprovidesanimprovedmethodfordetectingoutliersthatcausepoorprecision,especiallypoorbetweenlaboratoryagreement.Second,afteroutlierdetectionthenewstandardprovidestwooptions;(1)outlierdeletionor(2)outlierreplacement.Whenoutliersaredeletedtherevisedstandardprovidesawaytoretainthenon-outlierlaboratorydata.Thisallowsforabroaderdatabaseforprecisioncalculation.ThecurrentASTMCommitteeE11comput...
