Designation:E178−08AnAmericanNationalStandardStandardPracticeforDealingWithOutlyingObservations1ThisstandardisissuedunderthefixeddesignationE178;thenumberimmediatelyfollowingthedesignationindicatestheyearoforiginaladoptionor,inthecaseofrevision,theyearoflastrevision.Anumberinparenthesesindicatestheyearoflastreapproval.Asuperscriptepsilon(´)indicatesaneditorialchangesincethelastrevisionorreapproval.1.Scope1.1Thispracticecoversoutlyingobservationsinsamplesandhowtotestthestatisticalsignificanceofthem.Anoutlyingobservation,or“outlier,”isonethatappearstodeviatemark-edlyfromothermembersofthesampleinwhichitoccurs.Inthisconnection,thefollowingtwoalternativesareofinterest:1.1.1Anoutlyingobservationmaybemerelyanextrememanifestationoftherandomvariabilityinherentinthedata.Ifthisistrue,thevalueshouldberetainedandprocessedinthesamemannerastheotherobservationsinthesample.1.1.2Ontheotherhand,anoutlyingobservationmaybetheresultofgrossdeviationfromprescribedexperimentalproce-dureoranerrorincalculatingorrecordingthenumericalvalue.Insuchcases,itmaybedesirabletoinstituteaninvestigationtoascertainthereasonfortheaberrantvalue.Theobservationmayevenactuallyberejectedasaresultoftheinvestigation,thoughnotnecessarilyso.Atanyrate,insubsequentdataanalysistheoutlieroroutlierswillberecognizedasprobablybeingfromadifferentpopulationthanthatoftheothersamplevalues.1.2Itisourpurposeheretoprovidestatisticalrulesthatwillleadtheexperimenteralmostunerringlytolookforcausesofoutlierswhentheyreallyexist,andhencetodecidewhetheralternative1.1.1above,isnotthemoreplausiblehypothesistoaccept,ascomparedtoalternative1.1.2,inorderthatthemostappropriateactioninfurtherdataanalysismaybetaken.Theprocedurescoveredhereinapplyprimarilytothesimplestkindofexperimentaldata,thatis,replicatemeasurementsofsomepropertyofagivenmaterial,orobservationsinasupposedlysinglerandomsample.Nevertheless,thetestssuggesteddocoverawideenoughrangeofcasesinpracticetohavebroadutility.2.ReferencedDocuments2.1ASTMStandards:2E456TerminologyRel...
