Scribd Logo
Upload

Welcome to Scribd!

  • 81 views

    Tolerance Intervals Nist

    Uploaded by

    Jeiel França
    This document discusses tolerance intervals for a normal distribution. It defines tolerance intervals as covering a fixed proportion of a population with a stated confidence, unlike confidence intervals which cover population parameters. It provides the formulas for calculating two-sided and one-sided tolerance interval factors (k) for a normal distribution using sample data, examples, and Dataplot commands.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Tolerance Intervals Nist

    Uploaded by

    Jeiel França
    81 views4 pages
    This document discusses tolerance intervals for a normal distribution. It defines tolerance intervals as covering a fixed proportion of a population with a stated confidence, unlike confidence intervals which cover population parameters. It provides the formulas for calculating two-sided and one-sided tolerance interval factors (k) for a normal distribution using sample data, examples, and Dataplot commands.

    Original Description:

    Definição de intervalos de tolerância e sua distinção para intervalos de confiança.

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document discusses tolerance intervals for a normal distribution. It defines tolerance intervals as covering a fixed proportion of a population with a stated confidence, unlike confidence intervals which cover population parameters. It provides the formulas for calculating two-sided and one-sided tolerance interval factors (k) for a normal distribution using sample data, examples, and Dataplot commands.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    81 views4 pages

    Tolerance Intervals Nist

    Uploaded by

    Jeiel França
    This document discusses tolerance intervals for a normal distribution. It defines tolerance intervals as covering a fixed proportion of a population with a stated confidence, unlike confidence intervals which cover population parameters. It provides the formulas for calculating two-sided and one-sided tolerance interval factors (k) for a normal distribution using sample data, examples, and Dataplot commands.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 4

    17/02/2015

    7.2.5.3.Toleranceintervalsforanormaldistribution

    7.ProductandProcessComparisons
    7.2.Comparisonsbasedondatafromoneprocess
    7.2.5.Whatintervalscontainafixedpercentageofthepopulationvalues?

    7.2.5.3. Toleranceintervalsforanormal
    distribution
    Definitionof
    atolerance
    interval

    Aconfidenceintervalcoversapopulationparameterwitha
    statedconfidence,thatis,acertainproportionofthetime.
    Thereisalsoawaytocoverafixedproportionofthe
    populationwithastatedconfidence.Suchanintervalis
    calledatoleranceinterval.Theendpointsofatolerance
    intervalarecalledtolerancelimits.Anapplicationof
    toleranceintervalstomanufacturinginvolvescomparing
    specificationlimitsprescribedbytheclientwithtolerance
    limitsthatcoveraspecifiedproportionofthepopulation.

    Difference
    between
    confidence
    andtolerance
    intervals

    Confidencelimitsarelimitswithinwhichweexpectagiven
    populationparameter,suchasthemean,tolie.Statistical
    tolerancelimitsarelimitswithinwhichweexpectastated
    proportionofthepopulationtolie.Confidenceintervals
    shrinktowardszeroasthesamplesizeincreases.Tolerance
    intervalstendtowardsafixedvalueasthesamplesize
    increases.

    Notrelatedto
    engineering
    tolerances

    Statisticaltoleranceintervalshaveaprobabilistic
    interpretation.Engineeringtolerancesarespecifiedouter
    limitsofacceptabilitywhichareusuallyprescribedbya
    designengineeranddonotnecessarilyreflectacharacteristic
    oftheactualmeasurements.

    Threetypesof
    tolerance
    intervals

    Threetypesofquestionscanbeaddressedbytolerance
    intervals.Question(1)leadstoatwosidedinterval
    questions(2)and(3)leadtoonesidedintervals.
    1. Whatintervalwillcontainppercentofthepopulation
    measurements?
    2. Whatintervalguaranteesthatppercentofpopulation
    measurementswillnotfallbelowalowerlimit?
    3. Whatintervalguaranteesthatppercentofpopulation
    measurementswillnotexceedanupperlimit?

    Tolerance
    intervalsfor
    measurements
    froma

    Forthequestionsabove,thecorrespondingtolerance
    intervalsaredefinedbylower(L)andupper(U)tolerance
    limitswhicharecomputedfromaseriesofmeasurementsY1,
    ...,YN:

    http://www.itl.nist.gov/div898/handbook/prc/section2/prc253.htm

    1/4

    17/02/2015

    normal
    distribution

    7.2.5.3.Toleranceintervalsforanormaldistribution

    1.
    2.
    3.

    wherethekfactorsaredeterminedsothattheintervalscover
    atleastaproportionpofthepopulationwithconfidence, .
    Calculationof
    kfactorfora
    twosided
    tolerance
    limitfora
    normal
    distribution

    Ifthedataarefromanormallydistributedpopulation,an
    approximatevalueforthefactorasafunctionofpand fora
    twosidedtoleranceinterval(Howe,1969)is

    where
    isthecriticalvalueofthechisquare
    distributionwithdegreesoffreedom,N1,thatisexceeded
    withprobability and
    isthecriticalvalueofthe
    normaldistributionwhichisexceededwithprobability(1
    p)/2.
    Exampleof
    calculation

    Forexample,supposethatwetakeasampleofN=43silicon
    wafersfromalotandmeasuretheirthicknessesinorderto
    findtolerancelimitswithinwhichaproportionp=0.90of
    thewafersinthelotfallwithprobability =0.99.

    Useoftables
    incalculating
    twosided
    tolerance
    intervals

    Valuesofthekfactorasafunctionofpand aretabulatedin
    sometextbooks,suchasDixonandMassey(1969).Touse
    thetablesinthishandbook,followthestepsoutlinedbelow:
    1. Calculate =(1p)/2=0.05
    2. Gotothetableofuppercriticalvaluesofthenormal
    distributionandunderthecolumnlabeled0.05find
    =1.645.
    3. Gotothetableoflowercriticalvaluesofthechi
    squaredistributionandunderthecolumnlabeled0.99
    intherowlabeleddegreesoffreedom=42,find
    =23.650.
    4. Calculate

    Thetolerancelimitsarethencomputedfromthesample
    mean, ,andstandarddeviation,s,accordingtocase(1).
    Important

    Thenotationforthecriticalvalueofthechisquare

    http://www.itl.nist.gov/div898/handbook/prc/section2/prc253.htm

    2/4

    17/02/2015

    7.2.5.3.Toleranceintervalsforanormaldistribution

    note

    distributioncanbeconfusing.Valuesastabulatedare,ina
    sense,alreadysquaredwhereasthecriticalvalueforthe
    normaldistributionmustbesquaredintheformulaabove.

    Dataplot
    commandsfor
    calculating
    thekfactor
    foratwo
    sided
    tolerance
    interval

    TheDataplotcommandsare:
    letn=43
    letnu=n1
    letp=.90
    letg=.99
    letg1=1g
    letp1=(1+p)/2
    letcg=chsppf(g1,nu)
    letnp=norppf(p1)
    letk=nu*(1+1/n)*np**2
    letk2=(k/cg)**.5

    andtheoutputis:
    THECOMPUTEDVALUEOFTHECONSTANTK2=0.2217316E+01

    Anothernote

    ThenotationfortailprobabilitiesinDataplotistheconverse
    ofthenotationusedinthishandbook.Therefore,inthe
    exampleaboveitisnecessarytospecifythecriticalvaluefor
    thechisquaredistribution,say,aschsppf(1.99,42)and
    similarlyforthecriticalvalueforthenormaldistribution.

    Direct
    calculationof
    tolerance
    intervals
    using
    Dataplot

    Dataplotalsohasanoptionforcalculatingtoleranceintervals
    directlyfromthedata.Thecommandsforproducing
    toleranceintervalsfromtwentyfivemeasurementsof
    resistivityfromaqualitycontrolstudyataconfidencelevel
    of99%are:
    read100ohm.datcrwafermodayhminophum...
    probetempyswdf
    tolerancey

    Automaticoutputisgivenforseverallevelsofcoverage,and
    thetoleranceintervalfor90%coverageisshownbelowin
    bold:
    2SIDEDNORMALTOLERANCELIMITS:XBAR+K*S
    NUMBEROFOBSERVATIONS=25
    SAMPLEMEAN=97.069832
    SAMPLESTANDARDDEVIATION=0.26798090E01
    CONFIDENCE=99.%
    COVERAGE(%)LOWERLIMITUPPERLIMIT
    50.097.0424297.09724
    75.097.0230897.11658
    90.097.0029997.13667
    95.096.9902097.14946
    99.096.9652297.17445
    99.996.9362597.20341

    Calculation

    Thecalculationofanapproximatekfactorforonesided

    http://www.itl.nist.gov/div898/handbook/prc/section2/prc253.htm

    3/4

    17/02/2015

    foraone
    sided
    tolerance
    intervalfora
    normal
    distribution

    7.2.5.3.Toleranceintervalsforanormaldistribution

    toleranceintervalscomesdirectlyfromthefollowingsetof
    formulas(Natrella,1963):

    where

    isthecriticalvaluefromthenormaldistribution

    thatisexceededwithprobability1pand
    isthecritical
    valuefromthenormaldistributionthatisexceededwith
    probability1 .
    Dataplot
    commandsfor
    calculating
    thekfactor
    foraone
    sided
    tolerance
    interval

    Fortheexampleabove,itmayalsobeofinterestto
    guaranteewith0.99probability(or99%confidence)that
    90%ofthewafershavethicknesseslessthananupper
    tolerancelimit.Thisproblemfallsundercase(3),andthe
    Dataplotcommandsforcalculatingthefactorfortheone
    sidedtoleranceintervalare:
    letn=43
    letp=.90
    letg=.99
    letnu=n1
    letzp=norppf(p)
    letzg=norppf(g)
    leta=1((zg**2)/(2*nu))
    letb=zp**2(zg**2)/n
    letk1=(zp+(zp**2a*b)**.5)/a

    andtheoutputis:
    THECOMPUTEDVALUEOFTHECONSTANTA=0.9355727E+00
    THECOMPUTEDVALUEOFTHECONSTANTB=0.1516516E+01
    THECOMPUTEDVALUEOFTHECONSTANTK1=0.1875189E+01

    Theupper(onesided)tolerancelimitistherefore97.07+
    1.8752*2.68=102.096.

    http://www.itl.nist.gov/div898/handbook/prc/section2/prc253.htm

    4/4

    You might also like