mba course material

Upload: ankitrohilla

Post on 29-May-2018

232 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/8/2019 Mba Course Material

    1/12

  • 8/8/2019 Mba Course Material

    2/12

  • 8/8/2019 Mba Course Material

    3/12

  • 8/8/2019 Mba Course Material

    4/12

  • 8/8/2019 Mba Course Material

    5/12

    In quota samplinz, the population is first segmented into mutually exclusive sub-groups, just as

    in stratified sampling. Then judgment is used to select the subjects or units from each segment based

    on a specified proportion. For example, an interviewer may be told to sample 200 females and{

    00

    males between the age of 45 and 60.

    It is this second step which makes the technique one of non-probabili ty sampling. In quota sampling,

    the selection of the sample is non-random unlike random sampling and can often be found unreliable.

    For example interviewers might be tempted to interview those people in the street who look most

    helpful, or may choose to use accidental sampling to question those which are closest to them, for

    time-keeping sake. The problem is that these samples may be biased because not everyone gets a

    chance of selection. This non-random element is i ts greatest weakness and quota versus probability

    has been a matter of controversy for many years.

    Quota sampling is useful when time is limited, sampling frame is not available, research budget is

    very tight or when detailed accuracy is not important. You can also choose how many of each

    category is selected.

    A quota sample is a convenience sample with an effort made to insure a certain distribution of

    demographic variables. Subjects are recruited as they arrive and the researcher will assign them to

    demographic groups based on variables like age and gender. When the quota for a given

    demographic group is filled, the researcher will stop recruiting subjects from that particular group.

    This is the non probability version of stratified sampling. Subsetsare chosen and then either

    convenience or judgment sampling is used to choose people from each subset.

    Stratified sampling is probably the most commonly used probability method. Subsets of the population

    are created so that each subset has a common characteristic, such as gender. Random sampling

    chooses a number of subjects from each subset.

    Simple random sampleFrom Wikipedia, the free encyclopedia

    This article does not cite any references or sources.Please help improve this article by adding citations to reliable sources. Unsourced material may

    be challenged and removed. (April 2007)

    In statistics, a simple random sample is a subset ofindividuals (a sample) chosen from a larger set

    (a population). Each individual is chosenrandomly and entirely by chance, such that each individual

    has the same probability of being chosen at any stage during the sampling process, and each subset

    ofk individuals has the same probabili ty of being chosen for the sample as any other subset

    ofk individuals (Yates, Daniel S.; David S. | oore, Daren S. Starnes (2008). The Practice of Statistics,

  • 8/8/2019 Mba Course Material

    6/12

    3rd

    Ed..} ~eeman. ISBN97

    7

    7

    73

    9

    .

    .

    i

    processand techniue is

    nownassimple

    random sampling, andshouldnot

    econfusedwithandom Sampling.

    Insmall populationsandoften in largeones, suchsampling is tpicall

    done "withoutreplacement"

    'S

    S

    '

    , i.e., onedeli

    eratelavoidschoosinganymemberof thepopulationmore thanonce.

    Althoughsimplerandomsamplingcanbeconductedwithreplacement instead, this is lesscommon

    andwouldnormallybedescribedmore fullyassimplerandomsamplingwith

    replacement'S

    S

    '

    . Samplingdonewithout replacement isno longer independent, but still

    satisfiesexchangeability, hencemanyresultsstill hold.urther, forasmall sample froma large

    population, samplingwithout replacement isapproximately thesameassamplingwithreplacement,

    since theoddsofchoosing thesamesample twice is low.

    Anunbiasedrandomselectionof individuals is important so that in the longrun, thesample

    represents thepopulation. However, thisdoesnot guarantee that aparticularsample isaperfect

    representationof thepopulation. Simplerandomsamplingmerelyallowsone todrawexternallyvalid

    conclusionsabout theentirepopulationbasedon thesample.

    onceptually, simplerandomsampling is thesimplest of theprobabilitysampling techni

    ues. It

    requiresacompletesampling frame, whichmaynot beavailableorfeasible toconstruct forlarge

    populations. Even ifacomplete frame isavailable, moreefficient approachesmaybepossible ifother

    useful information isavailableabout theunits in thepopulation.

    Advantagesare that it is freeofclassificationerror, and it requiresminimumadvancenowledgeof

    thepopulationotherthan the frame. Itssimplicityalsomakes it relativelyeasy to interpret data

    collectedvia S

    S.orthesereasons, simplerandomsamplingbest suitssituationswherenot much

    information isavailableabout thepopulationanddatacollectioncanbeefficientlyconductedon

    randomlydistributed items, orwhere thecost ofsampling issmall enough tomakeefficiency less

    important thansimplicity. If theseconditionsarenot true,stratifiedsamplingorclustersamplingmay

    beabetterchoice.

    [edit] istinctionbetweenasystematicrandomsampleandasimple

    randomsample

    Inasimplerandomsample, onepersonmust takearandomsample fromapopulation, andnot have

    anyorder inwhichonechooses thespecific individual.

    Let usassumeyouhadaschool with students, dividedequally intoboysandgirls, andyou

    wanted toselect

    of them forfurtherstudy. Youmight put all theirnames inabucket and thenpull

    namesout. Not onlydoeseachpersonhaveanequal chanceofbeingselected, wecanalso

    easilycalculate theprobabilityofagivenpersonbeingchosen, sinceweknow thesamplesie

    n)

    and thepopulation

    N)and it becomesasimplematterofdivision:

    n/N or

    /

    =.

    %)

  • 8/8/2019 Mba Course Material

    7/12

  • 8/8/2019 Mba Course Material

    8/12

    Scale social sciences) rom

    ikipedia, the freeencyclopedia

    In thesocial sciences, scaling is theprocessofmeasuringororderingentitieswithrespect to

    quantitativeattributesor traits.orexample, ascaling techniquemight involveestimating individuals'

    levelsofextraversion, or theperceivedqualityofproducts.ertainmethodsofscalingpermit

    estimationofmagnitudesonacontinuum, whileothermethodsprovideonly forrelativeorderingof the

    entities.

    Seelevel ofmeasurement foranaccount ofqualitativelydifferent kindsofmeasurement scales.

    Contents

    [hide]

    1 Comparati e and noncomparati e scaling

    2 Composite measures

    3 Data types

    4 Scale construction decisions

    5 Comparati e scaling techniques

    6 Non-comparati e scaling techniques

    7 Scale evaluation

    8 See also

    9 References

    10 Lists of related topics

    11 Links

    [edit] omparativeandnoncomparativescaling

    ithcomparativescaling, the itemsaredirectlycomparedwitheachother

    example :

    oyou

    preferPepsior

    oke?). Innoncomparativescalingeach item isscaled independentlyof theothers

    example : Howdoyou feel about

    oke?).

    [edit] ompositemeasures

    ompositemeasuresofvariablesarecreatedbycombining twoormore

    separateempiricalindicators intoasinglemeasure.

    ompositemeasuresmeasurecomplexconcepts

    moreadequately thansingle indicators, extend therangeofscoresavailableandaremoreefficient at

    handlingmultiple items.

  • 8/8/2019 Mba Course Material

    9/12

    Inaddition toscales, thereare twoothertypesofcompositemeasures. Indexesaresimilartoscales

    except multiple indicatorsofavariablearecombined intoasinglemeasure. The indexofconsumer

    confidence, forexample, isacombinationofseveral measuresofconsumerattitudes. A typology is

    similartoan indexexcept thevariable ismeasuredat thenominal level.

    Indexesareconstructedbyaccumulatingscoresassigned to individual attributes, whilescalesare

    constructed through theassignment ofscores topatternsofattributes.

    hile indexesandscalesprovidemeasuresofasingledimension, typologiesareoftenemployed to

    examine the intersectionof twoormoredimensions. Typologiesareveryuseful analytical toolsand

    canbeeasilyusedasindependent variables, althoughsince theyarenot unidimensional it isdifficult

    touse themasadependent variable.

    [edit] ata types

    The typeof informationcollectedcan influencescaleconstruction.

    ifferent typesof informationare

    measured indifferent ways.

    . Somedataaremeasuredat thenominal level. That is, anynumbersusedaremere labels :

    theyexpressnomathematical properties. Examplesare SKU inventorycodesand UP

    bar

    codes.

    . Somedataaremeasuredat theordinal level. Numbers indicate therelativepositionof items,

    but not themagnitudeofdifference. Anexample isapreferenceranking.

    3. Somedataaremeasuredat theinterval level. Numbers indicate themagnitudeofdifference

    between items, but there isnoabsoluteeropoint. Examplesareattitudescalesandopinion

    scales.

    . Somedataaremeasuredat theratio level. Numbers indicatemagnitudeofdifferenceand

    there isa fixederopoint.

    atioscanbecalculated. Examples include: age, income, price,

    costs, salesrevenue, salesvolume, andmarket share.

    [edit]Scaleconstructiondecisions

    hat level ofdata is involved

    nominal, ordinal, interval, orratio)?

    hat will theresultsbeused for?

    Shouldyouuseascale, index, or typology?

    hat typesofstatistical analysiswouldbeuseful?

    Shouldyouuseacomparativescaleoranoncomparativescale?

    Howmanyscaledivisionsorcategoriesshouldbeused

    to

    ;

    to 7; 3 to +3)?

    Should therebeanoddorevennumberofdivisions?

    ddgivesneutral centervalue; even

    forcesrespondents to takeanon-neutral position.)

    hat should thenatureanddescriptivenessof thescale labelsbe?

  • 8/8/2019 Mba Course Material

    10/12

    hat should thephysical formorlayout of thescalebe?

    graphic, simple linear, vertical,

    hori

    ontal)

    Shouldaresponsebe forcedorbe left optional?

    [edit] omparativescaling techniques

    Pairwisecomparisonscale arespondent ispresentedwith two itemsat a timeandasked to

    select one

    example :oyoupreferPepsi or

    oke?). This isanordinal level techniquewhena

    measurement model isnot applied. Krusand Kennedy

    977)elaborated thepairedcomparison

    scalingwithin theirdomain-referencedmodel. The BradleyTerryLuce

    BTL)model

    Bradleyand

    Terry, 9

    ; Luce,9

    9)canbeapplied inorder toderivemeasurementsprovided thedata

    derived frompairedcomparisonspossessanappropriatestructure. Thurstone'sLawof

    comparative judgmentcanalsobeapplied insuchcontexts.

    aschmodelscaling respondents interact with itemsandcomparisonsare inferredbetween

    items from theresponses toobtainscalevalues.

    espondentsaresubsequentlyalsoscaledbasedon theirresponses to itemsgiven the itemscalevalues. The aschmodel hasaclose

    relation to the BTL model.

    Rank-orderscale arespondent ispresentedwithseveral itemssimultaneouslyandasked to

    rank them

    example :

    ate the followingadvertisements from

    to

    .). This isanordinal level

    technique.

    Bogardussocial distancescale measures thedegree towhichaperson iswilling toassociate

    withaclassor typeofpeople. It askshowwilling therespondent is tomakevariousassociations.

    Theresultsarereduced toasinglescoreonascale. Therearealsonon-comparativeversionsof

    thisscale.

    Q-Sortscale Up to

    itemsaresorted intogroupsbasedarank-orderprocedure.

    Guttmanscale This isaprocedure todeterminewhetheraset of itemscanberank-orderedon

    aunidimensional scale. It utilies the intensitystructureamongseveral indicatorsofagiven

    variable. Statementsare listed inorderof importance. Therating isscaledbysummingall

    responsesuntil the first negativeresponse in the list. The Guttmanscale isrelated to

    asch

    measurement; specifically,aschmodelsbring the Guttmanapproachwithinaprobabilistic

    framework.

    Constantsum scale arespondent isgivenaconstant sumofmoney, script, credits, orpoints

    andasked toallocate these tovarious items example : Ifyouhad Yen tospendon food

    products, howmuchwouldyouspendonproduct A, onproduct B, onproduct , etc.). This isan

    ordinal level technique.

    Magnitudeestimationscale Inapsychophysicsprocedure inventedbyS. S. Stevenspeople

    simplyassignnumbers to thedimensionof judgment. Thegeometricmeanof thosenumbers

    usuallyproducesapower lawwithacharacteristicexponent. Incross-modalitymatching instead

    ofassigningnumbers, peoplemanipulateanotherdimension, suchas loudnessorbrightness to

  • 8/8/2019 Mba Course Material

    11/12

    match the items. Typically theexponent of thepsychometric functioncanbepredicted from the

    magnitudeestimationexponentsofeachdimension.

    [edit]Non-comparativescaling techniques

    Continuousrating scale

    alsocalled thegraphicratingscale) respondentsrate itemsby

    placingamarkona line. The line isusually labeledat eachend. Therearesometimesaseriesof

    numbers, calledscalepoints,

    say, fromero to

    )under the line. Scoringandcodification is

    difficult.

    Likertscale espondentsareasked to indicate theamount ofagreement ordisagreement

    fromstronglyagree tostronglydisagree)ona five- tonine-point scale. Thesame format isused

    formultiplequestions. Thiscategorical scalingprocedurecaneasilybeextended toamagnitude

    estimationprocedure that uses the full scaleofnumbersrather thanverbal categories.

    Phrasecompletionscales

    espondentsareasked tocompleteaphraseonan

    -point

    responsescale inwhich

    represents theabsenceof the theoretical construct and

    representsthe theori

    edmaximumamount of theconstruct beingmeasured. Thesamebasic format isused

    formultiplequestions.

    Semantic differential scale espondentsareasked torateona 7 point scalean itemon

    variousattributes. Eachattributerequiresascalewithbipolarterminal labels.

    Stapel scale This isaunipolarten-point ratingscale. It ranges from + to andhasnoneutral

    eropoint.

    Thurstonescale This isascaling technique that incorporates the intensitystructureamong

    indicators.

    Mathematically derived scale esearchers inferrespondentsevaluationsmathematically.

    Twoexamplesaremulti dimensional scalingandconjoint analysis.

    [edit]Scaleevaluation

    Scalesshouldbe tested forreliability, generali

    ability, andvalidity. Generali

    ability is theability to

    make inferences fromasample to thepopulation, given thescaleyouhaveselected.eliability is the

    extent towhichascalewill produceconsistent results. Test-retest reliabilitycheckshowsimilarthe

    resultsare if theresearch isrepeatedundersimilarcircumstances. Alternative formsreliabilitychecks

    howsimilartheresultsare if theresearch isrepeatedusingdifferent formsof thescale. Internal

    consistencyreliabilitycheckshowwell the individual measures included in thescaleareconverted

    intoacompositemeasure.

    Scalesand indexeshave tobevalidated. Internal validationchecks therelationbetween the individual

    measures included in thescale, and thecompositescale itself. External validationchecks therelation

    between thecompositescaleandotherindicatorsof thevariable, indicatorsnot included in thescale.

    ontent validation

    alsocalled facevalidity)checkshowwell thescalemeasureswhat issupposed to

    measure.riterionvalidationcheckshowmeaningful thescalecriteriaarerelative tootherpossible

  • 8/8/2019 Mba Course Material

    12/12

    criteria.

    onstruct validationcheckswhat underlyingconstruct isbeingmeasured. Thereare three

    variantsofconstruct validity. Theyareconvergent validity, discriminant validity, andnomological

    validity

    ampbell and

    iske,

    9

    9; Krusand Ney,

    97

    ). Thecoefficient ofreproducibility indicates

    howwell thedata from the individual measures included in thescalecanbereconstructed from the

    compositescale.