STATISTICSDEFINITION&TYPES
Introduction:
Decisionmakersmakebetterdecisionswhentheyuseallavailable
informationinaneffectiveandmeaningfulway.Theprimaryroleof
statisticsistotoprovidedecisionmakerswithmethodsforobtainingand
analyzinginformationtohelpmakethesedecisions.Statisticsisusedto
answerlongrangeplanningquestions,suchaswhenandwheretolocate
facilitiestohandlefuturesales.
Definition:
Statisticsisdefinedasthescienceofcollecting,organizing,presenting,
analyzingandinterpretingnumericaldataforthepurposeofassistingin
makingamoreeffectivedecision.
TypesofStatistics:
Therearetwotypesofstatistics
1.DescriptiveStatisticsisconcernedwithsummarycalculations,graphs,
chartsandtables.
2.InferentialStatisticsisamethodusedtogeneralizefromasampletoa
population.Forexample,theaverageincomeofallfamilies(the
population)intheUScanbeestimatedfromfiguresobtainedfromafew
hundred(thesample)families.
StatisticalPopulation:
Isthecollectionofallpossibleobservationsofaspecifiedcharacteristicof
interest.AnexampleisallofthestudentsinBUSA3101courseinthis
term.Notethatasampleisasubsetofthepopulation.
Variable:
Avariableisanitemofinterestthatcantakeonmanydifferentnumerical
values.
TypesofVariablesorData:
1.QualitativeVariablesarenonnumericvariablesandcan'tbemeasured.
Examplesincludegender,religiousaffiliation,stateofbirth.
2.QuantitativeVariablesarenumericalvariablesandcanbemeasured.
Examplesincludebalanceinyourcheckingaccount,numberofchildrenin
yourfamily.Notethatquantitativevariablesareeitherdiscrete(whichcan
assumeonlycertainvalues,andthereareusually"gaps"betweenthe
values,suchasthenumberofbedroomsinyourhouse)orcontinuous
(whichcanassumeanyvaluewithinaspecificrange,suchastheair
pressureinatire.)
TypesofQuantitativeData:
Therearefour(4)typesofquantitativedata:
1.NominalData:Theweakestdatameasurement.Numbersareusedto
representanitemorcharacteristic.Examplesinclude:acollegemay
designatemajorsbynumbers,i.e.,BBAinaccounting=1,BBAin
management=04,ormale=1andfemale=2.Notethatsuchdatashouldnot
betreatedasnumerical,sincerelativesizehasnomeaning.
2.OrdinalorRankData:Numbersareusedtorank.Anexampleiswind
forcesatsea.Agentlebreezeisratedat3,astrongbreezeat6.Simple
arithmeticoperationsarenotmeaningfullyappliedtoordinaldata.
Anotherexampleisexcellent,good,fairandpoor.
Themaindifferencebetweenordinaldataandnominaldataisthat
ordinaldatacontainbothanequality(=)andagreaterthan(>)
relationship,whereasthenominaldatacontainonlyanequality(=)
relationship.
3.IntervalData:Ifwehavedatawithordinalproperties(>&=)andcan
alsomeasurethedistancebetweentwodataitems,wehaveaninterval
measurement.
Intervaldataarepreferredoverordinaldatabecause,withthem,decision
makerscanpreciselydeterminethedifferencebetweentwoobservations,
i.e.,distancesbetweennumberscanbemeasured.Forexample,frozenfood
packagershavedailycontactwithacommonintervalmeasurement
temperature.
4.RatioData:Isthehighestlevelofmeasurementandallowsforallbasic
arithmeticoperations,includingdivisionandmultiplication.Data
measuredonaratioscalehaveafixedornonarbitraryzeropoint.
Examplesincludebusinessdata,suchascost,revenueandprofit.
SourcesofData:
1.SecondaryData:Datawhicharealreadyavailable.Anexample:
statisticalabstractofUSA.Advantage:lessexpensive.Disadvantage:may
notsatisfyyourneeds.
2.PrimaryData:Datawhichmustbecollected.
MethodsofCollectingPrimaryData:
1.FocusGroup;2.TelephoneInterview;3.MailQuestionnaires;4.Doorto
DoorSurvey;5.MallIntercept;6.NewProductRegistration;7.Personal
Interview;and8.Experimentsaresomeofthesourcesforcollectingthe
primarydata.
SamplingMethods:
Therearemanywaystocollectasample.Themostcommonlyused
methodsare:
A.StatisticalSampling:
1.SimpleRandomSampling:Isamethodofselectingitemsfroma
populationsuchthateverypossiblesampleofspecificsizehasanequal
chanceofbeingselected.Inthiscase,samplingmaybewithorwithout
replacement.
2.StratifiedRandomSampling:Isobtainedbyselectingsimplerandom
samplesfromstrata(ormutuallyexclusivesets).Someofthecriteriafor
dividingapopulationintostrataare:Sex(male,female);Age(under18,18
to28,29to39);Occupation(bluecollar,professional,other).
3.ClusterSampling:Isasimplerandomsampleofgroupsorclusterof
elements.Clustersamplingisusefulwhenitisdifficultorcostlyto
generateasimplerandomsample.Forexample,toestimatetheaverage
annualhouseholdincomeinalargecityweuseclustersampling,because
tousesimplerandomsamplingweneedacompletelistofhouseholdsin
thecityfromwhichtosample.Tousestratifiedrandomsampling,we
wouldagainneedthelistofhouseholds.Alessexpensivewayistoleteach
blockwithinthecityrepresentacluster.Asampleofclusterscouldthenbe
randomlyselected,andeveryhouseholdwithintheseclusterscouldbe
interviewedtofindtheaverageannualhouseholdincome.
B.NonstatisticalSampling:
1.JudgementSampling:Inthiscase,thepersontakingthesamplehas
directorindirectcontroloverwhichitemsareselectedforthesample.
2.ConvenienceSampling:Inthismethod,thedecisionmakerselectsa
samplefromthepopulationinamannerthatisrelativelyeasyand
convenient.
3.QuotaSampling:Inthismethod,thedecisionmakerrequiresthesample
tocontainacertainnumberofitemswithagivencharacteristic.Many
politicalpollsare,inpart,quotasampling.
