《社会研究方法》艾尔巴比-英文ppt-Ch16

Chapter16StatisticalAnalysisChapterOutlineIntroductionDescriptiveStatisticsInferentialStatisticsOtherMultivariateTechniquesDescriptiveStatisticsStatisticalcomputationsthatdescribethecharacteristicsofasampleortherelationshipamongvariablesinasample.Inferentialstatisticsmakeinferencesaboutthelargerpopulationfromwhichthesampleobservationsweredrawn.PartialRawDataMatrixHypotheticalRawDataonEducationandPrejudiceMeasuresofAssociationDescriptivestatisticssummarizingtherelationshipsbetweenvariables.Manymeasuresofassociationarebasedonaproportionatereductionoferror(PRE)model.ProportionateReductionofError(PRE)Alogicalmodelforassessingthestrengthofarelationshipbyaskinghowmuchknowingvaluesononevariablewouldreduceourerrorsinguessingvaluesontheother.Example:Ifweknowhowmucheducationpeoplehave,wecanimproveourabilitytoestimatehowmuchtheyearn,thusindicatingthereisarelationshipbetweenthetwovariables.ProportionateReductionofError(PRE)Basedonacomparisonof:1.Thenumberoferrorswewouldmakeinattemptingtoguesstheattributesofagivenvariableforeachofthecasesunderstudy-ifweknewnothingbutthedistributionofattributesonthatvariable.2.Thenumberoferrorswewouldmakeifweknewthejointdistributionoverallandweretoldforeachcasetheattributeofonevariableeachtimewewereaskedtoguesstheattributeoftheother.Question______________isalogicalmodelforassessingthestrengthofarelationship.A.MRIB.DVUC.OMGD.PREAnswer:DPREisalogicalmodelforassessingthestrengthofarelationship.NominalVariablesIftwovariablesconsistofnominaldata(gender,religiousaffiliation,race),lambda(l)wouldbeoneappropriatemeasure.Lambdaisbasedonyourabilitytoguessvaluesononeofthevariables:thePREachievedthroughknowledgeofvaluesontheothervariable.OrdinalVariablesIfthevariablesbeingrelatedareordinal(socialclass,religiosity,alienation),gamma(g)isoneappropriatemeasureofassociation.Lambdaisbasedonguessingexactvalues,gammaisbasedonguessingtheordinalarrangementofvalues.GammaComputedfromtwoquantities:1.Thenumberofpairshavingthesamerankingonthetwovariables.2.Thenumberofpairshavingtheoppositerankingonthetwovariables.Forpairshavingthesameranking:Thefrequencyofeachcellinthetableismultipliedbythesumofallcellsbelowandtotherightofit,withallproductssummed.QuestionA.Ifthevariablesbeingrelatedareordinal,_________isanappropriatemeasureofassociation.A.GammaB.LambdaC.RhoD.chisquareAnswer:AIfthevariablesbeingrelatedareordinal,gammaisanappropriatemeasureofassociation.IntervalorRatioVariablesIfintervalorratiovariables(age,income,gradepointaverage,andsoforth)arebeingassociated,oneappropriatemeasureofassociationisPearson’sproduct-momentcorrelation(r).rreflectshowcloselyyoucanguessthevalueofonevariablethroughyourknowledgeofthevalueofanother.RegressionAnalysisAmethodofdataanalysisinwhichtherelationshipsamongvariablesarerepresentedintheformofanequation,calledaregressionequation.LinearregressionanalysisAformofstatisticalanalysisthatseekstheequationforthestraightlinethatbestdescribestherelationshipbetweentworatiovariables.QuestionIfintervalorratiovariablesarebeingassociated,oneappropriatemeasureis____________.A.gammaB.lambdaC.Pearson’sproductD.noneofthesechoicesAnswer:CIfintervalorratiovariablesarebeingassociated,oneappropriatemeasureisPearson’sproduct.SimpleScattergramofValuesofXandYAScattergramoftheValuesofTwoVariableswithRegressionLineAddedMultipleRegressionAnalysisAformofstatisticalanalysisthatseekstheequationrepresentingtheimpactoftwoormoreindependentvariablesonasingledependentvariable.PartialRegressionAnalysisAformofregressionanalysisinwhichtheeffectsofoneormorevariablesareheldconstant,similartothelogicoftheelaborationmodel.CurvilinearRegressionAnalysisAformofregressionanalysisthatallowsrelationshipsamongvariablestobeexpressedwithcurvedgeometriclinesinsteadofstraightones.InferentialStatisticsThebodyofstatisticalcomputationsrelevanttomakinginferencesfromfindingsbasedonsampleobservationstosomelargerpopulation.NonsamplingErrorThoseimperfectionsofdataqualitythatarearesultoffactorsotherthansamplingerror.Examples:misunderstandingsofquestionsbyrespondents,erroneousrecordingsbyinterviewersandcoders,andkeypuncherrors.StatisticalSignificanceAgeneraltermreferringtothelikelihoodthatrelationshipsobservedinasamplecouldbeattributedtosamplingerroralone.TestsofStatisticalSignificanceAclassofstatisticalcomputationsthatindicatethelikelihoodthattherelationshipobservedbetweenvariablesinasamplecanbeattributedtosamplingerroronly.TheLogicofStatisticalSignificance1.Assumptionsregardingtheindependenceoftwovariablesinthepopulationstudy.2.Assumptionsregardingtherepresentativenessofsamplesselectedthroughconventionalprobability-samplingprocedures.3.Theobservedjointdistributionofsampleelementsintermsofthetwovariables.Question_________________indicatethelikelihoodthattherelationshipobservedbetweenvariablesinasamplecanbeattributedtosamplingerroronly.A.expostfactohypothesizingB.testsofstatisticalsignificanceC.disconfirmationD.allofthesechoicesAnswer:BTestsofstatisticalsignificanceindicatethelikelihoodthattherelationshipobserve