市场调查方法(英文版)第十五章

PowerPointPresentationbyCharlieCookTheUniversityofWestAlabamaWilliamG.ZikmundBarryJ.BabinPart5AnalysisandReporting3rdEdition©2007Thomson/South-Western.Allrightsreserved.Chapter15DifferencesBetweenGroupsandRelationshipsAmongVariables©2007Thomson/South-Western.Allrightsreserved.15–2LEARNINGOUTCOMES1.Understandwhatmultivariatestatisticalanalysisinvolvesandknowthetwotypesofmultivariateanalysis2.Interpretresultsfrommultipleregressionanalysis.3.Interpretresultsfrommultivariateanalysisofvariance(MANOVA)4.InterpretbasicexploratoryfactoranalysisresultsAfterstudyingthischapter,youshouldbeableto©2007Thomson/South-Western.Allrightsreserved.15–3WhatIstheAppropriateTestofDifference?•TestofDifferencesAninvestigationofahypothesisthattwo(ormore)groupsdifferwithrespecttomeasuresonavariable.Behavior,characteristics,beliefs,opinions,emotions,orattitudes•BivariateTestsofDifferencesInvolveonlytwovariables:avariablethatactslikeadependentvariableandavariablethatactsasaclassificationvariable.Differencesinmeanscoresbetweengroupsorincomparinghowtwogroups’scoresaredistributedacrosspossibleresponsecategories.©2007Thomson/South-Western.Allrightsreserved.15–4EXHIBIT15.1ChoosingtheRightStatistic©2007Thomson/South-Western.Allrightsreserved.15–5EXHIBIT15.1ChoosingtheRightStatistic(cont’d)©2007Thomson/South-Western.Allrightsreserved.15–6TypeofMeasurementDifferencesbetweentwoindependentgroupsDifferencesamongthreeormoreindependentgroupsOrdinalMann-WhitneyU-testWilcoxontestKruskal-WallistestCommonBivariateTestsNominalZ-test(twoproportions)Chi-squaretestChi-squaretestIntervalandratioIndependentgroups:t-testorZ-testOne-wayANOVA©2007Thomson/South-Western.Allrightsreserved.15–7Cross-TabulationTables:Theχ2TestforGoodness-of-Fit•Cross-Tabulation(Contingency)TableAjointfrequencydistributionofobservationsontwomorevariables.•χ2DistributionProvidesameansfortestingthestatisticalsignificanceofacontingencytable.Involvescomparingobservedfrequencies(Oi)withexpectedfrequencies(Ei)ineachcellofthetable.Capturesthegoodness-(orcloseness-)of-fitoftheobserveddistributionwiththeexpecteddistribution.©2007Thomson/South-Western.Allrightsreserved.15–8Example:PapaJohn’sRestaurantsUnivariateHypothesis:PapaJohn’srestaurantsaremorelikelytobelocatedinastand-alonelocationorinashoppingcenter.BivariateHypothesis:Stand-alonelocationsaremorelikelytobeprofitablethanareshoppingcenterlocations.©2007Thomson/South-Western.Allrightsreserved.15–9Chi-SquareTestiiiE)?E(Oχ²χ²=chi-squarestatisticOi=observedfrequencyintheithcellEi=expectedfrequencyontheithcellnCREjiijRi=totalobservedfrequencyintheithrowCj=totalobservedfrequencyinthejthcolumnn=samplesize©2007Thomson/South-Western.Allrightsreserved.15–10DegreesofFreedom(d.f.)(R-1)(C-1)=(2-1)(2-1)=1d.f.=(R-1)(C-1)©2007Thomson/South-Western.Allrightsreserved.15–11Thet-TestforComparingTwoMeans•IndependentSamplest-TestAtestforhypothesesstatingthatthemeanscoresforsomeinterval-orratio-scaledvariablegroupedbasedonsomelessthanintervalclassificatoryvariable.meansrandomofyVariabilit2MeanSample-1MeanSamplet2121XXSt©2007Thomson/South-Western.Allrightsreserved.15–12Thet-TestforComparingTwoMeans(cont’d)•Determiningwhenanindependentsamplest-testisappropriate:Isthedependentvariableintervalorratio?Canthedependentvariablescoresbegroupedbaseduponsomecategoricalvariable?Doesthegroupingresultinscoresdrawnfromindependentsamples?Aretwogroupsinvolvedintheresearchquestion?©2007Thomson/South-Western.Allrightsreserved.15–13Thet-TestforComparingTwoMeans(cont’d)•PooledEstimateoftheStandardErrorAnestimateofthestandarderrorforat-testofindependentmeansthatassumesthevariancesofbothgroupsareequal.21212222111121121nnnnSnSnSXX))(©2007Thomson/South-Western.Allrightsreserved.15–14Thet-TestforComparingTwoMeans(cont’d)©2007Thomson/South-Western.Allrightsreserved.15–15EXHIBIT15.2IndependentSamplest-TestResults©2007Thomson/South-Western.Allrightsreserved.15–16WhatIsANOVA?•AnalysisofVariance(ANOVA)Ananalysisinvolvingtheinvestigationoftheeffectsofonetreatmentvariableonaninterval-scaleddependentvariableAhypothesis-testingtechniquetodeterminewhetherstatisticallysignificantdifferencesinmeansoccurbetweentwoormoregroups.Amethodofcomparingvariancestomakeinferencesaboutthemeans.ANOVAtestswhether“grouping”observationsexplainsvarianceinthedependentvariable.©2007Thomson/South-Western.Allrightsreserved.15–17SimpleIllustrationofANOVA•Howmuchcoffeerespondentsreportdrinkingeachdaybasedonwhichshifttheywork(GYstandsforGraveyardshift).Day1Day3Day4Day0Day2GY7GY2GY1GY6Night6Night8Night3Night7Night6©2007Thomson/South-Western.Allrightsreserved.15–18EXHIBIT15.3IllustrationofANOVALogic©2007Thomson/South-Western.Allrightsreserved.15–19PartitioningVarianceinANOVA•TotalVariabilityGrandmeanThemeanofavariableoverallobservations.SSTThetotalobservedvariationacrossallgroupsandindividualobservationsSST=Totalof(observedvalue-grandmean)2©2007Thomson/South-Western.Allrightsreserved.15–20PartitioningVarianceinANOVA•Between-groupsVarianceThesumofdifferencesbetweent