Bài giảng Business Research Methods - Chapter 19: Multivariate Analysis: An Overview

Chapter 19Multivariate Analysis: An OverviewMcGraw-Hill/IrwinCopyright © 2011 by The McGraw-Hill Companies, Inc. All Rights Reserved. Learning ObjectivesUnderstand . . .How to classify and select multivariate techniques.That multiple regression predicts a metric dependent variable from a set of metric independent variables. That discriminant analysis classifies people or objects into categorical groups using several metric predictors.2Learning ObjectivesUnderstand . . .How multivariate analysis of variance assesses the relationship between two or more metric dependent variables and independent classificatory variables.How structural equation modeling explains causality among constructs that cannot be directly measured.3Learning ObjectivesUnderstand . . . How conjoint analysis assists researchers to discover the most importance attributes and the levels of desirable features.How principal components analysis extracts uncorrelated factors from an initial set of variables and exploratory factor analysis reduces the number of variables to discover the underlying constructs.4Learning ObjectivesUnderstand . . . The use of cluster analysis techniques for grouping similar objects or people. How perceptions of products or services are revealed numerically and geometrically by multidimensional scaling.5Wonder and Curiosity“Wonder, connected with a principle ofrational curiosity, is the source of allknowledge and discovery . . . but wonderwhich ends in wonder, and is satisfied with wonder, is the quality of an idiot.”Samuel Horsley English scientist and fellowRoyal Society6PulsePoint: Research Revelation60The percent of workers on four continents who trust their organization’s senior leaders.7Classifying Multivariate TechniquesInterdependencyDependency8Multivariate Techniques9Multivariate Techniques10Multivariate Techniques11Right Questions. Trusted Insight.When using sophisticated techniques you want to rely on the knowledge of the researcher. Harris Interactive promises you can trust their experienced research professionals to draw the right conclusions from the collected data.12Dependency TechniquesMultiple RegressionDiscriminant AnalysisMANOVAStructural Equation Modeling (SEM)Conjoint Analysis 13Uses of Multiple RegressionDevelop self-weightingestimating equation topredict values for a DVControl for confounding VariablesTest and explain causal theories14Generalized Regression Equation15Multiple Regression Example16Selection MethodsForwardBackwardStepwise17Evaluating and Dealing with MulticollinearityChoose one of the variables and delete the otherCreate a new variable that is a composite of the others CollinearityStatisticsVIF1.0002.2892.2892.7483.0253.06718Discriminant AnalysisPredicted SuccessActual GroupNumber of Cases01UnsuccessfulSuccessful 011515 13 86.70% 3 20.00% 2 13.30% 12 80.00%Note: Percent of “grouped” cases correctly classified: 83.33% Unstandardized StandardizedX1X1X1Constant .36084 2.61192 .53028 12.89685 .65927 .57958 .97505A.B.19MANOVA20MANOVA Output21Bartlett’s Test22MANOVA Homogeneity-of-Variance 23Multivariate Tests of Significance24Univariate Tests of Significance25Structural Equation Modeling (SEM)Model SpecificationEstimationEvaluation of FitRespecification of the ModelInterpretation and Communication26Structural Equation Modeling (SEM)27Concept Cards for Conjoint Sunglasses Study28Conjoint AnalysisBrandBolleHobbiesOakleySki OptiksStyle*ABCABCAAFlotationYesNoYesYesYesPrice$100$72$60$40$100$72$60$40$100$72$60$40$100$72$60$40* A = multiple color choices for frames, lenses, and temples. B = multiple color choices for frames, lenses, and straps (no hard temples. C = limited colors for frames, lenses, and temples.29Conjoint Results Participant 8 in Sunglasses Study30Conjoint Results for Sunglasses Study31Interdependency TechniquesFactor AnalysisCluster AnalysisMultidimensional Scaling32Factor Analysis33Factor Matrices AUnrotated FactorsBRotated FactorsVariableIIIh2IIIABCDEFEigenvaluePercent of varianceCumulative percent0.700.600.600.500.600.602.1836.336.3-.40-.50-.350.500.500.601.3923.259.50.650.610.480.500.610.720.790.750.680.060.130.070.150.030.100.700.770.8534Orthogonal Factor Rotations35Correlation Coefficients, Metro U MBA StudyVariableCourseV1V2V3V10V1V2V3V4V5V6V7V8V9V10Financial Accounting Managerial AccountingFinanceMarketingHuman BehaviorOrganization DesignProductionProbabilityStatistical InferenceQuantitative Analysis1.000.560.17-.14-.19-.21-.440.30-.05-.010.561.00-.220.05-.26-.00-.110.060.060.06.017-.221.00-.48-.05-.56-.040.07-.320.42-.010.060.42-.10-.23-.05-.08-.100.061.0036Factor Matrix, Metro U MBA StudyVariableCourseFactor 1Factor 2Factor 3CommunalityV1V2V3V4V5V6V7V8V9V10EigenvaluePercent of varianceCumulative percentFinancial Accounting Managerial AccountingFinanceMarketingHuman BehaviorOrganization DesignProductionProbabilityStatistical InferenceQuantitative Analysis0.410.010.89-.600.02-.43-.110.25-.430.251.8318.3018.300.710.53-.170.21-.24-.09-.580.250.430.041.5215.2033.500.23-.160.370.30-.22-.36-.03-.310.500.350.959.5043.000.730.310.950.490.110.320.350.220.620.1937Varimax Rotated Factor MatrixVariableCourseFactor 1Factor 2Factor 3V1V2V3V4V5V6V7V8V9V10Financial Accounting Managerial AccountingFinanceMarketingHuman BehaviorOrganization DesignProductionProbabilityStatistical InferenceQuantitative Analysis0.840.53-.01-.11-.13-.08-.540.410.07-.020.16-.100.90-.24-.14-.56-.11-.020.020.42-.060.14-.370.65-.27-.02-.22-.240.790.0938Cluster AnalysisSelect sample to clusterDefine variablesCompute similaritiesSelect mutually exclusive clustersCompare and validate cluster39Cluster Analysis40Cluster Membership________Number of Clusters ________FilmCountryGenreCase5432Cyrano de BergeracIl y a des JoursNikitaLes Noces de PapierLeningrad Cowboys . . . Storia de Ragazzi . . .Conte de PrintempsTatie DanielleCrimes and Misdem . . .Driving Miss DaisyLa Voce della LunaChe Hora EAttache-MoiWhite Hunter Black . . . Music BoxDead Poets SocietyLa Fille aux All . . .Alexandrie, Encore . . .DreamsFranceFranceFranceCanadaFinlandItalyFranceFranceUSAUSAItalyItalySpainUSAUSAUSAFinlandEgyptJapanDramaComDramaComDramaComDramaComComedyComedyComedyComedyDramaComDramaComDramaComDramaComDramaComPsyDramaPsyDramaPsyDramaPsyDramaDramaComDramaCom14561913237912141510811181617111122223333344445511112222333334444331111222233333333333111122222222222222241Dendogram42Similarities Matrix of 16 Restaurants43Positioning of Selected Restaurants44Key TermsAverage linkage methodBackward eliminationBeta weightsCentroidCluster analysisCollinearityCommunalityConfirmatory factor analysisConjoint analysisDependency techniquesDiscriminant analysisDummy variableEigenvalueFactor analysis45Key Terms (cont.)FactorsForward selectionHoldout sampleInterdependency techniquesLoadingsMetric measuresMulticollinearityMultidimensional scaling (MDS)Multiple regressionMultivariate analysisMultivaria analysis of variance (MANOVA)Nonmetric measuresPath analysis46Key Terms (cont.)Path diagramPrincipal components analysisRotationSpecification errorStandardized coefficientsStepwise selectionStress indexStructural equation modelingUtility score47