多元数据分析

章节摘录

版权页：插图：Missing data are termed missing at random （MAR） if the missing values of Y depend on X,  but not on Y In other words, the observed Y values represent a random sample of the actual Y  values for each value of X, but the observed data for Ydo not necessarily represent a truly ran-  dom sample of all Y values. Even though the missing data process is random in the sample, its  values are not generalizable to the population. Most often, the data are missing randomly  within subgroups, but differ in levels between subgroups. The researcher must determine the  factors determining the subgroups and the varying levels between groups.For example, assume that we know the gender of respondents （the X variable） and are  asking about household income （the Y variable）. We find that the missing data are random for  both males and females but occur at a much higher frequency for males than females. Even  though the missing data process is operating in a random manner within the gender variable,  any remedy applied to the missing data will still reflect the missing data process because gen-  der affects the ultimate distribution of the household income values.A higher level of randomness is termed missing completely at random （MCAR）. In these  instances the observed values of Y are truly a random sample of all Y values, with no underly-  ing process that lends bias to the observed data. In simple terms, the cases with missing data  are indistinguishable from cases with complete data.From our earlier example, this situation would be shown by the fact that the missing  data for household income were randomly missing in equal proportions for both males and  females. In this missing data process, any of the remedies can be applied without making  allowances for the impact of any other variable or missing data process.Diagnostic Tests for Levels of Randomness. As previously noted, the researcher must ascer-tain whether the missing data process occurs in a completely random manner. When the data set issmall, the researcher may be able to visually see such patterns or perform a set of simple calculations（such as in our simple example at the beginning of the chapter）. However, as sample size and the num-ber of variables increases, so does the need for empirical diagnostic tests. Some statistical programsadd techniques specifically designed for missing data analysis （e.g., Missing Value Analysis in SPSS）,which generally include one or both diagnostic tests.The first diagnostic assesses the missing data process of a single variable Y by forming twogroups: observations with missing data for Y and those with valid values of Y. Statistical testsare then performed to determine whether significant differences exist between the two groupson other variables of interest. Significant differences indicate the possibility of a nonrandommissing data process.Let us use our earlier example of household income and gender. We would first form twogroups of respondents, those with missing data on the household income question and thosewho answered the question. We would then compare the percentages of gender for each group.If one gender （e.g., males） was found in greater proportion in the missing data group, we wouldsuspect a nonrandom missing data process. If the variable being compared is metric （e.g., anattitude or perception） instead of categorical （gender）, then t-tests are performed to determinethe statistical significance of the difference in the variable's mean between the two groups. Theresearcher should examine a number of variables to see whether any consistent patternemerges. Remember that some differences will occur by chance, but either a large number or asystematic pattern of differences may indicate an underlying nonrandom pattern.A second approach is an overall test of randomness that determines whether the missing datacan be classified as MCAR.

内容概要

Joseph F. Hair 于1971获得佛罗里达大学市场营销博士学位，现为肯尼索州立大学市场营销系教授。他出版了四十多本书，包括《Marketing》、《Marketing Essentials》等。他是美国市场营销协会、市场营销科学学会、西南市场营销协会和南方市场营销学会委员。2004年他被美国市场营销科学学会授予杰出教育奖，2007年被市场管理协会授予创新性市场营销人才。

William C. Black 于1980年获得德州大学奥斯汀分校博士学位，现为路易斯安那州立大学工商管理学院市场营销系教授。他的研究兴趣包括多元统计、应用信息技术，以及与电子商务相关的市场原理的进展。他是《Journal of Business Research》编审委员会成员。

Barry J. Babin 于1991年获得路易斯安那州立大学工商管理学博士学位，现为路易斯安那理工大学市场营销与定量分析学教授、商学院Max P. Watson教授。他主要研究零售的各个方面和服务管理。他还曾被美国市场营销科学研究院和市场营销学会评为杰出研究员。

Rolph E. Anderson 拥有佛罗里达大学博士学位，现为Drexel大学工商管理学院Royal H. Gibson Sr教授。他曾两次获得Drexel大学优秀教师奖，并获得过《Journal of Personal Selling & Sales Management》杰出评论奖、Drexel大学商学院科研成就奖等。

书籍目录

CONTENTS

Preface iii

About the Authors v

Chapter 1 Introduction: Methods and Model Building 1

What Is Multivariate Analysis? 3

Multivariate Analysis in Statistical Terms 4

Some Basic Concepts of Multivariate Analysis 4

The Variate 4

Measurement Scales 5

Measurement Error and Multivariate Measurement 7

Statistical Significance Versus Statistical Power 8

Types of Statistical Error and Statistical Power 9

Impacts on Statistical Power 9

Using Power with Multivariate Techniques 11

A Classification of Multivariate Techniques 11

Dependence Techniques 14

Interdependence Techniques 14

Types of Multivariate Techniques 15

Principal Components and Common Factor Analysis 16

Multiple Regression 16

Multiple Discriminant Analysis and Logistic Regression 16

Canonical Correlation 17

Multivariate Analysis of Variance and Covariance 17

Conjoint Analysis 18

Cluster Analysis 18

Perceptual Mapping 19

Correspondence Analysis 19

Structural Equation Modeling and Confirmatory Factor Analysis 19

Guidelines for Multivariate Analyses and Interpretation 20

Establish Practical Significance as Well as Statistical

Significance 20

Recognize That Sample Size Affects All Results 21

Know Your Data 21

Strive for Model Parsimony 21

Look at Your Errors 22

Validate Your Results 22

A Structured Approach to Multivariate Model Building 22

Stage 1: Define the Research Problem, Objectives,

and Multivariate Technique to Be Used 23

Stage 2: Develop the Analysis Plan 23

Stage 3: Evaluate the Assumptions Underlying the Multivariate Technique 23

Stage 4: Estimate the Multivariate Model and Assess Overall Model Fit 23

Stage 5: Interpret the Variate(s) 24

Stage 6: Validate the Multivariate Model 24

A Decision Flowchart 24

Databases 24

Primary Database 25

Other Databases 27

Organization of the Remaining Chapters 28

Section I: Understanding and Preparing For Multivariate Analysis 28

Section II: Analysis Using Dependence Techniques 28

Section III: Interdependence Techniques 28

Section IV: Structural Equations Modeling 28

Summary 28 . Questions 30 . Suggested Readings 30

References 30

SECTION I Understanding and Preparing For Multivariate Analysis 31

Chapter 2 Cleaning and Transforming Data 33

Introduction 36

Graphical Examination of the Data 37

Univariate Profiling: Examining the Shape of the Distribution 38

Bivariate Profiling: Examining the Relationship Between Variables 39

Bivariate Profiling: Examining Group Differences 40

Multivariate Profiles 41

Missing Data 42

The Impact of Missing Data 42

A Simple Example of a Missing Data Analysis 43

A Four-Step Process for Identifying Missing Data and Applying Remedies 44

An Illustration of Missing Data Diagnosis with the Four-Step Process 54

Outliers 64

Detecting and Handling Outliers 65

An Illustrative Example of Analyzing Outliers 68

Testing the Assumptions of Multivariate Analysis 70

Assessing Individual Variables Versus the Variate 70

Four Important Statistical Assumptions 71

Data Transformations 77

An Illustration of Testing the Assumptions Underlying Multivariate Analysis 79

Incorporating Nonmetric Data with Dummy Variables 86

Summary 88 . Questions 89 . Suggested Readings 89

References 90

Chapter 3 Factor Analysis 91

What Is Factor Analysis? 94

A Hypothetical Example of Factor Analysis 95

Factor Analysis Decision Process 96

Stage 1: Objectives of Factor Analysis 96

Specifying the Unit of Analysis 98

Achieving Data Summarization Versus Data Reduction 98

Variable Selection 99

Using Factor Analysis with Other Multivariate Techniques 100

Stage 2: Designing a Factor Analysis 100

Correlations Among Variables or Respondents 100

Variable Selection and Measurement Issues 101

Sample Size 102

Summary 102

Stage 3: Assumptions in Factor Analysis 103

Conceptual Issues 103

Statistical Issues 103

Summary 104

Stage 4: Deriving Factors and Assessing Overall Fit 105

Selecting the Factor Extraction Method 105

Criteria for the Number of Factors to Extract 108

Stage 5: Interpreting the Factors 112

The Three Processes of Factor Interpretation 112

Rotation of Factors 113

Judging the Significance of Factor Loadings 116

Interpreting a Factor Matrix 118

Stage 6: Validation of Factor Analysis 122

Use of a Confirmatory Perspective 122

Assessing Factor Structure Stability 122

Detecting Influential Observations 123

Stage 7: Additional Uses of Factor Analysis Results 123

Selecting Surrogate Variables for Subsequent Analysis 123

Creating Summated Scales 124

Computing Factor Scores 127

Selecting Among the Three Methods 128

An Illustrative Example 129

Stage 1: Objectives of Factor Analysis 129

Stage 2: Designing a Factor Analysis 129

Stage 3: Assumptions in Factor Analysis 129

Component Factor Analysis: Stages 4 Through 7 132

Common Factor Analysis: Stages 4 and 5 144

A Managerial Overview of the Results 146

Summary 148 . Questions 150 . Suggested Readings 150

References 150

SECTION II Analysis Using Dependence Techniques 153

Chapter 4 Simple and Multiple Regression 155

What Is Multiple Regression Analysis? 161

An Example of Simple and Multiple Regression 162

Prediction Using a Single Independent Variable:

Simple Regression 162

Prediction Using Several Independent Variables:

Multiple Regression 165

Summary 167

A Decision Process for Multiple Regression Analysis 167

Stage 1: Objectives of Multiple Regression 169

Research Problems Appropriate for Multiple Regression 169

Specifying a Statistical Relationship 171

Selection of Dependent and Independent Variables 171

Stage 2: Research Design of a Multiple Regression Analysis 173

Sample Size 174

Creating Additional Variables 176

Stage 3: Assumptions in Multiple Regression Analysis 181

Assessing Individual Variables Versus the Variate 182

Methods of Diagnosis 183

Linearity of the Phenomenon 183

Constant Variance of the Error Term 185

Independence of the Error Terms 185

Normality of the Error Term Distribution 185

Summary 186

Stage 4: Estimating the Regression Model and Assessing Overall Model Fit 186

Selecting an Estimation Technique 186

Testing the Regression Variate for Meeting the Regression Assumptions 191

Examining the Statistical Significance of Our Model 192

Identifying Influential Observations 194

Stage 5: Interpreting the Regression Variate 197

Using the Regression Coefficients 197

Assessing Multicollinearity 200

Stage 6: Validation of the Results 206

Additional or Split Samples 206

Calculating the PRESS Statistic 206

Comparing Regression Models 206

Forecasting with the Model 207

Illustration of a Regression Analysis 207

Stage 1: Objectives of Multiple Regression 207

Stage 2: Research Design of a Multiple Regression Analysis 208

Stage 3: Assumptions in Multiple Regression Analysis 208

Stage 4: Estimating the Regression Model and Assessing Overall Model Fit 208

Stage 5: Interpreting the Regression Variate 223

Stage 6: Validating the Results 226

Evaluating Alternative Regression Models 227

A Managerial Overview of the Results 231

Summary 231 . Questions 234 . Suggested Readings 234

References 234

Chapter 5 Canonical Correlation 235

What Is Canonical Correlation? 237

Hypothetical Example of Canonical Correlation 238

Developing a Variate of Dependent Variables 238

Estimating the First Canonical Function 238

Estimating a Second Canonical Function 240

Relationships of Canonical Correlation Analysis to Other Multivariate Techniques 241

Stage 1: Objectives of Canonical Correlation Analysis 242

Selection of Variable Sets 242

Evaluating Research Objectives 242

Stage 2: Designing a Canonical Correlation Analysis 243

Sample Size 243

Variables and Their Conceptual Linkage 243

Missing Data and Outliers 244

Stage 3: Assumptions in Canonical Correlation 244

Linearity 244

Normality 244

Homoscedasticity and Multicollinearity 244

Stage 4: Deriving the Canonical Functions and Assessing Overall Fit 245

Deriving Canonical Functions 246

Which Canonical Functions Should Be Interpreted? 246

Stage 5: Interpreting the Canonical Variate 250

Canonical Weights 250

Canonical Loadings 250

Canonical Cross-Loadings 250

Which Interpretation Approach to Use 251

Stage 6: Validation and Diagnosis 251

An Illustrative Example 252

Stage 1: Objectives of Canonical Correlation Analysis 253

Stages 2 and 3: Designing a Canonical Correlation Analysis and Testing the Assumptions 253

Stage 4: Deriving the Canonical Functions and Assessing Overall Fit 253

Stage 5: Interpreting the Canonical Variates 254

Stage 6: Validation and Diagnosis 257

A Managerial Overview of the Results 258

Summary 258 . Questions 259 . References 260

Chapter 6 Conjoint Analysis 261

What Is Conjoint Analysis? 266

Hypothetical Example of Conjoint Analysis 267

Specifying Utility, Factors, Levels, and Profiles 267

Gathering Preferences from Respondents 268

Estimating Part-Worths 269

Determining Attribute Importance 270

Assessing Predictive Accuracy 270

The Managerial Uses of Conjoint Analysis 271

Comparing Conjoint Analysis with Other Multivariate Methods 272

Compositional Versus Decompositional Techniques 272

Specifying the Conjoint Variate 272

Separate Models for Each Individual 272

Flexibility in Types of Relationships 273

Designing a Conjoint Analysis Experiment 273

Stage 1: The Objectives of Conjoint Analysis 276

Defining the Total Utility of the Object 276

Specifying the Determinant Factors 276

Stage 2: The Design of a Conjoint Analysis 277

Selecting a Conjoint Analysis Methodology 278

Designing Profiles: Selecting and Defining Factors and Levels 278

Specifying the Basic Model Form 283

Data Collection 286

Stage 3: Assumptions of Conjoint Analysis 293

Stage 4: Estimating the Conjoint Model and Assessing Overall Fit 294

Selecting an Estimation Technique 294

Estimated Part-Worths 297

Evaluating Model Goodness-of-Fit 298

Stage 5: Interpreting the Results 299

Examining the Estimated Part-Worths 300

Assessing the Relative Importance of Attributes 302

Stage 6: Validation of the Conjoint Results 303

Managerial Applications of Conjoint Analysis 303

Segmentation 304

Profitability Analysis 304

Conjoint Simulators 305

Alternative Conjoint Methodologies 306

Adaptive/Self-Explicated Conjoint: Conjoint with

a Large Number of Factors 306

Choice-Based Conjoint: Adding Another Touch of Realism 308

Overview of the Three Conjoint Methodologies 312

An Illustration of Conjoint Analysis 312

Stage 1: Objectives of the Conjoint Analysis 313

Stage 2: Design of the Conjoint Analysis 313

Stage 3: Assumptions in Conjoint Analysis 316

Stage 4: Estimating the Conjoint Model and Assessing Overall Model Fit 316

Stage 5: Interpreting the Results 320

Stage 6: Validation of the Results 324

A Managerial Application: Use of a Choice Simulator 325

Summary 327 . Questions 330 . Suggested Readings 330

References 330

Chapter 7 Multiple Discriminant Analysis and Logistic Regression 335

What Are Discriminant Analysis and Logistic Regression? 339

Discriminant Analysis 340

Logistic Regression 341

Analogy with Regression and MANOVA 341

Hypothetical Example of Discriminant Analysis 342

A Two-Group Discriminant Analysis: Purchasers Versus Nonpurchasers 342

A Geometric Representation of the Two-Group Discriminant Function 345

A Three-Group Example of Discriminant Analysis: Switching Intentions 346

The Decision Process for Discriminant Analysis 348

Stage 1: Objectives of Discriminant Analysis 350

Stage 2: Research Design for Discriminant Analysis 351

Selecting Dependent and Independent Variables 351

Sample Size 353

Division of the Sample 353

Stage 3: Assumptions of Discriminant Analysis 354

Impacts on Estimation and Classification 354

Impacts on Interpretation 355

Stage 4: Estimation of the Discriminant Model and Assessing Overall Fit 356

Selecting an Estimation Method 356

Statistical Significance 358

Assessing Overall Model Fit 359

Casewise Diagnostics 368

Stage 5: Interpretation of the Results 369

Discriminant Weights 369

Discriminant Loadings 370

Partial F Values 370

Interpretation of Two or More Functions 370

Which Interpretive Method to Use? 373

Stage 6: Validation of the Results 373

Validation Procedures 373

Profiling Group Differences 374

A Two-Group Illustrative Example 375

Stage 1: Objectives of Discriminant Analysis 375

Stage 2: Research Design for Discriminant Analysis 375

Stage 3: Assumptions of Discriminant Analysis 376

Stage 4: Estimation of the Discriminant Model and Assessing Overall Fit 376

Stage 5: Interpretation of the Results 387

Stage 6: Validation of the Results 390

A Managerial Overview 391

A Three-Group Illustrative Example 391

Stage 1: Objectives of Discriminant Analysis 391

Stage 2: Research Design for Discriminant Analysis 392

Stage 3: Assumptions of Discriminant Analysis 392

Stage 4: Estimation of the Discriminant Model and Assessing Overall Fit 392

Stage 5: Interpretation of Three-Group Discriminant Analysis Results 404

Stage 6: Validation of the Discriminant Results 410

A Managerial Overview 412

Logistic Regression: Regression with a Binary Dependent Variable 413

Representation of the Binary Dependent Variable 414

Sample Size 415

Estimating the Logistic Regression Model 416

Assessing the Goodness-of-Fit of the Estimation Model 419

Testing for Significance of the Coefficients 421

Interpreting the Coefficients 422

Calculating Probabilities for a Specific Value of the Independent Variable 425

Overview of Interpreting Coefficients 425

Summary 425

An Illustrative Example of Logistic Regression 426

Stages 1, 2, and 3: Research Objectives, Research Design, and Statistical Assumptions 426

Stage 4: Estimation of the Logistic Regression Model and Assessing Overall Fit 426

Stage 5: Interpretation of the Results 432

Stage 6: Validation of the Results 433

A Managerial Overview 434

Summary 434 . Questions 437 . Suggested Readings 437

References 437

Chapter 8 ANOVA and MANOVA 439

MANOVA: Extending Univariate Methods for Assessing Group Differences 443

Multivariate Procedures for Assessing Group Differences 444

A Hypothetical Illustration of MANOVA 447

Analysis Design 447

Differences from Discriminant Analysis 448

Forming the Variate and Assessing Differences 448

A Decision Process for MANOVA 449

Stage 1: Objectives of MANOVA 450

When Should We Use MANOVA? 450

Types of Multivariate Questions Suitable for MANOVA 451

Selecting the Dependent Measures 452

Stage 2: Issues in the Research Design of MANOVA 453

Sample Size Requirements—Overall and by Group 453

Factorial Designs—Two or More Treatments 453

Using Covariates—ANCOVA and MANCOVA 455

MANOVA Counterparts of Other ANOVA Designs 457

A Special Case of MANOVA: Repeated Measures 457

Stage 3: Assumptions of ANOVA and MANOVA 458

Independence 458

Equality of Variance–Covariance Matrices 459

Normality 460

Linearity and Multicollinearity Among the Dependent Variables 460

Sensitivity to Outliers 460

Stage 4: Estimation of the MANOVA Model and Assessing Overall Fit 460

Estimation with the General Linear Model 462

Criteria for Significance Testing 463

Statistical Power of the Multivariate Tests 463

Stage 5: Interpretation of the MANOVA Results 468

Evaluating Covariates 468

Assessing Effects on the Dependent Variate 468

Identifying Differences Between Individual Groups 472

Assessing Significance for Individual Dependent Variables 474

Stage 6: Validation of the Results 475

Summary 476

Illustration of a MANOVA Analysis 476

Example 1: Difference Between Two Independent Groups 477

Stage 1: Objectives of the Analysis 478

Stage 2: Research Design of the MANOVA 478

Stage 3: Assumptions in MANOVA 479

Stage 4: Estimation of the MANOVA Model and Assessing the Overall Fit 480

Stage 5: Interpretation of the Results 482

Example 2: Difference Between K Independent Groups 482

Stage 1: Objectives of the MANOVA 483

Stage 2: Research Design of MANOVA 483

Stage 3: Assumptions in MANOVA 484

Stage 4: Estimation of the MANOVA Model and Assessing Overall Fit 485

Stage 5: Interpretation of the Results 485

Example 3: A Factorial Design for MANOVA with Two Independent Variables 488

Stage 1: Objectives of the MANOVA 489

Stage 2: Research Design of the MANOVA 489

Stage 3: Assumptions in MANOVA 491

Stage 4: Estimation of the MANOVA Model and Assessing Overall Fit 492

Stage 5: Interpretation of the Results 495

Summary 496

A Managerial Overview of the Results 496

Summary 498 . Questions 500 . Suggested Readings 500

References 500

SECTION III Analysis Using Interdependence Techniques 503

Chapter 9 Grouping Data with Cluster Analysis 505

What Is Cluster Analysis? 508

Cluster Analysis as a Multivariate Technique 508

Conceptual Development with Cluster Analysis 508

Necessity of Conceptual Support in Cluster Analysis 509

How Does Cluster Analysis Work? 510

A Simple Example 510

Objective Versus Subjective Considerations 515

Cluster Analysis Decision Process 515

Stage 1: Objectives of Cluster Analysis 517

Stage 2: Research Design in Cluster Analysis 518

Stage 3: Assumptions in Cluster Analysis 526

Stage 4: Deriving Clusters and Assessing Overall Fit 527

Stage 5: Interpretation of the Clusters 538

Stage 6: Validation and Profiling of the Clusters 539

An Illustrative Example 541

Stage 1: Objectives of the Cluster Analysis 541

Stage 2: Research Design of the Cluster Analysis 542

Stage 3: Assumptions in Cluster Analysis 545

Employing Hierarchical and Nonhierarchical Methods 546

Step 1: Hierarchical Cluster Analysis (Stage 4) 546

Step 2: Nonhierarchical Cluster Analysis (Stages 4, 5, and 6) 552

Summary 561 . Questions 563 . Suggested Readings 563

References 563

Chapter 10 MDS and Correspondence Analysis 565

What Is Multidimensional Scaling? 568

Comparing Objects 568

Dimensions: The Basis for Comparison 569

A Simplified Look at How MDS Works 570

Gathering Similarity Judgments 570

Creating a Perceptual Map 570

Interpreting the Axes 571

Comparing MDS to Other Interdependence Techniques 572

Individual as the Unit of Analysis 573

Lack of a Variate 573

A Decision Framework for Perceptual Mapping 573

Stage 1: Objectives of MDS 573

Key Decisions in Setting Objectives 573

Stage 2: Research Design of MDS 578

Selection of Either a Decompositional (Attribute-Free)

or Compositional (Attribute-Based) Approach 578

Objects: Their Number and Selection 580

Nonmetric Versus Metric Methods 581

Collection of Similarity or Preference Data 581

Stage 3: Assumptions of MDS Analysis 584

Stage 4: Deriving the MDS Solution and Assessing Overall Fit 584

Determining an Object’s Position in the Perceptual Map 584

Selecting the Dimensionality of the Perceptual Map 586

Incorporating Preferences into MDS 587

Stage 5: Interpreting the MDS Results 592

Identifying the Dimensions 593

Stage 6: Validating the MDS Results 594

Issues in Validation 594

Approaches to Validation 594

Overview of Multidimensional Scaling 595

Correspondence Analysis 595

Distinguishing Characteristics 595

Differences from Other Multivariate Techniques 596

A Simple Example of CA 596

A Decision Framework for Correspondence Analysis 600

Stage 1: Objectives of CA 601

Stage 2: Research Design of CA 601

Stage 3: Assumptions in CA 602

Stage 4: Deriving CA Results and Assessing Overall Fit 602

Stage 5: Interpretation of the Results 603

Stage 6: Validation of the Results 604

Overview of Correspondence Analysis 604

Illustrations of MDS and Correspondence Analysis 605

Stage 1: Objectives of Perceptual Mapping 606

Identifying Objects for Inclusion 606

Basing the Analysis on Similarity or Preference Data 607

Using a Disaggregate or Aggregate Analysis 607

Stage 2: Research Design of the Perceptual Mapping Study 607

Selecting Decompositional or Compositional Methods 607

Selecting Firms for Analysis 608

Nonmetric Versus Metric Methods 608

Collecting Data for MDS 608

Collecting Data for Correspondence Analysis 609

Stage 3: Assumptions in Perceptual Mapping 610

Multidimensional Scaling: Stages 4 and 5 610

Stage 4: Deriving MDS Results and Assessing Overall Fit 610

Stage 5: Interpretation of the Results 615

Overview of the Decompositional Results 616

Correspondence Analysis: Stages 4 and 5 617

Stage 4: Estimating a Correspondence Analysis 617

Stage 5: Interpreting CA Results 619

Overview of CA 621

Stage 6: Validation of the Results 622

A Managerial Overview of MDS Results 622

Summary 623 . Questions 625 . Suggested Readings 625

References 625

SECTION IV Structural Equations Modeling 627

Chapter 11 SEM: An Introduction 629

What Is Structural Equation Modeling? 634

Estimation of Multiple Interrelated Dependence Relationships 635

Incorporating Latent Variables Not Measured Directly 635

Defining a Model 637

SEM and Other Multivariate Techniques 641

Similarity to Dependence Techniques 641

Similarity to Interdependence Techniques 641

The Emergence of SEM 642

The Role of Theory in Structural Equation Modeling 642

Specifying Relationships 642

Establishing Causation 643

Developing a Modeling Strategy 646

A Simple Example of SEM 647

The Research Question 647

Setting Up the Structural Equation Model for Path Analysis 648

The Basics of SEM Estimation and Assessment 649

Six Stages in Structural Equation Modeling 653

Stage 1: Defining Individual Constructs 655

Operationalizing the Construct 655

Pretesting 655

Stage 2: Developing and Specifying the Measurement Model 656

SEM Notation 656

Creating the Measurement Model 657

Stage 3: Designing a Study to Produce Empirical Results 657

Issues in Research Design 658

Issues in Model Estimation 662

Stage 4: Assessing Measurement Model Validity 664

The Basics of Goodness-of-Fit 665

Absolute Fit Indices 666

Incremental Fit Indices 668

Parsimony Fit Indices 669

Problems Associated with Using Fit Indices 669

Unacceptable Model Specification to Achieve Fit 671

Guidelines for Establishing Acceptable and Unacceptable Fit 672

Stage 5: Specifying the Structural Model 673

Stage 6: Assessing the Structural Model Validity 675

Structural Model GOF 675

Competitive Fit 676

Comparison to the Measurement Model 676

Testing Structural Relationships 677

Summary 678 . Questions 680 . Suggested Readings 680

Appendix 11A: Estimating Relationships Using Path Analysis 681

Appendix 11B: SEM Abbreviations 683

Appendix 11C: Detail on Selected GOF Indices 684

References 685

Chapter 12 Applications of SEM 687

Part 1: Confirmatory Factor Analysis 693

CFA and Exploratory Factor Analysis 693

A Simple Example of CFA and SEM 694

A Visual Diagram 694

SEM Stages for Testing Measurement Theory Validation with CFA 695

Stage 1: Defining Individual Constructs 696

Stage 2: Developing the Overall Measurement Model 696

Unidimensionality 696

Congeneric Measurement Model 698

Items per Construct 698

Reflective Versus Formative Constructs 701

Stage 3: Designing a Study to Produce Empirical Results 702

Measurement Scales in CFA 702

SEM and Sampling 703

Specifying the Model 703

Issues in Identification 704

Avoiding Identification Problems 704

Problems in Estimation 706

Stage 4: Assessing Measurement Model Validity 707

Assessing Fit 707

Path Estimates 707

Construct Validity 708

Model Diagnostics 711

Summary Example 713

CFA Illustration 715

Stage 1: Defining Individual Constructs 716

Stage 2: Developing the Overall Measurement Model 716

Stage 3: Designing a Study to Produce Empirical Results 718

Stage 4: Assessing Measurement Model Validity 719

HBAT CFA Summary 727

Part 2: What Is a Structural Model? 727

A Simple Example of a Structural Model 728

An Overview of Theory Testing with SEM 729

Stages in Testing Structural Theory 730

One-Step Versus Two-Step Approaches 730

Stage 5: Specifying the Structural Model 731

Unit of Analysis 731

Model Specification Using a Path Diagram 731

Designing the Study 735

Stage 6: Assessing the Structural Model Validity 737

Understanding Structural Model Fit from CFA Fit 737

Examine the Model Diagnostics 739

SEM Illustration 740

Stage 5: Specifying the Structural Model 740

Stage 6: Assessing the Structural Model Validity 742

Part 3: Extensions and Applications of SEM 749

Reflective Versus Formative Measures 749

Reflective Versus Formative Measurement Theory 749

Operationalizing a Formative Construct 750

Distinguishing Reflective from Formative Constructs 751

Which to Use—Reflective or Formative? 753

Higher-Order Factor Analysis 754

Empirical Concerns 754

Theoretical Concerns 756

Using Second-Order Measurement Theories 756

When to Use Higher-Order Factor Analysis 757

Multiple Groups Analysis 758

Measurement Model Comparisons 758

Structural Model Comparisons 763

Measurement Bias 764

Model Specification 764

Model Interpretation 765

Relationship Types: Mediation and Moderation 766

Mediation 766

Moderation 770

Longitudinal Data 773

Additional Covariance Sources: Timing 773

Using Error Covariances to Represent Added Covariance 774

Partial Least Squares 775

Characteristics of PLS 775

Advantages and Disadvantages of PLS 776

Choosing PLS Versus SEM 777

Summary 778 . Questions 781 . Suggested Readings 781

References 782

Index 785

编辑推荐

《多元数据分析(英文版)(第7版)》特色：以循序渐进方式（流水线方式）组织内容：在内容组织上，各章集中概述一个论题，每章均从基础开始并讨论应用。后面各章逐步深入。扩展各种方法应用：对"经验法则"给出解释，包括像样本容量这类重要问题。重新组织结构方程建模这一重要内容，包括结构方程建模概述、验证-性因素分析、估计和检验结构模型的相关问题，以及验证性因素分析和结构方程建模的一些高级主题，如检验更高阶因子模型、群组模型、调节变量与中间变量。

作者简介

本书是一本面向应用的经典多元数据分析教材，自1979年出版第1版至今，深受读者好评。本书循序渐进地介绍了各种多元统计分析方法，并通过丰富的实例演示了这些方法的应用。书中不仅涵盖多元数据分析的基本方法，而且还介绍了一些新方法，如结构方程建模和偏最小二乘法等。

本书特色


以循序渐进方式（流水线方式）组织内容：在内容组织上，各章集中概述一个论题，每章均从基础开始并讨论应用，后面各章逐步深入。


扩展各种方法应用：对“经验法则”给出解释，包括像样本容量这类重要问题。


重新组织结构方程建模这一重要内容，包括结构方程建模概述、验证性因素分析、估计和检验结构模型的相关问题，以及验证性因素分析和结构方程建模的一些高级主题，如检验更高阶因子模型、群组模型、调节变量与中间变量。

图书封面

---

 多元数据分析下载 更多精彩书评

---