Modern Analysis of Customer Surveys

Edited by RON S. KENETT, KPA Ltd., Raanana, Israel, University of Turin, Italy, and NYU-Poly, Center for Risk Engineering, New York, USA SILVIA SALINI, Department of Economics, Business and Statistics, University of Milan, Italy

• Foreword xviiPreface xixContributors xxiiiPart I Basic Aspects of Customer Satisfaction Survey Data Analysis1 Standards and Classical Techniques in Data Analysis of Customer Satisfaction Surveys 3Silvia Salini and Ron S. Kenett1.1 Literature on customer satisfaction surveys 41.2 Customer satisfaction surveys and the business cycle 41.3 Standards used in the analysis of survey data 71.4 Measures and models of customer satisfaction 121.4.1 The conceptual construct 121.4.2 The measurement process 131.5 Organization of the book 151.6 Summary 17References 172 The ABC Annual Customer Satisfaction Survey 19Ron S. Kenett and Silvia Salini2.1 The ABC company 192.2 ABC 2010 ACSS: Demographics of respondents 202.3 ABC 2010 ACSS: Overall satisfaction 222.4 ABC 2010 ACSS: Analysis of topics 242.5 ABC 2010 ACSS: Strengths and weaknesses and decision drivers 272.6 Summary 28References 28Appendix 293 Census and Sample Surveys 37Giovanna Nicolini and Luciana Dalla Valle3.1 Introduction 373.2 Types of surveys 393.2.1 Census and sample surveys 393.2.2 Sampling design 403.2.3 Managing a survey 403.2.4 Frequency of surveys 413.3 Non-sampling errors 413.3.1 Measurement error 423.3.2 Coverage error 423.3.3 Unit non-response and non-self-selection errors 433.3.4 Item non-response and non-self-selection error 443.4 Data collection methods 443.5 Methods to correct non-sampling errors 463.5.1 Methods to correct unit non-response errors 463.5.2 Methods to correct item non-response 493.6 Summary 51References 524 Measurement Scales 55Andrea Bonanomi and Gabriele Cantaluppi4.1 Scale construction 554.1.1 Nominal scale 564.1.2 Ordinal scale 574.1.3 Interval scale 584.1.4 Ratio scale 594.2 Scale transformations 604.2.1 Scale transformations referred to single items 614.2.2 Scale transformations to obtain scores on a unique interval scale 66Acknowledgements 69References 695 Integrated Analysis 71Silvia Biffignandi5.1 Introduction 715.2 Information sources and related problems 735.2.1 Types of data sources 735.2.2 Advantages of using secondary source data 735.2.3 Problems with secondary source data 745.2.4 Internal sources of secondary information 755.3 Root cause analysis 785.3.1 General concepts 785.3.2 Methods and tools in RCA 815.3.3 Root cause analysis and customer satisfaction 855.4 Summary 87Acknowledgement 87References 876 Web Surveys 89Roberto Furlan and Diego Martone6.1 Introduction 896.2 Main types of web surveys 906.3 Economic benefits of web survey research 916.3.1 Fixed and variable costs 926.4 Non-economic benefits of web survey research 946.5 Main drawbacks of web survey research 966.6 Web surveys for customer and employee satisfaction projects 1006.7 Summary 102References 1027 The Concept and Assessment of Customer Satisfaction 107Irena Ograjenšek and Iddo Gal7.1 Introduction 1077.2 The quality–satisfaction–loyalty chain 1087.2.1 Rationale 1087.2.2 Definitions of customer satisfaction 1087.2.3 From general conceptions to a measurement model of customer satisfaction 1107.2.4 Going beyond SERVQUAL: Other dimensions of relevance to the B2B context 1127.2.5 From customer satisfaction to customer loyalty 1137.3 Customer satisfaction assessment: Some methodological considerations 1157.3.1 Rationale 1157.3.2 Think big: An assessment programme 1157.3.3 Back to basics: Questionnaire design 1167.3.4 Impact of questionnaire design on interpretation 1187.3.5 Additional concerns in the B2B setting 1197.4 The ABC ACSS questionnaire: An evaluation 1197.4.1 Rationale 1197.4.2 Conceptual issues 1197.4.3 Methodological issues 1207.4.4 Overall ABC ACSS questionnaire asssessment 1217.5 Summary 121References 122Appendix 1268 Missing Data and Imputation Methods 129Alessandra Mattei, Fabrizia Mealli and Donald B. Rubin8.1 Introduction 1298.2 Missing-data patterns and missing-data mechanisms 1318.2.1 Missing-data patterns 1318.2.2 Missing-data mechanisms and ignorability 1328.3 Simple approaches to the missing-data problem 1348.3.1 Complete-case analysis 1348.3.2 Available-case analysis 1358.3.3 Weighting adjustment for unit nonresponse 1358.4 Single imputation 1368.5 Multiple imputation 1388.5.1 Multiple-imputation inference for a scalar estimand 1388.5.2 Proper multiple imputation 1398.5.3 Appropriately drawing imputations with monotone missing-data patterns 1408.5.4 Appropriately drawing imputations with nonmonotone missing-data patterns 1418.5.5 Multiple imputation in practice 1428.5.6 Software for multiple imputation 1438.6 Model-based approaches to the analysis of missing data 1448.7 Addressing missing data in the ABC annual customer satisfaction survey: An example 1458.8 Summary 149Acknowledgements 150References 1509 Outliers and Robustness for Ordinal Data 155Marco Riani, Francesca Torti and Sergio Zani9.1 An overview of outlier detection methods 1559.2 An example of masking 1579.3 Detection of outliers in ordinal variables 1599.4 Detection of bivariate ordinal outliers 1609.5 Detection of multivariate outliers in ordinal regression 1619.5.1 Theory 1619.5.2 Results from the application 1639.6 Summary 168References 168Part II Modern Techniques in Customer Satisfaction Survey Data Analysis10 Statistical Inference for Causal Effects 173Fabrizia Mealli, Barbara Pacini and Donald B. Rubin10.1 Introduction to the potential outcome approach to causal inference 17310.1.1 Causal inference primitives: Units, treatments, and potential outcomes 17510.1.2 Learning about causal effects: Multiple units and the stable unit treatment value assumption 17610.1.3 Defining causal estimands 17710.2 Assignment mechanisms 17910.2.1 The criticality of the assignment mechanism 17910.2.2 Unconfounded and strongly ignorable assignment mechanisms 18010.2.3 Confounded and ignorable assignment mechanisms 18110.2.4 Randomized and observational studies 18110.3 Inference in classical randomized experiments 18210.3.1 Fisher’s approach and extensions 18310.3.2 Neyman’s approach to randomization-based inference 18310.3.3 Covariates, regression models, and Bayesian model-based inference 18410.4 Inference in observational studies 18510.4.1 Inference in regular designs 18610.4.2 Designing observational studies: The role of the propensity score 18610.4.3 Estimation methods 18810.4.4 Inference in irregular designs 18810.4.5 Sensitivity and bounds 18910.4.6 Broken randomized experiments as templates for the analysis of some irregular designs 189References 19011 Bayesian Networks Applied to Customer Surveys 193Ron S. Kenett, Giovanni Perruca and Silvia Salini11.1 Introduction to Bayesian networks 19311.2 The Bayesian network model in practice 19711.2.1 Bayesian network analysis of the ABC 2010 ACSS 19711.2.2 Transport data analysis 20111.2.3 R packages and other software programs used for studying BNs 21011.3 Prediction and explanation 21111.4 Summary 213References 21312 Log-linear Model Methods 217Stephen E. Fienberg and Daniel Manrique-Vallier12.1 Introduction 21712.2 Overview of log-linear models and methods 21812.2.1 Two-way tables 21812.2.2 Hierarchical log-linear models 22012.2.3 Model search and selection 22212.2.4 Sparseness in contingency tables and its implications 22312.2.5 Computer programs for log-linear model analysis 22312.3 Application to ABC survey data 22412.4 Summary 227References 22813 CUB Models: Statistical Methods and Empirical Evidence 231Maria Iannario and Domenico Piccolo13.1 Introduction 23113.2 Logical foundations and psychological motivations 23313.3 A class of models for ordinal data 23313.4 Main inferential issues 23613.5 Specification of CUB models with subjects’ covariates 23813.6 Interpreting the role of covariates 24013.7 A more general sampling framework 24113.7.1 Objects’ covariates 24113.7.2 Contextual covariates 24313.8 Applications of CUB models 24413.8.1 Models for the ABC annual customer satisfaction survey 24513.8.2 Students’ satisfaction with a university orientation service 24613.9 Further generalizations 24813.10 Concluding remarks 251Acknowledgements 251References 251Appendix 255A program in R for CUB models 255A.1 Main structure of the program 255A.2 Inference on CUB models 255A.3 Output of CUB models estimation program 256A.4 Visualization of several CUB models in the parameter space 257A.5 Inference on CUB models in a multi-object framework 257A.6 Advanced software support for CUB models 25814 The Rasch Model 259Francesca De Battisti, Giovanna Nicolini and Silvia Salini14.1 An overview of the Rasch model 25914.1.1 The origins and the properties of the model 25914.1.2 Rasch model for hierarchical and longitudinal data 26314.1.3 Rasch model applications in customer satisfaction surveys 26514.2 The Rasch model in practice 26714.2.1 Single model 26714.2.2 Overall model 26814.2.3 Dimension model 27214.3 Rasch model software 27714.4 Summary 278References 27915 Tree-based Methods and Decision Trees 283Giuliano Galimberti and Gabriele Soffritti15.1 An overview of tree-based methods and decision trees 28315.1.1 The origins of tree-based methods 28315.1.2 Tree graphs, tree-based methods and decision trees 28415.1.3 CART 28715.1.4 CHAID 29315.1.5 PARTY 29515.1.6 A comparison of CART, CHAID and PARTY 29715.1.7 Missing values 29715.1.8 Tree-based methods for applications in customer satisfaction surveys 29815.2 Tree-based methods and decision trees in practice 30015.2.1 ABC ACSS data analysis with tree-based methods 30015.2.2 Packages and software implementing tree-based methods 30315.3 Further developments 304References 30416 PLS Models 309Giuseppe Boari and Gabriele Cantaluppi16.1 Introduction 30916.2 The general formulation of a structural equation model 31016.2.1 The inner model 31016.2.2 The outer model 31216.3 The PLS algorithm 31316.4 Statistical interpretation of PLS 31916.5 Geometrical interpretation of PLS 32016.6 Comparison of the properties of PLS and LISREL procedures 32116.7 Available software for PLS estimation 32316.8 Application to real data: Customer satisfaction analysis 323References 32917 Nonlinear Principal Component Analysis 333Pier Alda Ferrari and Alessandro Barbiero17.1 Introduction 33317.2 Homogeneity analysis and nonlinear principal component analysis 33417.2.1 Homogeneity analysis 33417.2.2 Nonlinear principal component analysis 33617.3 Analysis of customer satisfaction 33817.3.1 The setting up of indicator 33817.3.2 Additional analysis 34017.4 Dealing with missing data 34017.5 Nonlinear principal component analysis versus two competitors 34317.6 Application to the ABC ACSS data 34417.6.1 Data preparation 34417.6.2 The homals package 34517.6.3 Analysis on the ‘complete subset’ 34617.6.4 Comparison of NLPCA with PCA and Rasch analysis 35017.6.5 Analysis of ‘entire data set’ for the comparison of missing data treatments 35217.7 Summary 355References 35518 Multidimensional Scaling 357Nadia Solaro18.1 An overview of multidimensional scaling techniques 35718.1.1 The origins of MDS models 35818.1.2 MDS input data 35918.1.3 MDS models 36218.1.4 Assessing the goodness of MDS solutions 36918.1.5 Comparing two MDS solutions: Procrustes analysis 37118.1.6 Robustness issues in the MDS framework 37118.1.7 Handling missing values in MDS framework 37318.1.8 MDS applications in customer satisfaction surveys 37318.2 Multidimensional scaling in practice 37418.2.1 Data sets analysed 37518.2.2 MDS analyses of overall satisfaction with a set of ABC features: The complete data set 37518.2.3 Weighting objects or items 38118.2.4 Robustness analysis with the forward search 38218.2.5 MDS analyses of overall satisfaction with a set of ABC features: The incomplete data set 38318.2.6 Package and software for MDS methods 38418.3 Multidimensional scaling in a future perspective 38618.4 Summary 386References 38719 Multilevel Models for Ordinal Data 391Leonardo Grilli and Carla Rampichini19.1 Ordinal variables 39119.2 Standard models for ordinal data 39319.2.1 Cumulative models 39419.2.2 Other models 39519.3 Multilevel models for ordinal data 39519.3.1 Representation as an underlying linear model with thresholds 39619.3.2 Marginal versus conditional effects 39719.3.3 Summarizing the cluster-level unobserved heterogeneity 39719.3.4 Consequences of adding a covariate 39819.3.5 Predicted probabilities 39919.3.6 Cluster-level covariates and contextual effects 39919.3.7 Estimation of model parameters 40019.3.8 Inference on model parameters 40119.3.9 Prediction of random effects 40219.3.10 Software 40319.4 Multilevel models for ordinal data in practice: An application to student ratings 404References 40820 Quality Standards and Control Charts Applied to Customer Surveys 413Ron S. Kenett, Laura Deldossi and Diego Zappa20.1 Quality standards and customer satisfaction 41320.2 ISO 10004 guidelines for monitoring and measuring customer satisfaction 41420.3 Control Charts and ISO 7870 41720.4 Control charts and customer surveys: Standard assumptions 42020.4.1 Introduction 42020.4.2 Standard control charts 42020.5 Control charts and customer surveys: Non-standard methods 42620.5.1 Weights on counts: Another application of the c chart 42620.5.2 The χ2 chart 42720.5.3 Sequential probability ratio tests 42820.5.4 Control chart over items: A non-standard application of SPC methods 42920.5.5 Bayesian control chart for attributes: A modern application of SPC methods 43220.5.6 Control chart for correlated Poisson counts: When things become fairly complicated 43320.6 The M-test for assessing sample representation 43320.7 Summary 435References 43621 Fuzzy Methods and Satisfaction Indices 439Sergio Zani, Maria Adele Milioli and Isabella Morlini21.1 Introduction 43921.2 Basic definitions and operations 44021.3 Fuzzy numbers 44121.4 A criterion for fuzzy transformation of variables 44321.5 Aggregation and weighting of variables 44521.6 Application to the ABC customer satisfaction survey data 44621.6.1 The input matrices 44621.6.2 Main results 44821.7 Summary 453References 455Appendix an Introduction to R 457Stefano Maria IacusA.1 Introduction 457A.2 How to obtain R 457A.3 Type rather than ‘point and click’ 458A.3.1 The workspace 458A.3.2 Graphics 458A.3.3 Getting help 459A.3.4 Installing packages 459A.4 Objects 460A.4.1 Assignments 460A.4.2 Basic object types 462A.4.3 Accessing objects and subsetting 466A.4.4 Coercion between data types 469A.5 S4 objects 470A.6 Functions 472A.7 Vectorization 473A.8 Importing data from different sources 475A.9 Interacting with databases 476A.10 Simple graphics manipulation 477A.11 Basic analysis of the ABC data 481A.12 About this document 496A.13 Bibliographical notes 496References 496Index 499