Nancy V. Wünderlich Acceptance of Remote Services
GABLER RESEARCH Applied Marketing Science / Angewandte Marketingfor...
42 downloads
952 Views
2MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Nancy V. Wünderlich Acceptance of Remote Services
GABLER RESEARCH Applied Marketing Science / Angewandte Marketingforschung Editorial Board: Prof. Dr. Dieter Ahlert, Universität Münster Prof. Dr. Heiner Evanschitzky, University of Strathclyde/UK Dr. Josef Hesse, Schäper Sportgerätebau GmbH Prof. Dr. Gopalkrishnan R. Iyer, Florida Atlantic University/USA Prof. Dr. Hartmut H. Holzmüller, Universität Dortmund Prof. Dr. Gustavo Möller-Hergt, Technische Universität Berlin Prof. Dr. Lou Pelton, University of North Texas/USA Prof. Dr. Arun Sharma, University of Miami/USA Prof. Dr. Florian von Wangenheim, Technische Universität München Prof. Dr. David Woisetschläger, Universität Dortmund
The book series ”Applied Marketing Science / Angewandte Marketingforschung“ is designated to the transfer of top-end scientific knowledge to interested practitioners. Books from this series are focused – but not limited – to the field of Marketing Channels, Retailing, Network Relationships, Sales Management, Brand Management, Consumer Marketing and Relationship Marketing / Management. The industrial focus lies primarily on the service industry, consumer goods industry and the textile / apparel industry. The issues in this series are either edited books or monographs. Books are either in German or English language; other languages are possible upon request. Book volumes published in the series ”Applied Marketing Science / Angewandte Marketingforschung“ will primarily be aimed at interested managers, academics and students of marketing. The works will not be written especially for teaching purposes. However, individual volumes may serve as material for marketing courses, upper-level MBA- or Ph.D.-courses in particular.
Nancy V. Wünderlich
Acceptance of Remote Services Perception, Adoption, and Continued Usage in Organizational Settings
With a foreword by Prof. Dr. Florian von Wangenheim
RESEARCH
Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de.
Dissertation Technische Universität München, 2009
1st Edition 2009 All rights reserved © Gabler | GWV Fachverlage GmbH, Wiesbaden 2009 Editorial Office: Claudia Jeske | Sabine Schöller Gabler is part of the specialist publishing group Springer Science+Business Media. www.gabler.de No part of this publication may be reproduced, stored in a retrieval system or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior written permission of the copyright holder. Registered and/or industrial names, trade names, trade descriptions etc. cited in this publication are part of the law for trade-mark protection and may not be used free in any form or by any means even if this is not specifically marked. Cover design: KünkelLopka Medienentwicklung, Heidelberg Printed on acid-free paper Printed in Germany ISBN 978-3-8349-1957-1
Foreword The way services are conceived, developed, and delivered has changed considerable in view of the recent advances in information and communication technologies. New "intelligent products" contain IT in the form of microchips, software, and sensors and are able to collect, process, and produce information. The continuous data flow from embedded IT-applications enables seamless services delivered in real-time and directed at connected objects. In this environment, remote services are quickly emerging as a new class of fascinating technology-mediated services. The number of remote service offerings has grown enormously in recent years and is expected to be the fastest growing technology-driven service type over the next decade. Remote services are highly complex, depend on mediating technologies, and require human-to-human interaction. One of the greatest challenge in this realm has proven to be the interplay between the drivers and barriers of customer adoption and acceptance, especially since remote services are predominantly found in heterogeneous B2B-environments. Theories of technology adoption that are well established in literature tend to focus mostly on the technology itself as the primary determinant of adoption and usage. This was sufficient for more basic types of services such as self-, or e-services but falls short with more advanced services in organizational settings. Thus, this dissertation is very important from a theoretical perspective. Nancy Wünderlich developed and validated a new model – the ITSUM – that forms a sound theoretical base for explaining intended and actual use of interactive remote services and can be used to predict actual service usage. The ITSUM introduces additional constructs including trust in and control of the ’service counterpart’, and aspects of customer co-production behavior like role clarity, ability and intrinsic motivation. By incorporating the ’human element’ into the model, Nancy Wünderlich contributes to the underlying theory and increases overall understanding of the phenomenon. She also shows that the predictors of remote service usage vary across groups, depending on whether the respondent’s company is in the early stages (pre-adopter) or already a user of remote services (continued usage). A major strength of the dissertation is its conceptual, theoretical, and qualitative work that precedes the rigorous quantitative testing of the ITSUM model. The model is very well supported by the data, but equally important it is also strongly supported by an extensive, inter-disciplinary
VI
Foreword
literature review and a careful, detailed, and deep qualitative interview study conducted in Germany, USA and China. This work is also critically important from a practical perspective. Helping organizations to understand the underlying drivers of customer acceptance and adoption of new types of services is of paramount interest not only in competitive dynamic markets but also to advance the organization itself. Nancy Wünderlich derives clear and concise managerial implications for remote service providers on how to increase remote service acceptance among their customers and facilitate the export of remote services. In sum, this is a remarkable thesis that substantially enhances the theoretical understanding of remote services as well as serving as a guide for managerial practice. Nancy Wünderlich has already been honored with several national and international awards – e.g., IMS & AMA SERVSIG Dissertation Proposal Award 2009, Doctoral Proposal Award of the Society for Marketing Advances 2008, Young Career in Service Science Award of the BMBF 2008, ASU/Liam Glynn Scholarship Award 2007 – for her dissertation proposal. I highly recommend this book to academics and practitioners who are interested in the management and marketing of innovative, technology-based services. Florian v. Wangenheim
Acknowledgements Foremost, my gratitude goes to my advisors, Professor Florian v. Wangenheim and Professor Mary Jo Bitner, for their endless support, enthusiasm, guidance, and inspiration. Their knowledge and insight were paramount to the success of this dissertation. Florian v. Wangenheim has provided the ideal environment for my work. He not only allowed me great freedom to pursue independent work, but from the beginning he encouraged me to participate in the international research community. Florian’s rigor, intelligence, and kindness have been invaluable not only to my development as a researcher, but also to my path as a human being. I am deeply grateful to Mary Jo Bitner for the long discussions that helped shape the direction of this work, and of my career. Mary Jo has always been there to listen and to offer indispensable advice. She has shown faith in my work from the start and has been a ceaseless advocate for me throughout the project and beyond. My truly memorable time with her, as a visiting PhD scholar at the marketing department of Arizona State University, will have a lasting impact on me. I want to thank the members of the service science community for creating a stimulating and friendly atmosphere that widened my scientific understanding in many respects. In particular, I thank Professors Ruth Bolton, Stephen Brown, Michael Ghoul, Hartmut Holzmüller, Amy Ostrom, Kay Lemon, and Ralf Reichwald for the lively discussions and for their insightful and encouraging comments. I gratefully acknowledge the institutional support that I have received while working on my dissertation. My study was conducted in the context of the project "EXFED - Export ferngelenkter Dienstleistungen" (FKZ: 01HQ0553), which was funded by the German Federal Ministry of Education (BMBF) and Research, and was supported by the German Aerospace Center (DLR). My gratitude also extends to my colleagues for their tremendous support during the time of my dissertation. In particular, I wish to thank Jan H. Schumann and Markus Wübben, whose sense of humor propelled me through the ups and downs of life as a PhD candidate. Jan and I undertook some of the most fun-filled and adventurous field trips one can ever hope to make.
VIII
Acknowledgements
Also, I want to thank my colleagues and friends at the marketing department of Technische Universität München who supported me in countless ways: Sebastian Ackermann, Armin "Raj" Arnold, Marcus Demmelmair, Christian Heumann, Clemens Hiraoka, Michael Lödding, Sabine Mayser, Anne Scherer, and Marcus Zimmer. I also received valuable input from practitioners. In particular, I acknowledge the assistance of Florian Bornemann, Veselin Panshef, Michael Pfeffer, and Weiwei Wang. Not only did their expert knowledge provide a continuous stream of insights, but they also put me in touch with the printing companies in Germany, USA, and China, which were central to my work. My deepest thanks go to my husband, Robin Wünderlich, for his unfailing love, encouragement, support, and kind indulgence with my mood and temper – especially as the deadline loomed. Without him this dissertation would not have been possible. Last but not least, I credit my cat Peppers for amazing me every day with his valiant efforts to distract me from typing. Nancy V. Wünderlich
Short Table of Contents Foreword
V
Acknowledgements List of Figures
VII XVIII
List of Tables
XIX
List of Abbreviations
XXI
1
Introduction
1
2
Conceptual Framework: Remote Services in Context of Technology-Mediated Services
7
3
Theoretical Framework for Remote Service Adoption and Continued Usage
31
4
Methodological Superstructure and Empirical Setting
85
5
Qualitative Exploratory Interview Study
93
6
Hypotheses Development
131
7
Quantitative Studies
149
8
Summary and Conclusions
201
References
209
A Additional Tables and Figures
255
Table of Contents Foreword
V
Acknowledgements List of Figures
VII XVIII
List of Tables
XIX
List of Abbreviations
XXI
1
2
Introduction
1
1.1
Motivation and Goals of the Thesis . . . . . . . . . . . . . . . . . . . . . . . .
1
1.2
Research Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4
1.3
Structure of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4
Conceptual Framework: Remote Services in Context of Technology-Mediated Services
7
2.1
Emerging Technology-Mediated Service Types . . . . . . . . . . . . . . . . .
7
2.1.1
E-Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7
2.1.2
Self-Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
9
2.1.3
Mobile Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12
2.1.4
Industry Specific Technology-Mediated Services . . . . . . . . . . . .
13
2.2
2.3
2.1.4.1
Teleservices in Engineering and Manufacturing Industries . .
13
2.1.4.2
Telematics in the Automotive Industry . . . . . . . . . . . .
15
2.1.4.3
Telemedicine in Health Care . . . . . . . . . . . . . . . . .
16
2.1.4.4
Services in the IT-Sector . . . . . . . . . . . . . . . . . . . .
18
Classification of Remote Services . . . . . . . . . . . . . . . . . . . . . . . .
19
2.2.1
Definition of Remote Services . . . . . . . . . . . . . . . . . . . . . .
19
2.2.2
Characteristics of Remote Services . . . . . . . . . . . . . . . . . . .
20
2.2.3
Benefits of Remote Services . . . . . . . . . . . . . . . . . . . . . . .
23
Classification of Interactive Remote Services . . . . . . . . . . . . . . . . . .
24
2.3.1
24
Definition of Interactive Remote Services . . . . . . . . . . . . . . . .
XII
TABLE OF CONTENTS
2.4 3
2.3.2
Characterization and Demarcation of Interactive Remote Services . . .
2.3.3
Positioning of Interactive Remote Services . . . . . . . . . . . . . . .
26
Conclusions and Implications . . . . . . . . . . . . . . . . . . . . . . . . . . .
29
Theoretical Framework for Remote Service Adoption and Continued Usage 3.1
31
3.1.1
Behavioral Theories from Social Psychology and Sociology . . . . . .
32
3.1.1.1
Innovation Diffusion Theory and Variants . . . . . . . . . .
32
3.1.1.2
The Theory of Reasoned Action and Variants . . . . . . . . .
34
3.1.1.3
The Theory of Planned Behavior and Variants . . . . . . . .
36
3.1.1.4
3.1.3
The Decomposed Theory of Planned Behavior . . . . . . . .
37
Models in IT-Adoption Based on Behavioral Theories . . . . . . . . .
39
3.1.2.1
The Technology Acceptance Model and Variants . . . . . . .
39
3.1.2.2
The Motivational Model and Variants . . . . . . . . . . . . .
41
3.1.2.3
The Unified Theory of Acceptance and Use of Technology .
42
3.1.2.4
Compeau and Higgins’ Model based on Social Cognitive Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
44
Theoretical Foundations of Continued Use of Technology . . . . . . .
45
3.1.3.1
Importance of Prior Experience . . . . . . . . . . . . . . . .
45
3.1.3.2
Studies On Continued Usage . . . . . . . . . . . . . . . . .
46
3.1.3.3
Comparison of Adoption and Continuance Drivers . . . . . .
48
Summary and Overview of Models in Technology Adoption . . . . . .
50
Theoretical Foundations of Interaction in the Service Encounter . . . . . . . .
58
3.2.1
Perceptions of Service Providers’ Employee Behavior . . . . . . . . .
58
3.2.1.1
Importance of Employee Behavior in the Service Encounter .
58
3.2.1.2
Customer Orientation of Employees . . . . . . . . . . . . .
60
3.2.1.3
Role of Employee Behavior in Service Quality Assessments .
60
3.2.1.4
Employee Behavior in Technology-Mediated Service Encoun-
3.1.4
ters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2
3.2.3
62
Customer Integration in the Service Process . . . . . . . . . . . . . . .
65
3.2.2.1
Research on Customer Co-Production . . . . . . . . . . . .
65
3.2.2.2
Drivers of Customer Co-Production . . . . . . . . . . . . . .
66
Customer Beliefs Regarding the Interaction with Service Technology .
69
3.2.3.1
Consumer Readiness as Driver of Technology-Mediated CoProduction . . . . . . . . . . . . . . . . . . . . . . . . . . .
69
Technology Readiness as a Driver of Technology Usage in Services . . . . . . . . . . . . . . . . . . . . . . . . . . . .
69
Transcending Concepts of Trust and Control across Disciplines . . . . . . . . .
71
3.2.3.2 3.3
31
Theoretical Foundations of Technology Adoption . . . . . . . . . . . . . . . .
3.1.2
3.2
25
3.3.1
Importance of Trust and Trustworthiness . . . . . . . . . . . . . . . .
71
3.3.2
Importance of Control Beliefs . . . . . . . . . . . . . . . . . . . . . .
73
TABLE OF CONTENTS 3.3.3 3.4
3.5 4
5
XIII
The Trust-Control Nexus . . . . . . . . . . . . . . . . . . . . . . . . .
75
Technology-Intensive Service Adoption in B2B contexts . . . . . . . . . . . .
77
3.4.1
. . . . . . . . . . . . . . . . . . . . .
77
3.4.2
Decision Making and the Adoption Process in Organizations . . . . . .
78
3.4.3
Organizational Adoption Drivers . . . . . . . . . . . . . . . . . . . . .
81
Summary of the Theoretical Foundations of Remote Services . . . . . . . . . .
82
Business Service Relationships
Methodological Superstructure and Empirical Setting
85
4.1
Methodological Superstructure . . . . . . . . . . . . . . . . . . . . . . . . . .
85
4.2
Empirical Setting of the Employed Studies . . . . . . . . . . . . . . . . . . . .
87
4.2.1
Selection of the Printing Industry . . . . . . . . . . . . . . . . . . . .
87
4.2.2
Printing Machine Manufacturing . . . . . . . . . . . . . . . . . . . . .
88
4.2.3
The Printing Industry . . . . . . . . . . . . . . . . . . . . . . . . . . .
89
Qualitative Exploratory Interview Study
93
5.1
Motivation and Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
93
5.2
Qualitative Research Methodology . . . . . . . . . . . . . . . . . . . . . . . .
94
5.2.1
Semi-Structured Interviews as Means of Data Collection . . . . . . . .
94
5.2.2
Qualitative Content Analysis as Means of Data Analysis . . . . . . . .
94
5.2.3 5.3
5.4
Validity and Reliability . . . . . . . . . . . . . . . . . . . . . . . . . .
95
Field Phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
97
5.3.1
Sample Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
97
5.3.2
Interview Situation and Questionnaire Design . . . . . . . . . . . . . . 102
5.3.3
Category Development and Coding . . . . . . . . . . . . . . . . . . . 105
Results of the Qualitative Interview Study . . . . . . . . . . . . . . . . . . . . 105 5.4.1
Assessment of Intercoder Reliability . . . . . . . . . . . . . . . . . . . 105
5.4.2
Structure of Results Presentation . . . . . . . . . . . . . . . . . . . . . 106
5.4.3
Technology Beliefs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
5.4.4
Relational Beliefs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 5.4.4.1
Trust in the Remote Service Technician . . . . . . . . . . . . 110
5.4.4.2
Trust in the Remote Service Provider Company . . . . . . . 113
5.4.5
Process Control Beliefs . . . . . . . . . . . . . . . . . . . . . . . . . . 114
5.4.6
Economic Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
5.4.7
Participation Beliefs . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.4.8
Cultural Differences in the Customer’s Willingness to Collaborate . . . 122
5.4.9
Prior Experiences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
5.4.10 Organizational Factors . . . . . . . . . . . . . . . . . . . . . . . . . . 124 5.4.11 Contextual Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126 5.4.12 Discussion of the Results . . . . . . . . . . . . . . . . . . . . . . . . . 127
XIV 6
TABLE OF CONTENTS
Hypotheses Development 6.1
6.1.1
6.1.2
Counterpart Beliefs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 6.1.1.1
Controllability of the Counterpart’s Actions . . . . . . . . . 132
6.1.1.2
Trustworthiness of the Counterpart . . . . . . . . . . . . . . 134
Technology Beliefs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 6.1.2.1
Trust in Technology . . . . . . . . . . . . . . . . . . . . . . 136
6.1.2.2
Ease of Use . . . . . . . . . . . . . . . . . . . . . . . . . . 136
6.1.3
Perceived Usefulness . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
6.1.4
Participation Beliefs . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
6.1.5
7
131
Development of the ITSUM . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
6.1.4.1
Role Clarity . . . . . . . . . . . . . . . . . . . . . . . . . . 139
6.1.4.2
Role Ability . . . . . . . . . . . . . . . . . . . . . . . . . . 140
6.1.4.3
Intrinsic Motivation . . . . . . . . . . . . . . . . . . . . . . 140
Organizational Characteristics . . . . . . . . . . . . . . . . . . . . . . 141 6.1.5.1
Subjective Norms . . . . . . . . . . . . . . . . . . . . . . . 141
6.1.5.2
Company Size and Respondent’s Function . . . . . . . . . . 142
6.2
Link Between Usage Intention and Actual Usage Behavior . . . . . . . . . . . 143
6.3
Hypotheses Development for Group Comparisons . . . . . . . . . . . . . . . . 143
Quantitative Studies
149
7.1
Motivation and Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
7.2
Methods and Techniques Employed . . . . . . . . . . . . . . . . . . . . . . . 150 7.2.1 7.2.2
Survey Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 Structural Equation Modeling . . . . . . . . . . . . . . . . . . . . . . 151 7.2.2.1
Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . 151
7.2.2.2
Assessment of Reliability and Validity . . . . . . . . . . . . 153
7.2.2.3
Assessment of Model Fit and Data Quality . . . . . . . . . . 154
7.2.2.4
Dependent Categorical Variables . . . . . . . . . . . . . . . 156
7.2.2.5
Multi-Group Comparison . . . . . . . . . . . . . . . . . . . 156
7.3
Study Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
7.4
General Outline of the Questionnaires . . . . . . . . . . . . . . . . . . . . . . 160
7.5
Operationalization of the Constructs . . . . . . . . . . . . . . . . . . . . . . . 161
7.6
Quality of the Questionnaire and Pre-Test . . . . . . . . . . . . . . . . . . . . 166
7.7
7.8
t1 -Study: Results of ITSUM Validation . . . . . . . . . . . . . . . . . . . . . . 167 7.7.1
Sample Structure and Description . . . . . . . . . . . . . . . . . . . . 167
7.7.2
Data Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
7.7.3
Measurement Validity . . . . . . . . . . . . . . . . . . . . . . . . . . 171
7.7.4
Assessing Common Method Variance . . . . . . . . . . . . . . . . . . 174
7.7.5
Validation of the ITSUM (n=717) . . . . . . . . . . . . . . . . . . . . 175
Multi-Group Comparison: Adoption vs. Continued Usage . . . . . . . . . . . 178
TABLE OF CONTENTS
7.9
XV
7.8.1
Description of the Groups . . . . . . . . . . . . . . . . . . . . . . . . 178
7.8.2 7.8.3
Assessing Measurement Invariance . . . . . . . . . . . . . . . . . . . 179 Results for Organizations in the Pre-Adoption Phase . . . . . . . . . . 184
7.8.4 Results for Organizations in the Continued Usage Phase 7.8.5 Comparison of Group Parameters . . . . . . . . . . . . t2 -Study: Intention - Behavior Link . . . . . . . . . . . . . . . . 7.9.1 Sample Description . . . . . . . . . . . . . . . . . . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
. . . .
186 189 192 192
7.9.2 Logistic Regression Results . . . . . . . . . . . . . . . . . . . . . . . 194 7.10 Discussion of the Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 8
Summary and Conclusions 201 8.1 Summary of the Central Results . . . . . . . . . . . . . . . . . . . . . . . . . 201 8.2 Managerial Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203 8.3 Implications for Future Research . . . . . . . . . . . . . . . . . . . . . . . . . 206
References
209
A Additional Tables and Figures
255
A.1 Interview Guideline of the Exploratory Qualitative Study . . . . . . . . . . . . 255 A.2 First Pages of the Online Survey t1 and t2 -study . . . . . . . . . . . . . . . . . 257 A.3 Exploratory Factor Analysis Results . . . . . . . . . . . . . . . . . . . . . . . 259 A.4 Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260 A.5 Calculation of Moderating Effects . . . . . . . . . . . . . . . . . . . . . . . . 261
List of Figures 1.1
Structure of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5
2.1
The Continuum from eService to eCommerce . . . . . . . . . . . . . . . . . .
9
2.2
Categories and Examples of SST . . . . . . . . . . . . . . . . . . . . . . . . .
11
2.3
Application Fields of Teleservices . . . . . . . . . . . . . . . . . . . . . . . .
15
2.4
Categories of Telematic Services . . . . . . . . . . . . . . . . . . . . . . . . .
17
2.5
Activity Portfolio of Customer and Provider Actions . . . . . . . . . . . . . .
22
2.6
Features of Remote Services . . . . . . . . . . . . . . . . . . . . . . . . . . .
23
2.7
Benefits and Challenges of Remote Services . . . . . . . . . . . . . . . . . . .
24
2.8
Features of Interactive Remote Services Compared to Remote Services . . . . .
27
2.9
Technology-Interaction-Service Matrix . . . . . . . . . . . . . . . . . . . . . .
28
3.1
The Innovation Decision Process . . . . . . . . . . . . . . . . . . . . . . . . .
33
3.2
The Theory of Reasoned Action . . . . . . . . . . . . . . . . . . . . . . . . .
34
3.3
The Theory of Planned Behavior . . . . . . . . . . . . . . . . . . . . . . . . .
37
3.4
The Decomposed Theory of Planned Behavior . . . . . . . . . . . . . . . . . .
38
3.5
The Technology Acceptance Model 1 and 2 . . . . . . . . . . . . . . . . . . .
40
3.6
The Unified Theory of Acceptance and Use of Technology . . . . . . . . . . .
43
3.7
Compeau and Higgins’ (1995) Model Based on SCT . . . . . . . . . . . . . .
45
3.8
IT Continuance Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
47
3.9
Adoption and Continuance Model of Karahanna, Straub, and Chervany (1999) .
49
3.10 The Customer Oriented-Skills of an Employee (COSE) . . . . . . . . . . . . .
61
3.11 Generic Dimensions to Evaluate Service Quality . . . . . . . . . . . . . . . . .
61
3.12 B-A-I Framework of Technology-Mediated Customer Service . . . . . . . . .
63
3.13 CSR Characteristics and their Effects on Customer Satisfaction . . . . . . . . .
64
3.14 Levels of Customer Participation across Different Services . . . . . . . . . . .
66
3.15 Key Predictors of Consumer Trial of Self-Service Technologies . . . . . . . . .
70
3.16 The Integrated Framework of Trust, Control, and Risk in Strategic Alliances . .
76
3.17 Key Differences between B2C and B2B Marketing . . . . . . . . . . . . . . .
77
3.18 Triangle Model and Square Model . . . . . . . . . . . . . . . . . . . . . . . .
79
3.19 The Innovation Process in Organizations . . . . . . . . . . . . . . . . . . . . .
82
XVIII
LIST OF FIGURES
4.1
Major Differences between Qualitative and Quantitative Research . . . . . . .
85
4.2
The Print Production Process . . . . . . . . . . . . . . . . . . . . . . . . . . .
91
5.1 5.2 5.3
Inductive Approach of Qualitative Content Analysis . . . . . . . . . . . . . . . 95 Coding Categories of the Qualitative Interview Study – Part 1 . . . . . . . . . . 103 Coding Categories of the Qualitative Interview Study – Part 2 . . . . . . . . . . 104
5.4 5.5
Conceptual Framework Resulting From Qualitative Study . . . . . . . . . . . . 107 Factors Influencing Remote Service Perception . . . . . . . . . . . . . . . . . 129
6.1
The Extended Interactive Technology-Mediated Service Usage Model . . . . . 132
7.1 7.2 7.3
Proposed Procedure for Assessing Measurement Invariance . . . . . . . . . . . 157 Company Size in the Printing Industry vs. Overall Sample . . . . . . . . . . . 168 Distribution of Respondent’s Age in the Overall Sample (n=717) . . . . . . . . 168
7.4
Self-Reported Classification of Business Segments . . . . . . . . . . . . . . . 169
7.5 7.6 7.7
Distribution of Respondent’s Function in the Sample . . . . . . . . . . . . . . 169 Results of the ITSUM (n=717) . . . . . . . . . . . . . . . . . . . . . . . . . . 176 Distribution of Respondent’s Age Across Groups . . . . . . . . . . . . . . . . 180
7.8 Distribution of Respondent’s Gender Across Groups . . . . . . . . . . . . . . 180 7.9 Distribution of Company Size Across Groups . . . . . . . . . . . . . . . . . . 181 7.10 Distribution of Respondent’s Function Across Groups . . . . . . . . . . . . . . 181 7.11 Results of the ITSUM: Pre-Adopter Group (n=364) . . . . . . . . . . . 7.12 Results of the ITSUM: Continued User Group (n=353) . . . . . . . . . 7.13 Distribution of Respondent’s Age Across Samples (t1 and t2 -Study) . . 7.14 Distribution of Respondent’s Gender Across Samples (t1 and t2 -Study) . 7.15 Distribution of Respondent’s Function Across Samples (t1 and t2 -Study)
. . . . .
. . . . .
. . . . .
. . . . .
185 187 193 193 194
A.1 First Page of the t1 -Study Questionnaire . . . . . . . . . . . . . . . . . . . . . 257 A.2 First Page of the t2 -Study Questionnaire . . . . . . . . . . . . . . . . . . . . . 258
List of Tables 3.1
Relevant Empirical Studies on Technology-Intensive Services and IT Adoption
51
5.1
List of Interviewees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
99
5.2
Inter-Coder Judgement Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . 106
6.1
Summary of Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
7.1
Operationalization of the Constructs . . . . . . . . . . . . . . . . . . . . . . . 161
7.2
Crosstable: Respondent’s Function / Number of Employees . . . . . . . . . . . 170
7.3
Fit Statistics for the Measurement Model (n=717) . . . . . . . . . . . . . . . . 172
7.4
Statistics of the Measurement Model . . . . . . . . . . . . . . . . . . . . . . . 173
7.5
Correlation of ITSUM Variables (n=717) . . . . . . . . . . . . . . . . . . . . 175
7.6
Results of the ITSUM (n=717) . . . . . . . . . . . . . . . . . . . . . . . . . . 176
7.7
ITSUM (n=717): Mediating Effects . . . . . . . . . . . . . . . . . . . . . . . 177
7.8
ITSUM (n=717): Moderating Effects . . . . . . . . . . . . . . . . . . . . . . . 178
7.9
Model Fit Statistics for Pre-Adopter Group and Continued User Group . . . . . 182
7.10 Assessing Measurement Invariance: Model Fits . . . . . . . . . . . . . . . . . 182 7.11 Pre-Adopter Group (n=364): Direct Effects . . . . . . . . . . . . . . . . . . . 185 7.12 Pre-Adopter Group (n=364): Mediating Effects . . . . . . . . . . . . . . . . . 186 7.13 Pre-Adopter Group (n=364): Moderating Effects . . . . . . . . . . . . . . . . 187 7.14 Continued User Group (n=353): Direct Effects . . . . . . . . . . . . . . . . . 188 7.15 Continued User Group (n=353): Mediating Effects . . . . . . . . . . . . . . . 188 7.16 Continued User Group (n=353): Moderating Effects . . . . . . . . . . . . . . . 189 7.17 Comparison of Path Coefficients Across Groups . . . . . . . . . . . . . . . . . 190 7.18 Comparison of Factor Means Across Groups . . . . . . . . . . . . . . . . . . . 191 7.19 Results of Hosmer-Lemeshow-Test . . . . . . . . . . . . . . . . . . . . . . . . 195 7.20 Classification Accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 7.21 Quantitative Results: Hypotheses Summary . . . . . . . . . . . . . . . . . . . 197 A.1 Structure Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259 A.2 AVE and Squared Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . 260 A.3 Correlation Matrix without CMV . . . . . . . . . . . . . . . . . . . . . . . . . 260
XX
LIST OF TABLES
A.4 ITSUM (n=717): Calculation of Moderating Effects . . . . . . . . . . . . . . . 261 A.5 Pre-Adopter Group (n=364): Calculation of Moderating Effects . . . . . . . . 262 A.6 Continued User Group (n=353): Calculation of Moderating Effects . . . . . . . 262
List of Abbreviations ACN . . . . . . . . . . . . . . . . . . . . . . . . Automatic Collision Notification AIC . . . . . . . . . . . . . . . . . . . . . . . . . Akaike Information Criterion AOL . . . . . . . . . . . . . . . . . . . . . . . . America Online APS . . . . . . . . . . . . . . . . . . . . . . . . Automated Phone System ATM . . . . . . . . . . . . . . . . . . . . . . . . Automatic Teller Machine ATT . . . . . . . . . . . . . . . . . . . . . . . . Attitude AVE . . . . . . . . . . . . . . . . . . . . . . . . Average Variance Extracted B . . . . . . . . . . . . . . . . . . . . . . . . . . . Behavior / Usage Behavior B-A-I . . . . . . . . . . . . . . . . . . . . . . . Belief Attitude Intention B2B . . . . . . . . . . . . . . . . . . . . . . . . Business-to-Business B2C . . . . . . . . . . . . . . . . . . . . . . . . Business-to-Consumer BD . . . . . . . . . . . . . . . . . . . . . . . . . Business Development BIC . . . . . . . . . . . . . . . . . . . . . . . . . Bayesian Information Criterion BVDM . . . . . . . . . . . . . . . . . . . . . . Bundesverband der Druck- und Medienunternehmen (German Printing and Media Industries Federation) CFA . . . . . . . . . . . . . . . . . . . . . . . . Confirmatory Factor Analysis CFI . . . . . . . . . . . . . . . . . . . . . . . . . Comparative Fit Index CI . . . . . . . . . . . . . . . . . . . . . . . . . . CMV . . . . . . . . . . . . . . . . . . . . . . . CONT . . . . . . . . . . . . . . . . . . . . . . COSE . . . . . . . . . . . . . . . . . . . . . . . CR . . . . . . . . . . . . . . . . . . . . . . . . . CRM . . . . . . . . . . . . . . . . . . . . . . .
Causal Inference Common Method Variance Control / Controllability Customer Oriented Skills of an Employee Composite / Construct Reliability Customer Relationship Management
CSR . . . . . . . . . . . . . . . . . . . . . . . . CTC . . . . . . . . . . . . . . . . . . . . . . . . df . . . . . . . . . . . . . . . . . . . . . . . . . . . DTPB . . . . . . . . . . . . . . . . . . . . . . . e-brokerage . . . . . . . . . . . . . . . . . . e-commerce . . . . . . . . . . . . . . . . . e-coupons . . . . . . . . . . . . . . . . . . .
Customer Service Representative Customer-Technology Contact Degrees of Freedom Decomposed Theory of Planned Behavior Electronic Brokerage Electronic Commerce Electronic Coupons
XXII
LIST OF ABBREVIATIONS
e-government . . . . . . . . . . . . . . . . Electronic Government e-health . . . . . . . . . . . . . . . . . . . . . Electronic Health e-learning . . . . . . . . . . . . . . . . . . . Electronic Learning e-payment . . . . . . . . . . . . . . . . . . . Electronic Payment e-tax . . . . . . . . . . . . . . . . . . . . . . . . Electronic Tax e-vendor . . . . . . . . . . . . . . . . . . . . Electronic Vendor e.g. . . . . . . . . . . . . . . . . . . . . . . . . . exempli gratia ECT . . . . . . . . . . . . . . . . . . . . . . . . Expectation Confirmation Theory EFA . . . . . . . . . . . . . . . . . . . . . . . . Exploratory Factor Analysis EOU . . . . . . . . . . . . . . . . . . . . . . . . et al. . . . . . . . . . . . . . . . . . . . . . . . . ETC . . . . . . . . . . . . . . . . . . . . . . . . etc. . . . . . . . . . . . . . . . . . . . . . . . . . ExFeD . . . . . . . . . . . . . . . . . . . . . .
Ease of Use et alii Electronic Toll Collection et cetera Export Ferngelenkter Dienstleistungen (Export of Remote Services)
EXP . . . . . . . . . . . . . . . . . . . . . . . . Expertise exp. . . . . . . . . . . . . . . . . . . . . . . . . . Expected f.e. . . . . . . . . . . . . . . . . . . . . . . . . . For Example FD . . . . . . . . . . . . . . . . . . . . . . . . . . FMS . . . . . . . . . . . . . . . . . . . . . . . . FTU . . . . . . . . . . . . . . . . . . . . . . . . GATF . . . . . . . . . . . . . . . . . . . . . . .
Factor Determinacy Flexible Manufacturing System Facilities Transformation Usage Framework Graphic Arts Technical Foundation
GM . . . . . . . . . . . . . . . . . . . . . . . . . General Manager I. . . . . . . . . . . . . . . . . . . . . . . . . . . . Interview Number i.e. . . . . . . . . . . . . . . . . . . . . . . . . . . id est ICT . . . . . . . . . . . . . . . . . . . . . . . . . IT Continuance Model IDT . . . . . . . . . . . . . . . . . . . . . . . . . Innovative Diffusion Theory INT . . . . . . . . . . . . . . . . . . . . . . . . . Intention IRCAD . . . . . . . . . . . . . . . . . . . . . Institute for Research of the Cancer and Digestive System IS . . . . . . . . . . . . . . . . . . . . . . . . . . Information System IT . . . . . . . . . . . . . . . . . . . . . . . . . . Information Technology ITSUM . . . . . . . . . . . . . . . . . . . . . Interactive Technology-Mediated Service Usage Model Jr. . . . . . . . . . . . . . . . . . . . . . . . . . . Junior KBA . . . . . . . . . . . . . . . . . . . . . . . . Koenig & Bauer AG LMS . . . . . . . . . . . . . . . . . . . . . . . . loc . . . . . . . . . . . . . . . . . . . . . . . . . . M.I. . . . . . . . . . . . . . . . . . . . . . . . . . MAN . . . . . . . . . . . . . . . . . . . . . . .
Latent Moderated Structural Equations Locus of Causality Modification Indices Maschinenfabrik Augsburg Nürnberg AG
LIST OF ABBREVIATIONS
XXIII
MIS . . . . . . . . . . . . . . . . . . . . . . . . Management Information System ML . . . . . . . . . . . . . . . . . . . . . . . . . Maximum Likelihood MLM . . . . . . . . . . . . . . . . . . . . . . . Maximum Likelihood Parameter Estimate M MLR . . . . . . . . . . . . . . . . . . . . . . . . MOTIV . . . . . . . . . . . . . . . . . . . . . n.s. . . . . . . . . . . . . . . . . . . . . . . . . . OLS . . . . . . . . . . . . . . . . . . . . . . . . P. . . . . . . . . . . . . . . . . . . . . . . . . . . .
Maximum Likelihood Parameter Estimate R Motivation Not Supported Ordinary Least Squares Participant Number
p. . . . . . . . . . . . . . . . . . . . . . . . . . . . Page p.s. . . . . . . . . . . . . . . . . . . . . . . . . . PA . . . . . . . . . . . . . . . . . . . . . . . . . . pb . . . . . . . . . . . . . . . . . . . . . . . . . . PC . . . . . . . . . . . . . . . . . . . . . . . . . . PDA . . . . . . . . . . . . . . . . . . . . . . . .
Partially Supported Proportional Agreement Prior Behavior Personal Computer Personal Digital Assistant
PL . . . . . . . . . . . . . . . . . . . . . . . . . . Perreault and Leigh Measure PRL . . . . . . . . . . . . . . . . . . . . . . . . Proportional Reduction in Loss PU . . . . . . . . . . . . . . . . . . . . . . . . . . Perceived Usefulness RA . . . . . . . . . . . . . . . . . . . . . . . . . Role Ability RC . . . . . . . . . . . . . . . . . . . . . . . . . Role Clarity RMSEA . . . . . . . . . . . . . . . . . . . . . Root Mean Squared Error of Approximation RRDM . . . . . . . . . . . . . . . . . . . . . . RS . . . . . . . . . . . . . . . . . . . . . . . . . . RSPC . . . . . . . . . . . . . . . . . . . . . . . RST . . . . . . . . . . . . . . . . . . . . . . . .
Remote Repair, Diagnosis and Maintenance Remote Service Remote Service Provider Company Remote Service Technician
s.d. . . . . . . . . . . . . . . . . . . . . . . . . . Standard Deviation s.e. . . . . . . . . . . . . . . . . . . . . . . . . . Standard Error SAT . . . . . . . . . . . . . . . . . . . . . . . . Satisfaction SCT . . . . . . . . . . . . . . . . . . . . . . . . Social Cognitive Theory SEM . . . . . . . . . . . . . . . . . . . . . . . . Structural Equation Modeling SME . . . . . . . . . . . . . . . . . . . . . . . . Small and Medium-Sized Enterprises SN . . . . . . . . . . . . . . . . . . . . . . . . . . Subjective Norms SRMR . . . . . . . . . . . . . . . . . . . . . . Standardized Root Mean Squared Residual SSTs . . . . . . . . . . . . . . . . . . . . . . . . Self-Service-Technologies T . . . . . . . . . . . . . . . . . . . . . . . . . . . Trust t-commerce . . . . . . . . . . . . . . . . . . Tele Commerce TAM . . . . . . . . . . . . . . . . . . . . . . . . TLI . . . . . . . . . . . . . . . . . . . . . . . . . TPB . . . . . . . . . . . . . . . . . . . . . . . . TR . . . . . . . . . . . . . . . . . . . . . . . . . .
Technology Acceptance Model Tucker-Lewis-Index Theory of Planned Behavior Technology Readiness
XXIV
LIST OF ABBREVIATIONS
TRA . . . . . . . . . . . . . . . . . . . . . . . . Theory of Reasoned Action TRAM . . . . . . . . . . . . . . . . . . . . . . Technology Readiness into Technology Acceptance Model TRI . . . . . . . . . . . . . . . . . . . . . . . . . Technology Readiness Index TT . . . . . . . . . . . . . . . . . . . . . . . . . . Trust in Technology TW . . . . . . . . . . . . . . . . . . . . . . . . . Trustworthiness URL . . . . . . . . . . . . . . . . . . . . . . . . Uniform Resource Locator USA . . . . . . . . . . . . . . . . . . . . . . . . United States of America UTAUT . . . . . . . . . . . . . . . . . . . . . Unified Theory of Acceptance and Use of Technology VDMA . . . . . . . . . . . . . . . . . . . . . . Verband Deutscher Maschinen- und Anlagenbauer (German Engineering Association) VIF . . . . . . . . . . . . . . . . . . . . . . . . . Variance Inflation Factor
Chapter 1 Introduction 1.1
Motivation and Goals of the Thesis "Service encounters are critical moments of truth in which customers often develop indelible impressions of a firm. ... Yet, across industries, technology is dramatically altering interpersonal encounter relationships." (Bitner, Brown, and Meuter 2000, pp. 139)
The increasing employment of information and communication technologies in companies and households has not only led to considerable changes in the way services are conceived, developed, and delivered, it has altered the nature of services themselves (Bitner, Brown, and Meuter 2000; Meuter et al. 2000). The convergence of technologies such as e-commerce, ubiquitous computing, and mobile communication is emerging as a promising new paradigm with the goal to provide services anytime, everywhere, and transparently to the user via devices embedded in the physical environment. New "intelligent products" contain IT in the form of microchips, software, and sensors and are able to collect, process, and produce information (Rijsdijk, Hultink, and Diamantopoulos 2007). Network technology embedded into such devices allows for connecting these objects to producers and customers enabling automatic identification, localization and remote sensor technologies (Jonsson, Westergren, and Holmström 2008; Lyytinen and Yoo 2002). This poses not only technical, social, and organizational challenges for product producers. It also has a strong impact on possibilities for service provision as the continuous data flow from embedded IT-applications enables seamless services delivered in real time, and directed at connected objects (Fano and Gershman 2002). In this environment, remote services are quickly emerging as a new class of fascinating interactive services. Remote services are predominantly applied as remote system administration, remote diagnosis, and remote repair of machines in organizational environments and high technology industries such as IT, automotive, and engineering (Biehl, Prater, and McIntyre 2004).
2
1. Introduction
An illustrative example in the field of mechanical engineering is the remote repair of a high volume printing machine. A service provider engineer located in Germany remotely accesses a printing machine in China to diagnose and solve a complex machine failure. He then interacts with a customer employee located at the machine in China to repair it remotely, jointly, and interactively while being thousands of kilometers apart. During the whole process, the service provider’s and customer’s employees are interacting and collaborating in a completely technology-mediated contact situation. The application fields of remote services are predicted to expand in scope and scale and become the fastest growing technology-driven IT-services within the next few years (Stiel 2004). The increasing application of remote services in business-to-business (B2B) settings foreshadows the tremendous impact this new technology will have on consumers as well. For example, remote surgeries have already been successfully conducted (Sila 2001). Just recently, Intel and General Electric announced a joint venture on telemedicine to market and develop applications to track the daily activities of patients in need of remote monitoring (The Wall Street Journal 2009). Remote control, repair, and diagnosis of car electronics are offered in the high-end luxury car segment (Chatterjee et al. 2001). Remote control of household amenities such as heating and water (Baker 2008) will change the way provider companies access our lives and raise new challenges in establishing security, trust, and protection of consumers’ privacy (Jonsson 2006). The implementation of remote services is expected to result in substantial efficiency gains on both the provider’s and the customer’s side, due to cost reductions, increased flexibility and time savings. For example, remote services in mechanical engineering help to substantially reduce travel and personnel cost up to 20–30% and the time of troubleshooting up to 10% (Borgmeier 2002). To maximize the benefits for the organization, it is crucial for remote service providers to increase the usage rates by attracting new customers and retaining users. Even though the opportunities are attractive for service providers and customers alike, remote services are associated with substantial challenges and barriers (Biehl, Prater, and McIntyre 2004; Wünderlich and Pfeffer 2007), for which neither practitioners nor academics have found an ideal solution. Even in lead industries like mechanical engineering, the acceptance rate is fairly low. Only about 28% of all companies used remote services in 1997; since then the acceptance rate has only marginally improved (Borgmeier 2002; Stolz 2006). In view of the increasing importance of remote services across industries, it is remarkable that there is no available research that goes beyond descriptive case studies of individual remote service applications. It is also significant that there is a lack of systematic research on the perception and acceptance of remote services. Further, studies on related services only capture a fraction of the relevant characteristics for understanding remote services. For example, studies on the effect of customer provider interaction and co-production on service perception are currently limited to services delivered via face-to-face encounters (e.g., Bendapudi and Leone
1.1 Motivation and Goals of the Thesis
3
2003; Bettencourt et al. 2002). Research on less complex and less demanding services like e-services and self-services mostly focuses on technology features as antecedents of (service) technology acceptance (e.g., Lin and Hsieh 2006; Featherman and Pavlou 2003). These partial approaches might be sufficient for their respective domains, but they fall short for interactive remote services where both the co-production with a service provider employee and the interaction through a mediating technology are essential for the service experience. Hence, to comprehensively analyze the remote service concept, this thesis links separate streams of literature from the fields of service marketing, inter-organizational relationship management, and information systems (IS) research.
The dissertation will contribute to theory by providing a holistic classification of remote services and interactive remote services. Moreover, this thesis explores the customer’s perception of these services and identifies drivers of organizational usage. A model to explain adoption and continuance, the ITSUM, is developed. Ultimately, this thesis aims to derive managerial implications for remote service providers on how to increase remote service acceptance among their customers.
Customers’ acceptance of remote services in a B2B setting does not only manifest in a first-time remote service usage decision instead it is embedded in business processes of repeated practice. Researchers criticize the extreme emphasis of acceptance (initial use) over continued usage in technology and technology-intensive service acceptance studies (Baron, Patterson, and Harris 2006; Bhattacherjee 2001). Baron, Patterson, and Harris (2006, p.111) call it "the inadequacy of a concentration on simple acceptance ... where technology is embedded in a ... community of practice." Support for this view comes from findings in relationship marketing, which also stress the need to retain existing customers (Grönroos 1990; 1996). Researchers claim that the importance of continuance is evident from the fact that acquiring new customers may cost as much as five times more than retaining existing ones given the costs of advertising, searching for new customers, setting up new accounts, and initiating new customers (Parthasarathy and Bhattacherjee 1998). Therefore, this thesis strives to provide a comprehensive approach of explaining both initial acceptance (adoption) and repeated, continued usage (continuance) of remote services in organizations.
From a methodological perspective this thesis aims at explaining organizational intention by linking it to perceptions of individual employees as a proxy. In addition, this doctoral research seeks evidence for the predictive power of organizational intention on actual usage behavior of organizations. Therefore, an empirical setting is chosen where intra-firm group decision making processes are minimal and individual personal attitudes and intentions can be related to organizational behavior.
4
1. Introduction
1.2
Research Questions
This doctoral research comprises different empirical studies that use both an exploratory and a confirmatory approach. The exploratory qualitative study is guided by the following fundamental research questions:
1
1. What benefits and obstacles of remote service stand out from a customer’s point of view? 2. How do customers perceive a remote service situation? 3. What factors determine the general acceptance of remote services? Within this thesis, I identify major belief groups that influence the intention of an organization to use interactive remote services. Building on these findings, my quantitative studies further explore the following research questions: 1. Do the identified beliefs affect an organization’s intention to use interactive remote services and, if they do, to which extent? 2. Do the identified beliefs affect an organization’s intention to continue to use interactive remote services and, if they do, to which extent? 3. How do the determinants of behavioral intention differ between organizations with few and organizations with more experience with interactive remote services? 4. Does the intention to use interactive remote services predict actual usage behavior of organizations and, if it does, to which extent?
1.3
Structure of the Thesis
This thesis employs a multi-methodological approach: it links conceptual, qualitative and quantitative research studies and aims at getting profound and accurate insights through triangulation. Therefore, the structure of this thesis follows the analytic procedure of the studies as shown in figure 1.1. Eight chapters comprise the dissertation. C HAPTER 2 is dedicated to the conceptualization and classification of remote services. Extant literature on new emerging technology-mediated services across industries is reviewed, remote services are characterized, and a demarcation between remote services and interactive remote services is given. Interactive remote services are discussed in comparison to self-services and face-to-face services. The reason why interactive remote services form a unique and distinct service type from both a theoretical and practical standpoint is outlined. 1
The research questions will be refined in the context of the individual empirical studies presented in chapter 5 and 7.
1.3 Structure of the Thesis
5
1
2
!!"-" "$. " $!
3
$"'") $! ") "!
")2, " " "# ) (""
4
" #"# ""#!)"#'"&"
5
% " $ "!
6
'"!! * )"+"# !) """!+!,
7
"#'!-"1"2,"#'.) "(") " "+"# !)"""!+!,
8
# ' !#"!) "!) ""!) #"# ! "!
Figure 1.1: Structure of the Thesis: Contents of the Chapters 1–8
Because literature on remote services is sparse, I approach the theoretical substantiation of the study by combining different research streams. In CHAPTER 3, I present the theoretical foundations of this thesis drawing from the fields of information systems, especially studies on ITand innovation adoption, concepts of personal customer-provider interaction and co-production from the service marketing field, and management literature on relationship management in strategic alliances. In addition, concepts of B2B-marketing and organizational innovation adoption are discussed. In CHAPTER 4 the methodological research structure and the triangulation through qualitative and quantitative studies are outlined. These are the building blocks of the analytical framework and guide the proceedings of this thesis. The rationale for choosing the empirical setting, which is the printing industry, is explained and background information on its structure in Germany, USA and China is given. The international qualitative exploratory study is described in CHAPTER 5 explaining the motivation, study design, and method in detail. The results of the qualitative study identify the relevant categories of beliefs that affect remote service attitudes. These are synthesized to develop a new framework for remote service perception.
6
1. Introduction
Based on the findings of the qualitative study and my conceptual work, I develop the Interactive Technology-Mediated Service Usage Model (ITSUM) for measuring the interactive remote service acceptance in CHAPTER 6. Hypotheses are derived for the ITSUM, the relationship between organizational usage intention and actual behavior, as well as for the comparison of organizations in the pre-adoption phase and in the continuance phase. The quantitative studies are described in CHAPTER 7. The first study measures remote service acceptance in the German printing industry and tests the formal hypotheses of the ITSUM derived in the previous chapter. Organizations in the pre-adoption phase are compared with organizations in the continued usage phase. A second quantitative study with the same subjects at a later point in time is conducted to validate the link between organizational usage intention and actual usage behavior. The chapter finishes with a synthesis of the quantitative results. C HAPTER 8 combines all individual results from the studies. It connects the insights gained and presents a comprehensive summary. Managerial implications for the provision of remote services are derived. This thesis concludes with an outlook towards future research with respect to the limitations of this work.
Chapter 2 Conceptual Framework: Remote Services in Context of Technology-Mediated Services New types of technology-mediated services such as e-services, mobile services, self-services, and recently remote services have become reality. In this chapter, these new service types will be defined, and their application in practice and recognition in scientific research will be discussed. This discussion will set the frame of reference for the main objects of this thesis: remote services and interactive remote services. A classification and demarcation will be given, and the unique service experience from a customer’s point of view will be examined.
2.1 2.1.1
Emerging Technology-Mediated Service Types E-Services
In e-services, the production, consumption, or provision of services takes place through electronic networks or the Internet. These characteristics form the core of most e-service definitions given in literature that see e-services as web-based services (Reynolds 2000), as interactive services that are delivered on the Internet (Boyer, Hallowell, and Roth 2002), or generally as information services because the primary value exchanged between the two parties is information (Rust and Lemon 2001). Despite the various attempts at defining e-services, no universal agreement has been reached. This thesis follows Colby and Parasuraman’s (2003, p.28) definition of e-services, because it reflects an understanding common to a majority of researchers: E - SERVICES are "all services delivered via an electronic medium (usually the Internet) and comprising transactions initiated and largely controlled by the customer."
8
2. Conceptual Framework: Remote Services in Context of Technology-Mediated Services
The application of e-services is multifaceted. Familiar e-services include but are not limited to: online banking (Bradley and Stewart 2003; Vatanasombut et al. 2008); online auctions such as www.ebay.com (Chan, Kadiyali, and Park 2007; Reynolds, Gilkeson, and Niedrich 2009; Clark and Ward 2008); online retailing (Griffith 2005; Haynes and Taylor 2006); e-learning such as classes being videostreamed online (Santhanam, Sasidharan, and Webster 2008; Rossiter 2007); e-health such as online medical advice (Burchert 2003; Lewis Jr. 2008); e-government such as e-taxes (Hsu and Chiu 2004; Hung, Chang, and Yu 2006; Fu, Farn, and Chao 2006); e-libraries providing electronic access to journal articles or book chapters (Padgett 2004; Kajikawa, Abe, and Noda 2006); and information and location-based services (Yee and Korba 2005). The usage of e-services grows continuously. For example, from 2005 to 2008 the usage rate of internet banking in the USA increased from 27% up to 39%, and nearly all American Internet users (93%) have at one time or another conducted e-commerce (Horrigan 2008). E-services are not limited to the domain of new economy companies. Established organizations are also augmenting their traditional offerings with e-services and approach their customers via multi-channel-strategies (Cassab 2009; Müller-Lankenau, Wehmeyer, and Klein 2005). For example, airlines offer ticket ordering via their websites, retrieval of flight information via call centers, automated phone systems (APS), or check-in via self-check-in kiosks (Lufthhansa AG 2007). E-service applications are often tied to e-commerce business models that sell goods, for example the purchasing of physical goods that are then delivered by traditional means. A prominent example is Amazon.com, where a book is purchased online, but delivered by mail to the buyers. Voss (2000) proposes that e-commerce and e-service are two ends of a continuum ranging from pure sales on the web, with little or no service content, to pure service, delivered free of service contracts, or as a part of a service contract (see figure 2.1). In between, a variety of business models are found. These include: services that sell information (e.g., newspapers selling their content online); value-added services (e.g., online travel agency offering travel insurances); or bundles of services and goods (e.g., online selling of personal computers combined with support services). Key themes in the e-services literature are e-service quality and its associated dimensions and measures (Parasuraman, Zeithaml, and Berry 1985). Other frequently addressed topics include: the elements of the web experience (Lin, Wu, and Tsai 2005; Novak, Hoffman, and Yung 2000); customer satisfaction (Ha and Janda 2008; Zhang, Prybutok, and Huang 2006); customer’s buying intention and loyalty (Chellappa and Kumar 2003; Herington and Weaven 2007); and service operations (Roth and Menor 2003). Service quality has been recognized as the key to reach additional strategic and operational objectives such as improved customer satisfaction, increased retention rates, enhanced operational efficiency, and profitability (Al-Hawari and Ward 2006; Cronin Jr. 2003; Rust, Zahorik, and Keiningham 1995). In comparison to traditional shopping channels, researchers have found that customers perceive information availability and content quality as more important when shopping online (e.g., Zei-
2.1 Emerging Technology-Mediated Service Types
9
!
Figure 2.1: The Continuum from eService to eCommerce Source: Own Illustration, based on Voss (2000, p. 21) thaml, Parasuraman, and Malhotra 2002). The shaping of these characteristics is an effective way to reduce the perception of uncertainty and risk (Featherman and Pavlou 2003; Rowley 2006). Many approaches stemming from traditional service quality measurement tools to measure e-service quality have been applied in various forms. Some of these approaches take the SERVQUAL measurement scale as a basis and extend it by web site and usability-specific quality dimensions (e.g., Parasuraman, Zeithaml, and Malhotra 2005; Zeithaml, Parasuraman, and Malhotra 2002). Researchers developed a number of web-specific scales such as WebQual (Loiacono, Watson, and Goodhue 2002), E-Qual (Kaynama and Black 2000), SITEQUAL (Yoo and Donthu 2001), e-SQ (Zeithaml, Parasuraman, and Malhotra 2002), ETailQ (Wolfinbarger and Gilly 2003), and E-S-QUAL (Parasuraman, Zeithaml, and Malhotra 2005) to capture eservice quality. Research on exploring e-service adoption has identified numerous antecedents. Service design and usability (Surjadjaja, Ghosh, and Antony 2003; Massey, Khatri, and Montoya-Weiss 2007), service delivery features (Dabholkar, Bobbitt, and Lee 2003; Iqbal, Verma, and Baran 2003) as well as relational factors such as trust (Dutton and Shepherd 2006), social presence (Gefen and Straub 2004), and internal communication (Surjadjaja, Ghosh, and Antony 2003; Walker and Johnson 2006) are relevant in explaining the adoption of e-services.
2.1.2
Self-Services
Advances in technology and high labor costs led to a surge in the provision of self-service and self-service technologies (Dabholkar, Bobbitt, and Lee 2003). These services feature a stronger active participation by customers in the service process (Bendapudi and Leone 2003; Zhu et al. 2007). According to Meuter et al. (2005, p.61), S ELF - SERVICE TECHNOLOGIES (SST) are technological interfaces that enable "customers to produce services for themselves without assistance from firm employees."
10
2. Conceptual Framework: Remote Services in Context of Technology-Mediated Services
There is no agreement in literature on the conceptual relationship between e-services and selfservices. Surjadjaja, Ghosh, and Antony (2003) argue that to use a self-service, a customer has to go to the technology (such as an ATM) to receive the service whereas in e-service, a customer can receive the service through the Internet at home or in other places of his choosing. They consider a self-service to be less flexible than e-service due to constraints of location. Other authors do not follow this distinction and do not explicitly distinguish between self-services and e-services, e.g., they consider online banking to be a self-service (Curran and Meuter 2007; Meuter et al. 2005). In this sense Rowley (2006), takes the position that all e-services are essentially self-services, whether they are delivered through a web page on a PC, a mobile device, or a kiosk. Dabholkar (1994) follows this view, but includes in her typology of self-services a different distinction based on the location of the technology where customers can access the self-services technology. She distinguishes between "provider-based self-services" and "customer-based selfservices." In provider-based self-services, the access technology is provided by the service provider who sets up certain machines such as check-in kiosks or package pick-up centers. Customer-based self-services on the other hand can be accessed using technological devices that are available at the customers’ homes via a PC connected to the internet. In practice, self-services are found in a wide variety of business scenarios: monetary transactions (e.g., using ATM or online banking), shopping (e.g., online booking of a trip), self-help (e.g., distance learning), fully automated phone systems (e.g., telephone banking), and customer services (e.g., hotel checkout). The provision and usage of technology-based self-services is growing at exponential rates all over the world (Shamdasani, Mukherjee, and Malhotra 2008). Beatson, Coote, and Rudd (2006) predict prolific advances in technology and expect that SST facilities will continue to evolve and will play an even more important role in service delivery than they currently do. For example, in 2007, consumers in the United States conducted 14.9 billion ATM transactions at 415,000 ATM terminals compared to 10.5 billion ATM transactions at 396,000 ATM terminals in 2005 (Mohsberg 2008). The same trend can be seen in the airline industry when in 2007 when 20% of all passengers of the German airline Lufthansa used online check-in options or self check-in kiosks. Lufthansa expects the quota of passengers using self check-in solutions to rise up to 65% in 2010 and even up to 90% in 2015 (Lufthansa AG 2007). Extensive research has been conducted on self-services in recent years. Meuter et al. (2000) identify four different forms of interfaces: telephone/interactive voice response; online/internet; interactive kiosks; and video/cd. They also distinguish three main purposes of SST that are perceived as beneficial by customers (see figure 2.2): service surrogates (such as order status viewing); transactions (such as hotel checkout); and self-help (such as distance learning).
2.1 Emerging Technology-Mediated Service Types
%)!('
79
%" '#"&$
76
7;
%(&'*#%'"&&2"% 3
79
87
!
:
;
&#&"#")""
;
;
&$#"
9
:
"# #,"',
7
7
%%',&&2 %#%%(&''3
76
96
%(&'""# #,
#%' "&!
9
9
$$%
%&'
Figure 5.3: Coding Categories of the Qualitative Interview Study – Part 2
5.4 Results of the Qualitative Interview Study
5.3.3
105
Category Development and Coding
Out of a total of 19 interview sessions, some of which involved more than one employee, 17 were audio taped and transcribed verbatim in English or German. In the case of the Mandarin, Wu, and Cantonese interviews, the English translation by the on-site translator was transcribed. Three participants declined the request to record interviews due to the sensitive nature of the subject, but detailed notes were taken during the interviews. The transcribed versions of the interviews, notes, and tapes including both the original language audio and the comments by the on-site translators constitute the material for the subsequent interpretation of meaning via qualitative content analysis (Mayring 2003). The source material was coded using an inductive approach of category development according to the qualitative content analysis (Mayring 2003). The categories were formed in an iterative process with constant revision of the coding and the tentative categories. The final coding scheme consists of nine main categories and 49 subcategories. In sum, 357 statements – text units that can include one or more sentences – from the source material were related to the subcategories. I used a qualitative data analysis software package, NViVo7, to support the text interpretation (Mayring 2000). This can, in turn, improve the rigor of the analysis by validating some of the researcher’s impressions (Welsh 2002). The hierarchical structure of the coding categories together with the number of statements attached to those categories as well as a description of those categories are shown in figures 5.2 and 5.3.
5.4 5.4.1
Results of the Qualitative Interview Study Assessment of Intercoder Reliability
Researchers generally agree that an estimate of intercoder reliability must correct for chance agreement among coders, register systematic coding errors, and determine the stability and quality of the data obtained (Hughes and Garrett 1990; Rust and Cooil 1994). Assigning statements during the coding process to mutually exclusive categories is sometimes not clear cut and to a certain degree "fuzzy" involving subjective judgement. In contrast, a categorization is reliable if coders can be shown to agree on the categories assigned to statements to a certain extent (Artstein and Poesio 2008). This study employs four intercoder reliability assessments metrics based on multiple judges assigning codes to the statements derived within the coding process such as proportional agreement (PA), Cohens κ (κ), Perreault and Leigh measure (PL), and the proportional reduction in loss (PRL) (Hughes and Garrett 1990; Rust and Cooil 1994). To confirm the inter-coder reliability of the content analysis, two judges (J=2) were used. The 357 statements (n=357) and 9 categories (K=9) were given to both judges, individuals with expertise in remote service themes, to independently assign each statement to one category.
106
5. Qualitative Exploratory Interview Study
The resulting Matrix Ω is shown in table 5.2. Every element on the main diagonal holds the frequency of statements, which both judges assigned to the same category. The sum of all diagonal elements, therefore, is the total number of agreements. The entries not on the main diagonal represent disagreements between the two judges. For example, the entry for judge A/category 6 and judge B/category 3 holds the number of statements assigned by judge A to category 6 and by judge B to category 3, in this case 1 statement. Using the matrix Ω and the total number of pairwise judgements11 , four intercoder reliability measures can be derived. The resulting proportional agreement (PA) of 0.882 is well beyond the recommended cutoff point of 80% proposed by Neuendorf (2002). Cohens κ, the most pessimistic measure, equals 0.863 and indicates a good reliability. This metric is generally considered adequate if it is above 0.4 (Artstein and Poesio 2008; Landis and Koch 1977). The theoretically most accurate measure are PL and PRL. In a two-judge case PL and PRL are equivalent12 , their value in this study is 0.931. Following Rust and Cooil (1994), this value can be considered good because it is higher than the suggested minimum of 0.8. In conclusion, intercoder reliability can be assumed.
5.4.2
Structure of Results Presentation
The major themes that emerged from the interviews with customers and suppliers form a conceptual framework for understanding the factors that influence the customer’s attitude toward remote services. The framework is shown in figure 5.4. It depicts the grouping and relationship between the relevant variables influencing remote service perception: relational beliefs; economic values; technology beliefs; former experiences; beliefs about customers’ participation Table 5.2: Inter-Coder Judgement Matrix Ω C ODER A: C ATEGORIES
1 2 3 4 5 6 7 8 9 ∑
11
12
C ODER B: C ATEGORIES 2 3 4 5 6 7 8
9
∑
21 0 0 0 0 0 0 0 3 0 54 0 3 0 0 0 0 0 0 0 4 0 0 0 0 0 0 0 0 0 30 0 0 0 0 1 0 0 0 0 48 3 0 0 0 3 0 1 1 0 39 0 4 3 0 0 0 0 0 0 12 2 0 0 0 0 0 2 2 0 60 4 4 2 0 0 1 1 1 1 47 28 56 5 34 51 45 13 67 58
24 57 4 31 51 51 14 68 57 357
1
In a two judge scenario the total number of pairwise judgements T: T = N, generally the total number of pairwise judgements is: T = J(J−1) 2 N For a prove see Rust and Cooil (1994).
5.4 Results of the Qualitative Interview Study
107
in a remote service; organizational factors; process control beliefs; and contextual factors such as the geographical distance. These themes represent the major results discovered through the process of inductive category development. The following sections explain in the individual results and the derivation of the conceptual framework. This chapter is structured by following each of the major themes that form the framework.
5.4.3
Technology Beliefs
Due to the high technology intensity during a remote service encounter, perceived attributes of the technology play a major role in the customers’ perception, e.g., the attitudes the customers have about the necessary network equipment. The findings from the qualitative interview study show that a high risk perception is a major issue in offering remote service to customers. One theme that arises in most of the interviews is the customers’ fear of third party attacks, e.g., that the remote service connection could be used by hackers to access the network and/or the machine. The following statements exemplify the range of potential dangers customers see: "[The threat] is reasonable. We must avoid that the hackers can access our network. If we would get some viruses due to attack from hackers, it would influence our operation on the machine. Therefore, it must be avoided. ... Could [the remote service provider] 100% guarantee that no virus can access my machine?" (P.28, I.18, customer, China) 13
Figure 5.4: Conceptual Framework Resulting From Qualitative Study 13
In the following the references to interviews are written in short-form: P. = participant no.; I. = interview no.; customer = remote service (potential) customer company; provider = remote service provider company; China/Germany/USA = country of interview conduction.
108
5. Qualitative Exploratory Interview Study
The importance of these perceived technology risks should not be underestimated. A remote service encounter is perceived as risky and one customer even explicitly stated that these risks are a major barrier to buy remote service offerings: "If so, we would like to pay the 2000 Yuan. Because if there is no virus, which can access this machine, we would like to connect to the remote service permanently, otherwise we are not interested." (P.28, I.18, customer, China) Another customer of remote services even explains, that he rejects using remote services because of his security concerns: "Secure or not? Because we are no experts in this area, we have worries about it. For example, we have a [company name] machine and they have remote services, too, but we never used it because we are worried about the security." (P.30, I.19, customer, China) Some customers did not have any serious security concerns and did not see any real issues with network security when using remote services. A certain technological affinity or trust in technology helps to reduce the fears. Also, customers’ own security measures like firewalls impact the customer in having a more positive attitude towards remote services: "From the aspect of technology, I like to use new technologies. I can also give some suggestions to our company or technical department." (P.27, I.17, customer, China) "[The remote service provider company] has done something to protect the data transmission, like data encryption. But internet is a public network, if we do it via a public network, it may be a little bit unsafe, however, normally it is acceptable." (P.29, I.19, customer, China) "I just have superficial knowledge regarding information technology, but I am quite sure [the remote service provider] can protect against it [hacker attacks]. I don’t think it is a real threat that somebody can access private data." (P.4, I.3, customer, Germany) "It is normal and natural that others worry about the security. But for us, we have a lot of servers in our company; we do not think this is a problem for us." (P.27, I.17, customer, China) Apart from unauthorized third party access such as hacker attacks, remote service customers are also concerned about the remote service provider company’s unauthorized access to their system. Some customers seem to not know exactly whether the remote service technician can access private information beyond their agreement and to what extent the provider can access confidential information. Other customers fear espionage and imagine that the remote service technician (RST) accesses sensitive information out of curiosity or in order to check up on them. Additionally, a number of customers handle sensitive information themselves, e.g., print jobs for
5.4 Results of the Qualitative Interview Study
109
their respective customers. Beyond the concerns about their own business data, customers fear that their customers’ information might get stolen, especially in pre-press and digital printing businesses: "The service provider promises to just take agreed-on actions, but maybe he does something else instead. [A remote service] is really risky and dangerous." (P.4, I.3, customer, Germany) "If I connect to the remote service, what I am doing on the machine, the content of my printjob could perhaps be watched by them [RSTs], to my mind it could be a reason [for not using remote services]." (P.30, I.19, customer, China) "Because there is a lot of information that you can put on that machine that you do not want people to know about. ... The thing is, as if I had something on there I didn’t want them to see, I will delete it. ... And then I can put it on my machine later. For me, it would just take the temptation away and there is no problem." (P.9, I.7, customer, USA) "Through the connecting, the engineers of [the remote service provider] can look at our processes. I think, they do this because they want to know something about other companies." (P.24, I.14, customer, China) It is fair to say, that especially in interviews with unexperienced users (P.4, P.5, P.28, P.29, P.30), potential risks were a resurfacing theme, which usually were perceived as a major draw-back for remote service usage. Experienced users such as P.24 and P.9, however, were aware of the risks, but assessed them to be no hinderance to use the service. In contrast to a self-service system where the system response is rather predictable and a worstcase scenario is that the service is simply unavailable, in a remote service encounter, the RST could actually damage the service object (printing machine) resulting in substantial downtime and costs. Customers expressed the potential danger as follows: "I am a little bit afraid of somebody remotely shutting down my machine." (P.1, I.1, customer, Germany) "I don’t worry about our work being spied on, I am afraid of that the machine is destroyed." (P.28, I.18, customer, China) "Such a remote service is not without danger. Just recently our printing configuration was erased in a flawed repair attempt [by the counterpart]. That cost us a huge amount of time." (P.7, I.5, customer, Germany) "Often something gets broken right one day after an on-site visit of an engineer. So, they [the technician] can really misconfigure something with harmful consequences. I guess the same could happen during a remote service." (P.5, I.4, customer, Germany)
110
5. Qualitative Exploratory Interview Study
In addition to risk, technology characteristics such as the ease of use, usability and convenience of using the technology necessary in the remote service provision seem to be relevant for the customers’ general evaluation of remote services. An owner of a German printing company emphasized the effect of inconvenience in using remote services: "I think remote services are pretty useful, but just recently I wanted to start the remote service procedure and I had problems with the technology and it was very difficult for me to handle this." (P.7, I.5, customer, Germany) Most of the customers, however, seemed very well aware of how to request/initiate a remote service and how to cooperate during the remote service process. The process of connecting the machine was perceived as very easy and the customers seemed to be well trained. "As they trained us last time, it was easy. It took ten and a few minutes to explain how to connect to the server, that’s all!" (P.28, I.18, customer, China) "It is very easy to operate." (P.22, I.13, customer, China) "Of course, it is easy, because what we do is just pressing buttons or connecting cables." (P.23, I.14, customer, China)
5.4.4
Relational Beliefs
The exploratory interviews highlight the importance for relational beliefs affecting the remote perception. Trust and trustworthiness related themes frequently emerged and were emphasized from both service provider and customer as important for the customer’s attitude toward remote services. It is striking that the object of trust beliefs are both the RST as the social interaction partner during a remote service and the remote service provider company (RSPC) on an organizational level. In the following sections I will distinguish between those two trust objects.
5.4.4.1
Trust in the Remote Service Technician
The statements show that personal relational factors to the RST have strong positive effects on the customer’s perception of the remote service. In addition, the majority of the customers seem to build a relationship with the company through individual relationships with the RSTs. The following statements underline the importance of the customer’s trust in the individual RST: "It really ... matters which engineer or technician is send to us. There are huge differences. It depends strongly on the individual person." (P.1, I.1, customer, Germany) "Of course the persons we know before are more trustworthy. Because we knew him, he has worked with us before, therefore, we trust him more." (P.28, I.18.
5.4 Results of the Qualitative Interview Study
111
customer, China) "Because we have had a good relationship with them [the RSTs] for a long time, we know each other and trust each other." (P.30, I.19, customer, China) "My trust towards [company name] is based on the relationships I have with the people, the technicians." (P.9, I.7, customer, USA) "We trust the technicians more than [company name] in general." (P.7, I.5, customer, Germany) The trustworthiness of the RST seems to be very important for the customers, especially if they are evaluating an interactive remote service with an active RST accessing their machine. The interviews indicate many aspects of trustworthiness comprising a number of slightly different accentuated beliefs such as benevolence or knowledge of the RST. Due to the high risk perception of remote services, the customers’ perception that the RST adhered to a set of agreed-on principles seemed to be important because this aspect was repeatedly emphasized in the interviews. This belief in the integrity of the RST was implicated by the following statements from managers in Germany and in the USA: "Because we have a good relationship I believe that they only take agreed-on actions, e.g., just checking the fixing unit .... and if I tell them that we don’t want them to check anything else then they wouldn’t do that." (P.3, I.2, customer, Germany) "This is my experience with [company name], how the work has well trained, very well mannered technicians, and if I ask them not to do something I’m sure that they will just not do it." (P.9, I.7, customer, USA) In general, the RSTs were attributed with high skills and competencies. Some customers ventured the opinion that the skills of an RST are higher than the skills of on-site technicians, because remote services enable them to get in contact with experts compared to locally available on-site technicians. The competence of the RSTs was seen as one of the major and essential reasons for a successful service outcome: "...It is similar in remote services, meaning that the guy who remotely logs into our machine is most likely the absolute specialist for this model. Not like a typical mechanic, who has to do 20 different machine models and just happens to be the guy located in the branch nearby." (P.1, I.1, customer, Germany) "... web remote service is only as good as the technician is. If you don’t have good people it becomes like calling AOL – America Online – they’ve got these guys that read from a script." (P.9, I.7, customer, USA) In only a few cases was the general trustworthiness of a RST seen negatively. One customer in Germany, who was dissatisfied in general with the remote service, attributed his feeling of disappointment to a lack of benevolent behavior of the RST. This underlines the importance of
112
5. Qualitative Exploratory Interview Study
trust-assuring behavior in gestalt of doing good, showing sensitivity to the needs of the other party, and not taking economic advantage of the other party. The customer stated his opinion of technicians with these words: "Sometimes I get the impression that they [the technicians] don’t have a clue what to do. They do not really help me. At least they don’t do everything in their power to help me and they do not think proactively and foresighted." (P.5, I.4, customer, Germany) Customers favored the personal contact of an remote service technician and liked to work with acquainted personnel. The interviews conveyed the need for social contact and interaction with the RST, e.g., via telephone during a repair of a machine. ".... Well, yes, I miss the personal contact, because I’ve liked everyone of the guys, they’ve sent in here. But, we got a business to run." (P.9, I.7, customer, USA) Customers tended to require a remote service technician with whom they were acquainted with, especially when there was an emergency case. Based on prior encounters and service experiences, customer employees and provider employees were able to build up a stable relationship. The concrete relationship between an RST and the customer is a key risk-levering issue, which nearly every interviewee mentioned. If the customer knew the remote service engineer personally, he felt reassured about the know-how and integrity of this person: "Especially the new recruited engineers might sometimes do something wrong." (P.28, I.18, customer, China) "If we know the roles and skills of the technician, we can require a suitable remote technician for the task. It is good to get a qualified technician, it improves our communication and it can help speed up fixing the problem." (P.30, I.19, customer, China) Nonetheless, the latent wish for personal contact can negatively affect the perception of a remote service when the remote service is perceived as not sufficiently providing social contact. The customers complained about the limited means of communication with the RST during the remote service process. A production manager from a small print shop stated: "I hope we could have more communication with the service technicians." (P.28, I.18, customer, China) Also, to improve the cooperation between customer and engineer, a personal dialogue is helpful as a customer employee from China stated: "There are many things which can only be understood after much communication with [the remote service provider]. Without face-to-face communication or phone calls many things are difficult to understand, because we are not in the same level, maybe sometime we can understand what he told us. Therefore, we
5.4 Results of the Qualitative Interview Study
113
need more [face-to-face or telephone] communication." (P.29, I.19, customer, China) Moreover, some customers chose local services with direct personal contact to a technician over a remote service based on their need for social interaction but demanded immediate remote service when the problem was urgent and severe. It seems likely that the incentives to use a remote service vs. a local service are not enough to convince customers to increase their usage of remote services, except in emergency situations as a machine operator from a Chinese printing company stated: "It depends on if production is urgent. If it is not urgent, then we let them come to us. If it is urgent, then we can use it to solve problems at once and then continue production." (P.23, I.14, customer, China)
5.4.4.2
Trust in the Remote Service Provider Company
The interviews show that both the remote service provider employees and the remote service customer employees consider trust itself as fundamental when a customer adopts remote services. In addition to the RST, who acts as a service counterpart for the customer employee during the service delivery, the trust the customers have in the remote service provider company [RSPC] is important in forming their attitudes towards remote services. A manager of a US service provider company expressed it this way: "Trust is one of the most important factors for our customers. If they do not trust our company we could not sell the remote services to them." (P.12, I.9, provider, USA) A Chinese Manager from a remote service provider company located in Shanghai assesses the role that trust in the RSPC plays for his customers: "90% of the customers trust the brand, they are very trusting." (P.20, I.12, provider, China) Customers referred to the RST as their trustee more frequently than they referred to the RSPC. The brand image and reputation of the provider company also seemed to influence their evaluation of the remote services those companies offer. Customers’ statements confirmed these views. Further, when it comes to security concerns towards remote services it is especially obvious that a trusting relationship between the RSPC and the customers helps to overcome barriers: "So long as [the RSPC] guarantees that, I could certainly throw away my worry." (P.28, I.18, customer, China)
114
5. Qualitative Exploratory Interview Study "We trust [company name] very much because we only use [company name] press machines up to now. ... The most important is we trust them and they can fix our problem." (P.27, I.17, customer, China) "No, there we do not have any concerns, because we have a good relation [to the RSPC] and trust him. He [the RSPC] must also trust us, especially in regard to "full-test" machines, as these are machines the customer is not allowed to touch. They [the RSPC] must trust us and we must trust them - mutually." (P.3, I.2, customer, Germany)
This understanding of mutual trust was mirrored by an RST, who also saw a need to establish trust in the customer employee: "For example if there is a new customer, a new operator, we worry about if we told them to check something, maybe he gets the wrong places, the wrong action, then we get a big problem. Still, we have to trust him." (P.15, I.10, provider, China) The trust the customers have in the RSPC strongly relates to the level of security a reputable company can guarantee and the experience it has with security issues. The brand name and a good reputation of an RSPC was evaluated by some customers as an additional security guarantee for the trustful behavior of the company as the customers assume that providers do not want to risk reputation loss. "With regard to security, we don’t worry about it. [The RSPC] is a big company, they would not set any bad options in my machine." (P.28, I.18, customer, China) "We do not think [hacker attacks] are a problem for us. .... Our system has installed a firewall. I think [company name] should have thought about this problem if they supply such a service." (P.27, I.17, customer, China) "In so far, that if a vendor like [company name x] or [company name y] .... would do something like that, it would make rounds in the business community. I believe the negative consequences would be catastrophic for the vendor." (P.1, I.1, customer, Germany)
5.4.5
Process Control Beliefs
Based on the interviews, I obtained evidence for the importance of process factors for the customers’ perceptions of remote services. Reoccurring statements touched upon themes such as the imagination of the process, the perception of control, transparency, and social presence. The customers mentioned the importance of control regarding the remote service process itself and also with regard to the actions of the RST. The importance of control beliefs also became
5.4 Results of the Qualitative Interview Study
115
clearer as the interviewees remarked on the need for control mechanisms to stop, abort, change, or direct the RST’s action; or that they should require such mechanisms. "Control plays a very important role. I want to decide what exactly is done with my machine." (P.4, I.3, customer, Germany) In an interactive remote service encounter, the importance of the employee who delivers the company’s services, usually the RST, is reinforced. An RST actively accesses the machine and sometimes asks the customer employee for support in the repair. As the RST’s actions are invisible and hard to control due to physical dislocation of customer and provider, customers lack sufficient evidence for service and may have difficulties evaluating the effort, quality, and security of the service process even while collaborating. The reduced observability of remote services triggers a mental simulation of the service production process in the customers’ minds, as one customer, a production manager from Germany, indicated: "Of course I’m thinking about what the technician is doing right now [during the remote service]. I can afterwards check the login-protocols but during the service process I have to essentially trust him." (P.7, I.5, customer, Germany) In the remote service situation where the stakes are high, control beliefs are likely to play an additional and substantial role in the remote service perception. Most customers view remote services as risky and search for tangible clues about the interactive collaboration process with the service provider, e.g., observable login-protocols or a real-time representation of the RST’s actions. "From the point of view of an operator like me, [the displayed information] certainly is useful." (P.23, I.14, customer, China) "One point, we must know what he is doing and make sure that his doings are allowed and does not disturb our production." (P.22, I.13, customer, China) Regarding the high risk perceptions of some customers, it is understandable that customers have a need to monitor the actions of the RST to prevent him from making (in their mind) mistakes and damaging the machine. Based on the customers’ statements it is fair to say that a high level of transparency and the availability of control mechanisms is definitely desirable during the complete remote service process. For example, an abort button displayed on screen and the possibility to terminate the remote service at any time is a feature that most customers said they appreciate: "I have complete control. If they’re doing something and I think that I don’t want them to do that, I can disconnect the service and they loose control over the machine." (P.9, I.7, customer, USA) "Yes, it’s important that there is some kind of "Power Off" button. .... Because first with the help of these two buttons we can check the status of the remote service; second, if some made a misoperation, we can correct in time, without the two
116
5. Qualitative Exploratory Interview Study buttons we cannot. " (P.30, I.19, customer, China)
Also, the communication with the RST is not only seen as possible social interaction but also as a tool to control the RST as a general manager of a German printing company stated: "I would like it better ... if there is always a simultaneous telephone dialog taking place to check on his actions." (P.4., I.3., customer, Germany) The fear of loosing control even seems to hinder customers to permanently connect their machines to the remote service provider to allow constant remote monitoring. Some did not like to imagine being remotely monitored: "There are software companies, which hide some clauses in the contractual fine print, so that they theoretically have free access to all your data at any time; something I find partly illegal." (P.1, I.1, customer, Germany) "[Constant remote monitoring] would probably be going to far and I find it suspect, I would rather expect the machine to first alert the machine operator here and he decides whether to allow the machine "phoning" home or not. This is all going too far... I want to be in control or our operator or somebody like his foreman, the manager of operations or the head of printing. They should decide, if its ok to use this service or if its not ok." (P.4, I.3., customer, Germany) In addition to tools that provide transparency and control, the interviews show that trust in the RST can decrease control beliefs. Therefore, control beliefs will be strongly affected by the degree of confidence the customer has in the remote service provider company or the RST in person. This can be seen in the following statements: "With [company name] I don’t worry, I trust their people and only afterwards I look if anything changed. But with another provider it’s different. I am on alert during the whole process and always think about what he [the remote service technician] might do just now." (P.7, I.15, customer, Germany) "It is necessary that we have the control during all operations because the engineer especially the new recruited engineer might sometimes do something wrong, too." (P.28, I.18, customer, China) Closely tied to trust and the way personal contact is presented in a remote service process is the degree of social presence of an RST. From the interviews, it can be derived that a higher degree of social presence helps to build trust and decrease mistrust and the longing for control of the customer. Verbal communication, video-conferencing or the provision of information about the engineer like a photo seemed to be appreciated and wished for by the customers: "We think telephone call is easier to communicate with them. We feel that they are more closely with us." (P.22, I.13, customer, China) "Sure, it is very useful [video conferencing]. That will be the best if you can talk
5.4 Results of the Qualitative Interview Study
117
face-to-face for this kind of service. But costs must be considered." (P.23, I.14, customer, China) "[Things we would like to know are:] Name, telephone number. It is very important if we know him [the RST]. But if we do not know him, we want to know his specialty, for which kind of problems he is good at. .... In the new version we can look the photos of engineers. We feel it is good." (P.22, I.13, customer, China) "If I know you and the next time you call or I call you, I have a face to put with the voice and it just makes it more .... Even if I knew some of the man I haven’t met or ladies, it makes it a little bit easier for me to make me feel comfortable with them." (P.9, I.7, customer, USA) The customers wished to get more information about the emergency cases and the reasons for the problems. They also wanted to have further documentation covering the status of the machine parts, reports on the health-status of tools, repairs, damage and how the issue was solved. Some customers emphasized their wish to learn how to conduct the repair steps from the RST. In the interviews, a need for process transparency became apparent. Transparency can be established by communication on the process or log-in mechanisms and protocols: "I wish the service has a documentation function. That means, it can record detailed information of each time, for example, when, where, which machine, the reason, how to fix, and so on. We record the information by ourselves, too. The function is just like case history of a patient. It is very important. " (P.27, I.17, customer, China) "We can only see who is connecting to the remote service. Sometimes a photo of a German engineer is shown on the screen. I can only see these two things. And I can also see on which pages are they operating, but I cannot see his conclusion." (P.30, I.19, customer, China) "I hope, the screen can show us both, components that are working well and components that have problem. It can also show us what the problem is or which potential problem some components have. In case of problem it can give us a result description." (P.27, I.17, customer, China)
5.4.6
Economic Values
In a business setting, the benefits of remote services and their relative advantage compared to face-to-face delivered services expectedly influence the customer employees’ perception of remote services. In general, the customers were very well aware of the advantages that come along with using remote services compared to using on-site services. The most prominent advantage in the view of the customer was the increased availability of the machine due to time
118
5. Qualitative Exploratory Interview Study
savings in emergency cases, especially when compared to the time needed for a technician to physically go to the customer: " And it’s much faster and easier for both sides, it’s a win-win-situation ... for both sides of the remote service. They can solve the problem faster and easier and less costly for both parties .... If I had the option to give up remote service and call for somebody, I wouldn’t. With the remote service they are going to answer the phone today. And they going to call me today. And we are going to work on that press." (P.9, I.7, customer, USA) "The biggest advantage [of remote services] is simply: time. The distances to the printing machines vendors are often very large, especially with [company name x] it’s over 600 km. And also you immediately have a competent counterpart that also has access to the machines controls. He can say where the problem is, they even can locate mechanical parts." (P.2, I.1, customer, Germany) "That’s the main aspect of the whole thing [remote service]. I find that we save costs and - foremost - time through this remote service, especially machine downtime can be reduced, as in many cases the problem can immediately be solved." (P.3, I.2, customer, Germany) Additionally, using remote services for scheduled/planned maintenance purposes is also seen as an advantage. Flexibility in scheduling maintenance has a positive effect on maintaining machines and can reduce downtime. "It can shorten the maintenance time. If they can fix the problem through remote service, the maintenance time will be reduced." (P.25, I.15, customer, China) As to be expected, customers regarded price as an important factor for the decision to buy or continue using remote services. Some of the interviewees, who had already used remote services, were still in the guarantee period, so that they hadn’t had to pay for these services yet or were preferred customers who got them for free. Considering the importance of the price criterion, it is striking that some potential customers had only a vague idea regarding the costs of remote services. When the customers who were uncertain about the price of remote services were asked to estimate they imagined the price as being "high" and as a barrier to using the remote service: "It is difficult to say. I think it depends on how much we must pay for it. If the price is high, then perhaps we will stop it." (P.25, I.15, customer, China) "My question is, it is free of charge in the guarantee period, after this period it will be charged, but how much it will be charged?" (P.28, I.18, customer, China) "Another reason [for not using remote services] are maybe the costs." (P.29, I.19, customer, China) "A lot of customers are worried about the costs in future." (P.15, I.10, provider,
5.4 Results of the Qualitative Interview Study
119
China) In addition to the full amount of costs, the pricing model influenced the customer’s rating of usefulness of remote services. The pricing models seem to vary within the printing industry, e.g., remote services can be bought in a contractual setting or just purchased at an hourly rate, as the following statements show: "With [company name] I think it [remote service] should be standard, included with the machine. In post press some providers offer it as an option. With others it is standard. There are different models. Some charge a hefty surcharge when buying [the machine] with a modem-connection, but this then includes a lifetime of remote support service. Others include the modem but charge for the usage and their effort in the remote service." (P.1, I.1, customer, Germany) "In press side, we have a maintenance contract. This is, we buy service hours for either on-site or remote services."(P.22, I.13, customer, China) Customers further stated that apart from the maintenance contracts with the RSPC, they saw additional costs associated with the remote service such as internet fees, network equipment, and cabling. The customers evaluated these costs as relevant and consider them in their evaluation on whether to buy remote services. "The cost includes Internet fee and the service fee [the RSPC] charged." (P.29, I.19, customer, China) The customer interviews indicate that often an on-site visit from technician has to follow a remote service because the problem could not be fully resolved. But until then, the remote diagnosis bears the advantage to gain information about the problem in advance to prepare the right tools and spare parts for the on-site technician. Customers interpreted this fact in different ways. Some customers felt safer if they got an accurate description of the situation through a quick remote service check. This allows them to assess the seriousness of the problem as well as the extent of the necessary repairs. Other customers saw the difference but did not value it as much: "Sometimes, we do not understand the showed problem, so we can ask for a remote service or call. Through remote service, they can direct check the errors showed in computer. It is better than telephone call because they can know our problem accurately." (P.23, I.14, customer, China) "If we really use the remote service, for example, they know what happened, such as some parts are broken, and then they can bring it along." (P.24, I.14., customer, China) "Only sometimes does the remote service allow to pinpoint mechanical problems. Only in rare cases the problems are due to software flaws, that could be fixed remotely. In so far the benefit of remote service should not be overestimated, even
120
5. Qualitative Exploratory Interview Study though I would not want to do without [remote services]." (P.1, I.1, customer, Germany) "The biggest advantage first: I can get help with any problem I have within a half hour. If I had to wait, if we did the phone based supporting and I could not resolve the issue with their help, I had to wait for a technician. I still may have to wait for a technician with the remote service. But when that technician shows up he has all what he needs to fix the machine." (P.9, I.7, customer, USA)
Customers also consider the likelihood of an emergency event in their overall evaluation of remote services, because they perceive the remote service as being useful only when there is a problem with the machine leading to down-time. Depending on the quality perception of the machine manufacturer, they judge the likelihood of these incidents and base their decision to use remote services on that evaluation. Because the customers only rarely had problems during the warranty period, they did not experience the benefits of remote services first hand. This leads to customers forecasting the actual usage too low, judging the benefits as too small, or to the customers feeling that the price is too high: "Yes. I think, in the future two or three years the possibility is very high that the condition of the machine is similar to now. That means, we just use it once per year. Then the price should be relative high. As a result we will not consider it." (P.25, I.15, customer, China)
5.4.7
Participation Beliefs
Interactive remote services are intensively co-produced by the business customer employee together with the remote service engineer. Therefore, the customer’s own attitude becomes an important driver of his service perception. This is supported by the interviews, which indicates that the motivation of the customer and his mental image of the collaboration with the RST affected his attitude towards interactive remote services. This factor is not so important in remote monitoring services but becomes more so in interactive remote services such as in a remote repair where the customer might support the RST by performing mechanical tasks. Nearly all interviewees were very positive regarding the co-production activities of the customers during a remote service, i.e., performing mechanical tasks on the machine, opening and checking the water cabinet, or changing replacement parts as a customer explained: "Yes, we know. For example, some problem occurs, like problem about water cabinet, we do not know how to solve the problem, after the engineer has checked they tell us, the reason is some switch is turned off. Then we will turn it on. It is just some easy operations like this." (P.24, I.14, customer, China) In general, customers were not too worried about the collaboration. They have had good expe-
5.4 Results of the Qualitative Interview Study
121
riences and appreciated the guidance they got from the RST, which provided them with a clear image about what was expected from them during the interactive remote service: "It’s very easy. And I’m very comfortable with the equipment. I have been running the equipment for a while, so I’m comfortable doing that. It’s easy. They take me through it step by step. They are not asking me to take the machine down into a lot of little pieces. But they are asking me to do the things that I should be able to do on this end." (P.9, I.7, customer, USA) "This is implemented in a very good manner, as if for example a technician or a software engineer logs into the machine - and I stand next to the machine’s screen - they can guide me through all the menus. With the mouse he directs me to click on item a or item b; he guides me exactly ... But without this guidance, a normal machine operator would not find it." (P.3, I.2, customer, Germany) "I think it is no problem, because only in case that we understand what they told us through internet, then we do some actions; if we do not understand, we do nothing." (P.27, I.17, customer, China) In contrast to the customers’, the remote service provider employees frequently mentioned the existence of customers with doubts about their own ability to perform in an interactive remote service. Managers of remote service provider companies in Germany and China emphasized the existence of self-efficacy doubts of the customer: "I believe that especially inexperienced employees of our customers have fears of using a remote service technology." (P.8, I.6., provider, Germany) "... we talk to the operator on the phone and tell him to do some job in the machine. This is why the operator worries about touching the machine, because he worries about he is not a electrician or a mechanician. .... I think, it is true, that the worries are acceptable because he is just an operator, he needs to take care of this engineering job on behalf of our company at this moment." (P.16, I.11, provider, China) It is striking that the participation of a customer in a remote service is potentially dangerous if he fails in performing his task. Remote service managers in China classified such failures as a disaster with respect to legal issues and consequences for the production process: "There is another safety issue, about which we also worry. If we are telling them [the remote service customer employees], I mean, to do the wrong step, unluckily or accidentally; if they really do the wrong thing then that would be really a disaster." (P.18, I.11, provider, China)
122
5.4.8
5. Qualitative Exploratory Interview Study
Cultural Differences in the Customer’s Willingness to Collaborate
Although outside the scope of this study, the point of cultural differences emerged during the coding and analysis process of the study. In my qualitative interview study, I could identify two extreme views of customer’s on their attitudes towards co-production in an interactive remote service, that were expressed by customers from China and Germany. The difference in their attitude towards collaboration is backed-up by the opinions of remote service provider employees from the United States. Customers from Germany and China seem to have opposing views on their own participation in an interactive remote service. A statement of a German machine operator represents a skeptical attitude: "Sometimes I think it is not fair that I should help the remote service engineer. It is his task - not mine - to repair the machine. He should come over to repair it." (P.5, I.4, customer, Germany) "I, for example, do not think that the exchange of spare parts is the task of a printer [printing machine operator]." (P.7, I.5, customer, Germany) Although most of the customers I interviewed had a professional and positive view on their support function in a joint remote repair, it is striking that in some Chinese customer interviews the collaboration process was tightly connected to an experience of knowledge sharing and worthwhile accomplishment. This is in contrast to statements of refusal from two German customers. In my interviews, most Chinese customers saw more than just the practical benefits in supporting the remote service engineer. The customers felt honored to help in the remote service repair process and thought that they could speed up the repair process by sharing their knowledge. In contrast to statements of customers in USA and in Germany, most customers in China mostly trusted the analysis of the RST and were willing to follow his instructions. "I like that I have to help the [remote service] engineer. I feel appreciated and I am happy that he values my support and knowledge. I think that without me, the remote repair would not be effective." (P.26, I.16, customer, China) "It is very important that we can share some knowledge through remote services." (P.22, I.13, customer, China) Just once in the interviews, the remote service provider employees mentioned a refusal of a Chinese machine operator to support the RST, but this was because of an unclear understanding of hiararchy: "You know, sometimes they [machine operators] refuse to do [support]. They only accept orders from their management, they have to do it. Because the Chinese, they still, you know, do follow the rules." (P.18, I.11, provider, China) In addition, three remote service provider employees from the USA saw cultural difference in
5.4 Results of the Qualitative Interview Study
123
terms of a willingness to co-produce between the customers in the USA and Germany. They especially referred to German customers as more unwilling to co-produce than customers from the US, which they considered to be more experienced with remote collaboration: "So, they [the American customer] will go much further than the German customers in order to try to resolve the press over the phone. .... We have multiple people like machine operators, electricians, and mechanics basically look at the screen and fuel their knowledge, resolve a mechanical lock-up of a protector device, which is a complicated device, and resolve it over the phone. I don’t think anybody has done this outside of the US. So this is the typical US-stuff, they always go further than everybody else. ..... it’s just their willingness to go the extra mile and they rely on their own knowledge or whatever." (P.11, I.9, provider, USA) "The customers in the US are a lot more used to working, actually working, with people over the phone to help troubleshoot their problems." (P.14, I.9, provider, USA) "It is integration of the customer because you have to have a relationship with him and he has to be willing to do stuff. And that’s I think the big difference between Germany and the US. The customers in the US are much more used to, for years have always done that." (P.12, I.9, provider, USA) The study of cultural differences was not a primary objective of this research, but the statements hint at differences in the attitude towards co-production between customers in different countries. Thus, in the interviews with Chinese customers, the role of appreciation and knowledge sharing stands out, whereas in the interviews with German customers, some customers emphasized their disapproval of providing support during a service for which they pay. American remote service providers see a difference in the willingness to collaborate between German and American customers mostly due to the fact that the latter have more experience with this type of collaboration. These findings raise interesting questions on cultural differences and could be a fruitful avenue for future research.
5.4.9
Prior Experiences
Experience seems to influence how customers evaluate the perceived usefulness of remote services. Customers who have had positive experiences tended to be more loyal and give a more positive evaluation of remote services. They also based their decision to continue using remote services on these positive experiences: "And we [note: the RSPC and the customer] have been successful .... And as long as that continues, we’ll be in good shape and we will stick to [RSPC]." (P.9, I.7, customer, USA)
124
5. Qualitative Exploratory Interview Study "We have been a "full-test" customer of [company name] for 10 to 15 years.... through that we have gained a lot of experience with how good remote service is. Based on that we decided to buy or rent remote services for all our machines." (P.3, I.2, customer, Germany)
If the customer had bad experiences, e.g., the remote service did not fully solve the problem, then he was skeptical about the general value of remote services or did not see the value at all: "This is a very practical problem. Maybe remote services has not let us experience its full benefit. .... this service can not let us get the feeling that our problem will be solved if we use it. Normally, we must connect with them for several times. At the same time a telephone call is also absolutely necessary. And at last, they have to come to us. I think, maybe because the limitation of their knowledge, they cannot always solve all problems if we have trouble." (P.22, I.13, customer, China) "From a remote diagnosis I expect that I get a definite statement if they can solve the problem or not, that they log in the control and remove the blockade, that the machine works again. But most of the time they cannot do that." (P5., I.4, customer, Germany) "The downside is that if somebody doesn’t have any problems, which is a good thing, then he doesn’t see the value. He thinks "All this year I didn’t have to use my remote service." It is just like anything else, if you don’t utilize it, you find the value lower." (P.10, I.8, provider, USA)
5.4.10
Organizational Factors
The interviews suggest that the tasks and responsibilities in remote services are performed by a small set of functional roles within the customer companies. These vary slightly in their terminology from company to company but commonly include: production managers; owners/managers; and machine operators. In the smaller companies multiple roles are taken by one individual, e.g. the owner/manager. Most interviewees had participated in a remote service and were included in the decision to use or buy a remote service contract: "I am involved in making the decision to use remote services." (P.3, I.2, customer, machine operator, Germany) "I have a little bit of, my opinion weighs pretty heavy for the owner. I am the production manager and he hired me specifically for my talents to work with the people and the equipment. And if I’m madly against it, he is going to pay a lot of attention to that. " (P.9, I.7, customer, production manager, USA)
5.4 Results of the Qualitative Interview Study
125
"Typically I think the guy who would call in is the supervisor, something like that, the production manager, not as often the press man." (P.11, I.9., provider, USA) "Normally they have a production manager, a guy to handle the operation. And then the two operators at the machine." (P.17, I.11, provider, China) "The foremen and department managers are the ones who are responsible for the remote service." (P.1, I.1, customer, Germany) Employees, especially from lower ranks, do not form their attitudes solely on their own perceptions. Instead, there seems to be a general understanding of attitudes towards remote services that is influenced by superiors. These attitudes towards remote services seems to be tied to a general openness on an organizational level towards new technologies itself. For instance, bosses saw colloborative behavior as a duty of their employees: "We are open to improvements and new technologies in our company. My boss always supported us in using remote services." (P.3, I.2, customer, Germany) "Of course they will, they must be willing to do that. If they don’t like it, it means they are irresponsible for their job ... ."(P.28, I.18, customer, China) In interviews with some Chinese companies, it was striking that some customer companies wanted to learn from remote services to make their own repair attempts and that the management advised them to observe and learn from a remote service. The management of customer companies sometimes not only wanted their employees to learn but also wanted to monitor their staff through the remote monitoring feature. They wished to find out who might be responsible for problems and exert higher pressure onto their personnel to become more efficient: "We can learn something through this [remote service] process. Then perhaps we can solve it by ourselves if the same problem occurs." (P.22, I.13, customer, China) "If there are some problems with the machines, feedback can be shown on the screen and they can tell us, for example, first, what went wrong; second, what we have done wrong; and what we should improve. Then, it should be better." (P.30, I.19, customer, China) "We hope that the machine can permanently connect with the remote service, like just said, and the engineer can periodically, let’s say once a month connect with our machine to check if our operators work with the machine regularly. If his work isn’t reasonable, he should tell him and let him improve, if he won’t change because of laze, he should tell his supervisor about that. Then we will let him change, so that the machine can work normally." (P.28, I.18. customer, China) The influence of a company’s general attitude toward technological advances on the decision to favor remote services instead of face-to-face services was mentioned briefly by a remote service
126
5. Qualitative Exploratory Interview Study
provider employee: "No, it’s just that in the press world they were slow to transitions into the computer age. Yes, it is, because it is very analog system that hasn’t really changed or revolved over the time and people get into a pattern of being used to one thing and having a service tech come in." (P.10, I.8, provider, USA)
5.4.11
Contextual Factors
From the results of the interview study, it is apparent that the infrastructure of the remote service customer companies influences the availability of these services. Although most customers had no problems with accessing the internet, two customers in China had issues with their internet connection and extended these negative associations to remote services: "In general, we feel they [remote services] are very good. But sometimes because of network problems, for example, the network is very slow, the efficiency will be affected." (P.22, I.13, customer, China) "I think another problem is hardware, in Beijing especially, hardware and the network situation here. Some customers, small customers, rent a house very simple in the countryside. Even for the phone line the telephone is a problem." (P.15, I.10, provider, China) When examining the customers’ internal network situation, one aspect, which might have had an effect on remote service usage, became obvious from visiting customer’s facilities and some off-the-record discussions. Usually, the printing machines need to be connected via a physical cable to the network. A number of facilities were not prepared to have the network access near the machine. The cable then was just lying across the workspace. This led to disconnects when somebody accidentally got caught in the cable and pulled it out of its socket, breaking the network connection. A Chinese customer put it into words: "It is ok when it is connected all the time. But we may take it away if the machine doesn’t work wrong because the cable is a hindrance to us if we are working. Therefore, we take it away." (P.28, I.18., customer, China) The same inconvenience factor also applies to a German customer, who complained about insufficient technological compatibility of the machines with standard network configuration. "What bothers me, ... and what is a huge problem with the remote access is that they [the providers] requires an analogue modem connection. This led to the addition of analogue ports to our telephone system, this meant a huge effort to do the cabling, instead of just using a wireless-lan card inside the machine. We already have wireless-lan everywhere in-house. I would imagine that to be
5.4 Results of the Qualitative Interview Study
127
much simpler, but here the machine vendors and providers are very reluctant and still work with analog modems." (P.1, I.1, customer, Germany) From the interviews, it can be derived that one of the reasons for not using a remote service is the availability of a traditional local service in the customer’s vicinity, especially in China and the USA. A customer from Hong Kong, who operates near a remote service provider company’s office in Hong Kong, put it like this: "Maybe, because we are very close to [company name’s] office, we do not need this kind of service. In case that we were very far way from them, if they can [remotely] identify the problem, for example, and if some parts are broken, they can sent them to us. Therefore we can save time. At this point it is helpful ... But we are nearby, if I suggest it [remote service] to him [contact at company x], he will refuse and will say "we can ask them to come quickly."" (P.24, I.14, customer, China)
5.4.12
Discussion of the Results
The qualitative interview study delivers unique insights, because it is the first study focusing on remote service perception. It identifies not only single beliefs groups, but also enables a holistic view of various remote service perceptions. Based on the interview analysis, I developed a comprehensive conceptual framework of factors that are relevant for remote service attitudes and perception. The framework summarizes the findings of the qualitative study and is presented in figure 5.5. The conceptual framework of factors influencing the perception of remote services comprises 44 different beliefs nested in eight main belief categories. The identified belief groups encompass the beliefs remote service customers directly or indirectly mention in this study. The relevant main influencing factors are discussed below: 1. Relational Beliefs: Relational factors such as trust towards both the remote service provider company and the remote service technician positively relates to the customers’ perception of a remote service. Customers’ trust a remote service provider company because of its good brand image and well-known reputation. In turn, this is taken as an assurance for not letting them down. The trustworthiness of the remote service technician seems to be very important for the customers, especially if they are evaluating an interactive remote service with an active RST accessing their machine. The behavior of an RST is perceived as trustworthy when he acts with integrity, has the interest of the customer in mind, and shows competence and effort. For customers who value personal contact, the perception of a remote service can be negatively evaluated if the remote service is perceived as not sufficiently providing social contact.
128
5. Qualitative Exploratory Interview Study
2. Process Factors: Customers want to have control over the remote service process. The interaction with a non-observable remote service technician becomes the object of control wishes, especially in interactive remote services. The customer’s imagination of the remote service process can be influenced by fears over misbehavior of the RST. Further, a high level of transparency and the availability of control mechanisms are positively remarked on because they enable the customer to stop, abort, change or direct the RST’s actions. In addition to perceived controllability, a high level of social presence of the remote service technician enabled via advanced communication technology such as video streaming or live chat enhances customer’s attitudes towards remote services. As much as customers wish to control the RST, there is also an equally strong fear of being controlled by the remote monitoring of their machines. 3. Participation Beliefs: Customers showed different views on their collaborative part in interactive remote services. Their perceptions ranged from seeing collaboration as an appreciation of their work and the sharing of knowledge, to uncertainties about whether the remote service technician had the power to direct the customer, to the outright refusal to take responsibility during the repair. The efforts of the remote service technician to guide the customer during the collaboration process were evaluated positively by a majority. Self-efficacy and ability beliefs were also important when the customer formed his attitude towards interactive remote services. 4. Technology Beliefs: During a remote service encounter, perceived attributes of the technology play a major role in the customers’ perceptions. A high risk perception of the customer regarding security issues negatively affects customer’s attitude towards remote services. It is important to note that especially interview partners with none or just few remote service experiences perceive the risk as a major drawback for remote services usage. Trust in the technology, the ease of use, and the convenience in using the technology, as well as a certain level of technology affinity foster a positive evaluation of remote services. 5. Economic Value: In a business setting the benefits of remote services and their relative advantage compared to face-to-face delivered services strongly influence the general perception of customer employees of remote services. Time savings, costs, pricing, and contractual models are factors that are perceived as important attributes of remote services. 6. Experience: Prior experience seems to influence how well the customers evaluate the usefulness of remote services. Customers who have had positive experiences tend to show loyal behavior, satisfaction, and a more positive evaluation of remote services. Customer’s who did not have any experience rely on their imagination of remote service scenarios. In contrast, customers with bad experiences are more reluctant to use remote services.
5.4 Results of the Qualitative Interview Study
129
7. Organizational Factors: Employees, especially from lower ranks, do not form their attitudes solely on their own perceptions. Instead, there are influenced by norms set by immediate superiors, management, and the organization as a whole. The attitude of the individual employee, the attitudes of superiors, management support, and the organizational openness for technology and services all influence the organizational decision towards remote services. 8. Contextual Factors: The technical infrastructure of the remote service customer companies, the distance to the service provider, and the size of the company impacts the availability, past experiences, and attitudes towards remote services. The findings from this qualitative study contribute to marketing literature by discovering the substantial role of the RST’s behavior on the customers’ perception of remote services. Also, it sheds new light on the interplay between trust and control, the relationship between interaction and technology, and the customer provider nexus in B2B service settings. Managerial implica-
%! $
! !'& 1%'' %$ 1!$%$ 1! %#%$/# 1!!!&$
!!* $ 1$!$ 1! ' 1%* ! # $ 1 %* 1#&$% !!*
1#&$%(!#% $$
1 %#%* 1!"% 1 '! 2!#% 1#$! ! %%
1#&$% 1"&%%! 1#
# +%! %!#$ 1!" * !#$ 1 %&""!#% 1!$/& %! 1" $$ 1$#!##
#!$$
! %#!%!#$
!%#' #"%!
1! %#!!'# 1# $"# * 1 %! 1! #$ 1#! !$#'
)"# $
! %)%&%!#$
1%$%! 1$$%$%! 1 !-'.)"# $
1!#"$% 1!%! 1 #$%#&% 1 %(!# 1 ,!"%%*
#%"%! $ 1%* 10* 1""#%! 1 !(#
1!#% 1& 1!(#%!#% 1$"! $%*
Figure 5.5: Factors Influencing Remote Service Perception
130
5. Qualitative Exploratory Interview Study
tions will be discussed in the end of this dissertation together with insights from the quantitative studies. The above results, foremost the conceptual framework, are the foundation for the hypotheses derivation and the model development discussed in the next chapter. There, the identified factors influencing remote service adoption and continuance will be put in the context of literature and the interdependencies are discussed. The resulting model will then be validated within the quantitative studies presented in chapter 7.
Chapter 6 Hypotheses Development In this chapter, I propose three major groups of hypotheses. First, I outline the hypotheses of the Interactive Technology-Mediated Service Usage Model (ITSUM) based on the results of the qualitative interview study (see chapter 5.4.12) and on the findings of prior research in marketing, IS, and management literature (see chapter 3). The ITSUM aims to clarify the causal relationships of the variables and to explain the customer companies’ intention to use remote services. Next, I will extend the ITSUM by proposing a link between a company’s intention to use interactive remote services and actual usage behavior. In a third step, I present my hypotheses regarding group differences between antecedents of adoption and continued usage of interactive remote services.
6.1
Development of the ITSUM
The ITSUM hypotheses are based on behavioral models such as the TRA (Fishbein and Ajzen 1975) and the TPB (Ajzen and Fishbein 1980). These models aim at explaining usage behavior14 via intention to use, which in turn is determined by individual beliefs and attitudes towards performing the behavior. This reference frame is adapted to the most salient interactive remote services belief groups that were identified in the qualitative study: 1. C OUNTERPART BELIEFS comprise trustworthiness and controllability beliefs directed at the remote service counterpart and his actions. They refer to the process control beliefs and relational beliefs identified in the qualitative study. 2. T ECHNOLOGY BELIEFS consist of ease of use and trust in technology. 3. U SEFULNESS BELIEFS comprise perceived usefulness as identified in the qualitative study as a part of economic value. 14
The link between the individual perception of intention, organizational intention, and actual organizational behavior is addressed by a follow-up study at a second point in time (see chapter 6.2 and chapter 4.1)
132
6. Hypotheses Development
ITSUM #
"
!
# $
#
#
#
!
!
Figure 6.1: The Extended Interactive Technology-Mediated Service Usage Model 4. PARTICIPATION BELIEFS comprise role clarity, role ability, and motivation of the remote service customer. 5. O RGANIZATIONAL CHARACTERISTICS comprise subjective norms, function, and company size as identified in the qualitative study. An overview of the final result in the form of a graphical representation of the ITSUM model is given in figure 6.1. The derivation of each hypothesis is described in the next sections.
6.1.1
Counterpart Beliefs
6.1.1.1
Controllability of the Counterpart’s Actions
The findings of the exploratory study show that customers wish for transparency of the RST’s actions and the possibility to control the actions of the RST. The RST’s actions are mediated through technology and hard to control due to the physical dislocation of customer and provider companies. In particular, the customers’ need to be able to control the actions of the service counterpart that directly affect the service object. Or as a customer put it: "Control plays a very important role. I want to decide what exactly is done with my machine." (P.4, I.3, customer, Germany).
6.1 Development of the ITSUM
133
Especially interview partners with few or no remote service experiences perceive potential risks as major barriers for remote services usage, e.g., they fear that the RST could actually damage the printing machine. In this context the absence of control has been described as undesirable, and the perception of controllability of a customer is closely tied to his risk evaluation and will likely influence his decision to use remote services: "One point, we must know what he is doing and make sure that his doings are allowed and does not disturb our production." (P.22, I.13, customer, China). These findings are supported by literature in the context of e-services adoption, as researchers found that service perception and intended service usage are adversely affected by risk perceptions (e.g., Featherman and Pavlou 2003; Hsu and Chiu 2004; Pavlou 2003). In this study, controllability beliefs refer to the degree to which the service customer believes that he is able to exert control over the RST’s behavior during the service process. Control over a (human) counterpart’s actions has not yet been measured in an adoption or continuation context. Even though this construct is new in the context of technology mediated services, controllability has been studied in other contexts such as control over one’s own behavior, control over service processes, and control over another organization in strategic alliances. In the TBP the perceived control over one’s own action is a usage antecedent (Ajzen 1985; 1991) and reflects a person’s perception of ease or difficulty toward implementing the behavior of interest (Ajzen 2002; 1991). Because some behaviors pose difficulties of execution that might limit volitional control, the TPB considers perceived behavioral control in addition to actual control. Control over one’s own behavior has been shown to positively affect consumer feeling and consumer satisfaction (Hui and Bateson 1991; Hui and Toffoli 2002; Namasivayam 2004). In the context of technology or technology-intensive service adoption, the perception of control is interpreted as the perceived control over facilitating conditions (e.g., the functionality of a website or an ATM) or as the perceived procedural control over the outcomes and processes when using a technology or technology-intensive service. For explaining usage of services, the effect of control on usage intention has been supported in studies on telemedicine (Chau and Hu 2002), online-banking (Lee and Allaway 2002; Liao and Shao 1999), e-commerce (Pavlou and Fygenson 2006), e-brokerage services (Bhattacherjee 2000), e- and mobile coupons (Dickinger and Kleijnen 2008; Kang et al. 2006), computing services (Taylor and Todd 1995b;a), and technology free services such as housing services (Christian, Armitage, and Abrams 2003). The view of control in organizational science and management literature which defines control as a process that regulates behaviors of organizational partners in alliances and cooperations (Bradach and Eccles 1989; Cardinal, Sitkin, and Long 2004; Das and Teng 2001) is closest to the form of control proposed in this thesis. Control increases the predictability of the partner’s future behaviors (Nooteboom 2002; Vlaar, Van den Bosch, and Volberda 2007). Therefore it reduces uncertainty and encourages risk taking (Mayer, Davis, and Schoorman 1995).
134
6. Hypotheses Development
The more control over the RST the customer perceives, the more likely he will use interactive remote services despite potential uncertainties and risks. For example, if a customer is able to terminate the remote service engineer’s access to the service object or is able to make decisions regarding the service process at any time, he will have a higher perceived controllability and, thus, would be more likely to accept services being performed remotely. The control does not need to actually be exerted. The possibility to control the provider’s proxy is often perceived as sufficient (Namasivayam 2004). Assuming that an individual customer employee in a B2B remote service situation participates in the decision to use remote services, the following is proposed: H1: The individual customer employee’s perception of the controllability of the service counterpart’s actions is related positively to the organizational intention to use the service.
6.1.1.2
Trustworthiness of the Counterpart
In the qualitative interview study, not only the control over the service counterpart emerged as a dominant topic, trust towards the service counterpart was also a major factor in forming customers’ attitudes about remote services. The customers lack sufficient "evidence for service" (Bitner 1993) because of the reduced observability and the perceived risks of interactive remote services. This triggers the imagination to run a "mental simulation" of the service production process in the customers’ minds (Taylor et al. 1998). For example, a production manager stated: "Of course I’m thinking about what the technician is doing right now [during the remote service]. I can afterwards check the login-protocols but during the service process I have to essentially trust him." (P.7, I.5, customer, Germany). Interview partners with few or no remote service experiences perceived the risk as a major barrier for remote services usage, e.g., they fear that the RST could actually damage the printing machine and worry about data espionage. These findings are supported by literature in the context of e-services adoption, as researchers found that service perception and intended service usage is adversely affected by risk perceptions (e.g., Featherman and Pavlou 2003; Hsu and Chiu 2004; Pavlou 2003). This is recognized by providers as well as their customers, or as one provider put it: "Trust is one of the most important factors for our customers. If they do not trust our company we could not sell the remote services to them." (P.12, I.9, provider, USA) The trustworthiness of the remote service engineer itself was often named by the customers as one of the fundamental considerations when adopting a remote service: "My trust towards [company name] is based on the relationships I have with the people, the technicians." (P.9, I.7, customer, USA) Customers trust remote service technicians if their service counterpart acts with integrity, has the interest of the customer in mind, and shows competence and effort. Due to the reduced
6.1 Development of the ITSUM
135
observability during an interactive remote service, customers must to a certain degree, simply have faith that the counterpart will act in a trustworthy and authorized way: "Sometimes I get the impression that they [the technicians] don’t have a clue what to do. They do not really help me. At least they don’t do everything in their power to help me and they do not think proactively and foresighted." (P.5, I.4, customer, Germany). The trustworthiness of the RST seems to be important for the customers, especially if they are evaluating an interactive remote service with an active RST accessing their machine. Trustworthiness of a service interaction partner has frequently been researched in marketing. A number of characteristics of trustworthiness have been identified including competence, customer orientation, helpfulness, sociability, friendliness, courtesy, empathy, credibility, and attentiveness (e.g., Gremler and Gwinner 2008; Hennig-Thurau 2004; Sundaram and Webster 2000; Surprenant and Solomon 1987). Service quality perceptions have been shown to be influenced by the service counterpart’s responsiveness, attentiveness, and empathy (Parasuraman, Zeithaml, and Berry 1985; Zeithaml, Berry, and Parasuraman 1988) as well as satisfaction (Froehle 2006). Trustworthiness beliefs are drivers of interpersonal relationship building (Crosby, Evans, and Cowles 1990; Hwang and Kim 2007) and encourage risk taking (Mayer, Davis, and Schoorman 1995). In inter-organizational relationships, trustworthiness of the partner organization has been shown to positively influence the engagement success (Costa and Bijlsma-Frankema 2007). The influence of interpersonal trust on technology-mediated service usage intention has not been researched explicitly before. In contexts where the object of trust is not a human service counterpart, trust has been researched, e.g., as "online-trust" towards e-vendors, e-service providers, or to websites. In this context, empirical evidence for the positive effect of trust on customer’s service usage intention and usage has been shown in studies on online banking (Suh and Han 2003), online recommendation systems (Wang and Benbasat 2005), web site usage (Bart et al. 2005), online shopping (Gefen, Karahanna, and Straub 2003b;a; Gefen and Straub 2004; Pavlou 2003) and e-services (Gefen and Straub 2004). In this study, the trustworthiness of a customer’s interactive remote service counterpart is understood as the customer’s beliefs in regard to his integrity, benevolence, and ability as defined by Gefen (2002b); Mayer, Davis, and Schoorman (1995). If the customer thinks that the RST acts in a competent and benevolent way and shows only agreed-on behavior, he is more likely to use remote services (Mayer, Davis, and Schoorman 1995). Therefore, assuming that an individual customer employee in a B2B remote service situation participates in the decision to use remote services, the following is proposed: H2: The individual customer employee’s perception of the trustworthiness of the service counterpart is related positively to the organizational intention to use the service.
136
6. Hypotheses Development
6.1.2
Technology Beliefs
6.1.2.1
Trust in Technology
One finding of the qualitative study is that customers may not perceive the remote service technology as trustworthy for a number of reasons. They fear data espionage by third parties, are afraid of hacker attacks, or might believe that use of the technology is more in the interest of the service provider rather than their own. A customer put it like this:"[The threat] is reasonable. We must avoid that the hackers can access our network. If we would get some viruses due to attack from hackers, it would influence our operation on the machine. Therefore, it [remote services] must be avoided. ... Could [the remote service provider] 100% guarantee that no virus can access my machine?" (P.28, I.18, customer, China) Trust in the technology counteracts a high risk perception and fosters the customer’s positive evaluation of remote services. Thus, customer employees may believe that benefits of the technology will outweigh its risks. A customer from China put it bluntly: "If so, we would like to pay the 2000 Yuan. Because if there is no virus, which can access this machine, we would like to connect to the remote service permanently, otherwise we are not interested." (P.28, I.18, customer, China) These findings are in line with trust and control theories that state that if a customer looses his trust in the technology he will be unlikely to take risks (Mayer, Davis, and Schoorman 1995). Bart et al. (2005) show that website privacy and security characteristics affect trust in a website. Johnson, Bardhi, and Dunn (2008) emphasize the influence of trust in technology on service satisfaction in self-service settings. Gefen, Karahanna, and Straub (2003b) and Pavlou (2003) identify trust in an e-vendor as a strong driver of the perceived usefulness of a website and a suppressor for perceived risk (Pavlou 2003). Trust in the recommendation systems influences a customer’s perceived usefulness of the system (Wang and Benbasat 2005). In this study, trust in technology refers to the trust the interactive remote service customer has in the technology necessary to deliver a remote service. The technology can be network and communication technologies or a remote service technology within the service object. Therefore, the following is proposed: H3: The individual customer employee’s perception of general trust in the service technology is related positively to his perceived usefulness of the service.
6.1.2.2
Ease of Use
Perceived ease of use (EOU) is one of the core factors of the TAM (Davis, Bagozzi, and Warshaw 1989). EOU was found to be a significant antecedent of perceived usefulness in several meta-analyses of the TAM (Lee, Kozar, and Larsen 2003; Ma and Liu 2004; Yousafzai, Foxall,
6.1 Development of the ITSUM
137
and Pallister 2007a;b). This effect has been supported by numerous studies on IS-acceptance and technology-intensive service acceptance including: computing services (Taylor and Todd 1995b;b); online shopping (Pavlou 2003); online recommendation systems (Wang and Benbasat 2005); and IS system usage (Davis, Bagozzi, and Warshaw 1989; Venkatesh and Bala 2008; Venkatesh and Davis 2000). The interview study supports the major role technology attributes play in the customers’ perceptions. In the course of the interviews, employees from remote service customer organizations explicitly stated that factors such as the perceived EOU of the technology and the convenience in using the technology influence their evaluation of general usefulness of the technology: "I think remote services are pretty useful, but just recently I wanted to start the remote service procedure and I had problems with the technology and it was very difficult for me to handle this." (P.7, I.5, customer, Germany). In an interactive remote service situation, EOU is the extent to which the interactive remote service customer believes that using technology to request and receive a remote service is free of effort. The less effort the remote service technology requires, the more it increases the perceived usefulness. Therefore, the following is proposed: H4: The individual customer employee’s perception of the ease of use of the service technology is related positively to his perceived usefulness of the service.
6.1.3
Perceived Usefulness
The exploratory study shows that the benefits of remote services and their relative advantage compared to face-to-face delivered services strongly influence the general perception of customer employees and the organizational decision to use these services. Reasons why customers perceive remote services as useful compared to on-site visits of service technicians are for example, time savings and increased flexibility. A remote service customer stated: "And it’s much faster and easier for both sides, it’s a win-win-situation I think for both sides of the remote service. They can solve the problem faster and easier and less costly for both parties .... If I had the option to give up remote service and call for somebody, I wouldn’t do. With the remote service they gonna answer the phone today. And they going to call me today. And we are going to work on that press." (P.9, I.7, customer, USA). Perceived usefulness is a core construct of TAM research, which proposes that employees form their intentions toward behaviors they consider useful (Davis, Bagozzi, and Warshaw 1989). Meta-studies validating the TAM show, that perceived usefulness is the strongest determinant of behavioral intention (Ma and Liu 2004; Lee, Kozar, and Larsen 2003; Yousafzai, Foxall, and Pallister 2007a;b). Perceived usefulness is closely related to achieving extrinsic rewards, for example through increases in time savings and efficiency increases (Davis, Bagozzi, and Warshaw 1992; Fagan, Neill, and Wooldridge 2008; Vroom 1964). Extrinsic motivation has been shown to influence usage behavior in IS usage (Fagan, Neill, and Wooldridge 2008; Venkatesh,
138
6. Hypotheses Development
Speier, and Morris 2002) and in self-service technologies trial (Meuter et al. 2005). Next to extrinsic motivation, perceived usefulness also strongly relates to the relative advantage construct from Rogers’ (2003) innovation characteristics. In a review and meta-analysis of 75 studies, Tornatzky and Klein (1982) assess and support the effect of relative advantage on adoption behavior. Perceived usefulness has been shown to positively influence IS usage (Davis 1989; Davis, Bagozzi, and Warshaw 1989; Venkatesh and Davis 2000; Venkatesh and Bala 2008; Venkatesh 2000); online shopping (Gefen, Karahanna, and Straub 2003b;a; Pavlou 2003); and online recommendations systems (Wang and Benbasat 2005). Empirical evidence suggests that perceived usefulness is a main driver of technology-intensive service usage including: e-services (Kang et al. 2006; Lin, Shih, and Sher 2007; Pavlou 2003); mobile commerce services (Nysveen, Pedersen, and Thorbjørnsen 2005); e-payment services (Featherman and Pavlou 2003); computing services (Taylor and Todd 1995b;a); and telemedicine technology (Hu et al. 1999; Chau and Hu 2002). Perceived usefulness has been shown to predict usage intention in both adoption and continuance scenarios (Gefen, Karahanna, and Straub 2003a; Hu et al. 2009; Venkatesh and Bala 2008; Venkatesh and Davis 2000). For example, Naidoo and Leonard (2007) identify perceived usefulness as the strongest predictor of continued usage. In the context of interactive remote services, perceived usefulness is the degree to which the customer believes that using an interactive remote service would be helpful to the organization. If the customer thinks that an interactive remote service would be useful, he is more likely to intend to use it. Therefore, assuming that an individual customer employee in a B2B remote service situation participates in the decision to use remote services, the following is proposed: H5: The individual customer employee’s perception of the usefulness of the service is related positively to the organizational intention to use the service. Perceived usefulness is not only proposed to have a direct effect on intention to use interactive remote services, but also to mediate the effects of trust in technology and ease of use on intention to use remote services. This is based on the findings of the meta-analysis by Lee, Kozar, and Larsen (2003), who show that perceived ease of use is an unstable measure for predicting behavioral intention. Featherman and Pavlou (2003); Kang et al. (2006); Pavlou (2003); and (Wang and Benbasat 2005) identify a mediating effect of perceived ease of use on intention. Pavlou (2003) found that the effect of trust in web retailing on behavioral intention was mediated by perceived usefulness. Therefore, assuming that an individual customer employee in a B2B remote service situation participates in the decision to use remote services, the following is proposed: H5a: Perceived usefulness mediates the relationship between trust in technology and intention to use the service.
6.1 Development of the ITSUM
139
H5b: Perceived usefulness mediates the relationship between ease of use and intention to use the service.
6.1.4
Participation Beliefs
The customer’s attitude towards the collaboration with an RST is an important driver of his intention to request a interactive remote service. The interviews indicate that the motivation of the customer and his mental image of the collaboration with the RST, his role clarity, and his self-efficacy affect his attitude towards interactive remote services. These attitudes towards the co-production in an interactive remote service are in line with findings on customer participation in human resources research and industrial psychology (Bowen 1986; Schneider and Bowen 1985; Vroom 1964). Bowen (1986) identifies a set of determinants comprising role clarity, ability, and motivation as drivers of customers’ co-producing behavior that is frequently used in explaining customer participation (Lengnick-Hall, Claycomb, and Inks 2000; Dellande, Gilly, and Graham 2004; Meuter et al. 2005).
6.1.4.1
Role Clarity
According to Bowen (1986) customers’ behavior is shaped by how they understand what is expected from them. A clear understanding results in high role clarity. In interactive remote service settings, role clarity reflects the customer’s knowledge and understanding of how and when he needs to participate in order to support the RST. The rationale is that if customers know what to do and how they are expected to perform, they are more likely to collaborate with the remote service technician and to use an interactive remote service. For example, a customer employee, who does not have a clear vision of how to support the RST in an interactive remote service may fear this uncertainty and decide to favor an on-site visit of an technician. The efforts of the remote service technician to guide the customer during the collaboration process were evaluated positively by customers in the interview study, emphasizing the importance of role clarity perceived by the customer. Research on customer compliance in health care supports the effect of role clarity on behavior (Dellande, Gilly, and Graham 2004) and on self-service trial (Meuter et al. 2005). Meuter et al. (2005) developed the term "customer readiness" for a combination of constructs including role clarity, role ability, and motivation. Ho and Ko (2008) developed a single construct called "customer readiness" (with items pertaining to role clarity) and proved the positive effect of the customer readiness construct on intention to continue using online banking. Assuming that an individual customer employee in a B2B remote service situation participates in the decision to use remote services, the following is proposed:
140
6. Hypotheses Development H6: The individual customer employee’s role clarity is related positively to the organizational intention to use the service.
6.1.4.2
Role Ability
In addition to role clarity, the existence of customers’ doubts about their own ability to perform in an interactive remote service is frequently mentioned in the qualitative study. Role ability refers to the customer employee’s perception of whether or not he possesses the required skills and confidence to complete the tasks necessary during an interactive remote service. Effective co-production requires customers who are capable of making useful and timely contributions to support the remote service technician. It is reasonable to assume that if a customer considers himself as able to collaborate, he will less likely perceive uncertainties and is more motivated to show his abilities by using an interactive remote service. Research on health care services, self-services, and online banking (Dellande, Gilly, and Graham 2004; Meuter et al. 2005) supports the effect of role ability on behavioral intention and actual behavior, just as it did with role clarity. A customer’s perception of his ability to perform is similar to the self-efficacy concept (see chapter 3.1.2.4). Self-efficacy has been shown to be a driver of information system usage (Agarwal, Sambamurthy, and Stair 2000; Compeau and Higgins 1995a; Hwang and Yi 2002; Venkatesh 2000) and e-service acceptance (Hsu and Chiu 2004). Assuming that an individual customer employee in a B2B remote service situation participates in the decision to use remote services, the following is proposed: H7: The individual customer employee’s role ability is related positively to the organizational intention to use the service.
6.1.4.3
Intrinsic Motivation
The decision to participate in an interactive remote service is dependent on the customers’ willingness to co-produce. They must not only know what to do and be able to perform these tasks, they must also be willing to make direct contributions to various organizational activities (Etgar 2008). This is reflected in the qualitative interview study as remote service customers hold different views on their collaborative part. Their motivational beliefs range from understanding collaboration as an appreciation of their work, a sharing of knowledge, to the outright refusal to take responsibility during the repair. One customer likes to use remote services and feels appreciated: "I like that I have to help the [remote service] engineer. I feel appreciated and I am happy that he values my support and knowledge. I think that without me the remote repair would not be effective." (P.26, I.16, customer, China). While another outright rejects his participation in a remote service collaboration: "Sometimes I think it is not fair that I should help the
6.1 Development of the ITSUM
141
remote service engineer. It is his task - not mine - to repair the machine. He should come over to repair it." (P.5, I.4, customer, Germany). Therefore, this study proposes the rationale that the more motivated a customer is to collaborate the more likely he is to seek a situation, e.g. use an interactive remote service, which requires his collaboration. Intrinsic motivation15 refers to the pleasure and inherent satisfaction derived from a specific activity (Vallerand 1997), e.g., a customer’s feeling of appreciation when he can share his knowledge during an interactive remote service. Davis, Bagozzi, and Warshaw (1992) show that extrinsic and intrinsic motivation are key drivers of an individual’s intention to use technology. Meuter et al. (2005) provide empirical support for the overall effectiveness and relative strength of intrinsic motivation beliefs on self-services usage. Ho and Ko (2008) demonstrate evidence for intrinsic motivation beliefs (within the customer readiness construct) on online banking usage and Dellande, Gilly, and Graham (2004) support the effect of motivation on compliance behavior in health care. It is proposed that the higher the customer is motivated by intrinsic beliefs, the more likely he is to use an interactive remote services. Or in other words, if the customer does not feel appreciated, maybe even utilized, this will negatively affect his intention to use an interactive remote service. Assuming that an individual customer employee in a B2B remote service situation participates in the decision to use remote services, the following is proposed: H8: The individual customer employee’s intrinsic motivation to co-produce is related positively to the organizational intention to use the service.
6.1.5
Organizational Characteristics
6.1.5.1
Subjective Norms
The results of the qualitative interview study suggest that organizational norms towards interactive remote service usage are very likely to influence the companies intention to use remote services. Employees do not form their attitudes solely on their own perceptions, instead there seems to be a general understanding of attitudes towards remote services inside a customer company, which is influenced by peers, colleagues, competitors and superiors. In literature this general understanding is summarized under the term subjective norms (Mathieson 1991) and has been identified as a direct determinant of intention in behavioral theories such as TRA (Fishbein and Ajzen 1975), TPB (Ajzen 1991) as well as in adoption models such as TAM2 (Venkatesh and Davis 2000), TAM3 (Venkatesh and Bala 2008), and UTAUT (Venkatesh et al. 2003). Subjective norms have been found to be an antecedent of intention in e-brokerage services (Bhattacherjee 2000), e-coupons (Kang et al. 2006), computing services 15
In comparison extrinsic motivation refers to valued outcomes such as improved job or organizational performances and is included in the ITSUM within the perceived usefulness construct (see 6.1.3).
142
6. Hypotheses Development
(Taylor and Todd 1995b;a), mobile services (Nysveen, Pedersen, and Thorbjørnsen 2005), online shopping (Hansen, Jensen, and Solgaard 2004; Yoh et al. 2003), t-commerce (Yu et al. 2005), and general technology usage (Karahanna, Straub, and Chervany 1999; Venkatesh and Davis 2000). Assuming that an individual customer employee in a B2B remote service situation is part of a remote service customer organization the following is proposed: H9: The individual customer employee’s perception of subjective norms is related positively to the organizational intention to use the service.
6.1.5.2
Company Size and Respondent’s Function
The impact of an individual’s personal beliefs on the company’s behavioral intention is dependent on characteristics of the organization. Premkumar and Roberts (1999) and Palvia, Means, and Jackson (1994) show the effect of company size on organizational adoption of IT technology in small and middle-sized companies. Furthermore, studies show that in smaller companies, informal working environments enable stronger employee involvement in management decisions (Wilkinson, Dundon, and Grugulis 2007). In this study, it is assumed that with increasing size of an organization, the impact of an individual’s opinion on a decision at an organizational level decreases (Premkumar and Roberts 1999). This leads to the proposition of a moderating effect of company size: H10: The company size has a moderating effect on the structural relationships between the antecedents of intention to use the service and the intention to use the service. The function of the respondent, whether he is the ultimate decision maker or an employee having to get approval of his superiors, will affect how strongly personal beliefs influence organizational decision. The owner may tend to base his strategies on personal desires and backgrounds as opposed to selecting the best-fit strategy based on rational analysis (Brouthers, Andriessen, and Nicolaes 1998). Compared to this, his employees are often forced to convince a number of different peers and colleagues across functional units and hierarchical levels to influence future organizational behavior. Therefore, it is hypothesized that the effect of the different beliefs on intention to use interactive remote services proposed in H1-H9 will be stronger, if the respondent has a decision making role in the organization: H11: The function of an individual customer employee has a moderating effect on the structural relationships between the antecedents of intention to use the service and intention to use the service.
6.2 Link Between Usage Intention and Actual Usage Behavior
6.2
143
Link Between Usage Intention and Actual Usage Behavior
The role of intention as a predictor of individual behavior is critical and has been well-established in social psychology within the TRA and TPB (Fishbein and Ajzen 1975; Ajzen and Fishbein 1980). Both suggest that the proximal determinant of behavior is one’s intention to engage in that behavior. Behavioral intentions represent a person’s motivation in the sense of his conscious plan or decision to exert effort to enact the behavior (Conner and Armitage 1998). Also, intentions and behavior are considered to be strongly related when measured at the same level of specificity (Fishbein and Ajzen 1975). In IS adoption literature, which is based on behavior models, the link between intention to use and usage behavior is supported by several studies such as Hwang and Yi (2002), Venkatesh and Davis (2000), Sheppard, Hartwick, and Warshaw (1988); and Taylor and Todd (1995b). Sheppard, Hartwick, and Warshaw (1988) report a mean correlation of 0.53 for predicting behavior from intention in their meta-analysis on 87 studies. Van den Putte (1993) provides a meta-analysis based on 113 studies and reports a mean multiple correlation of 0.62 for predicting behavior from intention. In this thesis, the assumption is made that the individual perceptions ultimately affect the decision taken on the company level. Also, it is assumed that if the organizational intention, as perceived by an individual, to use remote services increases it becomes more likely that the organization actually uses a remote service, instead of an on-site service. Drawing upon TRA’s theoretical rationale and previous empirical evidence, the following thesis proposes an extension to the ITSUM regarding actual behavior: H12: The organizational intention to use the service will positively affect actual usage behavior of the organization. This hypothesis is validated through a longitudinal study design that includes a follow-up survey at a second point in time (t2 -study).
6.3
Hypotheses Development for Group Comparisons
The results of the qualitative interview study suggest that the more experienced a company and its employees are with remote services technology and the service itself, the more trustful and self-confident the employees feel and the more they have an optimistic view of remote services. For example, a long time user of remote services stated: "We have been a "full-test" customer of [company name] for 10 to 15 years.... through that we have gained a lot of experience with how good remote service is. Based on that we decided to buy or rent remote services for all our machines." (P.3, I.2, customer, Germany).
144
6. Hypotheses Development
Especially inexperienced customers perceive a remote services situation as risky (see chapter 5.4.3). The customer employee’s fear triggers a negative mental simulation of remote service processes, which they haven’t truly experienced yet: "This is a very practical problem. Maybe remote services has not let us experience its full benefit. .... this service can not let us get the feeling that our problem will be solved if we use it. Normally, we must connect with them for several times. At the same time telephone call is also absolutely necessary. And at last, they have to come to us. I think, maybe because the limitation of their knowledge, they cannot always solve all problems if we have trouble." (P.22, I.13, customer, China) Attitude is formed based on three general classes of information: information concerning past behavior; affective information; and cognitive information (Zanna and Rempel 1988). The past expertise of an organization was found to be an influence factor for explaining technology adoption (Palvia, Means, and Jackson 1994; Premkumar and Roberts 1999). In case of the continued usage of remote services it seems to be plausible that the earlier evaluations will affect later evaluations, because knowledge gained from experience is certainly a critical piece of information for decision making (Bolton 1998; Hogarth and Einhorn 1992; Hu et al. 2009). According to Karahanna, Straub, and Chervany (1999), beliefs in an organization’s adoption phase are formed primarily based on indirect experience while post-adoption usage beliefs are formed based on past experience. Prior experience and past behavior have been shown to influence service usage and service usage intention (Dickinger and Kleijnen 2008; Kang et al. 2006; Yoh et al. 2003) or moderate the effect of drivers of usage intention (Venkatesh and Davis 2000; Venkatesh et al. 2008). Nevertheless only few studies explicitly compared experienced and inexperienced users to derive insights in how far adoption drivers differ from continuance drivers (see Karahanna, Straub, and Chervany (1999)). In order to explore the effect of past experiences on the formation of attitudes and intentions about remote services, I compare organizations with little or no experience to organizations with considerable remote service experience. I also distinguish between pre-adopter organizations (pre-adopter group) and organizations that already show continued usage of remote services (continued user group). This distinction is in line with Rogers’ (2003) two main stages of adoption: an initiation phase, where organizations have not yet adopted interactive remote services but may have tested them; and an implementation phase, where organizations have already implemented the usage of interactive remote services in their business processes on a regular basis. The employees in pre-adoption organizations may feel higher uncertainty surrounding the adoption decision compared to a continued usage decision in a more experienced organization. One would expect the former to have a richer, more complex set of behavioral beliefs. Karahanna, Straub, and Chervany (1999) state that it is likely that customers focus on a wider set of beliefs when uncertainty is high. Karahanna, Straub, and Chervany (1999) test attitudes towards using services for non-users compared to users and found that only two beliefs significantly form the
6.3 Hypotheses Development for Group Comparisons
145
users attitude, whereas five beliefs underlie the attitudes of pre-adopters. Klonglan and Coward Jr. (1970) found that sociological variables may be more important in explaining mental acceptance of innovations, whereas economic variables may be more important in explaining use. In an IS adoption context, this is supported in the comparison of usage antecedents over time with the UTAUT framework. The influence of performance expectancy (including perceived usefulness) increased over time, whereas the influence of facilitating conditions (including control), social influence (including subjective norms), and effort expectancy (including ease of use) decreased over time (Klonglan and Coward Jr. 1970). Nooteboom (2002) and Vlaar, Van den Bosch, and Volberda (2007) show that with a lower risk perception, trustworthiness and control perceptions decline in importance. This coincides with results from the qualitative study that suggests that after companies routinely use interactive remote services, uncertainties become less relevant, because the fears are reduced and more rational considerations such as the relative advantages of remote services become more important. Further, inexperienced customers seem to have vague ideas about the security of a remote service and feel uncertainty when they think about letting an RST access their machine remotely. Based on the assumption that prior evaluations and past use affects subsequent evaluation, the importance of beliefs regarding the service counterpart, such as perceived trustworthiness and controllability, will decrease since the level of uncertainty declines as organizations move from the pre-adoption phase to continued usage of remote services: H13: The individual customer employee’s perception of controllability of the remote service technician will have a stronger impact on intention to use the service for organizations in the pre-adoption phase than for organizations in the continued usage phase. H14: The individual customer employee’s perception of trustworthiness of the remote service technician will have a stronger impact on intention to use the service for organizations in the pre-adoption phase than for organizations in the continued usage phase. With regard to the customer employees’ participation beliefs, much of the same reasoning can be applied. The qualitative study shows that a customer employee feels even more assured about his own ability when he frequently experiences a remote service. It is reasonable to assume that in organizations that have already implemented interactive remote services in their business processes, employees have already formed a clearer understanding of their participating role in an interactive remote service. Therefore, it can be assumed that the impact of the customer participation variables on the organizational intention to use interactive remote services will decrease within an increasing experience level. This view is backed by research on habitual individual behavior (Bargh 1989; 1994; Logan 1989; Ouellette and Wood 1998; Triandis 1971; 1980), which states that over time new behav-
146
6. Hypotheses Development
iors become routine and an individual returns to a state of habitual behavior without requiring conscious processing (Bargh et al. 2001; Jasperson, Carter, and Zmund 2005; Kim, Malhotra, and Narasimhan 2005). As a result, interactive remote service usage may occur automatically without the process of establishing associated goals. Therefore, it is assumed that the influence of the employee’s participation beliefs on usage intention will decrease as organizations become more experienced: H15: The individual customer employee’s perception of his role clarity will have a stronger impact on intention to use the services for organizations in the pre-adoption phase than for organizations in the continued usage phase. H16: The individual customer employee’s perception of his role ability will have a stronger impact on intention to use the service for organizations in the pre-adoption phase than for organizations in the continued usage phase. H17: The individual customer employee’s motivation to co-produce with the remote service technician will have a stronger impact on intention to use the service for organizations in the pre-adoption phase than for organizations in the continued usage phase. Subjective norms in terms of normative pressure from supervisors and peers to adopt the innovation reduces the individual risk of adoption and uncertainty because it provides strong reassurance of the legitimacy and appropriateness of the adoption decision (Karahanna, Straub, and Chervany 1999). Since the level of uncertainty declines as the organization gets more experienced with remote services, analogous to the trustworthiness and control assumption, I assume that the perceptions of subjective norms will have a stronger impact on usage intention in organizations in the pre-adoption phase than in organizations that already use remote services: H18: The individual customer employee’s perception of subjective norms will have a stronger impact on intention to use the service for organizations in the preadoption phase than for organizations in the continued usage phase. In line with the findings of the qualitative study are the findings of Klonglan and Coward Jr. (1970) that emphasize that economic variables may be more important in explaining continued usage than adoption. Also, findings from Hu et al. (2009) show that in a group of users of website services, the effect of perceived usefulness on usage intention increases over time. These results are commonly supported by numerous studies in the IS and technology-intensive services field. For example, Davis, Bagozzi, and Warshaw (1989); Karahanna, Straub, and Chervany (1999); Venkatesh et al. (2003); Venkatesh and Davis (2000); Gefen, Karahanna, and Straub (2003b); and Venkatesh (2000) show that the influence of perceived usefulness on intention increases with growing experience. In line with these findings, I propose the following: H19: The individual customer employee’s perception of the usefulness of remote
6.3 Hypotheses Development for Group Comparisons services will have a weaker impact on intention to use the service for organizations in the pre-adoption phase than for organizations in the continued usage phase. All hypotheses are summarized in table 6.1.
147
148
6. Hypotheses Development
Table 6.1: Summary of Hypotheses Direct Effects H
Independent Variable
Dependent Variable
Expected Effect
1 2 3 4 5 6 7 8 9 12
Controllability Trustworthiness Trust in Technology Ease of Use Perceived Usefulness Role Clarity Role Ability Motivation Subjective Norms Intention to use
Intention to Use Intention to Use Perceived Usefulness Perceived Usefulness Intention to use Intention to Use Intention to Use Intention to Use Intention to Use Actual Usage
Positive Positive Positive Positive Positive Positive Positive Positive Positive Positive
Mediating Effect H
Mediator
Path
Expected Effect
5a 5b
Perceived Usefulness Perceived Usefulness
Trust in Technology → Intention Ease of Use → Intention
Mediation Mediation
Moderating Effect H
Mediator
Path
Expected Effect
10a 10b 10c 10d 10e 10f 10g 11a 11b 11c 11d 11e 11f 11g
Company Size Company Size Company Size Company Size Company Size Company Size Company Size Function Function Function Function Function Function Function
Perceived Usefulness → Intention Controllability → Intention Trustworthiness → Intention Role Clarity → Intention Role Ability → Intention Motivation → Intention Subjective Norms → Intention Perceived Usefulness → Intention Controllability → Intention Trustworthiness → Intention Role Clarity → Intention Role Ability → Intention Motivation → Intention Subjective Norms → Intention
Attenuated Attenuated Attenuated Attenuated Attenuated Attenuated Attenuated Strenghened Strenghened Strenghened Strenghened Strenghened Strenghened Strenghened
Group Comparison H
Path
Expected Effect
13 14 15 16 17 18 19
Controllabilty → Intention Trustworthiness → Intention Role Clarity → Intention Role Ability → Intention Motivation → Intention Subjective Norms → Intention Perceived Usefulness → Intention
Stronger for pre-adopters Stronger for pre-adopters Stronger for pre-adopters Stronger for pre-adopters Stronger for pre-adopters Stronger for pre-adopters Weaker for pre-adopters
Legend: H: Hypothesis.
Chapter 7 Quantitative Studies 7.1
Motivation and Goals
In this chapter, I discuss the validation of the ITSUM and present a multi-group comparison to understand organizational intention to use interactive remote services in the pre-adoption phase and in the continuance phase. Moreover, I analyze the link between behavioral intention and actual behavior. The quantitative data offers insights into the relative strength of remote service adoption and continuance drivers as well as their interactions. The research is guided by the following research questions: 1. Do the identified perceptions of a customer’s individual employee affect the organization’s intention to use interactive remote services and, if they do, to what extent? In particular, (a) Do controllability and trustworthiness beliefs regarding the service counterpart influence the organization’s intention to use interactive remote services, and if they do, to what extent? (b) Do usefulness beliefs influence the organization’s intention to use interactive remote services and, if they do, to what extent? (c) Does the perception of technology characteristics, including ease of use and trustworthiness of the technology, influence the organization’s intention to use interactive remote services and, if they do, to what extent? (d) Do participation beliefs including role clarity, role ability, and intrinsic motivation influence the organization’s intention to use interactive remote services and, if they do, to what extent? (e) Do subjective norms influence the organization’s intention to use interactive remote
150
7. Quantitative Studies services and, if they do, to what extent?
2. Does the strength of the proposed antecedents for usage intention differ for organizations in the pre-adoption phase compared to organizations in the continuance phase and, if they do, to what extent? 3. Do organizational characteristics such as company size and the respondent’s function interact with the identified antecedents of usage and, if they do, to what extent? 4. Does the individual perception of organizational intention lead to actual organizational behavior and, if it does, to what extent? The survey methodology, structural equation modeling techniques, the multi-group analysis approach, and ordered logistic regressions will be described in the first part of this chapter. The second part lays out the questionnaire design, operationalization of constructs, and the data collection. The final part presents and discusses the obtained results.
7.2 7.2.1
Methods and Techniques Employed Survey Research
Over the last decades survey research has been one of the most used and vital techniques in social sciences (Malhotra and Birks 2007). Rindfleisch et al. (2008) show that of the 636 empirical articles published in the Journal of Marketing and the Journal of Marketing Research, between 1996 and 2005, approximately 30% use survey methods. Survey research techniques are based upon the use of structured questionnaires given to a sample of a population. They may be classified by mode of administration as telephone interviews, personal interviewing, and mail or electronic interviewing. Biases reduce the accuracy and the quality of the data gained from a survey study. To ensure the validity of survey research techniques and the ability to generalize the findings, researchers such as Podsakoff et al. (2003), Lindell and Whitney (2001), or Rindfleisch et al. (2008) urge colleagues to address potential biases early in the survey design. The following brief discussion of potential biases and guidelines serves as an overview. A detailed examination will be presented in latter sections. Common method variance (CMV) refers to the amount of spurious covariance shared among variables because of the common method used in collecting data (Buckley, Cote, and Comstock 1990). Common method biases are problematic because the actual phenomenon under investigation becomes hard to differentiate from measurement artifacts (Hufnagel and Conca 1994; Malhotra, Kim, and Patil 2006). CMV is caused by a systematic method error due to the use of a single rater or a single source (Rindfleisch et al. 2008).
7.2 Methods and Techniques Employed
151
Cross-sectional survey design in general is prone to potential CMV (Podsakoff et al. 2003) even though Malhotra, Kim, and Patil (2006) find that common method biases in the IS domain are less serious than in other disciplines. Podsakoff et al. (2003, p. 882) identify potential sources of common method bias including common rater effects, item characteristics and context effects, and measurement context effects. The latter refer to the fact that measures of different constructs obtained at the same point in time, at the same location, or with the same medium might produce artificial covariance independent of the content of the constructs themselves. Rater effects, (or response bias) refers to the tendency of the respondent to answer a question in "a particular and unique systematic way that distorts their answers and true thoughts" (Hair Jr., Bush, and Ortinau 2009, p. 240). For example, respondents may give socially desirable responses. Respondent’s tendencies in replying to a survey (e.g., response styles) can also result in CMV. Not only the responses themselves can be biased, but also a systematic non-response bias can occur when the final sample differs from the planned sample (Hair Jr., Bush, and Ortinau 2009, p. 240). Causal inference (CI) describes the ability to infer causation from observed empirical relations. Especially cross-sectional research is widely viewed as being incapable of causal insights (Rindfleisch et al. 2008). In this thesis, I follow the guidelines set forth by Lindell and Whitney (2001), Podsakoff et al. (2003), and Rindfleisch et al. (2008) to minimize potential biases in all phases of the study design, operationalization, and data collection to data analysis. The strategies employed include aiming at a concise questionnaire, multiple waves or reminders, building credibility, offering incentives, assuring anonymity, multiple scale formats, minimal question ambiguity, multiple respondents, multiple periods, multiple data sources, coherence evaluation of causality, and the CMV proxy marker method. The implementation of these guidelines will be addressed in detail in the respective sections on study design (chapter 7.3), operationalization (chapter 7.5), questionnaire design (chapter 7.4), pre-test (chapter 7.6), sample characteristics (chapter 7.7.1), and within data analysis (chapter 7.7.4).
7.2.2
Structural Equation Modeling
7.2.2.1
Methodology
Structural equation modeling (SEM) is a statistical methodology with a confirmatory approach to the analysis of a structural theory and provides a flexible framework for testing complicated models involving latent and observed variables (Byrne 2001, p. 3). It comprises the concept of latent variables in psychometrics, path models in sociology, and structural models in econometrics (e.g., Bollen 1989; Cheung 2008; Jöreskog and Sörbom 1996; Muthen and Muthen 1998-2007). It is a collection of statistical techniques including factor analysis and multiple regression. SEM allows the researcher to examine relationships between several continuous or
152
7. Quantitative Studies
discrete independent variables and several continuous or discrete dependent variables (Malhotra and Birks 2007, p. 604). A full latent variable model comprises a measurement model and a structural model related by a set of linear equations whose unknown coefficients can then be estimated (Bollen 1989, p. 11 ; Malhotra and Birks 2007, p. 604). The measurement model relates to the measurement of latent variables and hypothetical constructs that are not directly observed. It relies on a multivariate regression model that describes the relationships between a set of observed dependent variables and a set of continuous latent variables. The observed variables serve as indicators of the underlying construct that they are presumed to represent (Byrne 2001, p. 3). The observed variables are referred to as factor indicators and the continuous latent variables are referred to as factors. It is important to note that observed variables are exposed to a certain measurement error, which may comprise random error and systematic error (Bagozzi, Yi, and Phillips 1991). The structural model encompasses the relations among the latent variables (Byrne 2001, p. 3) and are the "causal" relations between variables (Bollen 1989, p.11). The structural model describes three types of relationships in one set of multivariate regression equations: the relationship among factors; the relationships among observed variables; and the relationships between factors and observed variables that are not factor indicators. In this thesis, the Mplus notation of structural equation models is used (Muthen 1998-2004). Equation 7.1 describes the measurement model in regard to the p-dimensional latent response variable y∗ . y∗i = ν + Ληi + Kxi + εi
(7.1)
The latent variables (factors or constructs) are denoted by the m-dimensional vector η, x is a q-dimensional vector of independent variables, ε is a p-dimensional vector of residual errors. Means are represented by a p-dimensional vector ν, Λ is a p × m parameter matrix of factor loadings λ , and K is a p × q parameter matrix of regression coefficients. The residual errors are uncorrelated by default. The matrix K allows for direct effects of independent variables x on the latent response variables y. In other formulations this matrix is assumed to be zero. The structural part of the model, equation 7.2, is a regression of the latent variables on each other and the q-dimensional vector x of independent variables. B is an m × m parameter matrix of regression slopes β . Γ is a m × q slope parameter matrix for regression of the latent variables on the independent variables. ζ is the m-dimensional vector of residuals. α is an m-dimensional parameter vector. Additionally it is assumed that B has zero diagonal elements and that [I−B]−1 exists. ηi = α + Bηi + Γxi + ζi
(7.2)
7.2 Methods and Techniques Employed
153
The hypothesized model can be tested statistically in a simultaneous analysis of the entire system of variables to determine the extent to which it is consistent with the data. In the covariance based SEM approach (as chosen in this study), this is achieved via the parameter estimation of the covariance matrix based on the observed measures and the comparison of this estimated covariance matrix with the covariance matrix predicted by the theoretical model. This approach attempts to minimize the difference between the sample’s actual covariances and the theoretical model’s predicted covariances (Chin and Newsted 1999). A model may be estimated by using different estimators. The most widely used fitting function for general structural equation modeling is the maximum likelihood (ML) function (Bollen 1989, p. 8), which is efficient with large samples. If the observed covariance matrix resembles the model’s matrix, the model’s goodness of fit is adequate and indicates plausibility of postulated relations among variables (Byrne 2001, p. 3). Estimation is computationally expensive and usually performed by dedicated software packages on workstations. In this thesis, the software Mplus 5.1 is used, which also allows for distribution-free estimators for continuous variables or weighted/unweighted least squares estimators for categorical variables. It accommodates for continuous, dichotomous, and categorical data; allows for confirmatory factor analysis (CFA), exploratory factor analysis (EFA), SEM, latent class, growth modeling, and Monte Carlo Simulation; and offers a wide range of estimators.
7.2.2.2
Assessment of Reliability and Validity
The quality of a full model is assessed by the reliability and validity of the measurement models as well as the overall model fit. For reflective measurement models, reliability is defined as "the degree to which measures are free from random error and thus reliability coefficients estimate the amount of systematic variance in a measure" (Peter and Churchill Jr. 1986, p. 4). Reliability is established when a good proportion of the variance of the observed variables is explained through the factor (Homburg and Giering 1998, p. 116). Validity is established "...when the differences in observed scores reflect true differences on the characteristics one is attempting to measure and nothing else..." (Churchill Jr. 1979, p. 65). Validity comprises content validity, convergent validity, discriminant validity, and nomological validity. Content validity reflects the degree to which the indicators of a measurement model reflect the semantic meaning of a construct. It can be assessed via qualitative expert opinions on the scales (Homburg and Giering 1998, p. 117). Convergent validity is the degree to "which multiple attempts to measure the same concept are in agreement" (Bagozzi and Phillips 1982, p. 468). The idea is that two or more measures of the same thing should highly covary if they are valid measures of the concept. Discriminant validity is the degree to which measures of different concepts are distinct. The notion is that if two or more concepts are unique, then valid measures
154
7. Quantitative Studies
of each should not correlate too highly (Bagozzi, Yi, and Phillips 1991, p. 425). Nomological validity represents the degree to which predictions based on a concept are confirmed within the context of larger theory, for example through grounding hypothesis development in general adoption and service theory (see chapter 6). The term "first-generation" tests of reliability refers to measures such as factor loadings of the observed variables, explained variance of the observed variables, item-to-total correlation, and Cronbach’s α (Homburg and Giering 1998, p. 119). These measures require interpretation which is often simplified by using cut-off values to assess the measure in question. There is considerable variance on the threshold values employed in literature. Because of this, I avoid extreme opinions and adhere to the most commonly given recommendations: Homburg and Giering (1996, p. 8) consider factors loadings ≥ 0.4 as acceptable; Bearden, Netemeyer, and Teel (1989, p. 475) regard an item-to-total correlation ≥ 0.5 as sufficient; according to Peter (1997, p. 180) the cut-off value for explained variance of an indicator ≥ 0.5 is to be used; Cronbach’s α should be equal to or exceed 0.7 according to Nunally and Bernstein (1978, p. 245). Usually the criteria for convergent validity are assessed within a confirmatory factor analysis. It includes tests for construct reliability (CR; or factor or composite reliability), average variance extracted (AVE), factor determinacy (FD), and significance test of the factor loadings. The thresholds used in this thesis to assume convergent validity are: construct reliability≥ 0.6 as stated by Bagozzi and Yi (1988, p. 82); factor determinacy should be near one (Muthen and Muthen 1998-2007, p. 586); and the AVE should be equal to or greater than 0.4 (Bagozzi and Baumgartner 1994, p. 402) to indicate that at least 40% of the variance in a construct is due to the hypothesized underlying items. The Fornell-Larcker-criterium can be used to judge discriminant validity (Fornell and Larcker 1981, p. 46). If the average variances extracted by the correlated latent variables is greater than the square of the correlation between the latent variables then discriminant validity requirement is satisfied.
7.2.2.3
Assessment of Model Fit and Data Quality
For assessing the overall model fit, several ways to measure goodness-of-fit measures are available (Hu and Bentler 1999; Marsh, Hau, and Wen 2004). The most popular ways to evaluate fit are those that involve the χ 2 -goodness-of-fit statistic. The χ 2 -statistic is sensitive to sample size (n > 200) and distributional misspecification; therefore, some researchers recommend supplementing the χ 2 -test statistic with alternative measures of absolute and incremental fit (Bentler and Bonett 1980). Absolute fit indices assess how well an a-priori model reproduces the sample data, e.g., the Root Mean Squared Error of Approximation (RMSEA) and the Standardized Root Mean Squared Residual (SRMR). Incremental fit indices measure the proportionate improvement in fit by
7.2 Methods and Techniques Employed
155
comparing a target model with a more restricted, nested baseline model, e.g., the Tucker-LewisIndex (TLI), Comparative Fit Index (CFI) for the ML method. The CFI is one of the most reliable incremental fit indices and it is the most reported measure of fit in the literature (McDonald and Ho 2002). The CFI avoids TLI’s problems concerning underestimation of fit and considerable sampling variability in small samples (Yu 2002). For model comparison, the Akaike Information Criterion (AIC) serves as an indicator for a better fit. The absolute value of AIC has relatively little meaning, rather the focus is on its relative size. The rationale is that if two models are compared then the model with the smaller AIC should be preferred (Bühner 2006, p. 352). This thesis uses the following values recommended by literature to assess model fit: ratio χ 2 /d f ≤ 3.0 (Homburg and Giering 1996, p. 13); RMSEA ≤ 0.06 (Hu and Bentler 1999, p. 1); SRMR ≤ 0.08 (Hu and Bentler 1999, p. 1). The incremental fit thresholds to assume model fit are: TLI ≥ 0.9 (Janssens et al. 2008, p. 296) and CFI ≥ 0.9 (Janssens et al. 2008, p. 296). Multicollinearity describes a state of high inter-correlations among the latent exogenous constructs (Grewal, Cote, and Baumgartner 2004; Hair Jr., Bush, and Ortinau 2009). Multicollinearity can result in several problems such as SEM estimates far from the true parameters and large standard errors of the estimates (Jagpal 1982). Criteria for assessing multicollinearity are the variance inflation factor (VIF) and the tolerance value. Rule of thumbs recommend values under 4.0 for VIF and values above 0.10 for the tolerance statistic (Moosmüller 2004, p. 131; Myers 1993, p. 369). Grewal, Cote, and Baumgartner (2004, p. 526) show that SEM estimations should be doubted when multicollinearity is extremely high (correlation > 0.8). Even when multicollinearity is between 0.4 and 0.8, one should be cautious if composite reliability is low ( 0.10. The logit coefficient for a binary dependent variable can be directly interpreted as the proportional change in the log-odds given a unit change in the independent variable.
7.2.2.5
Multi-Group Comparison
This thesis aims at understanding the heterogeneity between respondents and validating the ITSUM in a multi-group framework. An essential prerequisite to gain valid insights into group differences is to establish an invariant measurement model. Differences in observed measures can only be attributed to differences between groups, if the measures of the ITSUM are equivalent across the groups. The general hierarchical procedure to assess measurement invariance as proposed by Steenkamp and Baumgartner (1998a) is illustrated in figure 7.1. The level of invariance required depends on the goal of the study. For examining structural relationships with other constructs across groups, full or partial metric invariance has to be established. Scalar invariance is required if
7.2 Methods and Techniques Employed
157
Figure 7.1: Proposed Procedure for Assessing Measurement Invariance Source: Own Illustration, based on Steenkamp and Baumgartner (1998a, p. 83)
158
7. Quantitative Studies
the comparison of absolute means across the groups is in the focus of a study. Otherwise this step can be omitted from the procedure (Steenkamp and Baumgartner 1998b, p. 414). If the researcher wants to compare standardized measures of association such as correlation coefficients and standardized regression coefficients across groups, factor variance invariance is required in addition to partial metric invariance. Lack of error variance invariance does not create a problem as long as differences in measurement error are explicitly taken into account, for example through latent variable modeling (Steenkamp and Baumgartner 1998a). The following forms of measurement invariance are necessary for comparing path coefficients: 1. Configural invariance is the most basic level of invariance. It is achieved if the pattern of salient (non-zero) and nonsalient (zero) factor loadings is equal across groups and the model of interest fits across the groups. Although the model is the same across groups, the unknown parameters of the model are assumed to be different across the groups. 2. Metric invariance requires a stronger test of invariance compared to configural invariance. Loadings are constrained to be the same across groups, which implies equal metrics or scale intervals across groups. If an item satisfies the requirement of metric invariance, different scores on the item can be meaningfully compared across groups, and these observed item differences are indicative of similar cross-group differences in the underlying construct. 3. Scalar invariance is required if a mean comparison across groups is meant to be meaningful. Even if metric invariance is satisfied, scores on the items can be biased. By constraining the vector of item intercepts across groups, equality of measurement intercepts is achieved. 4. Factor variance and covariance invariance is required in addition to metric invariance if one wants to compare standardized regression coefficients across groups. If both the factor variance and the covariances are invariant, the correlations between the latent constructs are invariant. In addition to full invariance tests, individual constraints on different parameters can be relaxed to explore for subtle group differences leading to partial invariance assumptions (Byrne, Shavelson, and Muthen 1989). The assessment of configural, metric, scalar, and factor (co)variance invariance as well as individual parameter tests are conducted based on χ 2 -difference tests that compares less restricted models to progressively more restricted models (see chapter 7.8.2). If the models are not significantly different, invariance of the models in the constrained parameters is assumed. In addition to the χ 2 -difference test, Steenkamp and Baumgartner (1998a) suggest to include the CFI, TLI, AIC, and the RMSEA in assessment of the nested models. Steenkamp and Baumgartner (1998a; 1998b), and van Birgelen et al. (2002) showed, that the assessment of the fit indices can outweight significant χ 2 -difference test results when large sample sizes are estimated(Steenkamp and Baumgartner 1998a) and form the basis for invariance
7.3 Study Design
159
assessment.
7.3
Study Design
The study aims to measure interactive remote service usage intention and actual usage within the German printing industry. The study is split into two parts, with surveys at two different points in time. The first part entails the survey conducted in May and June 2008 and contains the full measurement instrument to validate the ITSUM derived in chapter 6. It will be referred to as the t1 -study in the further course of this thesis. The second part is a follow-up survey and was conducted in March 2009. It comprises a shortened questionnaire with a focus on the respondent’s usage behavior displayed since June 2008 to explore the relationship between intention and behavior. This study will be called the t2 -study. Addresses and electronic contact details could be obtained for 7,139 companies from the printing industry in Germany. The 7,139 companies represent over 65% of all printing companies in Germany according to numbers published by the Bundesverband der Druck und Medienunternehmen (BVDM) – the German Printing and Media Industries Federation (BVDM 2008). The survey phase of the t1 -study started with a postal announcement sent out to 4,000 randomly chosen companies to increase response rate (May 2008). Also an electronic email invitation to the survey was sent to all 7,139 companies two weeks later (June 2008). In both cases the respondents could choose to participate electronically through an electronic survey system or through a traditional fill-in questionnaire to be sent in by traditional mail. A professional internet based online survey software (EFS Survey) was used for the electronic data collection. The electronic survey system’s URL was personalized for each respondent so he could be identified for the t2 -study. The questionnaire was accessible from May 15, 2008 – July 8, 2008 (through http://www.unipark.de/uc/Drucker/40ed/individualkey). The last written questionnaire sent by traditional mail came in on the August 4, 2008. The invitations were personalized and targeted at those individuals within the organizations who have the most influence on the decision to use remote services and, at the same time, are involved in operational processes. The printing industry in Germany is dominated by small companies, where the owner/general manager fulfills a multitude of functional roles simultaneously and often even operates the machines himself. The owner/general manager was targeted in small companies while in mid-sized or larger companies the production manager was addressed. The specific structure of the industry suggests a strong link between the targeted individual’s perceptions and organizational decision making.16 If no name was available, the survey was sent to the companies main address with the pledge to forward the mail to a person who is responsible for decisions with regard to the purchase, usage, or discontinuance of main16
For further details, see the discussion on the empirical setting in chapter 4.
160
7. Quantitative Studies
tenance services and other printing machine related services. Only one respondent was targeted per organization and identified across the t1 and t2 -studies. The t2 -study — conducted in March 2009 — was only sent to respondents, who participated in the first survey in June 2008 and whom completely filled-in the survey and agreed to be contacted for the second survey study. In total, an invitation to the follow-up survey was sent to 567 companies. The questionnaire was accessible from March 1, 2009 – April 1, 2009. The main goal of this follow-up survey was to explore the actual behavior of the organizations — the usage of interactive remote services — after a time lag and relate it to the perceptions and intention variables of the same respondent in the t1 -study.
7.4
General Outline of the Questionnaires
The questionnaires were developed to comprise a specific set of questions that operationalize and measure the constructs proposed. They ultimately serve to test the hypotheses of the ITSUM (see chapter 6) and allow the collection of demographical data and characteristics of the respondents and the company he presents. The questionnaires at t1 and t2 were developed according to Malhotra and Birks’ (2007, pp. 375) guidelines for questionnaire design. The opening questions are supposed to raise interest and be simple and non-threatening. A description of a typical interactive remote service is presented at the beginning of the questionnaire. The opening question refers to this example and enquires whether the respondent has ever had experience with this kind of service. Difficult, sensitive, or complex questions are placed late in the sequence. The final questions cover personal data. The first page (see appendix, figure A.1 and figure A.2) of both surveys provided information on the author and the institution conducting the survey, as well as the reason for approaching the potential respondent, the aim of the research, the approximated duration for completing the questionnaire, and an expression of gratitude in advance for completing the questionnaire. Also, the respondents were reassured that: • All answers, no matter if the respondents had extensive, little, or even no experiences with interactive remote services, were a valuable contribution to the research; • There were no right or wrong answers; and • All data would be analyzed anonymously. Response rates can be increased by offering incentives (Malhotra and Birks 2007, p. 446; Hair Jr., Bush, and Ortinau 2009, p. 265). Because of this, a lottery to win a portable electronic music device and shopping coupons were offered on the first page of the survey. If the
7.5 Operationalization of the Constructs
161
respondent wished to participate in this lottery, he was asked to fill in his contact information at the end of the survey. In both cases anonymity was assured.
7.5
Operationalization of the Constructs
The questionnaire of the t1 -survey aims at capturing the constructs of the ITSUM. Seven-point scales were used for all of the constructs’ measurements, with 1 being the positive end of the scale and 7 being the negative end of the scale. Controllability was measured with seven categories within a semantic differential; the items of the intention scale were measured with sevenpoint-Likert-scales with the end points of 1 indicating "very likely" and 7 indicating "very unlikely." The other constructs were measured with seven-point-Likert scales verbally anchored with 1 equaling "I strongly agree" and 7 meaning "I strongly disagree." The items for the latent variables and their origins are shown in table 7.1 and will be discussed below. Table 7.1: Operationalization of the Constructs Perceived Controllability Please express how you, as a customer, would feel about the behavior of the RST. That is, the behaviors of the RST make you feel: Item No.
Items
adapted from
CONT1
controlling – controlled dominant – submissive influential – influenced staying on top of things – kept in the dark confident – helpless
Poon et al. (2004) Poon et al. (2004) Poon et al. (2004) Qualitative Study Poon et al. (2004)
CONT2 CONT3 CONT4
—a
Trustworthiness of the RST How would you evaluate the following statements regarding the RST? Item No.
Items
adapted from
TW1
An RST is experienced in performing necessary tasks during a remote service. I expect our needs and wishes to be important to an RST. The intentions of an RST are benevolent. An RST is competent. I do not doubt the honesty of an RST.
Gefen (2002)
TW2 TW3 TW4 TW5
to be continued on the next page. . .
Gefen (2002) Gefen (2002) Gefen (2002) Gefen (2002)
162
7. Quantitative Studies
Perceived Ease Of Use How strongly do you agree with the following statements about remote technology? Item No.
Items
adapted from
EOU1
Learning how to use and operate a remote service would be easy. I think the interaction with the remote service technology would be clear and understandable. I would find the remote service technology to be flexible to interact with the service provider company.
Davis (1989)
EOU2
—b
Davis (1989) Davis (1989)
Perceived Usefulness How do you evaluate the advantages and disadvantages of remote services compared to face-to-face delivered services? Item No.
Items
adapted from
PU1
Remote services provide greater convenience than face-toface delivered services. Remote services save time. Remote services make it easier for us to get the services we want. Remote services improve service quality. Using remote services enhances our effectivity. Remote services are useful for my individual job tasks.
Qualitative Study
PU2 PU3
PU4 PU5
—b
Davis (1989) Davis (1989) Davis et al. (1989) Davis (1989) Davis (1989)
Role Clarity Please focus on the collaboration with the RST. How strongly do you agree with the following statements? Item No.
Items
adapted from
RC1
I feel certain about how to effectively collaborate during a remote service. The steps in the process of using a remote service are clear to me. I know what is expected of me in a remote service situation. I believe there are clear directions regarding how to collaborate during a remote service.
Meuter et al. (2005)
RC2
RC3
—a
Meuter et al. (2005) Meuter et al. (2005) Meuter et al. (2005)
Role Ability Please focus on the collaboration with the RST. How strongly do you agree with the following statements? Item No.
Items
adapted from
RA1
I am fully capable of supporting the RST during a remote service. I am confident in my ability to cooperate during a remote service. I feel I am qualified for the task of collaborating with the RST.
Meuter et al. (2005)
RA2
RA3
to be continued on the next page. . .
Meuter et al. (2005) Meuter et al. (2005)
7.5 Operationalization of the Constructs
163
Intrinsic Motivation The collaboration with the RST) during a remote service would . . . Item No.
Items
adapted from
MOTIV1
. . . provide me with personal feelings of appreciation of my knowledge. . . . allow me to have increased confidence in my skills. . . . allow me to feel innovative in how I interact with a service provider . . . provide me with feelings of worthwhile shared accomplishment.
Qualitative Study
MOTIV2 MOTIV3
MOTIV4
Meuter et al. (2005) Meuter et al. (2005) Meuter et al. (2005)
Intention to Use Remote Services Please rate the likeliness of your future activities below. Item No.
Items
adapted from
INT1
We intend to use remote services in the next 6 months. We plan to use remote services in case of the next emergency incident.
Venkatesh et al. (2003) Venkatesh et al. (2003)
INT2
Legend: a : deleted during pre-test; b : deleted during scale improvement.
In practice, the term "remote services" and "interactive remote services" are used rather freely. There are no clear cut definitions used throughout the industry, and the understanding of remote services even varies from individual to individual. For this reason and to reduce other sources of ambiguity and confusion to the respondents, I presented an example at the beginning of the questionnaire to illustrate interactive remote services and provide a common reference base. For the same reason, the term "remote services" was used, instead of the technically more accurate term "interactive remote services" throughout the survey. The following explanation of interactive remote services was given to the respondents: When performing a remote service an employee of the service provider (remote service technician) logs onto the customer’s printing machine remotely through the internet or a modem connection. During a remote diagnosis and repair of a printing machine, the remote service technician remotely monitors and might even change certain parameters. The customer has to collaborate with the remote service technician in order to perform a remote repair. This collaboration includes mechanical tasks, for example the opening and changing of parts, which cannot be performed remotely by the remote service technician. In the following, you will be asked about your personal experience as a customer of this kind of service regardless of specific provider companies. For clarification throughout the questionnaire, this example was available to participants who were unfamiliar with this type of remote service. The questions either directly referred to this
164
7. Quantitative Studies
example, stressing the interactive and collaborative characteristics of the scenario, or to comparable experiences that the customers might have had. The survey was mostly sent to business addresses where it was likely to be answered by employees at their place of work. To reach a maximum fill-in rate and to avoid aborters, especially due to respondents’ limited time at work, I aimed for a parsimonious, concise, and not too timeconsuming questionnaire. Accordingly, original scales with six or more reflective items were reduced to four or five item scales, if possible, and if the omission did not change the meaning of the latent variable. Bergkvist and Rossiter (2007) and Rossiter (2002) state that "when in the mind" of the survey respondents 1) the object of the construct is "concrete singular", meaning that it consists of one object that is easily and uniformly imagined, and 2) the attribute of the construct is concrete, again meaning that it is easily and uniformly imagined, a measurement with multiple reflective items is not necessary and even single items can suffice. Perceived controllability (CONT) of the RST’s actions was measured by adapting the seven seven-point semantic differential scales from Poon, Hui, and Au (2004) based on the "dominance" scale of Mehrabian and Russell (1974, appendix B) and the "helplessness" concept of Glass and Singer (1972). The items were adapted to measure the expected perceived control felt by customer employees with respect to the behavior and actions of the RST during a remote service. The semantic differential item "staying on top of things" to "kept in the dark" was newly added to the scale based on the findings of the qualitative study (see chapter 5.4.5). The trustworthiness perception (TW) of an RST scale was conceived as a single construct combining beliefs in ability, integrity, and benevolence. Trustworthiness was measured as a single scale according to the studies of Gefen (2002a) and Jarvenpaa, Tractinsky, and Vitale (1999). This thesis follows this approach and measures the perception of trustworthiness of the RST as a latent variable with five items stemming from Gefen’s (2002) seven-items scale. The items were adapted with regard to the context of the trustworthiness of an interaction partner in an interactive remote service. Ease of use (EOU) was measured using three items from Davis’ (1989) scale for perceived ease of use. The wording of all items was adapted to the remote service setting. Perceived usefulness (PU) was measured using four items from Davis’ (1989) scale for perceived usefulness. To adapt the scale to the special characteristics of remote services one item was added from Davis, Bagozzi, and Warshaw’s (1989) perceived usefulness scale. Also, one item was newly developed based on the results from the qualitative study: "Remote services provide greater convenience than face-to-face-services" (see chapter 5). The wording of all items was adapted to the remote service setting and contrasted face-to-face services. The construct of role clarity (RC) was measured by adapting Meuter et al.’ (2005) scale used to study a consumer’s trial behavior of self-service technology. All four items from the scale were applied and adapted to the remote services context. Also, role ability (RA) was measured with
7.5 Operationalization of the Constructs
165
three items stemming from the Meuter et al.’ (2005) scale. The items were selected with respect to their applicability in the interactive remote service scenario and verbally adapted to capture the customer’s perceived ability to collaborate in a remote service situation. Intrinsic motivation was measured with three items also stemming from Meuter et al.’ (2005) instrumentality scale for intrinsic motivation. Based on the qualitative interviews, the scale was extended with one item referring to the "feelings of appreciation" of the customer’s knowledge (see chapter 5.4.8)
The organization’s intention to use interactive remote services (INT) was measured by using the individual employee’s perception of the organization’s intention as a proxy. This is done under the assumption that in the printing industry, the decision is made by few people, mostly the owner, general managers or production manager. Intention was measured with a two-item scale adapted from Venkatesh et al. (2003).
Trust in technology (TT), subjective norms (SN), and organizational remote service experience were measured by single questions. This was done to cut back the time needed for respondents to answer the questionnaire while at their place of business. Trust in technology was measured using a single-item — "In general, remote service technology is trustworthy." from the Johnson, Bardhi, and Dunn (2008) scale. Subjective norms were measured with the item — "My peers or superiors expect me to use remote services." — referring to the subjective norms scales used in studies on IT-adoption such as in Venkatesh and Davis (2000), but adapted for the B2Bsetting.
In addition, respondents were asked to specify the number of employees and their own function within the company. The number of employees is used as a categorical variable indicating company size as done in Premkumar and Roberts (1999). The categories of company sizes were chosen according to the BVDM as 1–9, 10–19, 20–49, 50–99, 100–499, 500–999, >1000 employees (BVDM 2008).
To measure the level of remote service implementation in the organization according to Rogers’ (2003) classification of innovation adoption in organizations the single-item "We have already strongly implemented remote services in our business processes." was used. In the t2 -study, the organizational usage behavior was measured with the question: "How often did your company use remote services since the last questionnaire in June 2008?" The respondents could choose between the following categories: "never", "once", "2–3 times", or "four and more times." For the t2 -study, the same respondents, who answered the t1 -study were contacted to capture the behavior of the organization since the completion of the t1 -survey.
166
7.6
7. Quantitative Studies
Quality of the Questionnaire and Pre-Test
A multistep procedure was employed to ensure the quality of the t1 and t2 questionnaires’. It includes the following steps: 1. Established and existing scales for most of the constructs were adapted for the survey instrument. They were extended, if necessary, based on the results of the qualitative study (see chapter 7.5). 2. The survey was reviewed by two employees of a remote service provider company and five experts in the field of remote service research to verify content validity, applicability, and linguistic clarity of all constructs and items. 3. The survey was pre-tested with a convenience sample of ten employees from printing companies to assess its clarity. The pre-test was conducted in early May 2008 to identify and eliminate problems with the questionnaire. Ten employees from printing companies in Germany were asked to fill-in the online questionnaire and to check for wording, comprehensibility, logical flow, and overall gestalt of the questionnaire. During this process, the wording and ambiguous questions were clarified to reach maximum comprehensibility of the target respondents, as required by Rossiter (2002). Adjustments were made to the controllability and role clarity constructs. One item in each construct was eliminated due to ambiguous interpretations of the pre-testers (see table 7.4). The survey design was chosen carefully to address and prevent potential biases. Several of the procedural remedies that Podsakoff and colleagues (2003, pp. 887) recommend were used: 1. The measurement of the behavioral intention and actual behavior was temporally separated by introducing a time lag to avoid measurement context effects and response bias; 2. Different scale formats were used such as semantic differential and Likert-scales to avoid common rater effects; 3. Respondents anonymity was protected and evaluation apprehension was avoided through the assurances that each survey would be submitted anonymously and that no identifying information would be collected in the survey (to address response bias); 4. Respondents were assured that no particular answer was encouraged or discouraged, i.e. there are no right or wrong answers to these questions (to adress response bias); 5. Careful wording was used and pre-tests of the questionnaires were conducted to reduce method bias; and 6. Scales were improved by eliminating ambiguous items (e.g., complicated syntax).
7.7 t1 -Study: Results of ITSUM Validation
167
To further control for common method bias, a marker variable (Lindell and Whitney 2001) was used in the analysis of the model.
7.7 7.7.1
t1-Study: Results of ITSUM Validation Sample Structure and Description
The t1 -survey was sent out to 7,139 companies in the German printing industry. The questionnaire was answered by 1,076 participants reflecting a response rate of 15.07%. From the sample with 1,076 cases, all responses were selected that met the following conditions: All intention items had to be fully filled-in, all organizational characteristics had to be fully filled-in, and the percentage of missing values per case had to be lower than 5%. Based on these constraints, 359 cases were deleted from the sample resulting in a sample of 717 questionnaires and an effective response rate of 10%. This is somewhat low overall, but not unusual for B2B-research of comparable industries (Rauyruen and Miller 2007). Responses from organizations who declined to participate revealed that a lack of time and business pressures were the main reason for not completing the survey. In the following text, the sample of 717 respondents is referred to as the "overall sample." The distribution of companies of different sizes within the sample and the distribution of companies of different sizes in the industry according to the official industry statistics (BVDM 2008) is illustrated in figure 7.2. The categories on the x-axis reflect the number of employees; the dark bars denote the percentages in the industry, and the light bars show the percentages in the sample. Companies with 19 employees and less form the majority of the sample with 67.09% — as expected from the industry structure. SME represent over 98% of all companies sampled. The overall sample distribution according the company size reflects the distribution of the whole industry fairly well, even though larger companies are slightly over represented. The gender distribution of the overall sample respondents includes 87% male and 12% female. Distribution of age groups indicates that majority of respondents are between 41–60 years old (see figure 7.3). The respondents classified their respective organization according to its business segment. They were able to choose between pre-press, press, and post-press. Multiple answers were possible. The self-reported classification is shown in figure 7.4. Within the sample, 86% of 717 companies reported themselves operating in the core segment — press — of the printing industry. The functions of the respondents are illustrated in figure 7.5. Within the overall sample, over 90% of all respondents are owners, general managers, or production managers. These are the key actors who decide whether a remote service is used in a company (Herdler 2006).
168
7. Quantitative Studies
Figure 7.2: Company Size (Number of Employees) in the Printing Industry vs. in the Overall Sample in %
Figure 7.3: Age Distribution in the Overall Sample in % (n=717)
7.7 t1 -Study: Results of ITSUM Validation
169
Figure 7.4: Self-Reported Classification of Business Segments in % (multiple answers possible)
$
!$
$
$
$
$
!$ $
$
Figure 7.5: Distribution of Respondents’ Function in the Sample in % (n=717))
170
7. Quantitative Studies
Crosstable 7.2 shows the distribution of the respondent’s function over company size. In the smaller companies, the general manager is more often the respondent when compared to larger companies. The sample is divided in to an early and a late group according to the point in time they answered the questionnaires. The sample is split at June 10, 2008 and a test was performed on the two groups to check for a potential non-response bias. The results of the χ 2 -tests17 indicated no response bias in terms of respondent’s function (χ 2 (2) = 4.341), respondent’s age (χ 2 (2) = 3.468), and gender of the respondent (χ 2 (1) = 1.141). A further examination of the validity across time will be conducted in the t2 -study (see chapter 7.9). Although the overall targeting was successful, further validation is needed to reduce sources of heterogeneity. Tests for the influence of company size, function of the respondent, and expertise of the respondent (EXP) will be employed further in data analysis.
7.7.2
Data Quality
The data was checked for sufficient sample size, missing values, and normality of data distribution. The overall sample of 717 respondents consists of 33 variables per case, resulting in a 22:1 (cases:variables) ratio that signals a sufficient sample size according to Bentler and Bonett (1980) and Hu and Bentler (1999). For the constructs used to estimate the ITSUM, the number of missing values for the 717 cases were very low (< 4% per variable). They were imputed with an expectation-maximizationbayesian (EMB) algorithm provided by AMELIA software (Honaker, King, and Blackwell 2007). This software allows users to appropriately impute incomplete data sets so that estimations, requirering complete observations can use all the information present in a dataset with missing cases. It also allows researchers to avoid biases and inefficiencies that can result from dropping all partially observed observations from the analysis. Multivariate normality is required by ML estimation, which is the predominant method in SEM for estimating structure (path) coefficients (Bollen 1989, p. 8). Multivariate normality is the assumption that all variables and all combinations of variables are normally distributed. In a first step, univariate normality was assessed by inspecting skewness and kurtosis values. SkewTable 7.2: Crosstable: Respondent’s Function / Number of Employees
Owner / GM Production Manager Machine Operator Others 17
1–9
10–19
20–49
50–99
100–499
500–999
>1000
263 41 0 21
121 27 1 7
73 21 2 17
48 12 0 8
15 18 0 11
3 1 0 3
0 3 0 1
Categories with less than 5 cases were not included in the tests.
7.7 t1 -Study: Results of ITSUM Validation
171
ness values for the ITSUM items range from -0.05 up to 1.62, kurtosis values range from -1.4 up to 2.23 and, thus, lay under the cut-off values of 2.0 for skewness and 7.0 for kurtosis recommended by Finch, West, and MacKinnon (1997). To further assess multivariate normality Mardia’s statistic (Mardia, Kent, and Bibby 1979, p. 21) was used to assess the joint multivariate normality of the collection of ITSUM items. The average multivariate kurtosis value across the sample is 837.697 (s.d. = 3.033), returning a Z-test value of 1124.407 (ρ < 0.0001). The average multivariate skewness value across the sample is 33.892 (s.d. = 0.87), returning a Z-test value of 114.271 (ρ < 0.0001). Because the null hypothesis of both tests is that the data arise from a joint multivariate normal population distribution, these results suggest non-normality of the ITSUM items’ distributions. Therefore, for the confirmatory factor analysis and the structural equation modeling, the Mplus MLM estimator was chosen. The MLM estimator delivers maximum likelihood parameter estimates with standard errors and a mean-adjusted χ 2 -test statistic that is robust to non-normality (Muthen and Muthen 1998-2007, p. 484). The MLM χ 2 -statistics in MPlus equals the Satorra-Bentler-χ 2 , which adjusts the ML χ 2 estimate for kurtosis (Satorra and Bentler 2001). In the Satorra-Bentler χ 2 -statistic, the usual normal-theory χ 2 -statistic is divided by a scaling correction to better approximate χ 2 under non-normality. The Satorra-Bentler χ 2 -statistic has been shown to outperform other asymptotic robust test statistics in nearly all conditions of sample size and distribution (Chou, Bentler, and Satorra 1991). Accordingly, model fit indices that depend on χ 2 , the statistic will be scaled.
7.7.3
Measurement Validity
To assess the general validity of the measurement model, an EFA, a CFA and reliability tests were conducted on the overall sample. The 30 variables that form the eight latent variables are measured on a seven-point Likert-scale, which, strictly speaking, have an ordinal measurement level. Yet, in practice they are often treated as being interval-scaled because of the assumption of equal appearing intervals (Janssens et al. 2008, p. 255). The Bartlett’s test of sphericity could not support the hypothesis that the variables in the ITSUM are uncorrelated (ρ < 0.000). Also, the "Kaiser-Meyer-Olkin measure of sampling adequacy" statistic equals 0.908 and indicates a sufficient correlation because it is over the recommended cut-off value of 0.5 (Janssens et al. 2008, p.255). Both criteria indicate that a factor analysis is meaningful. A principal component factor analysis was carried out to confirm the expected factor structure of the ITSUM (see structure matrix of the EFA in Appendix, figure A.3). Because of crossloadings (> 0.5) on different factors, two variables, one from the ease of use scale and one from the perceived usefulness scale, were omitted from the study leaving 28 dependent variables assigned to 8 latent variables in the ITSUM. As a result, all variables of the ITSUM show significant and strong loadings (> 0.7) on their respective factors. The measurement model with eight latent factors was confirmed in the exploratory factor analy-
172
7. Quantitative Studies
sis. It contains the factors trustworthiness, controllability, perceived usefulness, perceived ease of use, role clarity, role ability, intrinsic motivation, and intention to use remote services. In line with the work of Anderson and Gerbing (1988), the confirmatory factor analysis corroborates the convergent validity and reliability for the scale items of all constructs. The fit statistics for the eight-factor measurement model (χ 2 (322) = 812.158) indicate that the measurement model fits the data well with a χ 2 /d f < 3.0 and CFI, TLI, RMSEA, and SRMR values well above the recommended cut-off criteria as shown in table 7.3. The χ 2 -statistic is significant (ρ < 0.001), but this was expected because of the sensitivity of the χ 2 -statistic to large sample sizes (Marsh, Balla, and McDonald 1988). Complete results of the confirmatory factor analysis, including factor loadings, Cronbach’s α coefficients, composite reliability, factor determinacy and AVE values are provided for each variable (item) in table 7.4 and discussed in the following section. In addition, multicollinearity statistics are presented. All measures included in the analysis are reliable. All loadings on hypothesized factors are highly significant (p 2.4, at least at ρ < 0.05). As Lindell and Whitney (2001, p. 18) note, "this suggests that the results cannot be accounted for by CMV." Similar to the approach of Agustin and Singh (2005), Grayson, Johnson, and Chen (2008), and Malhotra, Kim, and Patil (2006), the proxy for CMV was included in the structural equations analysis (CMV PROXY). Incorporating this factor in a structural equation model offers the advantage of accounting for measurement error and hypothesized structural relationships in 19
The variable was included for purposes outside this thesis.
7.7 t1 -Study: Results of ITSUM Validation
175
Table 7.5: Correlation of ITSUM Variables (n=717)
PU CONT RC RA TW INT MOTIV EOU SN TT EXP
PU
CONT
RC
RA
TW
INT
MOTIV
EOU
SN
1.000 0.242 0.283 0.255 0.325 0.410 0.249 0.373 0.209 0.584 0.123
1.000 0.442 0.287 0.356 0.317 0.371 0.223 0.221 0.402 0.208
1.000 0.720 0.572 0.483 0.485 0.499 0.365 0.400 0.438
1.000 0.433 0.261 0.459 0.521 0.206 0.340 0.300
1.000 0.445 0.604 0.377 0.342 0.516 0.220
1.000 0.249 0.308 0.578 0.408 0.397
1.000 0.261 0.288 0.404 0.123
1.000 0.227 0.427 0.229
1.000 0.336 0.402
TT
1.000 0.169
EXP
1.000
addition to CMV, a benefit that the matrix adjustment approach does not provide (Malhotra, Kim, and Patil 2006).
7.7.5
Validation of the ITSUM (n=717)
The general validity of the ITSUM is assessed using the overall sample of 717 cases. The SEM was estimated using the MLM estimator in Mplus 5.1, which yielded the results shown in table 7.6. The path diagram with all path estimates β is illustrated in figure 7.6. The CFI is at a satisfying level of 0.928. The TLI is at a level of 0.914. Both values indicate a good model fit. The RMSEA is fairly good at 0.051 and the SRMR is well under the cut-off value of 0.08 with 0.073 (Hu and Bentler 1999). The χ 2 /d f ratio is 2.877, indicative of a good fit. One can conclude that the data fits the research model ITSUM, judging by the fit indicators. All relationships in the ITSUM, except the effect of control on intention to use, are strong and significant. Six of nine hypotheses regarding the direct effects of the ITSUM are supported. Trustworthiness has a strong significant effect on intention to use, which supports H2. Also, the effects of ease of use and trust in technology on perceived usefulness are significant and strong (β = 0.152, ρ < 0.01; β = 0.349, ρ < 0.001) supporting H4 and H3. As postulated, perceived usefulness is a significant predictor of usage intention which supports hypothesis H5. In addition, role clarity has the strongest significant effect on intention (β = 0.259, ρ < 0.001), supporting H6. Role ability and motivation have significant negative effects on intention, therefore H7 and H8 could not be supported for the overall sample. Also, the effect of controllability on intention is not significant, meaning that H1 cannot be supported for the ITSUM model for the full sample of 717 cases. Path analysis reveals that subjective norms has a significant effect of β = 0.168 (ρ < 0.001) on intention, thus, H9 is supported. Overall, the antecedents explain 52% of the variance of intention to use interactive remote services. To test the mediated relationship between trust in technology and ease of use on perceived usefulness and perceived usefulness on intention (H5a and H5b), the product of coefficients
176
7. Quantitative Studies
Table 7.6: Results of the ITSUM (n=717) Direct Effects H
Independent Variable
Dependent Variable
Expected Effect
Standardized Path Coefficients
ρ-value
1 2 3
Controllability Trustworthiness Trust in Technology Ease of Use
Intention to use Intention to use Perceived Usefulness Perceived Usefulness Intention to use
Positive Positive Positive
0.060 (0.043) 0.180 (0.050) 0.349 (0.025)
0.160 0.000 0.000
n. s. supported supported
Positive
0.152 (0.044)
0.001
supported
Positive
0.237 (0.028)
0.000
supported
Intention to use Intention to use Intention to use Intention to use
Positive Positive Positive Positive
(0.064) (0.045) (0.045) (0.015)
0.000 0.007 0.005 0.000
supported n. s. n. s. supported
0.001 (0.021) 0.056 (0.019)
0.977 0.003
4 5 6 7 8 9 C C
Perceived Usefulness Role Clarity Role Ability Motivation Subjective Norms CMV Proxy
0.259 −0.131 −0.128 0.168
Intention to use Intention to use
EXP
Result
Model Fit Statistics (nn = 7 1 7 ) χ 2 (df) ratio
χ 2/df
SCALING
1185.385 (412) 2.877 1.197
RMSEA SRMR
r2 (INT)
0.051 0.073 0.517
CFI TLI
r2 (PU)
0.928 0.914 0.359
Legend: Values in parentheses are standard errors if not otherwise denoted; C: Control Variable; H: Hypothesis; n. s.: not supported.
!
!
!
'*%"(&)
!
**
Figure 7.6: Results of the ITSUM (n=717)
$+%"%( $$+%"%& $$+%"%%& ""*#
7.7 t1 -Study: Results of ITSUM Validation
177
method was applied (MacKinnon et al. 2002). Indirect effects, their standard errors, and a test of significance are estimated in Mplus based on the Sobel test (Baron and Kenny 1986; MacKinnon et al. 2002; MacKinnon and Fairchild 2009; Sobel 1986).20 The indirect effect of trust in technology on intention to use over perceived usefulness is significant (β = 0.083, ρ < 0.001). Also, the indirect effect of ease of use on intention to use over perceived usefulness is significant (β = 0.036, ρ < 0.01). Therefore, it can be concluded that hypotheses 5a and 5b are supported (see table 7.7). According to the hypotheses 10a-g, the interaction effects of company size with the proposed direct antecedents of intention to use are tested. In line with Baron and Kenny’s (1986) definition of moderation, I used the latent moderated structural equations (LMS) estimation procedure, as implemented in Mplus 5.1, to test interaction effects between variables within a SEM (Klein and Moosbrugger 2000). This procedure is viewed as a reliable and valid method to explore moderating effects (Marsh, Hau, and Wen 2004). I compared adjacent nested models (model with a moderating effect / model without a moderating effect) with a χ 2 -difference test based on loglikelihood values and scaling correction factors obtained with the MLR estimator.21 It delivers scale-adjusted loglikelihood values. Scaled values cannot be used for χ 2 - difference testing of nested models because a difference between two scaled loglikelihood values for nested models is not distributed as χ 2 . Therefore, a χ 2 -difference test was computed based on loglikelihood values and scaling correction factors obtained with the MLR estimator according to Satorra (2000). Interaction effects are identified for two pairs of constructs: company size and perceived usefulness (path estimate = -0.181, ρ < 0.001), and company size and company norms (path estimate = -0.088, ρ < 0.001). This is indicated by significant effect sizes of the interactions, an inTable 7.7: ITSUM (n=717): Mediating Effects Mediating Effects H
Mediator
Path
exp. effect
Std. Path Coef.
ρ-value
5a
Perceived Usefulness Perceived Usefulness
Trust in Technology → Intention Ease of Use → Intention
mediation
0.083 (0.018)
0.000
supported
mediation
0.036 (0.015)
0.013
supported
5b
Result
Legend: Values in parentheses are standard errors if not otherwise denoted. 20
21
The more commonly used causal steps method by Baron and Kenny (1986) is not chosen for two reasons. The coefficient method of MacKinnon et al. (2002) is superior as it provides a direct estimate of the size of the indirect effect, whereas the causal steps method does not provide a statistical test for indirect effects. Also, the requirements for Baron and Kenny’s (1986) approach cancel out models in which the independent and dependent variables have no significant relation or effects with opposite signs. This is the case for the variables ease of use and trust in technology in this study. Furthermore, preliminary tests with both variables showed no significant effects on intention to use. The MLM estimator is not available for the LMS estimation within Mplus. The MLR estimator is chosen because it is robust to non-normality of observations and can be employed with continuous and categorical variables in the interaction effect calculation.
178
7. Quantitative Studies
Table 7.8: ITSUM (n=717): Moderating Effects Moderating Effects H
Moderator
Path
exp. effect
Path Coefficients
10a 10b 10c 10d 10e 10f 10g 11a 11b 11c 11d 11e 11f 11g
Company Size Company Size Company Size Company Size Company Size Company Size Company Size Function Function Function Function Function Function Function
Perceived Usefulness → Intention Controllability → Intention Trustworthiness → Intention Role Clarity → Intention Role Ability → Intention Motivation → Intention Subjective Norms → Intention Perceived Usefulness → Intention Controllability → Intention Trustworthiness → Intention Role Clarity → Intention Role Ability → Intention Motivation → Intention Subjective Norms → Intention
Attenuated Attenuated Attenuated Attenuated Attenuated Attenuated Attenuated Strenghened Strenghened Strenghened Strenghened Strenghened Strenghened Strenghened
−0.181*** not supported not supported not supported not supported not supported −0.088*** not supported not supported not supported not supported not supported not supported not supported
Moderating effects of function were tested on a reduced sample (n=649). H: Hypothesis; * : ρ < 0.05; ** : ρ < 0.01; *** : ρ < 0.001.
crease of the AIC value, and a significant χ 2 -difference test that show an improvement of the model fit with the interaction effect compared to the model without an interaction effect (see table A.4 in the Appendix). The results support the Hypotheses 10a and H10g and suggest that with increasing company size, the effect of perceived usefulness and subjective norms will be weaker as individual perceptions of an employee may be more unlikely to affect decisions of the organization. Furthermore, function as a categorical variable with the categories owner/general manager, production manager, and machine operator was tested. The category "others" was omitted and the sample size was reduced by the 68 respective respondents for this test. This left 649 individuals in the sample. Respondents’ function had no significant effect with any antecedent of intention to use interactive remote services. Therefore, H11a-g could not be supported (see table 7.8).
7.8
Multi-Group Comparison: Adoption vs. Continued Usage
7.8.1
Description of the Groups
It was hypothesized that the effects of an individual’s perceptions on organizational decisions to use interactive remote services are different for organizations that are mainly in a pre-adoption phase and organizations that are in a continued usage phase. To accurately detect differences in the effect of antecedents on usage intention between these two groups of organizations, a
7.8 Multi-Group Comparison: Adoption vs. Continued Usage
179
multi-group comparison was conducted. To yield groups big enough for the estimation and to get a conservative estimation of the pre-adopter group, a median split was performed on the measure "organization’s level of remote service implementation" according to Rogers’ (2003) classification of innovation adoption. The median of the measure is at 5, and the arithmetic middle at 4.52. Group 1 consists of organizations in the pre-adoption phase and includes cases with ratings from 1 to 4 on the implementation level scale (n=353). This group is referred to as the "pre-adopter group." Group 2 comprises organizations that are in a continued usage phase and includes cases with ratings from 5 to 7 on the implementation level scale (n=364). This group is referred to as the "continued user group." The split of the sample leads to two groups, which are similar in their demographic characteristics (see figure 7.7 and 7.8). Results of the χ 2 -test indicate no difference in the distribution of the respondent’s age (χ 2 (2) = 0.683).22 Figure 7.9 illustrates the distribution of company size in both groups. The pre-adopter group includes more small enterprises (62%) when compared to the continued user group (27%). The continued user group, however, contains more of the larger companies (> 20 employees). This finding suggests that remote services are more frequently used in middle sized and larger companies, whereas small enterprises still hesitate to adopt this new service type. This corresponds to the finding that in the pre-adopter group, the respondent is more often a general manager or managing owner (76%) than in the continued user group (69%) (see figure 7.10). A multi-group analysis and further tests for moderation of company size and function of the respondent on both groups are conducted below to gain insight into the interplay of these factors.
7.8.2
Assessing Measurement Invariance
To assess measurement invariance of the multi-group model, the ITSUM is separately fitted within each group (pre-adopter and continued user). The overall model fit indices for both groups are shown in table 7.9. The results of the ITSUM for organizations within the preadoption phase (n=364) exhibit a good fit. The CFI is at a satisfying level of 0.916. The TLI is at a level of 0.900. The RMSEA is at 0.056 and the SRMR is well under the cut-off value of 0.08 with 0.069 (Hu and Bentler 1999). The results of the structural test for organizations within the post-adoption phase (n=353) provide strong support for the proposed ITSUM. The CFI is at a satisfying level of 0.920, and the TLI is at a level of 0.905. The RMSEA is at 0.051. The SRMR is 0.079, which is under the cut-off value of 0.08. The variance of the dependent variable, intention to use interactive remote service, is explained on a good level of 43% in the continued user group. In the pre-adopter group, intention to use remote service is explained on an acceptable level (r2 = 0.360). 22
Categories with less than 5 cases were not included in the test.
180
7. Quantitative Studies
"
" "
" " "
" "
" "
"
"
" "
"
"
#
$
Figure 7.7: Distribution of Respondent’s Age Across Groups (in %)
Figure 7.8: Distribution of Respondent’s Gender Across Groups (in %)
7.8 Multi-Group Comparison: Adoption vs. Continued Usage
181
#
# #
#
#
#
# #
#
#
#
# #
#
#
# #
# #
#
#
#
$
Figure 7.9: Distribution of Company Size (Number of Employees) Across Groups (in %)
&* %*
%$"* $'"*
$* #* "* !* *
'#* "&*
&#*#*
*
!*$*
*
Figure 7.10: Distribution of Respondent’s Function Across Groups (in %)
182
7. Quantitative Studies
Table 7.9: Model Fit Statistics for Pre-Adopter Group and Continued User Group Pre-Adopter Group (nn = 3 6 4) χ 2 (df) ratio
χ 2/df
SCALING
883.078 (412) 2.140 1.114
RMSEA SRMR
r2 (INT)
0.056 0.069 0.360
0.916 0.900 r2 (PU) 0.312 CFI
TLI
Continued User Group (nn = 3 5 3 ) χ 2 (df) ratio
χ 2/df
SCALING
789.642 (412) 1.920 1.118
0.051 0.076 r2 (INT) 0.429 RMSEA
CFI
SRMR
TLI
r2 (PU)
0.920 0.905 0.296
Table 7.10: Assessing Measurement Invariance: Model Fits Invariance Configural regular χ 2 scaled χ 2 scaling corr. df CFI TLI RMSEA AIC
1485.482 1198.091 1.240 644 0.936 0.925 0.049 63516.017 Model 1
Legende:
p:
Metricp 1521.887 1230.159 1.237 663 0.935 0.926 0.049 63514.422 Model 2
Scalarp 1553.932 1259.147 1.234 674 0.933 0.925 0.049 63524.468 Model 3
Factor Variancep 1583.678 1283.139 1.234 671 0.931 0.922 0.050 63560.213 Model 4
Factor Co-Variancep 1585.363 1285.474 1.233 690 0.932 0.925 0.049 63523.899 Model 5
partial.
The comparison of the effects of antecedents on usage intention within the ITSUM across the pre-adopter and the continued user groups is only reliable and statistically justifiable if measurement invariance is assured across groups. This thesis follows the recommended procedure by Steenkamp and Baumgartner (1998a) and uses a hierarchical ordering of four nested models and a baseline model as introduced in chapter 7.2.2.5. In addition to the χ 2 -difference test, Steenkamp and Baumgartner (1998a) suggest that the CFI, TLI, AIC, and the RMSEA should be included in the assessment of the nested models. Steenkamp and Baumgartner (1998a; 1998b), and van Birgelen et al. (2002) also show that the assessment of the fit indices can outweight significant χ 2 -difference test results when large sample sizes are estimated. Within Mplus, the MLM estimator delivers scaled χ 2 -values. The scaled χ 2 cannot be used for χ 2 -difference testing of nested models because a difference between two scaled χ 2 -values for nested models is not χ 2 distributed. Therefore, a corrected χ 2 -difference test statistic is computed according to Satorra (2000) for the nested model comparison. The results were obtained by using Steenkamp and Baumgartner’s (1998a) procedure. All parameter estimates are shown in table 7.10 including the unscaled χ 2 -value, scaled χ 2 -value,
7.8 Multi-Group Comparison: Adoption vs. Continued Usage
183
scaling correction factor, degrees of freedom (df), CFI, TLI, RMSEA, and AIC. In sum, five models were estimated: 1. Model 1: Configural invariance is assessed by first testing a multi-group model without measurement invariance. The results indicate that the data fits well with the a priori hypothesized model (CFI = 0.936, TLI = 0.925, and RMSEA = 0.049). Configural variance can, therefore, be assumed. 2. Model 2: Metric invariance has to be satisfied to compare the scale intervals of items across groups. Because the factor loadings carry the information about how changes in latent scores relate to changes in observed scores, metric invariance can be tested by constraining the factor loadings to be equal across groups (Steenkamp and Baumgartner 1998a). A model with all factor loadings constrained did not pass the test for full metric invariance. Therefore, the procedure by Byrne, Shavelson, and Muthen (1989) for assessing partial invariance was followed. Byrne, Shavelson, and Muthen (1989) argue that full metric invariance is not necessary for further tests of invariance, provided that at least one item of a construct is metrically invariant. The modification indices (M.I.) indicated the item "role1" as being a cause for non-invariance (M.I. = 24.248). According to Steenkamp and Baumgartner (1998a), the invariance constraint for this loading was relaxed across groups for all of the subsequent tests. The model fit of this model (model 2) is very good (CFI = 0.935, TLI = 0.926, RMSEA = 0.049). All model fit values only differ very marginally if at all from the configural invariance model (model 1), the TLI and the AIC even improve. In terms of the χ 2 -difference test, the fit of this model is not highly significantly worse than the fit of the configural invariance model (corr. Δχ 2 (19) = 32.066, ρ = 0.031). Based on these results, partial metric invariance is assumed. 3. Model 3: To assess scalar invariance a third model was tested by additionally holding the measurement intercepts equal. This model did not pass the test for full scalar invariance. The modification indices indicated the items of the role clarity and motivation construct to be a cause for the non-invariance (M.I .> 18.248). According to Steenkamp and Baumgartner (1998a), the invariance constraints for these loadings and means were relaxed across groups, except for the marker items, to assess scalar invariance. The data fit is good with this hypothesized model (model 3, CFI = 0.933, TLI = 0.925, RMSEA = 0.049). Compared with model 2, the increase in the χ 2 -value is statistically significant — corrected Δχ 2 (11) = 30.427, ρ < 0.01 — but is very modest in magnitude. This might due to the sensitivity of the χ 2 statistic to large sample sizes. However, the inspection of the other fit indices such as AIC, CFI, TLI and RMSEA, which are less sensitive to sample-size, shows a less substantial decrease in fit - if any at all. Therefore, it is concluded that partial scalar invariance is supported. The means for all constructs except role clarity and motivation can be compared across groups. 4. Model 4: To assess factor variance invariance a forth model was tested while holding
184
7. Quantitative Studies the factor variances equal. The data fit satisfactorily with the hypothesized model (CFI = 0.931, TLI = 0.922, RMSEA = 0.05). The increase in the χ 2 -value between model 2 and the this model is statistically significant — corrected Δχ 2 (8)= 62.708, ρ < 0.001 — but is modest in magnitude. This might be affected by the sensitivity of the χ 2 -statistic to large sample sizes. The inspection of the other fit indices shows a less substantial decrease in fit. For example the RMSEA-value is nearly unchanged. Therefore, it is concluded that partial factor variance invariance is supported.
5. Model 5: A fifth model was tested to assess if the factor covariances were invariant across groups compared to model 2. The increase in corrected χ 2 between model 2 and the fifth model is significant: corrected Δχ 2 (27)= 55.937 (ρ < 0.001), but modest in magnitude. The fit of the fifth model is good (CFI = 0.932, TLI = 0.925, RMSEA = 0.049) and most model-fit values are nearly the same as in model 2, the RMSEA value is the same as in model 2. Therefore, partial factor covariance invariance is supported and it can be concluded that most factor correlations are identical across groups. Overall, I found support for configural invariance, partial metric invariance, partial scalar invariance, partial factor variance invariance, and partial factor covariance invariance. Therefore, it is possible to compare path coefficients and factor means across groups. The comparison of the path coefficients across groups is based on model 2. The comparison of the factor means across groups is based on model 3. The means for the constructs role clarity and motivation cannot be compared in a meaningful way.
7.8.3
Results for Organizations in the Pre-Adoption Phase
The estimation of the pre-adopter group model shows that the ITSUM hypotheses are supported to a large extent. Seven of the nine hypotheses regarding the direct effects are supported. All path coefficients for the ITSUM (pre-adopters) are summarized in table 7.11, and illustrated in figure 7.11. The analysis reveals that the proposed antecedents of behavioral intention regarding the actions of the service counterpart have a significant positive effect on intention to use interactive remote services. Controllability beliefs (β = 0.132, ρ < 0.05) and trustworthiness beliefs (β = 0.215, ρ < 0.01) directly affect the organization’s intention to use remote service, which supports hypotheses H1 and H2. Ease of use and trust in technology have been proven to be antecedents of perceived usefulness with strong significant effects (β = 0.142 and β = 0.326) supporting H3 and H4. Perceived usefulness has a significant positive effect on behavioral intention (β = 0.141, ρ