Customer retention in the residential Internet services market: the case of Thailand

Transcription

1 Customer retention in the residential Internet services market: the case of Thailand Paramaporn Thaichon Thesis submitted in fulfillment of the requirements for the degree of Doctor of Philosophy Faculty of Business and Law Swinburne University of Technology 2014

2 Abstract This thesis is based on the issues which have been debated in the marketing and psychological literature relating to consumer behaviour, services marketing and service quality. The two main research gaps are: first, the absence of a specific set of relevant dimensions which comprehensively measure the service quality of an Internet Service Provider (ISP) as perceived by its customers; second, the lack of academic research which addresses issues relating to customer retention and loyalty of an ISP. The conceptual framework for the current study is focused on three key areas: (1) service quality dimensions of an ISP; (2) cognitive and affective evaluations; and (3) outcomes or resultants. The specific ISP service quality dimensions in the conceptual framework consist of the core services offered by the ISP to the customer in the Internet service context. It has long been acknowledged that both cognitive and affective evaluations impact customer loyalty. The conceptual model includes a variety of such evaluations that represent the various ways customers assess and respond to the services of their ISPs, as well as the customers response to the outcomes of the ISP service quality dimensions by their ISPs. The final component of the model involves customers attitudes towards their ISPs as well as their future intentions. The resultants, hence, refer to the overall impact of the service encounters and projects undertaken on the long-term relationship between the ISP and its customers. The research design is a quantitative study, reflecting the scientific realism paradigm, using an online survey to obtain data on the eleven constructs in the proposed conceptual model. The respondents were those who were not locked in any contracts, therein being free to switch to other ISPs. Data was obtained from 3,803 Internet users in all the regions of Thailand. The relationships between the various constructs of the proposed conceptual model were tested using Exploratory Factor Analysis (EFA), Confirmatory Factor Analysis (CFA), and Structural Equation Modelling (SEM). The analyses also include segmenting ISP s customers on the basis of their usage pattern, age, and income, and evaluating their perceptions of Internet service quality dimensions as well as their cognitive and affective evaluations and loyalty. The findings confirm that service quality dimensions for Internet Service Providers are network quality, ii

3 customer service, information support and privacy. Overall service quality also impacts customer cognitive and affective evaluations, including satisfaction, value, trust and commitment, which in turn influence customer loyalty. The relationships between variables in this research vary across different groups of customers characterised by their Internet usage patterns and demographics factors. The findings of this study have both academic and practical implications. Firstly it enhances the knowledge and measurement of service quality in the home Internet services industry which is not adequately addressed by measures such as SERVQUAL and E-S-QUAL. Secondly, this study is an empirical attempt to provide insights into how service quality contributes to the way customers affectively and cognitively evaluate the service. Additionally, a robust model of customer loyalty was developed in the context of home Internet services, which contributes to the current loyalty literature. In terms of practical contributions, the findings demonstrate the benefits of investing in service quality with a view to developing a long term oriented relationship. This research also outlines an action plan which can be considered by ISPs to improve their overall service quality and as a result, to retain their existing customers. This is critical especially for the home Internet services market which has been experiencing a high churn rate. Keywords: Customer Retention, Attitudinal Loyalty, Behavioural Loyalty, Segmentation, Service Quality, Services Marketing, Internet Service Providers (ISP). iii

4 Acknowledgements I wish to express my appreciation and gratitude to all those who inspired, encouraged, supported, assisted and were patient with me during the three years that I have been undertaking this award. Associate Professor Antonio Lobo, my hero. You are a truly delightful supervisor whom I could not ask for more. Your insightful guidance and valuable direction have been immensely helpful throughout my candidature. Thank you for believing that this 21 year old could start and accomplish a PhD. Thank you for your patience throughout the years and for always having faith in me. I would not have had any publications without your kind assistance. Thank you very much! Dr Ann Mitsis for her insight, wisdom and feedback over the years. Thank you for the motivation and inspiration during the hard times. Thank you for introducing me to the research assistant job which I truly enjoyed and have learned a lot from. It is very much appreciated. Dr Catherine Prentice for her advice and recommendations for the quantitative analysis. Thank you for assisting me with my first A ranked journal article. The hard work has eventually paid off. Dr Civilai Leckie for coaching me on the statistical analysis. Dr Elena Verezub for working with me on the professional academic English improvement program. Naomi Levin for her valuable help editing and final proof reading. Dr Rowan Bedggood for introducing me to the teaching world and your kindness throughout the semesters. Dr Elizabeth Levin for starting the engine of my PhD. Thank you for the spiritual support before and during my PhD. iv

5 Vice-Chancellor, Professor Linda Kristjanson for the motivation and inspiration during the hard periods. Thank you very much for your time. Professor Heath McDonald, Professor Pamela Green and Anne Cain for the administrative support throughout my candidature. It is much appreciated. Swinburne Research, as well as Faculty of Business and Enterprise, Swinburne University of Technology for generously supporting me with resources and training throughout my candidature. My parents, Dr Panthai Thaichon and Vachira Thaichon, who worked tirelessly to provide me with the best possible educational opportunities. You supported me, encouraged me and inspired me to strive for excellence in all my endeavours. Thank you very much. I am so proud to be your son. I love you! My grandparents, Former Governor Suchart Rutkumthai and Lady Malinee Rutkumthai. Thank you for encouraging me and inspiring me. I love both of you very much. I promise that I will always be a good boy. Sara Quach, who has been always listening and encouraging me. It is much appreciated for the lovely breakfasts, lunches and dinners throughout the years. Thank you for being by my side. My colleagues and friends, Thank you for your support and understanding. v

6 Statement of declaration I hereby declare that this thesis is the presentation of my original research work and I am the sole author of it. It contains no material that has been accepted for the award to the candidate of any other degree or diploma. I certify that except where due acknowledgement has been made, this thesis is my own work and contains no material previously published or written by another person except where due reference is made. This is a true copy of the thesis, including any required final revisions, as accepted by my examiners. I understand that my thesis may be made electronically available to the public. The work was done under the supervision of Associate Professor Antonio Lobo and Dr Ann Mitsis from Swinburne University of Technology. Naomi Levin edited this thesis. The editing addressed only style and grammar and not its substantive content. Paramaporn Thaichon August 2014 vi

7 Publications arising from this thesis Refereed Journal Articles Thaichon, P., Lobo, A., Prentice, C. & Quach, T. N. (2014). The development of service quality dimensions for Internet Service Providers: retaining customers of different usage patterns. Journal of Retailing and Consumer Services. vol. 21, no. 6, pp (ABDC Ranked: A) Thaichon, P., Lobo, A. & Mitsis, A. (2014). An Empirical model of home Internet services quality in Thailand. Asia Pacific Journal of Marketing and Logistics, vol. 26, no. 2, pp (ABDC Ranked: B) Thaichon, P., Lobo, A. & Mitsis, A. (forthcoming). Antecedents to customer loyalty of Internet Service Providers. International Journal of Quality and Service Sciences. (ABDC Ranked: C) Conference Publications Thaichon, P., Lobo, A. & Mitsis, A. (2014). Using segmentation to reduce the churn rate: the case of an Internet Service Provider. Proceedings of the 2014 Australia New Zealand Marketing Academy Conference (ANZMAC), Brisbane, Australia, 1st to 3rd December, Thaichon, P., Lobo, A. & Mitsis, A. (2014). How can we retain customers of varying usage patterns? - The case of an Internet Service Provider. Proceedings of the 2014 Australia New Zealand Marketing Academy Conference (ANZMAC), Brisbane, Australia, 1st to 3rd December, Thaichon, P., Lobo, A. & Mitsis, A. (2014). Internet Service Providers service quality and its effect on customer loyalty. Proceedings of the 2014 Australia New Zealand Marketing Academy Conference (ANZMAC), Brisbane, Australia, 1st to 3rd December, vii

8 Thaichon, P., Lobo, A. & Mitsis, A. (2014). Evaluating brand loyalty of an Internet Service Provider. Proceedings of the 2014 Australia New Zealand Marketing Academy Conference (ANZMAC), Brisbane, Australia, 1st to 3rd December, Thaichon, P., Lobo, A. & Mitsis, A. (2014). Customer loyalty in High-Tech Internet services. Proceedings of the 2014 Global Marketing Conference (GMC), EMAC-GAMMA joint symposium, Marina Bay Sands, Singapore, 15 th to 18 th July, Thaichon, P., Lobo, A. & Mitsis, A. (2014). Evaluating specific service quality dimensions which impact on customers behavioural loyalty in high-tech Internet services. Proceedings of the 2014 Global Marketing Conference (GMC), Marina Bay Sands, Singapore, 15 th - 18 th July, Thaichon, P., Lobo, A. & Mitsis, A. (2013). Determinants of service quality and their influence on customers of Internet service providers in Thailand. Proceedings of the 2013 European Marketing Academy Regional Conference (EMAC). Saint Petersburg, Russia, 25 th to 27 th September, Thaichon, P., Lobo, A. & Mitsis, A. (2013). An empirical model to evaluate the loyalty of customers of Internet service providers. Proceedings of the 2013 Australian and New Zealand Marketing Academy Conference (ANZMAC) Poster Session. Auckland, New Zealand, 1 st to 4 th December, Thaichon, P., Quach, T. N. & Lobo, A. (2013). Marketing communications: factors influencing brand loyalty of Internet service providers. Proceedings of the 2013 Australian and New Zealand Marketing Academy Conference (ANZMAC). Auckland, New Zealand, 1 st to 4 th December, viii

9 Thaichon, P. (2013). Investigating the antecedents to attitudinal and behavioural loyalty of Internet service providers in Thailand. Paper presented at the Doctoral Colloquium held as part of the 2013 Australian and New Zealand Marketing Academy Conference (ANZMAC), Auckland, New Zealand, 1 st to 4 th December, Thaichon, P., Lobo, A. & Mitsis, A. (2012). Investigating the antecedents to loyalty of Internet service providers in Thailand: developing a conceptual model. Proceedings of the 2012 Australian and New Zealand Marketing Academy Conference (ANZMAC), Adelaide, Australia, 3 rd to 5 th December, ix

10 Table of contents Abstract... ii Acknowledgements... iv Statement of declaration... vi Publications arising from this thesis... vii Table of contents... x List of figures... xvi List of tables... xviii Chapter 1: Introduction Chapter overview Background Research gaps Service quality Customer retention and customer loyalty Research questions Contribution to knowledge Academic contribution Practical contribution Conceptual framework and research objectives Methodology Outline of the thesis Glossary Definitions Chapter summary Chapter 2: Literature Review Chapter overview Services marketing Evolution of the services marketing literature Nature of services Service quality x

11 2.3 Relationship marketing Relationship marketing and customer retention Benefits of relationship marketing associated with Internet services Customer loyalty Context of the study Overview of Thailand Historical overview of the Internet The Internet services market in Thailand Consumer buyer behaviour in Thailand Research questions and research objectives Benefits of researching service quality in the telecommunications industry Chapter summary Chapter 3: Development of the Theoretical Model and Related Hypotheses Chapter overview Overview of the development of a theoretical model Overview of an ISP s service quality dimensions Influence of service quality on the cognitive and affective evaluations of customers Overview of the dependant constructs Previously validated models Proposed theoretical model An ISP s service quality dimensions Network quality Customer service and technical support Information quality and website information support Privacy and security Cognitive and affective evaluations of customers Customer trust Customer satisfaction Customer commitment Customer value Dependant Constructs xi

12 3.7.1 Customer trust and loyalty Customer satisfaction and loyalty Customer commitment and loyalty Customer value and loyalty Chapter summary Chapter 4: Methodology Chapter overview The scientific realism paradigm The online survey Sampling Sample size error Sample size Sample selection Data collection procedures Non-Response bias Unit of analysis Pre-Testing Pilot study Data collection Tools for analysis Reliability and validity Reliability Validity Constructing reliable and valid online questionnaires Measurement development Network quality Customer service and technical support Information quality and website information support Privacy and security Customer trust Customer satisfaction Customer commitment xii

13 4.8.8 Customer value Customer loyalty Preparing the data Coding and editing Structural equation modelling Ethical considerations Chapter summary Chapter 5: Analysis and Results Chapter overview Data screening Missing data Response time bias Outliers Assessing normality and reliability Profiles of Respondents Gender Age Monthly household income Level of education Main areas of respondents education Employment status Areas of employment Location Details of Internet services used Switching experience Reasons for switching Frequency of Internet use Internet usage time Number of people in a household Number of people using the Internet in a household Types of Internet connection Home Internet speed xiii

14 5.4.9 Internet expenditure per month Additional descriptive statistics Switching experience versus gender Switching experience versus age Switching experience versus household income Reasons for switching versus gender Reasons for switching versus age Reasons for switching versus household income Segmentation based on Internet usage patterns General characteristics of Internet usage groups Service quality perceptions Complaints behaviour Switching behaviour Intentions to recommend Exploratory analysis and reliability tests Confirmatory Factor Analysis and measurement models Maximum Likelihood Model fit indices Congeneric measurement models Structural Equation Model (SEM) Convergent validity Discriminant validity The output of the structural model Descriptive statistics and intercorrelations Assessing the reliability of the final model Discussion of hypothesis tests Details of segmentation analysis using invariance testing Internet usage pattern Age Income Chapter summary xiv

15 Chapter 6: Discussion of Findings, Recommendations and Conclusion Chapter overview Discussion of the findings Discussion of findings relating to stage 1 of the conceptual framework: service quality dimensions Discussion of findings relating to stage 2 of the conceptual framework: customers cognitive and affective evaluations Discussion of findings relating to stage 3 of the conceptual framework: customers cognitive and affective evaluations on customer loyalty Segmentation analysis Internet usage Age Income Contribution of this study Academic contribution Practical contributions and managerial implications Action plans for the decision makers: improving service quality Limitations of this study and future research directions Concluding remarks Chapter summary References Appendix 1: Final questionnaire Appendix 2: Ethics approval letter Appendix 3: Results for the initial structural model Appendix 4: Standardised coefficients of individual measures Appendix 5: Squared Multiple Correlations xv

16 List of figures Figure 1.1: Chapter organisation... 2 Figure 1.2: Initial conceptual framework for the study Figure 1.3: Structure of the overall thesis Figure 2.1: Chapter organisation Figure 2.2: Structure of the overall thesis Figure 3.1: Chapter organisation Figure 3.2: Structural model from Durvasula et al. (2004) Figure 3.3: Structural model from Caceres and Paparoidamis (2007) Figure 3.4: Structural model from Vlachos and Vrechopoulos (2008) Figure 3.5: Proposed theoretical model Figure 3.6: Structure of the overall thesis Figure 4.1: Chapter organisation Figure 4.2: Structure of the overall thesis Figure 5.1: Chapter organisation Figure 5.2: Evaluating service quality: Mean Statistics Figure 5.3: Reasons to switch Figure 5.4: Reason to recommend the current ISP to others Figure 5.5: CFA for network quality measures Figure 5.6: CFA for customer service and technical support measures Figure 5.7: CFA for information quality measures Figure 5.8: A re-specified one factor model for the information quality construct Figure 5.9: CFA for security and privacy measures Figure 5.10: CFA for customer trust measures Figure 5.11: A re-specified one factor model for customer trust construct Figure 5.12: CFA for customer satisfaction measures Figure 5.13: CFA for customer commitment measures Figure 5.14: CFA for customer value measures Figure 5.15: CFA for attitudinal loyalty measures Figure 5.16: A re-specified one factor model for attitudinal loyalty construct Figure 5.17: CFA for behavioural loyalty measures Figure 5.18: The service quality dimensions of the conceptual model xvi

17 Figure 5.19: Customer cognitive and affective evaluation constructs Figure 5.20: Resultant constructs of the conceptual model Figure 5.21: Structural model of customer loyalty Figure 5.22: Re-specified final best fit model of customer loyalty Figure 5.23: Structure of the overall thesis Figure 6.1: Chapter organisation Figure 6.2: Initial conceptual framework for the study xvii

18 List of tables Table 1.1: Glossary Table 1.2: Definition of theoretical, analytical and methodological terms used Table 2.1: Thai Internet users Table 2.2: Internet usage behaviour Table 4.1: Network Quality scale Table 4.2: Customer service and technical support scale Table 4.3: Information quality and website information support scale Table 4.4: Privacy and security scale Table 4.5: Customer trust scale Table 4.6: Customer satisfaction scale Table 4.7: Customer commitment scale Table 4.8: Customer value scale Table 4.9: Attitudinal loyalty scale Table 4.10: Behavioural loyalty scale Table 4.11: Model fit indicators adopted in this study Table 5.1: Means and 5% trimmed means Table 5.2: Results of progressive reliability tests Table 5.3: Switching experience versus gender Table 5.4: Switching experience versus age Table 5.5: Switching experience versus household income Table 5.6: Reasons for switching versus gender Table 5.7: Reasons for switching versus age Table 5.8: Reasons for switching versus household income Table 5.9: Age groups among light, medium and heavy users Table 5.10: Employment status among light, medium and heavy users Table 5.11: Gender among light, medium and heavy users Table 5.12: Area of employment among light, medium and heavy users Table 5.13: Percentages of students among light, medium and heavy users Table 5.14: Relationships between customers perceptions of service quality, and intention to complain, beta coefficients Table 5.15: Switching experience xviii

19 Table 5.16: Reasons to switch Table 5.17: Relationships between customer perceptions of service quality, and intention to switch, beta coefficients Table 5.18: Intention to recommend Table 5.19: Reason to recommend the current ISP to others Table 5.20: Relationships between customer perceptions of service quality, and intention to recommend, beta coefficients Table 5.21: Statistical conditions Table 5.22: Exploratory Factor Analyses (EFA) results Table 5.23: The reliability test Table 5.24: Model fit indicators adopted in this study Table 5.25: Regression weights for network quality measures Table 5.26: Regression weights for customer service measurements Table 5.27: Regression weights information quality measurements Table 5.28: Regression weights for security and privacy measurements Table 5.29: Regression weights for customer trust measurements Table 5.30: Regression weights for customer satisfaction measurements Table 5.31: Regression weights for customer commitment measurements Table 5.32: Regression weights for customer value measurements Table 5.33: Regression weights for attitudinal loyalty measurements Table 5.34: Regression weights for behavioural loyalty measurements Table 5.35: Summary of the reliability of the measurement models Table 5.36: Factor pattern and structure coefficients for service quality dimensions Table 5.37: Factor pattern and structure coefficients for the customer cognitive and affective evaluations constructs Table 5.38: Factor pattern and structure coefficients for the resultant constructs Table 5.39: Descriptive statistics and correlations among the study variables Table 5.40: Summary of reliability of the final model scales Table 5.41: Results of hypotheses testing Table 5.42: Ranking of service quality dimensions based on their influence on service quality Table 5.43: Ranking of cognitive and affective evaluations based on the influence of service quality xix

20 Table 5.44: Ranking of cognitive and affective evaluations based on their influence on attitudinal and behavioural loyalty Table 5.45: Standardised indirect effects of overall service quality on loyalty Table 5.46: Summary of the hypotheses tests Table 5.47: Regression weights (Internet usage groups) Table 5.48: Chi-square difference test (Internet usage groups) Table 5.49: Squared Multiple Correlations (Internet usage groups) Table 5.50: Regression weights (Age groups) Table 5.51: Chi-square difference test (Age groups) Table 5.52: Squared Multiple Correlations (Age groups) Table 5.53: Regression weights (Income groups) Table 5.54: Chi-square difference test (Income groups) Table 5.55: Squared Multiple Correlations (Income groups) xx

21 Chapter 1: Introduction 1.1 Chapter overview Chapter 1 provides a rationale and context for this thesis, as well as a roadmap for other chapters. This chapter comprises seven sections. It opens with a short background overview of this thesis (section 1.2). The next section (section 1.3) presents research gaps which are divided into (a) the research gap related to service quality (section 1.3.1) and (b) the research gap related to customer retention and customer loyalty (section 1.3.2). Section 1.4 articulates the research questions. The contribution of this study is discussed from both an academic (section 1.5.1) and a practical perspective (section 1.5.2). Section 1.6 reviews the conceptual framework and research objectives of the study. The methodology and data collection are summarised in Section 1.7. Section 1.8 outlines the chapters of the thesis. Glossary and definition are provided in Section 1.9 and Section 1.10 respectively. Section 1.11 presents a brief conclusion. The organisation of this chapter is shown in Figure

22 1.1 Chapter overview 1.2 Background 1.3 Research gaps 1.4 Research questions 1.5 Contribution to knowledge 1.6 Conceptual framework 1.7 Methodology 1.8 Thesis outline 1.9 Glossary 1.10 Definitions 1.11 Chapter summary 1Figure 1.1: Chapter organisation 2

23 1.2 Background This thesis originated from the author s interest in creating and maintaining customers sense of belonging and loyalty toward their service providers. It was further motivated by a request received from a telecommunications company seeking to understand the success of building loyal customers. Supporting this view, Buil et al. (2013) indicate that while many consumers are pleased with their current service provider, they need incentives and motivations to increase their intention to stay with the company. The motivation of retaining existing customers further enhanced the author s interest in findings ways to improve customer satisfaction and make customers feel attached to a service provider without providing extra inducements or cohesions. This thesis examines the issue of customer retention in the telecommunications industry, specifically Internet service providers or ISPs. It does this by reviewing the literature and then translating this into an empirical study that examines the issues from the perspective of ISP consumers in Thailand. The Thai telecommunications industry is under pressure to maintain and increase its existing customer base. The Thai National Statistical Office (TNSO, 2010) reports that an average of 10% of home Internet customers in Thailand switched service providers each year during the period 2003 to In 2009, this figure increased to 12% (True, 2010). Despite this increase, there is limited information relating to the creation of customer loyalty in high-tech services. A search on Google Scholar for research publications exploring consumer retention and loyalty in the ISP industry (on March 10, 2014) revealed only eight results for all countries and none in relation to Thailand (excluding the publications that have resulted from this thesis). Given the large Internet usage and high churn rates within the ISP sector, this area is grossly under-researched. Therefore, it is necessary to investigate the antecedents to customer loyalty and customer retention within this market. Generally, researchers use SERVQUAL and E-S-QUAL to measure and manage service quality by evaluating the main quality components (He and Li, 2010). Whilst SERVQUAL is the traditional service measure, E-S-QUAL was introduced later, specifically in the e-services context. Despite their popularity in service quality management, SERVQUAL and E-S-QUAL do not take into account the specific issues relevant to customers of high-tech ISPs. Additionally, SERVQUAL and E-S-QUAL 3

24 focus on service providers that operate via the Internet platform (Vlachos and Vrechopoulos, 2008), but not on those that actually provide the Internet connection and platform for online Business-to-Business (B2B) and Business-to-Customer (B2C) activity. This thesis aims to investigate the specific service quality dimensions and attributes which influence the overall service quality of an ISP as perceived by its customers. It intends to develop a customised scale which could be effectively used to measure the service quality of an ISP globally. This thesis is based on the issues stated above, which have been debated in the marketing and psychological literature in the areas of consumer behaviour, including services marketing and service quality. The topic of customer loyalty is of current relevance as governments and organisations not only focus on attracting new customers, but also seek to maintain existing customers (Tam, 2012). This thesis aims to answer the following three issues: (1) Identify the specific service quality dimensions and attributes which influence the overall service quality of an ISP; (2) Clearly and unambiguously define each of the relationships between service quality, and customers cognitive and affective evaluations in the home Internet services market; (3) Identify the effects of customers cognitive and affective evaluations on loyalty in the home Internet services market. 1.3 Research gaps This thesis aims to address the gaps in this area in the current literature. The two main research gaps identified are: (1) The absence of a specific set of relevant dimensions to comprehensively measure the service quality of an ISP as perceived by its customers; (2) The lack of academic research examining issues relating to loyalty of an ISP s customers Service quality Superior service quality is the buzzword for all types of services and this is true for high-tech services, including home Internet and mobile phone services (He and Li, 2010). Although prior research has established the link between overall service quality and loyalty (Kuo et al., 2013), there have been limited academic studies investigating the effects of specific ISP service quality dimensions on customer loyalty in the high- 4

25 tech residential Internet services market (Vlachos and Vrechopoulos, 2008). With the increase in technology-enabled services, the attention of the services literature has shifted towards measurement and operationalisation of service quality elements in different types of services (Carlson and O Cass, 2011; Ganguli and Roy, 2010; Kurt and Atrek, 2012; Wolfinbarge and Gilly, 2003). SERVQUAL and E-S-QUAL have been developed by Parasuraman et al. (1985) in an attempt to fully capture service quality across a range of services. However, the area associated with the measurement of service quality of Internet services has been somewhat neglected (He and Li, 2010; Thaichon et al., 2014). High-tech telecommunications service quality cannot be effectively measured by SERVQUAL or E-S-QUAL (He and Li, 2010) as these scales lack the ability to address specific issues relevant in this particular context. SERVQUAL and E-S-QUAL focus on service providers that operate via an Internet platform (Vlachos and Vrechopoulos, 2008) but not on those who actually provide the Internet connection and platform for online B2B and B2C activities. The literature on service quality reveals that perceived service quality dimensions are not limited to those identified in SERVQUAL and E-S-QUAL (Ganguli and Roy, 2010; He and Li, 2010). Modifications of SERVQUAL and E-S-QUAL have been proposed to adapt these scales to specific service contexts (He and Li, 2010) and researchers have recently attempted to develop service quality measurement scales in different high-tech services. Some of these are Shamdasani et al. (2008) in self-service Internet technologies and Vlachos and Vrechopoulos (2008) in mobile telephony services. Nevertheless, none of them have comprehensively evaluated ISPs service quality dimensions. Several basic differences exist between Internet services and other telecommunications services. For example, mobile service quality takes into consideration value-added services (e.g. SMS, MMS, WAP, GPRS) or mobile devices (Santouridis and Trivellas, 2010), which are irrelevant in the case of ISPs. Moreover, as the nature of home Internet services is essentially Internet driven, technical support available to customers on websites is critical when assessing an ISP s service quality. However this feature might not be relevant for other telecommunication services, such as television transmission. Therefore, specific service quality dimensions and attributes which influence the overall service quality of an ISP deserves further investigation. 5

26 Service quality is an important differentiator in a competitive business environment and a driver to service-based businesses (Zhao and Benedetto, 2013). By enhancing service quality, businesses can influence customers perception of value (Lai et al., 2009), trust (Sabiote and Roman, 2009), and commitment (Fullerton, 2005). These customer-related outcomes are important for business success and long-term customer loyalty (Prentice, 2013). However, scant research has been undertaken in relation to aspects of customer loyalty for Internet service providers. In particular, very few studies have assessed how different aspects of an ISP s service quality would influence their customers perception of value, trust, and commitment (Vlachos and Vrechopoulos, 2008). ISPs would benefit from learning more about their customers perception of the overall service quality being provided. Such information would assist ISP brand managers to develop appropriate marketing strategies in order to achieve competitive advantage and long term profitability. This thesis attempts to address this important research gap by investigating the effects of an ISP s service quality on their customers perception of value, trust, satisfaction, commitment and loyalty in the high-tech Internet services of Thailand Customer retention and customer loyalty Building strong customer loyalty is vital for all businesses, especially for high-tech services such as home Internet and mobile phone services (He and Li, 2010). It is assumed that customers in long term relationships will pay more, buy more, and act as promoters for the service provider, which eventually leads to lower acquisition costs and greater customer retention (El-Manstrly and Harrison, 2013). Service providers need to demonstrate strength in service performance and delivery in order to retain their customers (Tam, 2012). There has been scarce academic research evaluating the ISP service quality dimensions that influence customer loyalty in the residential Internet services market (He and Li, 2010). It is more valuable for a service provider to maintain and develop long-term relationships with customers rather than focus on attracting short-term customers (Rafiq et al., 2013). Customer retention is important, especially during times of economic adversity and increasing competition (El-Manstrly and Harrison, 2013). It has been determined that customer acquisition costs approximately five times more than customer retention (Christodoulides and Michaelidou, 2010). Additionally, a 1% 6

27 increase in the customer retention rate generally results in a 5% decrease in the cost of customer acquisitions (Han et al., 2012). This also results in an increase of approximate 5% of an ISP s profit, which reduces the pressure on seeking and acquiring new customers (Spiller et al., 2007). Researchers have empirically demonstrated that customer loyalty is a key factor in improving a company s economic and competitive position in the market (Álvarez et al., 2010). Therefore, this research aims to investigate the antecedents to customer retention and customer loyalty of Internet service providers in Thailand. The broader context of this research is found in the telecommunications industry, which is characterised by rapid changes and developments in countries including Hong Kong, Korea, Taiwan, the United States of America, the United Kingdom, India and Thailand (Nemati et al., 2010). Thailand the focal region of this thesis is endowed with a wide variety of natural resources, a substantial population and a relatively strong economy (Srihirun, 2011). The generous investment in education has resulted in knowledge improvement and higher qualifications for Thai people (Srihirun, 2011). In the past decade or so, there have been extensive developments in the Thai telecommunications industry (Thaichon et al., 2013). Noticeably, during the period of , the average Internet user growth rate per year was 30% (Srihirun, 2011). This figure demonstrates that telecommunications plays a growing role in Thailand s economy, and this is especially true for broadband coverage, which increased annually at a rate of approximately 23% from 2000 to 2010 (Srihirun, 2011). A deeper investigation into Internet usage in Thailand indicates that limited research is available regarding the purchase behaviour of customers of home Internet services. It is therefore necessary to undertake further research to investigate customer retention and loyalty in the Thai home Internet services market. By way of comparison, it was reported that in the UK broadband market, approximately 10% switched service providers in 2012 (Kenny and Dennis, 2013). The Thai National Statistical Office (TNSO, 2010) reported that an average of 10% of home Internet customers in Thailand switched service providers each year during the period In 2009, this figure went up to 12% (True, 2010). The Thai telecommunications industry is under considerable pressure to retain and also to increase their existing 7

28 customer base. Research has been conducted in relation to customer loyalty of services in Hong Kong, Vietnam, Taiwan and China by Chen and Green (2009); Grossmanova et al. (2009); Healy (2007); Hong and John (2010); Shang et al. (2006); Sung and Kim (2010). However, none of these authors have addressed service quality issues in Thailand, especially those associated with the home Internet services market. It is therefore necessary to investigate customer loyalty and customer retention in the Thai market. It would be interesting to investigate how the findings of this thesis compare with those of previous studies done in other Asian countries. In order to do this, this thesis will provide an opportunity for Internet service providers in Thailand and other developing countries to evaluate determinants that influence customer loyalty, which in turn, can be employed to nurture long term customer retention in the competitive home ISP market. 1.4 Research questions In order to address some of the current information gaps, the primary research question of this thesis focuses on the measurement of the service quality of ISPs as perceived by customers. The secondary research question of this thesis focuses on the interrelationships of the various constructs that are antecedents to customer retention. Primary research question: What are the specific service quality dimensions and attributes which influence the overall service quality of an ISP? Secondary research question 1: What are the clear and unambiguous relationships between service quality and customers cognitive and affective evaluations in the home Internet services market? Secondary research question 2: What are the effects of customers cognitive and affective evaluations on customer loyalty in the home Internet services market? 8

29 1.5 Contribution to knowledge The findings of the research have both academic and practical implications. The findings will enhance the knowledge and measurement of service quality in the home Internet services industry. This is not adequately addressed by current measures such as SERVQUAL and E-S-QUAL. Additionally, a robust model of customer loyalty will be developed in the context of home Internet services, which contributes to the existing loyalty literature. In terms of practical contributions, this study provides managerial implications which can be considered by ISPs to improve their overall service quality and retain existing customers. This is critical especially for the home Internet services market, which is experiencing a high churn rate of its customers Academic contribution Primary research question: What are the specific service quality dimensions and attributes which influence the overall service quality of an ISP? As mentioned previously, high-tech Internet service quality cannot be effectively measured by SERVQUAL or E-S-QUAL due to the inability of these scales to cover all specific issues relevant to this particular context (He and Li, 2010; Thaichon et al., 2014). This thesis aims to investigate service quality dimensions specific to ISPs and identify their influences on customers perceptions of overall service quality. This thesis is among the first of its kind to investigate the dimensions of an ISP s service quality, and the effects on customer loyalty in high-tech services. It contributes to the body of knowledge relating to service quality and consumer loyalty in the home Internet services market. It evaluates the service quality dimensions which are identified in Chapter 3 as (1) network quality, (2) customer service and technical support, (3) information quality and website information support, (4) security and privacy. Secondary research question 1: What are the clear and unambiguous relationships between service quality and customers cognitive and affective evaluations in the home Internet services market? Businesses have realised that sustainability in the global economy requires them to develop and maintain a long term relationship with their customers (Morgan and Hunt, 9

30 1994). Trust, satisfaction, value, and commitment representing the cognitive and affective evaluations of customers are essential elements in a relationship (Ulaga and Eggert, 2006). However, there has been scant evidence of the causal relationship between service quality and customers cognitive and affective evaluations in the home Internet services context. Hence, this study aims at establishing a theoretical basis for evaluating the influences of customers perceptions of service quality on customers cognitive and affective evaluations. Secondary research question 2: What are the effects of customers cognitive and affective evaluations on customer loyalty in the home Internet services market? The telecommunications industry is under pressure to keep its customers and to acquire new customers (Spiller et al., 2007). Despite the importance of the Thai telecommunications market and current issues related to this industry, surprisingly few studies have been done on the home Internet services in Thailand and the Thai market is an under-researched one (Thaichon et al., 2014). Given the high usage and increasing churn rates in the ISP sector, it is justified to investigate the antecedents to customer loyalty, especially customer retention, within the Thai Internet services market Practical contribution The following section provides details of the practical contributions of this study, which are comprised in three sections, i.e. (a) the telecommunications industry, (b) ISP service quality, and (c) customer retention Telecommunications industry As mentioned earlier, the telecommunications industry has witnessed high switching rates (True, 2010). Many companies try to acquire new customers through price competition, which might lead to a vulnerable and imbalanced market where smaller companies with low capital cannot compete against major corporations (Buil et al., 2013). Such competition can also diminish the firms performance and services infrastructure. This study presents a solution to this issue by investigating customer loyalty s determinants, for example, service quality, with a view to achieving sustainable development of the telecommunications industry as a whole. 10

31 ISP service quality This thesis evaluates service quality dimensions in the high-tech Internet service industry. The results will enable marketers to focus on the key dimensions of an ISP s service quality, which in turn would lead to enhanced long term profitability. These findings would be beneficial to high-tech service providers in other countries with similar backgrounds to Thailand, for instance Indonesia, Vietnam, and India (Jahanzeb et al., 2011). It is vital for ISPs to obtain accurate information regarding their service quality as perceived by their customers. This information enables them to formulate appropriate marketing strategies which would work towards achieving competitive advantage and long term sustainability (Vlachos and Vrechopoulos, 2008). By enhancing service quality, firms can influence customers behavioural and attitudinal loyalty, which are critical for an ISP s success and long term sustainability (Thaichon et al., 2012) Customer retention This study seeks to identify determinants of attitudinal and behavioural loyalty in hightech services, and underscores the importance of management devoting resources to retain customers through influencing loyalty s antecedents. This will form a foundation for service providers in the home ISP market to develop new retention strategies. By making customers more central in company operations, these strategies can also potentially reduce the expenses associated with acquiring new customers (Spiller et al., 2007). As a result, ISPs will be able to reduce the current issues relating to their customers switching behaviour in the residential Internet services market in Thailand, and in other countries that have similar demographic profiles. 1.6 Conceptual framework and research objectives Several authors have claimed that it is easier and cheaper to retain existing customers than to attract new ones (Jandaghi et al., 2011; Mao, 2010; Mathur, 2011). Therefore it is critical to understand what customers value in a service provision relationship. This exploratory study seeks to understand the dimensions that influence customers to remain loyal to their Internet service providers, as well as the interrelationships between these dimensions. The overall aim is to develop and empirically test a model that is capable of predicting customer loyalty in the home Internet service provider context. A 11

32 secondary aim is to investigate specific service quality dimensions and identify their influences in an ISP context, which then might impact on customer perceptions of ISPs service quality. The conceptual framework for the current study is shown in Figure 1.2 and will focus on three key areas: (1) the specific service quality dimensions in an ISP context, (2) cognitive and affective evaluations, and (3) resultants. Service quality dimensions of an ISP Cognitive and affective evaluations Outcomes or Resultants 2Figure 1.2: Initial conceptual framework for the study The specific ISP service quality dimensions in this conceptual framework consist of the core services offered by the ISP to the customer in the Internet services context. It has long been acknowledged that both cognitive and affective evaluations impact customer loyalty. The conceptual model includes a variety of such evaluations that represent the various ways customers assess and respond to the services of their ISPs, as well as the customers cognitive and effective evaluations. The final component of the model reflects the impact of customers attitudes towards their ISPs, as well as their future intentions. Hence, the resultants refer to the overall impact of the service encountered and projects undertaken for long-term relationships between the ISP and its customer. On the basis of the research questions presented in Section 1.4, the research objectives of the current study are to: Identify the attributes which influence the overall service quality of an ISP. Clearly and unambiguously define each of the identified dimensions that contribute to the maintenance of ongoing relationships between customers and their ISPs. Develop and test a model to predict customers loyalty to their ISPs. 12

33 1.7 Methodology This study belongs to in the scientific realism paradigm which maintains that although the object of our inquiry exists independent of the human consciousness, it cannot be completely understood with certainty by our observations (Peter, 1992). In this respect, approximate truth is sought rather than pure truth (Hunt, 1991). Consistent with the philosophical approach, this study adopted quantitative research using an online survey to obtain data on the eleven constructs in the proposed conceptual model. The constructs of the conceptual model were operationalised using multi-item measurement scales, which were sourced from extant literature. Data was collected from residential Internet users in Thailand. The study used the customer database of well-established major ISPs in Thailand, which included customers from all over the country and was representative of the Thai population. The respondents were those who were not locked in any contracts, therein being free to switch to other brands. It was a requirement that the participants were over 18 years of age and they should have used home Internet services. The survey instrument was administered online using the university s Opinio platform, which was kept live for a period of three months. Scientific realists state that theoretical constructs can always exist without having any observable referents (Hunt, 1991). Accordingly, this study involves several latent constructs, which are non-observable. The multi-scale nature of the data and the use of ordinal scales require the use of polychoric correlation matrices of software programs (Hair et al., 1998). Hence, SPSS Statistics 20 (Statistical Package for the Social Sciences) and AMOS Version 20 (Analysis of Moment Structures) were used to analyse the data. Exploratory Factor Analysis (EFA) was employed to determine the factor structure and Confirmatory Factor Analysis (CFA) was performed to examine whether theoretical relationship between items and their hypothesised factors were supported by the data (Cunningham, 2010). Subsequently, Structural Equation Modeling (SEM) was used to estimate the proposed and competing models. SEM is popular in marketing and management research due to its ability to effectively test theoretical models, specifically those consisting of latent constructs (Anderson and Gerbing, 1988). Moreover, SEM is the best choice for analysing the confirmatory nature of the research question and simultaneous nature of multiple relationships between the key constructs (Hair et al., 1998). 13

34 1.8 Outline of the thesis The current chapter discusses the purpose and background to this thesis and provides the structure of the thesis. Chapter 2 examines the literature on customer loyalty and relationship marketing with a specific focus on services and B2C relationship marketing. It commences with a discussion on services marketing (section 2.2), which is subdivided into the evolution of services marketing, the nature of services, and service quality. The review of the services marketing literature contextualises related extant research undertaken in service quality and service quality dimensions of an ISP. Section 2.3 then presents a discussion on relationship marketing including customer retention and the benefits of relationship marketing in the telecommunications sector. It also discusses the issues pertaining to customer loyalty, which are subdivided into customer loyalty, behavioural loyalty, attitudinal loyalty and the benefits of customer loyalty. Section 2.4 discusses the business and marketing environment in Thailand, historical aspects of the Internet, provision of Internet services in Thailand, and consumer behaviour in Thailand. Finally the chapter addresses a section on the research questions and research objectives (section 2.5) and the benefits of researching service quality in the telecommunications industry (section 2.6). Chapter 3 presents the theoretical model for the research study, as well as the research hypotheses. It commences with a review of the literature relevant to key concepts underlying the development of the theoretical model (section 3.2). Section 3.2 is divided into an overview of an ISP s service quality dimensions; a cognitive and affective evaluation of customers; and discussion about the dependant or outcome constructs. Section 3.3 presents a discussion on previously validated models. The next section (section 3.4) presents an outline of the development of the proposed theoretical model. Section 3.5 presents a discussion on an ISP s service quality dimensions, which is subdivided into four sections. These sections include (1) network quality; (2) customer service and technical support; (3) information quality and website information support; and (4) privacy and security. The next section (section 3.6) presents the cognitive and affective evaluations of customers. Section 3.6 is divided into (a) ISPs overall service quality and customer trust; (b) ISPs overall service quality and customer satisfaction; (c) ISPs overall service quality and customer commitment; and (d) ISPs overall service quality and customer value. The chapter then addresses issues relating to 14

35 dependant constructs (section 3.7), which include concepts of customer trust and customer loyalty; customer satisfaction and customer loyalty; customer commitment and customer loyalty; and customer value and customer loyalty. Chapter 4 discusses the research design including justifications for the use of quantitative methods, online survey, sampling and Structural Equation Modelling (SEM). It commences with a discussion of the research paradigm (section 4.2), survey research (section 4.3) and sampling (section 4.4). Section 4.5 then presents a discussion of the data collection procedures. It is subdivided into the selection of the sample, nonresponse bias, unit of analysis, pre-testing and the pilot study. The next section (section 4.6) presents a justification for the analytical technique. This chapter also discusses the issues of validity and reliability and the steps taken to minimise any related errors (section 4.7). The sampling issues of selection and size are discussed along with response rates. The development of measures and scales used are discussed (section 4.8). Section 4.9 discusses the foundations of the steps involved in Structural Equation Modelling. Ethical considerations are provided in Section Chapter 5 presents the results of the data analysis. This chapter begins with data screening (section 5.2), a profile of respondents (section 5.3), details of internet services used (section 5.4), additional descriptive statistics (section 5.4) and segmentation based on internet usage patterns (section 5.6). Profiles of the participants are provided, followed by a description of the statistical measures adopted to assess each of the constructs of the conceptual model. The chapter proceeds to assess the validity and reliability of the constructs. This is followed by exploratory factor analysis (section 5.7), confirmatory factor analysis, building of measurement models (section 5.8) and full Structural Equation Model analysis (section 5.9). Sections 5.10 presents segmentation analysis using invariance testing done on the basis of internet usage patterns, age groups and income levels. The results from the study are integrated in Chapter 6. This chapter also presents a detailed discussion of each of the individual dimensions included in the conceptual model. Chapter 6 starts with a discussion of the implications of the best-fit Structural Equation Model (section 6.2). The service quality dimensions are reviewed, and the 15

36 relationship between service quality, customers cognitive and affective evaluations and customer loyalty in the context of home Internet services is presented. This is followed by a discussion of findings of the segmentation analysis (section 6.3). The overall theoretical and managerial implications, as well as action plans of the research are discussed (section 6.4). The chapter concludes with the limitations of the study, directions for future research (section 6.5) and concluding remarks (section 6.6). 1.9 Glossary Definitions for glossary terms that are frequently discussed in this thesis are presented in Table Table 1.1: Glossary B2B B2C B2B2C CFA CFI df EFA GFI AGFI IFI ISP RMSEA SEM SRMR TLI x2 x2/df Term Definition Business-to-Business Business-to-Consumer Business-to-Business-to-Customer Confirmatory Factor Analysis Comparative Fit Index Degrees of Freedom Exploratory Factor Analysis Goodness of Fit Adjusted Goodness of Fit Bollens Index Internet Service Provider Root Mean-Square Error of Approximation Structural Equation Modelling Standardised Root Mean Square Residual Tucker Lewis Index Chi-Square Normed Chi-Square 1.10 Definitions Definitions for theoretical and analytical terms that are frequently discussed in this thesis are presented in Table

37 2Table 1.2: Definition of theoretical, analytical and methodological terms used Term Attitudinal loyalty Behavioural loyalty Customer commitment Customer loyalty Customer satisfaction Customer trust Customer value Confirmatory Factor Analysis (CFA) Congeneric measurement model Construct Exploratory Factor Analysis (EFA) Service quality Structural Equation Modelling (SEM) Latent construct Definition A customer predisposition towards a brand, which is a function of psychological processes (Jacoby and Chestnut, 1978). A consumer s tendency to purchase and repurchase a product, which is shown through behaviours that can be measured and impact directly on product sales (Worthington et al., 2010). A customer s conviction and enduring desire to maintain a relationship that might produce functional and emotional benefits (Hur et al., 2010). A customer s deeply held emotional commitment to stay with a product, service, brand or provider consistently in the future, and a willingness to recommend the service provider to peers (Oliver, 1997). Customers feelings of happiness, fulfilment and pleasure towards a service provider and its services through their overall experience with the company (Parasuraman et al., 1985). A customer s perception of a service provider s attributes, including its ability, integrity, and benevolence (Deng et al., 2010). An exchange between what customers receive and what customers give in order to purchase a service (Tam, 2012). A model, where the number of dimensions (latent variables) is constructed in advance (Bollen, 1989). CFA is a theory-testing model where researchers begin with a hypothesis concerning the factor structure prior to analysis (Hair et al., 1998). A one-factor model demonstrating the regression of a set of observed variables on a single latent variable (Cunningham, 2010). As the latent construct in a congeneric model is unidimensional, the correlated error terms do not exist (Cunningham, 2010). Concept that the researcher can define in conceptual terms but cannot be directly measured (Hair et al., 1998). Exploratory factor analysis is a statistical method which can be utilised to identify interpretable factors and ensure that the indicators of the factors represent the latent construct (Hair et al., 2006). An indicator of the performance of a delivered service in comparison to customers expectations (Oliver, 1980). A multivariate technique that combines aspects of multiple regression and factor analysis to estimate a series of interrelated dependence relationships simultaneously (Hair et al., 1998). Operationalisation of a construct in SEM. A latent construct cannot be measured directly but can be represented by one or more variables (Hair et al., 1998). 17

38 1.11 Chapter summary This chapter has provided a rationale and context for this research study, as well as a roadmap for the thesis. It commenced with a short description of the background for the research (section 1.2). This was followed by the identification of research gaps in Section 1.3 which included (a) the research gap related to service quality, and (b) the research gap related to customer retention and customer loyalty. Section 1.4 articulated the research questions. Section 1.5 discussed the research contributions from both academic and practical perspectives. Section 1.6 demonstrated the conceptual framework and research objectives of the study. The methodology and data collection were presented in Section 1.7. Section 1.8 outlined the chapters of the thesis. Glossary and definitions were provided in Section 1.9 and Section 1.10 respectively. Figure 1.3 provides a roadmap to the structure of the overall thesis. 18

39 Chapter One An introduction to the thesis and overview of the chapters Next chapter Chapter Two Literature Review Chapter Three Development of the theoretical model and related hypotheses Chapter Four Methodology Chapter Five Findings and Results Chapter Six Discussion, Conclusions and Recommendations 3Figure 1.3: Structure of the overall thesis 19

40 Chapter 2: Literature Review 2.1 Chapter overview This chapter provides a review of the literature associated with services marketing, service quality, relationships marketing and customer loyalty. It also provides the context and focal point of the study. It commences with a discussion on services marketing (section 2.2), which is subdivided into the evolution of services marketing, the nature of services, and service quality. The review of the services marketing literature contextualises related extant research undertaken in service quality and service quality dimensions of an ISP. Section 2.3 then presents a discussion on relationship marketing and customer retention and the benefits of relationship marketing in the telecommunications sector. The issues pertaining to customer loyalty, behavioural loyalty, attitudinal loyalty and the benefits of customer loyalty are also discussed. Section 2.4 introduces the business and marketing environment in Thailand, historical aspects of the Internet, provision of Internet services in Thailand, and consumer behaviour in Thailand. Finally the chapter addresses a section on the research questions and research objectives (section 2.5) and the benefits of researching service quality in the telecommunications industry (section 2.6). The organisation of this chapter is depicted in Figure

41 2.1 Chapter Overview 2.2 Services marketing 2.3 Relationship marketing 2.4 Context of the study 2.5 Research questions 2.6 Benefits of service quality in the telecommunications industry 2.7 Chapter conclusion 4Figure 2.1: Chapter organisation 21

42 2.2 Services marketing This section provides a discussion on services marketing, which is sub-divided into the evolution of the services marketing literature, the nature of services, and service quality. The review of the services marketing literature contextualises related extant research undertaken in service quality and service quality dimensions of an ISP Evolution of the services marketing literature Over the past 40 years, services marketing has become a well-established area of research in the marketing discipline (Fisk et al., 1993; Furrer and Sollberger, 2007; Lovelock, 1983). Fisk et al. (1993) remark that since the 1970s, services marketing has grown and become an important area because of the tremendous increase in both the supply of and demand for services. There have been a multitude of new services in the last 40 years, in Business-to-Business (B2B), Business-to-Customer (B2C), and Business-to-Business-to-Customer (B2B2C). In fact, there is evidence that since the early 2000, services have become a forerunner to economic activity in developed countries and they continue to grow. This growth is taking place while the two traditional sectors associated with goods manufacturing and agriculture are in decline (Gummesson, 2007). Fisk et al. (1993), Furrer and Sollberger (2007) and Lovelock (1983) have investigated various stages in the evolution of the services marketing literature commencing from They write that there are three identifiable stages in the evolution of the services marketing literature from its embryonic beginnings in 1953 to its maturity in These stages are termed crawling out (1953 to 1979), scurrying about (1980 to 1985), and walking erect (1986 to 1993) and a discussion of these stages follows. The first stage, crawling out (1953 to 1979), is identified as the inception of the services marketing literature (Fisk et al., 1993; Furrer and Sollberger, 2007; Lovelock, 1983; Vargo and Lusch, 2004). There were numerous questions and arguments relating to the specific nature of services marketing. For example, the need for a dedicated services marketing literature and the differences between services and products marketing. Fisk et al. (1993) and Shostack (1977) state that the crawling out stage began when pioneer services marketing scholars struggled to publish their work justifying the differences between the marketing of services and goods. The associated 22

43 arguments led to the identification of the distinctive characteristics of services, such as intangibility, heterogeneity, inseparability, and perishability (IHIP), which eventually became the main underpinning of services marketing (Furrer and Sollberger, 2007). In 1964, Judd (1964, pp ) established the basis for the modern approach to services definition by discussing illustrative definitions and definitions by listing. Judd (1964, pp ) suggest defining services "by exclusion (from products)... pending the development of a positive definition," because, "together with a product definition, it exhausts the category of economic goods". Nevertheless, Judd remarks that this view failed to address the typical characteristics of services. Two years later, Rathmell (1966, p 32) claims that the majority of marketers have some idea of the meaning of the term goods, but services seem to be everything else. The author asserts that all goods and services could be grouped into the goods-services basket. The author also elicits 13 marketing characteristics of services, which include intangibility, non-inventoriability, and non-standardisation. The second stage, scurrying about (1980 to 1985) was a bridging period when the services versus products debate began declining with more people becoming aware of the special nature of services marketing and its needs (Solomon et al., 1985; Zeithaml et al., 1985). Fisk et al. (1993) state that there were fewer questions relating to the differences between services and product marketing at this stage. During this period, Lovelock (1983) reports that researchers put too much effort into identifying differences between products and services marketing and not enough was done to develop a good understanding of marketing practices in the services sector. Lovelock (1983) then proposes a number of services classifications corresponding to different types of marketing treatment. During the same period, Sasser et al. (1978) and Oliver (1980) commenced the debate on service quality primarily relating to perceived service quality (Berry et al., 1985; Parasuraman et al., 1985). Shostack (1984) introduces a new focus on service design and mapping, and Czepiel et al. (1985) lay the foundation for service encounters. The third stage, termed walking erect (1986 to 1993), was when scholars received positive feedback and services marketing became a recognised area of study within the 23

44 marketing discipline (Lovelock, 1991). Fisk et al. (1993), Swartz et al. (1992) and Furrer and Sollberger (2007) state there were a number of publications on related topics as the study of services marketing matured considerably. These topics included managing quality and service experience; designing and controlling intangible processes, managing supply and demand in capacity constrained services; and organisational issues resulting from the overlap in marketing and operational functions. For example, Parasuraman et al. (1991) investigate the pros and cons of alternative methodologies to measure service quality, followed by Cronin and Taylor (1992); Parasuraman et al. (1993) and Brown et al. (1993). Following the walking erect stage (1986 to 1993), a number of researchers have investigated various facets of service offerings. One of their main focuses has been identification of the four characteristics of services: intangibility, heterogeneity, inseparability, and perishability (IHIP). This area has been comprehensively researched by Grönroos (2000); Gummesson (2007); Gummesson et al. (2010); Rahman and Areni (2010). Another area of services marketing is service quality and its conceptualisation, as well as the benefits of service quality. Authors noted for their work in this area include Asubonteng et al. (1996); He and Li (2010); Morgan and Hunt (1994). Though services marketing has thrived over the last few decades, several researchers have raised some concerns on its development (Lovelock and Gummesson, 2004). Lovelock and Gummesson (2004) contend that the predominance of a single topic, such as service quality measurement, has hindered the growth of other areas. Moreover, it has been argued that current service concepts are not completely applicable to Internet services in this information age (Lovelock and Gummesson, 2004). As such, new paradigms and fresh perspectives are required to maintain the success of services marketing (Lovelock & Gummesson 2004). In the new marketing era, the focus is moving from tangibles toward intangibles, for example, skills, intelligence, and experience, as well as toward interactivity, connectivity, and longevity of relationships (Vargo & Lursch, 2004). Likewise, the academic attention is changing from the object exchanged to the exchange process (Vargo & Lursch, 2004). The next section provides a discussion on the nature of services. 24

45 Nature of services Since 1960, there have been several scholars who have defined and discussed services as intangible activities and processes undertaken by service providers aimed at delivering benefits to consumers (Edvardsson et al., 2005; Gummesson et al., 2010; Lovelock, 1991). Lovelock (1991) defines a service as a process, method or performance rather than an object. Solomon et al. (1985, p. 106) believe that services marketing refers to the marketing of activities and processes rather than objects. Similarly, Vargo and Lusch (2004) describe a service as the use of skills and knowledge through activities, processes, and performances for the benefit of all parties. Hence, services can be seen as intangible activities and processes by service providers meant to deliver benefits to consumers. Services can include legal advice, finance consultancy and housekeeping among others (Gummesson, 2007). Internet services are the main focus of this study. Therefore, along with the discussion on the general nature of services, this study attempts to relate these characteristics to services in an ISP context. When dealing with physical goods, customers expect to see, hold and make use of the products which emerge at the end of the production process (Rahman and Areni, 2010). However, the consumption of services is process rather than outcome driven (Edvardsson et al., 2005). Consumers perceive the production process as part of the service consumption, unlike the traditional marketing of physical goods where they only receive and see the outcomes of that process (Grönroos, 1998; Vargo and Lusch, 2004). For example, when dealing with Internet services, the signals are made available for customers at the moment they use the Internet; therefore during the process of usage, the performance of an ISP is assessed by their customers. In other words, products are expected to be seen and grabbed by customers, while services are an intangible activities and processes to deliver benefits and provide solutions to customer s problems (Gummesson, 2007). In terms of management, managing services is different from managing goods due to their differing natures (Gummesson et al., 2010). Characteristics that are unique to services include intangibility, heterogeneity, inseparability and perishability (Gummesson, 2007; Gummesson et al., 2010; Rust and Chung, 2006). Intangibility refers to consumers being unable to make physical contact with services (Rust and 25

46 Chung, 2006; Vargo and Lusch, 2004). For example, it is impossible for customers to see, feel or hold the Internet connection. The customers can only experience the Internet connection thought their Internet modem and equipment. Heterogeneity indicates it is rare for service providers to offer and present the same outcomes each time to their customers, due to the presence of human factors (Rahman and Areni, 2010; Zeithaml et al., 1985). For example, a car manufacturer can control the standard of its product as the production system is controlled by machines and automated systems. Services, which in part are delivered by the actions of human beings, generally result in service quality that is difficult to standardise and measure before purchase (Gummesson et al., 2010). However not all services are the same. For example, with Internet services the Internet speed, connection status and coverage can be partly controlled due to automated processes and systems. This type of service is entirely different when compared to the services provided by a hairdresser, where the human aspect is predominant. Inseparability specifies how customers work together with the services providers during the stages of production, delivery, and consumption of services (Lovelock, 1991; Vargo and Lusch, 2004). For example, instructions and skills are received by students at the same time as they are delivered by the teachers. However, customers need not necessarily be present during all the stages of production, delivery and consumption in order to receive the service (Gummesson et al., 2010). This also applies to ISPs, as customers do not make physical contacts with the ISPs while using the Internet at their home. However, in order to access the Internet, a customer must first obtain connection via their ISP. Perishability means that services cannot be saved, stored, resold or returned (Gummesson, 2007) For example, a business consultant may charge a client who was unable to attend a meeting because the value exists only at the point when the client should have been present. Similarly, in an ISP market, when a customer purchases an Internet package on a 24-month contract, the customer still has to pay their ISP each month until the contract ends, even though he or she may travel overseas and not use the 26

47 Internet for a period of time. To illustrate perishability, Zeithaml et al. (1985) cite an example of a hotel room which is unoccupied for a period of time. To summarise, intangibility means that consumers are unable to touch, see, taste or smell a majority of services. Heterogeneity refers to the unlikeliness that a service provider will offer the same service experience each time due to the presence of human factors. Inseparability indicates that customers must participate in at least one of the stages associated with production, delivery, and consumption, together with service providers, in order to receive the services. Perishability means services cannot be preserved for future occasions or claimed back once the time of potential service delivery has occurred (Gummesson, 2007; Gummesson et al., 2010; Rahman and Areni, 2010). Section provides an introduction and review of service quality and a discussion on a specific set of service quality dimensions which are associated with residential Internet services Service quality The shift from transactional marketing to relationship marketing places service at the centre of business efforts to improve competitive advantage and profitability (Tohidinia and Haghighi, 2011). Service providers offer and compete to some degree on the basis of providing first-rate services in order to maintain their profitability in a competitive marketplace (Lovelock et al., 2002; Nemati et al., 2010), as well as to increase the level of customer satisfaction and customer retention (Ojo, 2010; Seth et al., 2008). The first stage of evolution of the services marketing literature crawling out (1953 to 1979) did not deal with service quality. However, interest in service quality and its measurement increased significantly in the 1980s (Oliver, 1980; Parasuraman et al., 1985). During the second stage of services marketing scurrying about (1980 to 1985) Sasser et al. (1978) and Oliver (1980) mark the start of the service quality discussion with a focus on perceived service quality. This was followed by Berry et al. (1985), then Parasuraman et al. (1985). The third stage walking erect (1986 to 1993) saw numerous publications on service quality and its dimensions, including the pros and cons of alternative methodologies to measure service quality. This stage was started by Parasuraman et al. (1991) and further discussed in research by Brown et al. (1993); 27

48 Cronin and Taylor (1992) and Parasuraman et al. (1998). Subsequently, other aspects of service quality and its conceptualisation became the main focus of scholars (Asubonteng et al., 1996; Morgan and Hunt, 1994; Rust and Oliver, 1994; Sachdev and Verma, 2004) What is service quality? Service quality is an indicator of the performance of a delivered service in comparison to customers expectations (Oliver, 1980; Parasuraman et al., 1985). In line with this thinking, Crosby (1979); Oliver (1993) and Spreng and Mackoy (1996) state that service quality is a company s compliance with customers requirements, including both the delivery process and outcome. Following their method, service quality can be defined as the consumers overall impression of correlative lower or higher performance of a business and its services (Butt and de Run, 2009; Oliver, 1980; Parasuraman et al., 1985). In addition, Ojo (2010) and Oliver (1980) point out that customers evaluate the quality of a service based on their previous experience with the service or with similar services. For example, after switching to another ISP, customers tend to evaluate the current service based on the standards of their previous service. In other words, the new service is perceived to be of high quality if it performs at the same or higher level than the old one, especially when customers were dissatisfied with their previous experience and they expect a better outcome by switching to a new ISP (Lee et al., 2001). For this reason, customers expectations provide the foundation on which service quality is evaluated (Oliver, 1980). It can be seen that quality is considered to be low if performance does not meet customers expectations; in contrast, high quality refers to a performance that corresponds to or exceeds their expectations. In an attempt to generalise the definitions of service quality, Ghobadian et al. (1994) proposes five broad categories of quality : (1) transcendent (i.e. natural or innate excellence); (2) product led (i.e. the units of goodness packed into a product or service), (3) process or supply led (i.e. conformance to requirements), (4) customer led (i.e. satisfying a customer s requirements), and (5) value led (i.e. costs to the producer and price for the customer). Among these categories, the customer-led definition is 28

49 generally recognised and used since customers play an important role in services and have been placed at the centre of businesses (Ghobadian et al., 1994). However, telecommunications services characterised by technology standard compliance is more likely to place quality in the process led category. A common trend in the market place is for service providers to include both values and customers requirements in a quality definition, as these factors can potentially contribute towards a service provider s competitive advantages, for example improvements associated with technical problems (Kyriazopoulos et al., 2007). In summary, this study attributes service quality to the performance of a service and how this matches customers expectations, which can be formed as a result of previous experience with the same or similar services. A service can be considered to be of low or high quality. This definition falls within the domain of the customer-led category, and recognises the increasingly important role of customers specifically in services and in the business world in general (Ojo, 2010; Oliver, 1980) A review of service quality models Numerous studies on service quality measurement have been performed not only in academic research, but have also been extensively applied in the industry (Asubonteng et al., 1996). For example in health care (Clow et al., 1995); a dental school patient clinic (Carman, 1990); independent dental offices (McAlexander et al., 1994); AIDS service agencies (Fusilier and Simpson, 1955); physicians (Walbridge and Delene, 1993); large retail chains (Teas, 1993); fast-food restaurants (Cronin and Taylor, 1992); mobile phone service (Aydin and Özer, 2005; Huan et al., 2005; Lim et al., 2006; Santouridis and Trivellas, 2010); and Internet service providers (Cheng et al., 2008; Dwivedi et al., 2010; Johnson and Sirikit, 2002). The first service quality model called SERVQUAL was introduced by Parasuraman et al. (1985). SERVQUAL was developed and based on Parasuraman et al. s (1985) gap model between performance and expectations: as performance exceeds expectations, quality increases and vice versa. The central idea in this model views service quality as a function of the different scores or gaps between expectations and perceptions (Bloemer et al., 1999; Zeithaml et al., 1996). According to the disconfirmation model 29

50 (Cooper et al., 1989), a high quality service is at a level that meets or exceeds consumer expectation, as opposed to a low quality service, where the performance is below expectations. For example, Lee, Lee and Feick (2001) confirm that customer expectation and perceived performance of services were found to be the main antecedents of service quality in the French mobile phone service industry. Zeithaml, Berry and Parasuraman (1993) assert that the model offers an effective way to measure desired service levels, minimum service levels, and consumers perceptions of actual service. Extending this argument, Parasuraman (2004) propose the zone of tolerance concept indicating the difference between desired service (i.e. customer expected level of service) and sufficient service (perceived adequate level of service). This is justified as customers have a range of expectations rather than having only one perfect level. Parasuraman (2004) regards this range of expectations as the zone of tolerance, in which desired service is at the upper end, and adequate service is at the lower end. If the service performance stays within the tolerance zone or exceeds the desired level, customers are likely to be satisfied and be delighted (Parasuraman, 2004). However, if the service delivery goes below the zone of tolerance, customers will experience dissatisfaction and disappointment (Parasuraman, 2004). Brady and Cronin (2001), as well as Zeithaml et al. (1990), name five key dimensions of SERVQUAL including: (1) Tangibles: the appearance of physical facilities, equipment, personnel and communication materials; (2) Reliability: the capacity to execute the promised service responsibly and precisely; (3) Responsiveness: the willingness to help customers; (4) Assurance: the knowledge and courtesy of employees, and ability to increase customers trust and confidence; and (5) Empathy: the caring, individual attention provided to customers. However, many researchers have questioned the validity of this model since customers disconfirmation cannot be practically applied in continuously supplied services which observe passive customer expectations. Disconfirmations are also likely to happen only when service alterations occur out of experience-based standards (Blery et al., 2009). For example, broadband customers do not generally hold differing expectations towards the performance of common industry standards which all ISPs have to comply with. 30

51 As a result, another simple performance-based only measure termed SERVPERF was originated by Cronin and Taylor (1992). SERVPERF disposes of customer expectations in SERVQUAL and uses only customers perceptions of actual service performance as an indicator of service quality. Researchers have applied SERVQUAL in various areas of study, such as higher education (Abdullah, 2006; Cui et al., 2003), the retail sector (Mehta et al., 2000), the ceramic industries (Llusar and Zornoza, 2000), libraries (Nejati and Nejati, 2008), the automotive repair industry (Andronikidis, 2009) and fast food restaurants (Qin et al., 2010). Advocates for SERVPERF argue that customers generally manifest their expectations by rating service performance; therefore there is no point in asking about their expectations alone (Carrillat et al., 2007). Additionally, there are some arguments on the role of expectations as a comparison point with service performance. Teas (1993) contends that as Parasuraman et al. (1985) determine expectations as a type of attitude, they should be examined as ideal points. In other words, service quality reaches its peak when equaling expectations. This leads to a theoretical inconsistency in superiority perceptions, which is suggested as the point when performance exceeds expectations (Teas, 1993). Though this idea still needs to be discussed further, it suggests that including expectations as a standard in assessing service quality is problematic and unnecessarily complicated. Despite the controversy surrounding SERVQUAL and SERVPERF, Carrillat s et al. (2007) meta-analytic study asserts that both instruments are equivalently meaningful measures of overall service quality. Other researchers (Cronin et al., 2000; Zeithaml et al., 1996) state that SERVQUAL is a suitable tool for identifying service underperformance and assessing the variation of dependent conceptual elements, while SERVPERF is often considered to be a suitable method for forecasting after-effect factors, such as customer satisfaction and customer loyalty. The main indicators of service quality, identified by using different instruments such as SERVQUAL and SERVPERF, in the telecommunications sector may vary depending on the business environment, for example, culture. For example, in the Turkish telecommunications industry, network coverage, customer complaints handling, valueadded service, promotional activities and their accomplishment have been found to be important factors in service quality (Aydin and Özer, 2005). Connection strength and 31

52 availability are critical in evaluating the performance of the telecommunications service in China (Wang et al., 2004; Woo and Fock, 1999), and Korea (Kim and Yoon, 2004). These factors illustrate the importance of the dimensions mentioned earlier, including reliability, responsiveness, assurance and empathy. They are all intangible and are often not obvious to customers because customers usually cannot see the network architecture or infrastructure. Section provides an introduction and review of an ISP s service quality dimensions Service quality dimensions of an ISP There have been a large number of researchers who have focused on the mobile services sector and the related influence of service quality in the past (Abdolvand et al., 2006; Alshurideh, 2010; Huan et al., 2005; Lim et al., 2006). However, few studies can be found that specifically look at the Internet services industry (Biczók et al., 2010; Cheng et al., 2008; Dwivedi et al., 2010). Though sharing many common features of quality with mobile services, Internet services possess some unique characteristics that need to be investigated thoroughly, such as network speed (download and upload speed) and its consistency. This area therefore deserves further investigation. A search on Google Scholar for research publications exploring an ISP s service quality dimensions (on March 10th, 2014) revealed no results at all (except the articles published from this study). Hence, this study aims to investigate the specific service quality dimensions and attributes and their differing influence on the overall service quality of an ISP. The two widely used scales for evaluating service quality are SERVQUAL for generic services and E-S-QUAL for services associated with information technology (He and Li, 2010; Rafiq et al., 2012). However, neither of these scales take into account the specific issues relevant to customers of high-tech ISPs (Thaichon et al., 2014). While SERVQUAL measures service quality in general businesses, E-S-QUAL focuses on service providers that operate using the Internet platform (Vlachos and Vrechopoulos, 2008), but neglects those that actually provide the Internet connection and platform for online B2B and B2C activities. It is vital for Internet service providers to obtain accurate information regarding their service quality as perceived by their customers. This information would certainly enable them to formulate appropriate marketing 32

53 strategies, which would work in their favour towards achieving competitive advantage and long term sustainability. Ganguli and Roy (2010), as well as He and Li (2010), reveal that perceived service quality dimensions are not limited to those identified in SERVQUAL and E-S-QUAL. Researchers have recently attempted to develop service quality measurement scales in different high-tech contexts, for example Shamdasani et al. (2008) in self-service Internet technologies and Vlachos and Vrechopoulos (2008) with mobile telephony. Nevertheless, none of these researchers comprehensively evaluate an ISP s service quality dimensions. Several basic differences exist between Internet services and other telecommunications services. For example, mobile service quality includes value-added services (e.g. SMS, MMS, WAP, GPRS) or mobile devices (Santouridis and Trivellas, 2010), which are not applicable in case of Internet services. In-home Internet services technical support on websites is critical when assessing an ISP s service quality, but this might not be significant for other telecommunications services, such as television transmission. Such an inconsistent conceptualisation deserves further investigation. In summary, SERVQUAL and SERVPERF are the main instruments for measuring service quality. While SERVQUAL, developed by Parasuraman et al. (1985), determines consumer perceptions of service quality by comparing the difference between performance and expectations, SERVPERF is a performance-based only measure that assumes customer expectations are implied in their performance evaluation. This study does not intend to debate the efficacy of either of these two models; it aims to investigate the specific service quality dimensions and their related attributes as applicable to the telecommunications industry, specifically the home Internet services industry. SERVQUAL and SERVPERF do not take into account the specific issues relevant to customers of high-tech services, and in particular those of residential Internet services. 2.3 Relationship marketing The study of relationship marketing began in the 1960s and became an important area of research at the beginning of 1980s (Sheth and Parvatiyar, 2002). In the 1960s, the focus of relationship marketing was on customer needs and wants, and customer satisfaction 33

54 (Gummesson, 1999; Lambe et al., 2002). The objective of these early studies was to enhance the positive relationships between customers and the company by offering a higher value product at a lower cost (Sheth and Parvatiyar, 2002). In the 2000s, relationship marketing became an important area that received much attention and awareness (Bennett, 2001; Kyriazopoulos et al., 2007; Trkman et al., 2008). Relationship marketing can be construed as the art of attracting, maintaining, and enhancing relationships with all stakeholders who are involved with the company (Tohidinia and Haghighi, 2011). It includes customers, employees, employers, shareholders, local communities, government and the like. This view of relationship marketing has been supported by a number of researchers, such as Berry (2000); Gummesson (1994); Morgan and Hunt (1994) and Tohidinia and Haghighi (2011). Relationship marketing was introduced by Berry in 1983, for the purpose of establishing, maintaining, and enhancing the relationships between customers and other partners (Berry, 2000). Sheth (1994) describes relationship marketing as the understanding, explanation, and management of ongoing relationships between suppliers and customers. Similar to Sheth (1994), Zinkhan (2002) describes relationship marketing as an approach to start, preserve, and boost the long-term relationship with all stakeholders. Gummesson (1994); Hunt et al. (2006) and Morgan and Hunt (1994) agree that relationship marketing is an integrated attempt to recognise, keep, and build up a network between consumers and a company, which considers the benefits to both parties, through interactive, individualised and value added contacts over a long period of time (Shani and Chalasani, 1992). In line with this thinking, Morgan and Hunt (1994) remark that relationship marketing refers to all marketing activities directed towards establishing, developing, and maintaining successful relational exchanges (p. 22). This concept is applied in both B2B and B2C contexts with typical characteristics, such as the number of customers, the degree of customisation and customer turnover rate, which leads to a variety of approaches and ideas (Zinkhan, 2002). In short, relationship marketing is a strategy to create, maintain and develop relationships with all stakeholders who are involved with the company, in order to gain their trust, satisfaction and commitment to increase revenue for the company. A clear understanding of 34

55 customer motives is necessary for relationship marketing to be successful. However, within the scope of this study, relationship marketing and related areas are examined in the B2C context, with a special focus on customer retention and loyalty in the telecommunications industry, specifically in the residential Internet services market Relationship marketing and customer retention Customer retention is a critical area of research in the field of relationship marketing that is mainly concerned with maintaining customers in the long term (Grönroos, 1994; Nemati et al., 2010). The telecommunications market is rapidly developing (Trkman et al., 2008) and the telecommunications sector has become a dynamic key area for economic and technological development (Nemati et al., 2010). Intense competition exists between mobile and Internet services providers across markets, including in Hong Kong, Korea, Taiwan, United State, United Kingdom, India and Thailand (Nemati et al., 2010; Trkman et al., 2008). The Thai National Statistical Office (TNSO, 2010) reports that an average of 10% of home Internet customers in Thailand switched service providers each year between 2003 and In 2009, the figure was 12% (True, 2010). For this reason, ISPs need to understand reasons attributed to customer loyalty in the Internet services market. When there are high customer churn rates, businesses are required to spend large amounts of money in order to attract new customers (Jandaghi et al., 2011; Mao, 2010; Mathur, 2011). Hence, in a highly competitive business environment, customer relationship marketing is as important as customer acquisition, and companies should endeavour to retain existing customers and also to attract new customers at the same time (Bateson, 1995; Bennett, 2001; Trkman et al., 2008). Bateson (1995) and Jandaghi et al. (2011) demonstrate that customer acquisition costs are increasing, with fewer guarantees that the customers will stay with one company forever. Accordingly, losing a customer represents a loss of future profit from that customer, and an additional cost of attracting a new customer as a replacement (Flint et al., 2011). Hypothetically, a company might spend $50 to retain an existing customer, however it has been proven that attracting a new customer to replace a customer who has departed can cost 5 to 10 times more (i.e. $250 to $500 per new customer) (Flint et al., 2011; Reichheld and Sasser, 1990). In the home Internet services market, customers expect 35

56 lower prices, improved facilities and technology and higher service quality (Ahn et al., 2006). Therefore, ISPs have to provide extra benefits to customers in order to attract new customers and increase customer retention rates. These benefits can include, for example, free handsets (Ahn et al., 2006). Costs can be reduced if an ISP attempts to maintain and enhance its performance and relationship with existing customers (Madden et al., 1999). Madden et al. (1999) confirm that a large number of ISPs in Australia are unprofitable due to difficulty in securing sufficient revenues to offset their costs. Hence, customer relationship marketing is extremely important in these businesses as it is an excellent approach to create competitive advantage. It also allows for a positive flow of information, as well as the formation of a partnership with the customer (Bennett, 2001; Trkman et al., 2008). In return, this approach provides the service provider with an opportunity for a close connection with its customer, which leads to a higher level of customer retention (Trkman et al., 2008). As a result, the company is more likely to survive and be profitable in the long run (Tamosiuniene and Jasilioniene, 2007). Hence, it is important for an ISP s survival to retain its existing subscriber base because of the high costs of attracting new customers Benefits of relationship marketing associated with Internet services Relationship marketing provides companies with the ability to understand what customers really need and want (Tohidinia and Haghighi, 2011). It offers a higher level of service quality and customer satisfaction, which can lead to greater competitive advantage and long term profitability (Ahn et al., 2006; Dwivedi et al., 2010; Kyriazopoulos et al., 2007; Leahy, 2011). For example, Erevelles et al. (2003) demonstrate that as the Internet services market matures, ISPs that pay attention to building an effective relationship with customers will have a competitive advantage in the marketplace. From the buyers perspective, customers are likely to look for an ongoing relationship with the service provider to reduce risk and uncertainty. This view has been supported by several researchers, including Lovelock (1983); Morgan and Hunt (1994); Patterson and Smith (2001) and Shani and Chalasani (1992) and further discussion follows. 36

57 Risk minimisation Consumers are motivated to participate in relationships with service providers in order to reduce risk and uncertainty (Gounaris and Venetis, 2002; Morgan and Hunt, 1994; Tohidinia and Haghighi, 2011). Services are thought to have higher levels of risk than consumer goods due to their intangible, ephemeral and often interpersonal nature in delivery, producing, consumption and evaluation process (Patterson and Smith, 2001). Lovelock (1983) highlights that customers look for an ongoing relationship with a service provider in order to reduce perceived risk. For example, customers are most likely to return to a familiar hairdresser because they do not wish to risk their hairstyle with a new one. Additionally, consumers generally choose a service provider that they can trust to reduce the risks associated with relational exchange (Morgan and Hunt, 1994). Tohidinia and Haghighi (2011) suggest that having strong relationships with customers helps a company to gain a better understanding of their customers needs, and in return to enhance customers confidence in the firm. In summary, there are higher levels of risk and uncertainty in services as customers are unlikely to be able to evaluate a service prior to purchase (Gounaris and Venetis, 2002; Morgan and Hunt, 1994). Thus, customers have to rely on referrals and recommendations by others customer or a good relationship with the service provider to minimise risk (Gounaris and Venetis, 2002) Understanding customers needs and wants It is important for a business to gain a large number of new customers per month in order to grow (Jahanzeb et al., 2011; Wahab et al., 2011). However, a company would suffer a loss if it cannot retain those new customers in future months (Flint et al., 2011). As mentioned earlier by Jahanzeb et al. (2011), attracting a new customer as a replacement can cost 5 to 10 times more than the costs of retaining an existing customer (i.e. $250 to $500 per new customer). In order to maintain existing customers, the company should be able to understand their needs and wants (Bodey and Grace, 2006; Leahy, 2011; Schiffman and Kanuk, 1997; Tam, 2012). A clear understanding of customers needs, wants, and their motivations for maintaining the relationship with a firm is an advantage to businesses (Leahy, 2011; Schiffman and Kanuk, 1997). Strong relationships with customers leads to a better understanding of customers needs, enhances customers confidence in the firm and generates a higher return on investment 37

58 for businesses in the long run (Tohidinia and Haghighi, 2011). In the Indian mobile service industry, Tripathi and Siddiqui (2010) confirm that by knowing where consumer preferences and their values reside, companies can design services incorporating all the requisite components in order to ensure customer satisfaction and loyalty, thus strengthening their competitive position. With this in mind, Dowling (2002) states that relationship marketing has claimed to increase the efficiency and effectiveness of target marketing, customer satisfaction, and customer values for the company, while helping customers to establish a conversation with the company Repurchase and retention To be successful in the long run, businesses must be able to maintain and increase their repurchase and retention rates, because attracting new customers is more expensive and less beneficial than retaining existing customers (Flint et al., 2011; Mathur, 2011; Qian et al., 2011). Customer loyalty programs play a significant role in achieving customer loyalty and enhancing profitability in the Jordanian mobile service market (Wahab et al., 2011). Similarly, Ahn et al. (2006) demonstrate that membership card programs contribute to customer retention in the Korean mobile services (Xevelonakis, 2004). Acquiring new customers is more expensive, whereas better managing existing customers can lead to lower operating costs due to repeat purchase (Abdolvand et al., 2006). Shani and Chalasani (1992) suggest that marketing costs can be reduced by developing long-term relationships between businesses and their customers, which contribute to long-term profitability. Only the ISPs that effectively distinguish themselves on key strategic dimensions are likely to survive and be profitable in the long run (Erevelles et al., 2003). An ISP that differentiates itself from competitors by offering superior customer support, attractive incentives and reliable service will likely have a sustainable strategic advantage in the market place (Erevelles et al., 2003). For example, mobile service providers can give out free handsets to increase customer retention rates (Ahn et al., 2006). A study conducted by Jahromi et al. (2010) reports that relationship marketing has a direct impact on customer churn in the telecommunications industry. For example, superior customer support provided by ISPs to their customers leads to customer satisfaction in the United Kingdom (Dwivedi et al., 2010) and Korea (Kim et al., 2004). 38

59 Consequently, a company with effective relationship marketing could be in a better economic position because of a higher rate of repurchase and retention of its current customers (Qian et al., 2011; Wahab et al., 2011). In other words, there will be a lower rate of profit loss and less costs of attracting new customers as replacement. It is critical for marketers to understand these factors and develop the most effective strategies to attract those people who will become long-term profitable customers (Frow et al., 2011; Tamosiuniene and Jasilioniene, 2007). In order to successfully apply these concepts to the customers, it is important to understand the nature of services in the telecommunications industry Customer loyalty This section provides a review of the literature associated with customer loyalty, behavioural loyalty, attitudinal loyalty and benefits of customer loyalty What is customer loyalty? The concept of customer loyalty was introduced in the 1940s by Guest (1944) and defined in attitudinal terms (i.e. brand preference), and behavioural terms (i.e. share of the market) by Cunningham (2010). Subsequently, Day (1969) proposed a twodimensional concept of customer loyalty that included both attitude and behaviour. Three years later, Jacoby (1971) developed a definition of customer loyalty based on Day s (1969) study, which appears to be the basis of most customer loyalty research today, for example Dick and Basu (1994); Evanschitzky et al. (2006); Jahanzeb et al. (2011); Kabiraj and Shanmugan (2010); Qian et al. (2011); Tam (2012); Thaichon et al. (2014). Loyalty is the biased (non-random) behavioral response (purchase) expressed over time by some decision-making unit with respect to one or more alternative brands out of a set of brands and is a function of psychological processes (Jacoby, 1971, p25). Traditionally, loyalty was deemed to be a form of repeat purchase over time (Jaiswal & Niraj, 2011). However, Bagozzi (1981); Baldinger and Rubinson (1996); Day (1969); Jacoby and Chestnut (1789) argue that an attitudinal perspective for genuine customer loyalty exists. Day (1969); Dick and Basu (1994) and Jacoby and Chestnut (1978) state 39

60 that behaviour-based loyalty is inadequate in reflecting truly loyal customers. Dick and Basu (1994) claim that the behavioural approach lacks a conceptual basis and only focuses on an outcome-based view. Their view is supported by Jaiswal and Niraj (2011); Shirin and Puth (2011) and (Tam, 2012), who state that the motivations behind repurchase might be due solely to promotion rather than customer loyalty. The extent to which customers are loyal is determined, not only by the rate of their repurchase, but also by their preferences and attitudes towards the service provider (Flint et al., 2011; Jones and Taylor, 2007; Qian et al., 2011). For example, loyalty can be assessed based on a willingness to recommend the service provider to others, or feelings and emotional attachments to a product, service, or organisation (Flint et al., 2011; Tam, 2012). In addition, a tendency to consider a favourite service provider as a first choice among alternative providers is another meaningful indicator for evaluating loyalty (Jahanzeb et al., 2011; Kabiraj and Shanmugan, 2010; Nam et al., 2011). This study defines customer loyalty as a customer s deeply held emotional commitment to stay with a product, service, brand or provider consistently in the future, and a willingness to recommend the service provider to peers. This definition of customer loyalty is based on Oliver s (1997) work, which is the basis of most customer loyalty research today (Deng et al., 2010; Díaz et al., 2011; Jahanzeb et al., 2011; Mao, 2010; Wang and Wu, 2012). Customer loyalty comprises behavioural and attitudinal types of loyalty. Behavioural loyalty refers to the frequency of repeat purchases from the same provider (Díaz et al., 2011; Eshghi et al., 2007; Nam et al., 2011). Attitudinal loyalty refers to emotional commitment, including intentions to purchase and intentions to recommend, without necessarily being involved in repurchasing (Han et al., 2011; Jahanzeb et al., 2011; Nam et al., 2011). Flint et al. (2011) describe customer loyalty as the degree to which a customer exhibits repeat purchase tendencies with a service provider, possesses a positive attitudinal disposition toward the provider, and only considers this provider when a need for this service arises. In relation to hospitality services, Nam et al. (2011) describe customers loyalty as their intentions to visit, or willingness to recommend, a particular hotel or restaurant. Similarly, Pan et al. (2012) describe loyalty as the strength of a customer s emotional attachments to a service provider and the willingness to repurchase the 40

61 service provider consistently in the future. In addition, Qian et al. (2011) report that the easiest way to see if a customer is loyal or not is to assess their willingness to recommend the service provider and also how often they repeatedly patronise that provider. Based on the current literature as discussed, this study defines behavioural loyalty as a customer s repurchasing intentions. This study also identifies attitudinal loyalty as being related to customers preferences, emotions and advocacy to a service provider in that they are willing to pay more, and have exclusive consideration and identification with that service provider Behavioural loyalty One of the first studies in the area of behavioural loyalty was conducted by Cunningham (1956), who was followed by Day (1969) and Jacoby (1971). Behavioural loyalty reflects customer actions and implications of the past experience with a service provider, or the probabilities of future purchase given past purchase behaviours (Evanschitzky et al., 2006; Jones and Taylor, 2007; Tam, 2012). Ringham, Johnson and Morton (1994, p44) define behavioural loyalty as the tending of a customer to stick with a supplier (not switch) and can be thought of as the degree to which a customer prefers a supplier over the competition. Likewise, Worthington et al. (2010) suggest that behavioural loyalty is a consumer s tendency to purchase and repurchase a product, which is shown through behaviours that can be measured and impact directly on product sales. In terms of behavioural loyalty measures, there are four popular measurements used by researchers. These include proportion of purchase (Cunningham, 1956; Jahanzeb et al., 2011), purchase probability (Dekimpe et al., 1997), average purchase (Tucker, 1964), and purchase frequency (Díaz et al., 2011; Jahanzeb et al., 2011; Nam et al., 2011). Wahab et al. (2011) state that behavioural loyalty can be measured by the percentage of customer repurchase or frequency of purchase. Likewise, Tam (2012) mentions that behavioural loyalty can be assessed by considering repurchase rates for the same service provider over time. Díaz et al. (2011) also report that behavioural loyalty is based on the frequency of visits, purchases or the percentage of expense. In subscription-type markets, such as telecommunications or finance where the purchase is more continuous, retention is evaluated based on how often and how long a consumer subscribes to the service provider (East, 2000). 41

62 Nonetheless, behavioural loyalty does not distinguish between true loyalty and fake loyalty (Day, 1969; Jaiswal and Niraj, 2011). This is due to the fact that behavioural loyalty fails to identify customers who purchase only because of price differences, convenience and situational factors (Dick and Basu, 1994; Jaiswal and Niraj, 2011), for example, non-availability of a service (Shirin and Puth, 2011; Tam, 2012) or a sales promotion (Dick and Basu, 1994; Jaiswal and Niraj, 2011). Additionally, East (1997) states that situational factors may prevent customers from purchasing an intended product if there is a lack of product availability. In other words, customers usually form an intention to purchase a particular product; however, if the product is not available at the time of purchase, they may select another service provider (Shirin and Puth, 2011; Tam, 2012). It is possible that customers may have no choice but to stay with the service provider, because there are no alternatives in the market or the switching costs are high (Shirin and Puth, 2011). This explains why fake loyal customers, who have large volumes and high frequency of purchases, can quickly switch to other alternatives which they find more appealing (Dick and Basu, 1994; Shirin and Puth, 2011). This limitation can lead to different conclusions in determining customer loyalty towards different service providers. For example, a customer simultaneously uses mobile phone service provider A for 5 years and mobile phone service provider B for 8 years. When measuring this customer s loyalty based on his repeat purchase cycle, it is concluded that the customer is loyal to service provider A because of his higher repurchase rate. In contrast, using proportion of purchase as the measuring method, the customer is considered to be loyal to B rather than A, because he stayed longer with B. Therefore, customer loyalty towards a service provider in comparison with its competitor cannot be determined by a single indicator such as behavioural loyalty. Behavioural loyalty depends on external factors that influence customer behaviour, such as sales promotions (Cunningham, 1956). Clarifying this idea, Rothschild and Gaidis (1981) assert that the reinforcement approach refers to external reinforcers, such as rewards, discount coupons or punitive (i.e. for late payment). The view of reinforcement in loyalty suggests that customers purchase services when incentives are offered, but tend to lose interest as these incentives disappear (Rothschild and Gaidis, 1981). As a 42

63 result, Jacoby and Chestnut (1978) and Day (1969) conclude that although behavioural loyalty provides the most accurate representation of a customer s past behaviour, it cannot confirm that a customer is truly loyal to a service provider without considering other factors, such as attitudes towards the service provider. Hence, Dick and Basu (1994); Jahanzeb et al. (2011); Jones and Taylor (2007) recommend that a study examining both attitudinal and behavioural loyalty as a two-dimensional nature of the construct can provide a richer insight into customer loyalty. These two aspects of loyalty can be further conceptualised as an interaction of attitude and behaviour (Han et al., 2011; Nam et al., 2011; Tam, 2012) Attitudinal loyalty Attitudinal loyalty can be described as internal processes, including emotions, attitudes, values and beliefs, which consumers have toward a service provider (Day, 1969; Jacoby, 1971; Shirin and Puth, 2011; Worthington et al., 2010). Jaiswal and Niraj (2011) suggest that attitudinal loyalty can be viewed as customers preferences, intentions to purchase, and brand choice. Attitudinal loyalty also refers to emotional commitment, such as a sense of attachment and willingness to recommend a service provider without receiving any extra benefit from it (Nam et al., 2011). Worthington et al. (2010) further state that customers with a high level of attitudinal loyalty hold very strong positive beliefs and feelings about a service provider. Attitudinally loyal customers have a higher level of favourable attitudes and emotional attachment towards a service provider (Jaiswal and Niraj, 2011). They feel appreciated and pleased with their previous experience with the company and are willing to recommend the service provider and its services to other consumers (Jaiswal and Niraj, 2011; Ou et al., 2011; Tam, 2012; Wang and Wu, 2012). An emotionally loyal customer usually stays with the company longer and is less affected by situational factors than a behaviourally loyal customer (Jahanzeb et al., 2011). Tam (2012), as well as Wang and Wu (2012) state that these customers will most likely repurchase, make recommendations over time and help attract new customers, which makes them different from behavioural loyal customers, (Jahanzeb et al., 2011; Jandaghi et al., 2011). Attitudinal loyalty can be measured by observing customers intent and emotional commitment to purchase and repurchase a provider s service, as well as a willingness to 43

64 recommend the service provider to other customers (Jahanzeb et al., 2011; Jaiswal and Niraj, 2011; Tam, 2012; Wahab et al., 2011). Jacoby and Chestnut (1978) claim that attitudinal loyalty can be assessed by considering customers feelings and values, which create an individual s overall attachments to a service. Han et al. (2011) state that marketers can look at the beliefs and emotions a customer has toward a service provider after using its service. In addition, real customer commitment is established when customers are motivated to purchase or repurchase without any encouragement such as promotions (Jandaghi et al., 2011). Thus, Jacoby and Chestnut (1978) suggest it is important for marketers to investigate customer attitudes, as well as their needs and wants, in order to fully understand and manipulate customers behaviour, especially their switching behaviour. Attitudinal loyalty does have a shortcoming in that it lacks predictive power towards actual customer purchase behaviour (Jacoby and Chestnut, 1978). This limitation is the result of intervening influences from other factors affecting purchase behaviour, such as distribution issues and customers situational factors, for examples sales promotion and product availability (Baldinger and Rubinson, 1996; Dick and Basu, 1994; Jaiswal and Niraj, 2011; Shirin and Puth, 2011). Bennett (2001) also claims that a high value sales promotion offered by competitors possibly results in temporarily changed attitudes. Thus, a truly loyal customer may also purchase from other service providers on occasions. For this reason, researchers recommend that marketers consider a composite of both attitudinal and behavioural loyalty in the measurement of customer loyalty (Day, 1969; Dick and Basu, 1994; Jahanzeb et al., 2011; Jones and Taylor, 2007). Therefore, Tam (2012) describes customer loyalty as not only an outcome of repeat purchase behaviour, but consequences of an attitudinal process that reflects true loyalty to the service provider. In line with Tam (2012), Dick and Basu (1994) and Jacoby and Chestnut (1978) highlight that customer loyalty is the result of an interface between customers attitudes toward a service provider and their repeat purchase behaviour. In summary, the extent to which customers are loyal is determined not only in terms of repeat purchases from the same service provider over time, but in terms of their preferences and attitudes towards the service provider (Flint et al., 2011; Jones and Taylor, 2007; Qian et al., 2011; Tam, 2012). Nonetheless, customer retention plays an 44

65 important role for any organisation, as it is critical to business survival and profitability. In a highly competitive business environment, customer retention is as important as customer acquisition and companies should put effort into both maintaining current customers and gaining new customers Benefits of customer loyalty In today s competitive and changing market place, customer loyalty is considered a crucial factor for business success (Flint et al., 2011). Attracting new customers is considered more expensive and less beneficial than retaining existing customers (Flint et al., 2011; Kabiraj and Shanmugan, 2010; Mathur, 2011; Qian et al., 2011). Researchers suggest that loyal customers are a competitive asset (Díaz et al., 2011; Qian et al., 2011; Wang and Wu, 2012), and customer retention can be improved through secure and collaborative relationships between customers and service providers (Bowen and Chen, 2001; Bowen and Shoemaker, 1998; Kabiraj and Shanmugan, 2010). A discussion of the benefits follows: Profitability Intense competition has resulted in a growing focus on existing customers and methods to retain them (Jahanzeb et al., 2011). A study by Reichheld and Sasser (1990) demonstrates business performance improves when the number of loyal consumers increases. This is because a loyal customer exhibits higher repurchase rates, more positive word of mouth and less price sensitivity, which leads to greater returns on investment (Bowen and Chen, 2001; Jahanzeb et al., 2011; Jones and Taylor, 2007; Kabiraj and Shanmugan, 2010). Additionally, loyal customers are most likely to stay with their chosen provider and spread favourable word of mouth (Cheng et al., 2008). In other words, loyal customers tend to spend more money with the provider they are committed to and put more effort into promoting the company when compared to a service provider s new customer (Deng et al., 2010; Jahanzeb et al., 2011; Jaiswal and Niraj, 2011). By creating and maintaining customer loyalty, companies can gain more profit over the lifetime duration of the business relationship with customers (Chiou, 2004; Pan et al., 2012). Jandaghi et al. (2011) claim that loyal customers bring more revenue, while also reducing customer acquisition costs. As a result, the company s profitability increases. For example, Kim, Morris and Swait (2008) demonstrate that in 45

66 the designer sunglasses market, a 5% growth in customer retention results in a 40% to 95% increase in profits. In line with this thinking, Jandaghi et al. (2011) confirms that a 5% increase in customers loyalty leads to a 25% to 85% profit increment in the financial services sector Costs reduction Today, companies face costly challenges relating to customer acquisition (Abdolvand et al., 2006). Recruiting new customers does not guarantee long-term success for a company, especially when that company cannot retain those customers (Shirin and Puth, 2011). Therefore, these authors suggest that it is important for companies to pay equal importance to customer acquisition and customer retention in order to enhance their performance and maximise values in the long run. Existing loyal customers are usually less expensive to serve because of their familiarity with the processes involved in purchasing and using the service. They require less training and less assistance (Mathur, 2011). Jandaghi et al. (2011) report that customer acquisition often costs more than customer retention. Research conducted by Mathur (2011) confirms that, in general, service providers need to spend about four times more resources to attract new customers than to maintain existing loyal customers. In the financial services industries, Jandaghi et al. (2011) report that the cost to acquire a new customer is about 15 times higher than retaining an existing customer in the Asian insurance industries. Mao (2010) mentions that in any business, a company has to spend around 1% of its expenses on fostering customer loyalty; in return the company may receive up to 15% more profits, much of this coming via existing loyal customers. Likewise, a 1% increase in customer loyalty is the equivalent of a 10% cost reduction (Kim et al., 2008). Additionally, greater profitability can be achieved because loyal customers spend more money and put in more effort than non-loyal customers, which leads to higher sales and more efficient marketing programs, while simultaneously reducing perceived risk, marketing costs and operational costs (Flint et al., 2011; Jones and Taylor, 2007; Kabiraj and Shanmugan, 2010; Mao, 2010; Mathur, 2011). Loyal customers are less price sensitive (Han et al., 2011; Mao, 2010; Ou et al., 2011), have high purchase frequencies and require less resources for a company to service, for example, no additional acquisition or start-up costs (Díaz et al., 2011). Loyal customers may also 46

67 attract new consumers through their positive word of mouth and recommendations (Díaz et al., 2011; Reichheld and Sasser, 1990). Retaining a loyal customer is more profitable in the long term, when compared with acquiring a new customer (Díaz et al., 2011) Lower customer turnover A customer s loyalty would most likely increase, not just their commitment to a provider, but also their tolerance of errors in the provider s performance. It would also enhance the scope of their relationship with the company and contribute towards reduced customer turnover for the company (Biczók et al., 2010; Jahanzeb et al., 2011; Oliver, 1999). Kabiraj and Shanmugan (2010) state that loyal customers are less likely to switch to other service providers when they experience differences in price or similarity of product, as compared to non-loyal customers. Loyal customers remain mostly unaffected by competitors promotional efforts (Ou et al., 2011). Jahanzeb et al. (2011) point out that a high level of loyalty provides more reasons for consumers not to switch or to search for alternative service providers, due to psychological, physical and economic costs. In the services industry, consumers are motivated to engage in relationships with service providers in order to reduce risk and uncertainty (Gounaris and Venetis, 2002; Morgan and Hunt, 1994; Tohidinia and Haghighi, 2011). Services are thought to carry higher levels of risk when compared to consumer goods, due to their intangible, ephemeral and often interpersonal nature in production, delivery, consumption and evaluation process (Patterson and Smith, 2001). As a result, Lovelock (1983) identifies that customers look for an ongoing relationship with a service provider to reduce the level of their perceived risk Word of Mouth In terms of recommendations or word of mouth promotions, loyal customers are willing to endorse a service provider among their peers, and ignore negative messages promoted by others (Cheng et al., 2008; Jahanzeb et al., 2011; Mao, 2010; Mathur, 2011). This is one of the main reasons why customer loyalty is considered to be a major source of competitive advantage (Díaz et al., 2011; Jaiswal and Niraj, 2011; Tam, 47

68 2012). In contrast, customers who are not loyal can be easily influenced by negative information about a service (Deng et al., 2010; Jahanzeb et al., 2011). For example, if customers frequently complain about the poor quality of a service provider to their peers, it is likely that these individuals will not want to form a relationship with that particular service provider (Mao, 2010). Moreover, attitudinally loyal customers have a higher level of favourable attitude and emotional attachment towards a service provider (Jaiswal and Niraj, 2011). These customers feel appreciated by the provider and pleased with their previous interactions with the company. They are willing to recommend the provider and its services to other consumers (Jaiswal and Niraj, 2011; Tam, 2012; Wang and Wu, 2012). Deng et al. (2010) concludes that loyal customers always spread favourable messages about their service provider and recommend new customers to the company. In summary, Deng et al. (2010) claim that loyal customers increase revenue and reduce costs for all businesses, which leads to greater profitability. The reason for this is that loyal customers are more likely to repurchase and promote their favourite service provider among their peers, while at the same time they are less motivated to switch to alternate providers and are less price sensitive (Jaiswal and Niraj, 2011). They are usually inexpensive to serve because they demand less training and assistance, thanks to their familiarity with the processes involved (Díaz et al., 2011). It is for this reason that studies of customer loyalty and customer retention have become significant and companies have recognised the importance of loyalty and its role in retention of existing customers (Flint et al., 2011; Jaiswal and Niraj, 2011; Wahab et al., 2011). Consequently, the marketing focus has been shifting from customer acquisition to relationship marketing and customer retention (Abdolvand et al., 2006; Chiou, 2004). 2.4 Context of the study Thailand is ranked third in south-east Asia for residential Internet usage, with an estimated 17,483,000 Internet users in 2009 (CIA, 2013) and more than 24 million Internet users in 2012 (IWS, 2013). The number in 2012 represented more than onethird of the Thai population. The competition in Thailand among residential Internet service providers is intense. Currently there are 3 majors ISPs and 16 smaller ones across the country. In this highly competitive market, the churn rate of users was 48

69 approximately 12% in 2009 (True, 2010). This scenario poses huge challenges to ISPs in Thailand, especially in the area of customers repurchase intentions Overview of Thailand Thailand is an independent country that lies in the heart of south-east Asia. It is a constitutional monarchy with King Bhumibol Adulyadej, the ninth king of the House of Chakri, as head of state. He has reigned since 1946, making him the world s longest serving current head of state and the longest reigning monarch in Thailand s history (CIA, 2013). The capital city is Bangkok, which is the country s centre of political, commercial, industrial and cultural activities (CIA, 2013). Thailand is endowed with a wide variety of natural resources, a substantial population and a relatively strong economy (Srihirun, 2011). Thailand experienced rapid economic growth between 1985 and 1995 and is a newly industrialised country with a large tourism industry. It is home to well-known tourist destinations, such as Pattaya, Bangkok, Phuket, and Chiang Mai, and major agricultural exports also contribute significantly to the economy (Guardian, 2010; WorldBank, 2014). In 2010, Thailand was the world s 50th largest country in terms of total land area, with a surface area of approximately 513,000 km 2 (CIA, 2013; WorldBank, 2014). It is the 21st most populous country, with approximately 67 million people. About 75% of the population are Thai, 14% are Chinese, 3% are Malay and 8% are other minority groups, including Mons, Khmers and various hill tribes (CIA, 2013). The country s official language is Thai. Buddhism is the predominant religion in Thailand. The national religion is Theravada Buddhism; 94.6% of the total population are Buddhists of the Theravada tradition. Muslims comprise 4.6% of the population, and is the second largest religious group in Thailand (WorldBank, 2014). The southern tip of Thailand is populated by ethnic Malays, the majority of who are Sunni Muslims. Christians represent 0.5% of the population and there are some small communities of Sikhs, Hindus and Jews (CIA, 2013) Historical overview of the Internet The Internet had its birth in the early 1960s, after the US Department of Defence realised the need for a decentralised computer network that could provide the Pentagon with a command and control communications system in the event of contingencies 49

70 (Simsim, 2011). The Internet also provided a robust and fault-tolerant computer network for the United States military (Teo and Tan, 1998). This pioneering network was known as the Advanced Research Projects Agency network or ARPANET, and slowly appeared in university and government research laboratories owing to its several advantages (Simsim, 2011). These advantages included its ability to let individuals exchange electronic mail, tap into remote databases and operate supercomputers remotely (Teo and Tan, 1998). Teo and Tan (1998) state that the greatest improvement was the communications protocol that gave the Internet its name the Internet Protocol since it allowed numerous computer networks to link up and act as one. Further developments led to the commercialisation of the Internet as an international network in the mid-1990s (Simsim, 2011). By the end of September 2009, the estimated number of Internet users around the world reached billion, indicating that the Internet was being used by more than 25% of the world s population and growing on a daily basis (Simsim, 2011). Trkman et al. (2008) assert that economic growth, competition, educational level, English proficiency, level of democracy, and social networks are positively related to an increase in the number of ISPs around the world. Internet usage is growing across the globe. In the 2010s, electronic communications became key to many people s daily lives, and the Internet was also being regularly utilised for entertainment, commerce and business activities (Wahab et al., 2011). Butt and de Run (2009) state that rapid growth in telecommunications technology can improve economic development by providing employment, improving business efficiency and contributing towards international investments. Magin et al. (2003) point out that an ISP provides both B2B and B2C access to the Internet, which is a critical source of communications and it is linked to related services, such as media content, website building or virtual hosting (Magin et al., 2003) The Internet services market in Thailand The Internet was introduced to Thailand in 1991, however it was only in 1999 that Internet usage rates experienced major growth (NECTEC, 2014). Since 2000, significant investment in education has resulted in greater knowledge and higher 50

71 qualifications among Thai people (Srihirun, 2011). This partly explains the massive development of the telecommunications industry in Thailand (Srihirun, 2011). During the period of 2000 to 2010, the number of Internet users grew by 30% each year (Srihirun, 2011). Similarly, the National Electronics and Computer Technology Centre, (NECTEC, 2014) reported 84.68% growth from 2005 to 2009 (Table 2.1). These figures show that telecommunications play an important part in Thailand s economy, and this is especially true for broadband, which is projected to grow annually at 23% (Srihirun, 2011). 3Table 2.1: Thai Internet users Year Internet Users Year Internet Users 2010 (N/A) ,300, ,300, ,500, ,100, , ,416, , ,413, , ,909, , ,970, , ,000, , ,800, ,500, The market leader in Thailand s ISPs The Internet services market in Thailand is highly competitive and there are 19 Internet service providers (NBTC, 2014). Of these, three large ISPs account for 95% of the Thai home Internet services market. They are True Internet, TOT and 3BB (True, 2014). In 2012, TOT had the highest percentage more than 38% of the Thai total market share (True, 2014). True Internet accounted for approximately 35% and 3BB held 22% market share in 2012 (True, 2014). Although TOT has dominated the ISP market in Thailand since 2005, True Internet is the market leader in Bangkok with 75% market share and has started to expand into other major areas in Thailand, especially the southern regions of Hat Yai, Phuket and Suratani. As a result, True Internet increased its market share in southern Thailand from 2% to 25% between 2008 and 2010 (NECTEC, 2014). Over the same period, in this specific region 3BB slightly strengthened its position with 31% market share, while TOT Internet experienced a drop from 61% to 39% market share (NECTEC, 2014; True, 2014). 51

72 TOT Internet and 3BB TOT Internet is a Thai state-owned telecommunications company. It was established in 1954 and corporatised in 2002 (TOT, 2013). TOT Internet was the first to offer Internet and landline services in Thailand and targeted middle income customers. It has been the market leader in the country, apart from Bangkok, since 1991 (TOT, 2013). In 2002, TOT Internet sparked a low-cost broadband Internet war, which led to a rapid growth in market demand (TOT, 2013). NECTEC (2014) points out that the growing popularity of online gaming in 2005 sparked the beginning of increased demand for broadband Internet in Thailand. TOT Internet became the most profitable Thai ISP in 2004 and 2005 because it was the only company that offered free access to online games in Thailand at that time (TOT, 2013). In 2009, TOT Internet established a sister company called 3BB targeting young consumers (i.e. 15 to 28 years old) through its budget offerings, such as low cost high speed Internet, online gaming and ADSL services. 3BB s market shares increased by 7% in 2009, but decreased from 36% to 31% in 2010 (TOT, 2013) True Internet True Internet was established in 1990 as a subsidiary of the Charoen Pokphand group (True, 2014). It is the dominant player in the Internet services market in Bangkok, with 75% market share (True, 2010). The company has started to expand its business into other major areas in Thailand, especially southern Thailand since late It has serviced almost all major Thai cities since 2010 (True, 2014). True Internet provides high speed Internet, dial-up Internet cards, e-tv, e-commerce, e- auction and e-security services to various types of consumers. The company employs different strategies for different target markets and is especially well known for a large number of cross promotions with other strategic business units under the same corporate brand (i.e. True cable TV, True mobile network, True online games, True cafe, True Internet cafe and True fitness). It increased its market share from 12% to 25% by using cross promotional strategies between True online gaming and True Internet (i.e. free access to True online games for True Internet customers) (True, 2014). Additionally, cross promotions between True Internet, True cable TV and True fitness are often used to attract 20 to 40 year old customers (True, 2014). In 2006, True Internet acquired KSC 52

73 Internet, an upper-end market company with a high level of brand awareness, creditability, and reliability (NBTC, 2014). As a result, True Internet increased its market share by 200% in 2008 and However, its market share is still lower than TOT Internet and 3BB across the wider Thai Internet sector (NECTEC, 2014) Internet usage behaviour Internet usage in Thailand grew due to the uptake of entertainment-based Internet applications by young users. Table 2.2 shows that in 2009, the greatest number of Internet users, 69.14%, used the Internet for entertainment and gaming. The second largest group, 27.4% of users, used the Internet for social and business affairs. Interestingly only 1.75% of users visited government websites and just 1.71% were interested in educational content (NECTEC, 2014). 4Table 2.2: Internet usage behaviour Type of website % Entertainment Games Internet 7.5 Social/Personal 7.34 News 6.83 Business 3.72 Computer 2.01 Government 1.75 Education 1.71 The low popularity of government and education websites reflects the poor execution of e-government and e-education policies, both of which were important areas of the 2001 IT Policy approved by the Thai government, called IT This policy was followed by the IT Master Plan for Thailand , which prompted the Thai government to set up the Ministry of Information and Communications Technology (NECTEC, 2014) Consumer buyer behaviour in Thailand Consumer behaviour is different across the world because each country has unique social traits and demographic characteristics, as well as differing economic challenges (Kotler, 2003; Schiffman and Kanuk, 1997). Marketers need to understand consumer behaviour in particular geographic areas where a business operates in order to undertake 53

74 effective marketing activities (Kotler, 2003). According to Schiffman and Kanuk (1997), understanding consumer behaviour can help to explain how and why consumers make decisions to purchase certain products and services. A study of Thai consumer buyer behaviour by Thananuraksakul (2007) reveals that Thai consumers purchase decisions are determined by both extrinsic (i.e. service quality, quantity, price, promotion, and availability) and intrinsic factors, such as consumer self-image, self-respect and social-respect, demographics and Thai culture (Silayoi and Speece, 2007). Extrinsic factors are considered important to consumers in a big city, for example, Bangkok. This is because markets in big cities tend to be highly competitive with a great number of companies offering undifferentiated services and targeting the same segments (Thananuraksakul, 2007). Songpradit (2005) states that a salesperson s attitude and behaviour are positively related to a Thai consumer s purchase decisions. However, when it comes to mobile phone and Internet services in Thailand, salespeople have less of an impact on the consumer purchasing decision. In addition, Thai consumers, especially those who have low to middle income, tend to purchase services that give them the best value, as weighed up in cost-benefit terms, regardless of the reputation of the service provider (Thananuraksakul, 2007). Many Thai people also make purchase decision designed to enhance their self-image, self-respect, social-respect and sense of social belonging (Sukato and Elsey, 2009). As a result, they spend more time and money purchasing services pertaining to their health, beauty and lifestyle (Sukato and Elsey, 2009). In terms of customer lifestyle, Feeny and Vongpatanasin (1996) note that Thai consumers in larger towns spend more time in shopping malls and department stores, using them as a meeting place with friends and family. Lekagul (2002) also reports Thai consumers frequently go to shopping malls for recreation activities, such as window shopping, karaoke, seeing a film and bowling. Recent trends show that Thai companies and consumers are becoming more conscious regarding environmental issues (Sukato and Elsey, 2009). As a result, they show preferences for services offered by companies that are environmentally friendly. However, Sukato and Elsey (2009) state that while Thai consumers may be interested in green products, they are not willing to compromise on price and safety. 54

75 Another factor that affects Thai consumers decision making is their culture (Silayoi and Speece, 2007; Webster, 2000). In general, Thai culture is classified as being collectivist, which means that Thai people are influenced by friends, family and social groups to which they belong (Sukato and Elsey, 2009). For example, young people in similar age groups tend to prefer similar style, food and fashion (Lekagul, 2002). In terms of family influence, Silayoi and Speece (2007) point out that the father is most likely to pay the household bills, whereas mothers tend to make the final decision at the point of purchase. Also, wives in Thailand tend to have strong influence over their husbands decision-making (Webster, 2000). It is also common for Thai children to have a say in purchasing decisions when the household service relates to them (Webster, 2000). In summary, the factors that influence Thai consumer behaviour are quality, price, brand, packaging, advertising, promotion, the attending salesperson, distribution, consumer attitudes towards the service, self-image and normative influences (i.e. spouse, friends, family and colleague). It is important that marketers understand these factors in order to develop the most effective strategy for long-term customer retention. Section 2.5 elicits the research questions and research objectives. 2.5 Research questions and research objectives This section elicits the research questions that have arisen as a result of this chapter s literature review. As mentioned in Chapter 1, in order to address some of the current gaps in extant literature, this study focuses on three key research questions as follows: Primary research question: What are the specific service quality dimensions and attributes which influence the overall service quality of an ISP? As mentioned previously, service quality in high-tech telecommunications services cannot be effectively measured by SERVQUAL or E-S-QUAL, due to their lack of ability to address all the specific issues relevant in this particular context (He and Li, 2010). This study aims to investigate the specific service quality dimensions and identify their influences in the context of an ISP, which then might impact on customer perceptions of an ISP s service quality. This study is original, in that it is the first of its kind to attempt to investigate the dimensions of an ISP s service quality, as well as their 55

76 effects on customer loyalty in high-tech services. It contributes to the body of knowledge on service quality and consumer loyalty in the home Internet services market. It investigates the service quality dimensions identified in Chapter 4 as (1) network quality; (2) customer service and technical support; (3) information quality and website information support; (4) security and privacy. Secondary research question 1: What are the clear and unambiguous relationships between service quality and customers cognitive and affective evaluations in the home Internet services market? Relationships are critical to all aspects of life, including businesses. Trust, satisfaction, value, and commitment representing the cognitive and affective evaluations of customers are essential elements in a relationship (Ulaga and Eggert, 2006). However, there has been limited research investigating the relationships between service quality and customers cognitive and affective evaluations in the home Internet services market. Therefore, this study aims to examine the influence of an ISP s service quality on customers cognitive and affective evaluations, which are important in developing and sustaining relationships between ISPs and their customers. Secondary research question 2: What are the effects of customers cognitive and affective evaluations on customer loyalty in the home Internet services market? The telecommunications industry is under pressure to retain customers and acquire potential new customers (Spiller et al., 2007). The Thai National Statistical Office (TNSO, 2010) reported that an average of 10% of home Internet customers in Thailand switched service providers each year between 2003 and In 2009, this figure was 12% (True, 2010). Despite the importance of the Thai telecommunications market and current issues related to this industry, few studies have been done on home Internet services in Thailand and the Thai Internet services market is generally under-researched. Given the high usage and increasing churn rates in the ISP sector, it is justified to investigate the antecedents to customer loyalty, in particular customer retention, within the Thai Internet services market. 56

77 In order to address the three research questions, the research objectives of the current study are to: Identify the attributes which influence the overall service quality of an ISP. Clearly and unambiguously define each of the identified dimensions that contribute to the maintenance of ongoing relationships between customers and their ISPs. Develop and test a model to predict customer loyalty towards a chosen ISP. 2.6 Benefits of researching service quality in the telecommunications industry It is vital for service providers to obtain accurate information regarding service quality, as perceived by customers. This information enables marketers to formulate appropriate marketing strategies and would work in their favour to achieve competitive advantage and long term sustainability (Vlachos and Vrechopoulos, 2008). The findings of this study would be beneficial to telecommunications service providers in Thailand, as well as in other countries which have similar economic context, for example Indonesia, Vietnam, and India (Jahanzeb et al., 2011). Service quality in the telecommunication sector is considered a determinant for longterm profitability (Butt and de Run, 2009). A high level of service quality can help firms differentiate themselves from competitors. Research indicates that service quality can influence customer trust, satisfaction, word-of-mouth (WOM), and buying decisions (Butt and de Run, 2009; Cheng et al., 2008; Morgan and Hunt, 1994). Trust exists provided there is adequate confidence in a partner's trustworthiness and honesty (Morgan and Hunt, 1994). Both researchers and practitioners have agreed that high service quality is a necessary condition for durable business relationships (Crosby et al., 1990), and helps grow trust between parties (Moorman et al., 1992). When customers believe that their service provider is dependable, responsible and empathetic, they are likely to develop trust towards the company (Morgan and Hunt, 1994). Service quality is a direct antecedent of satisfaction, which develops when performance meets or exceeds customer expectations (Butt and de Run, 2009; Han et al., 2008; Oliver, 1980; Parasuraman et al., 1985). Customer satisfaction is one of the most 57

78 important factors in customer decision-making process (Sachdev and Verma, 2004), and leads to customer trust, loyalty and retention in telecommunications industry (Ahn et al., 2006; Asaari and Karia, 2003; Cheng et al., 2008; Gerpott et al., 2001; Lovelock et al., 2002). Service providers need to increase customer satisfaction through enhanced service quality in order to maintain or improve their market positions. Blery et al. (2009) state that a customer s expectations and a customer s perception of actual performance can change over time; therefore, a long-term strategy is necessary to impress and retain current customers. Sabiote and Roman (2009) note that service quality has a positive influence on customer word-of-mouth (WOM). This, in turn, has a significant impact on a company s credibility and ability to recruit and retain customers. Zeithaml et al. (1990) claim that high service quality results in favorable WOM (Sabiote and Roman, 2009). Parasuraman et al. (1991) and Boulding et al. (1993) concur with this view, stating that consumers are willing to recommend service providers to others when the service is perceived to be of high quality. In contrast, customers who are not happy about service quality tend to spread negative WOM (Sabiote and Roman, 2009). Research has shown dissatisfied customers usually relate their experience to three or more people (Ghobadian et al., 1994). In addition, dissatisfied customers may switch to a competitor to seek better service quality, and at the same time, their negative WOM can be a challenge for their former service provider in attracting new customers as a replacement. Service quality is a significant factor for the success and survival of all organisations. It can influence customer repurchase decisions, switching intentions and loyalty (Ahn et al., 2006; Dwivedi et al., 2010; Lovelock et al., 2002; Zeithaml et al., 1990). Zeithaml et al. (1990) also draw a positive relationship between service quality, customer repurchase behaviour, willingness to pay a premium for a service and intentions to remain loyal regardless of price. There is evidence in the Indian mobile telephony market that a high level of service quality can boost a company s competency, which results in better customer acquisition as well as lower customer turnover (Seth et al., 2008). Furthermore, research indicates that network quality has a direct impact on customer retention in the Chinese (Cheng et al., 2008) and Korean mobile services 58

79 industries (Ahn et al., 2006), as well as in the Australian ISP industry (Spiller et al., 2007). A study conducted by Huan et al. (2005) suggests that in the short term, it is very difficult for small mobile phone service providers to compete with well-established competitors due to greater network coverage. In this case, lower prices can appear unattractive to mobile subscribers (Huan et al., 2005; Woo and Fock, 1999). It is therefore important for companies to concentrate on service quality in order to gain new customers, as well as to retain current customers over the long term. It is important for telecommunications firms to shift their focus from price, which is currently the dominant factor in competition to quality of services (i.e. network coverage and quality) (Huan et al., 2005). In summary, a high level of service quality can increase customer trust, satisfaction, word-of-mouth (WOM), and repurchase decisions in the telecommunications market. By enhancing service quality, firms can influence customers behavioural and attitudinal loyalty, which are critical for a company s success and long-term sustainability (Hunt et al., 2006). In addition, service providers might also be able to reduce expenses associated with acquiring new customers by concentrating on improved service performance (Spiller et al., 2007). As a result, businesses could effectively develop strategies to overcome current issues relating to customer switching and churn rate in the telecommunications market in Thailand, as well as in other countries with similar demographic profiles. 59

80 2.7 Chapter summary This chapter provides a review of the literature associated with services marketing, service quality, relationships marketing, customer loyalty and the context of the study. It commenced with a discussion on services marketing (section 2.2), which was subdivided into the evolution of services marketing, the nature of services, and service quality. The review of the services marketing literature helped put this study into context by highlighting related research previously undertaken in service quality and service quality dimensions of an ISP. Section 2.3 then presented a discussion on relationship marketing including customer retention and the benefits of relationship marketing in the telecommunications sector. It also discussed the issues pertaining to customer loyalty, which was subdivided into customer loyalty, behavioural loyalty, attitudinal loyalty and the benefits of customer loyalty. Section 2.4 then presented an overview of business in Thailand, historical aspects of the Internet, the provision of Internet services in Thailand, and consumer behaviour in Thailand. Finally the chapter addressed a section on the research questions and research objectives (section 2.5) and the benefits of researching service quality in the telecommunications industry (section 2.6). The next chapter (Chapter 3) develops and presents the conceptual and theoretical frameworks for the research. It provides discussion relating to key concepts underlying the development of the conceptual framework, including network quality, customer service and technical support, information and website support, privacy and security, overall service quality, customer trust, customer satisfaction, customer commitment and customer value. Figure 2.2 provides a roadmap to the structure of the overall thesis. 60

81 Chapter One An introduction to the study and overview of the chapters Chapter Two Literature Review Next chapter Chapter Three Development of the theoretical model and related hypotheses Chapter Four Methodology Chapter Five Analysis and Results Chapter Six Discussion, Recommendations and Conclusion 5Figure 2.2: Structure of the overall thesis 61

82 Chapter 3: Development of the Theoretical Model and Related Hypotheses 3.1 Chapter overview This chapter presents the theoretical model for the research study, as well as the research hypotheses. It commences with a review of the literature relevant to key concepts underlying the development of the theoretical model (section 3.2). This section is sub-divided into an overview of an ISP s service quality dimensions; cognitive and affective evaluation of customers; and discussion about the dependant or outcome constructs. Section 3.3 presents a discussion on previously validated models. The next section (section 3.4) presents an outline of the development of the proposed theoretical model. Section 3.5 presents a discussion on an ISP s service quality dimensions, which is subdivided into four sections. These sections include (1) network quality; (2) customer service and technical support; (3) information quality and website information support; and (4) privacy and security. The next section (section 3.6) presents the cognitive and affective evaluations of customers. This section is divided into (i) ISPs overall service quality and customer trust; (ii) ISPs overall service quality and customer satisfaction; (iii) ISPs overall service quality and customer commitment; and (iv) ISPs overall service quality and customer value. The chapter then addresses issues relating to dependant constructs (section 3.7), which include concepts of customer trust and customer loyalty; customer satisfaction and customer loyalty; customer commitment and customer loyalty; and customer value and customer loyalty. The organisation of this chapter is shown in Figure

83 3.1 Chapter overview 3.2 Overview of the development of a theoretical model 3.3 Previously validated models 3.4 Proposed theoretical model 3.5 An ISP s service quality dimensions 3.6 Cognitive and affective evaluations of customers 3.7 Dependant constructs 3.8 Chapter summary 6 Figure 3.1: Chapter organisation 63

84 3.2 Overview of the development of a theoretical model This section provides a brief overview of an ISP s service quality dimensions; the cognitive and affective evaluations of customers; and the dependant constructs of the proposed theoretical model. Details about each hypothesis are provided in sections 3.4 to Overview of an ISP s service quality dimensions Various service quality measurement models are available in the marketing literature. The earliest one was introduced by Parasuraman et al. (1985) and is called SERVQUAL. This model includes tangibles: appearance of physical facilities, equipment, personnel and communication materials; reliability: capacity to execute the promised service responsibly and precisely; responsiveness: willingness to help customers; assurance: knowledge of and courtesy toward employees, and the ability to enhance customers trust and confidence; and empathy: caring, individualised attention provided to customers. SERVQUAL was developed based on Parasuraman et al. s (1985) gap model between performance and expectations, which assumes that as performance exceeds expectation, quality increases and vice versa. In other words, the central idea of the gap model was to view service quality as a function of the difference in scores or gaps between expectations and perceptions (Zeithaml et al., 1996). With the rise of technology-enabled services, the attention of the services literature shifted to measurement and operationalisation issues (Carlson and O Cass, 2011; Ganguli and Roy, 2010). Consequently, numerous empirical studies, such as those of Carlson and O Cass (2011); Kurt and Atrek (2012) and Wolfinbarge and Gilly (2003), endeavour to examine service quality in different types of services. To better understand service quality in the information age, E-S-QUAL was developed by Parasuraman et al. (2005). E-S-QUAL is the quality measurement scale for services delivered by websites. It includes 22 items under four dimensions: efficiency, fulfillment, system availability, and privacy. However, the area associated with the measurement of service quality of Internet services has been under researched (He and Li, 2010). High-tech telecommunications service quality cannot effectively be measured by SERVQUAL or E-S-QUAL (He and Li, 2010), as these scales lack the ability to address issues relevant to this particular context. Both the SERVQUAL and E-S-QUAL scales do not take into 64

85 account specific issues relevant to customers of high-tech ISPs. SERVQUAL and E-S- QUAL focus on the service quality of service providers operating via the Internet platform (Vlachos and Vrechopoulos, 2008), rather than those actually providing the Internet connection and platform for online B2B and B2C activities. This study aims to investigate the specific service quality dimensions and attributes that influence the overall service quality of an ISP. It is vital for an ISP to obtain accurate information regarding their service quality as perceived by their customers. This information likely enables them to formulate appropriate marketing strategies which work in their favour to achieving competitive advantage and long-term sustainability. This thesis proposes that ISP service quality dimensions are based on interactions between customers, customer service and customer support staff, and evaluations of the store settings and perceptions of an ISP s stability and trustworthiness (i.e. Internet connection and speed). The tangible dimension of SERVQUAL, in the context of services provided by ISPs includes infrastructure and equipment which facilitate an Internet connection. Despite this, customers of residential Internet services are unable to visualise the type of facilities owned or leased by their ISP. For them, the tangibles are associated with attributes of service delivery, such as downloading or uploading speed and the consistency and stability of the signal (Vlachos and Vrechopoulos, 2008). A dimension, referred to as network quality, has been proposed as a driver of overall service quality in the residential Internet services market (Thaichon et al., 2012). This dimension is associated with the core service performance and also accounts for reliability in this service industry. Network quality includes the number of errors, downloading and uploading speed, and system response time (Vlachos and Vrechopoulos, 2008). In the telecommunications industry, the quality of networks and phone calls are key drivers of customer satisfaction (Ahn et al., 2006). Moreover, service provider reliability not only refers to stable service performance, but also involves privacy and security. As the nature of an ISP s services is Internet related, privacy and security is often an issue of concern for customers. Privacy and security is associated with customers feelings of protection and safety during online transactions and general Internet usage, for example, feeling protected enough to provide personal information over the Internet (Vlachos and 65

86 Vrechopoulos, 2008). Customers are prone to attribute low risks in purchasing from service providers that are reputable in relation to their security practices (Roca et al., 2009). The quality of customer service and technical support ties in with responsiveness, assurance and empathy, which are dimensions of SERVQUAL. Customer service and technical support relate to the performance of customer service staff. Performance is reflected through staff s knowledge, enthusiasm, professionalism and timeliness. The attitudes and problem solving capabilities of technical and customer service staff is important when customers choose high-tech service providers, such as ISPs (Santouridis and Trivellas, 2010). In addition, unlike traditional businesses, customers of an ISP not only look for face-to-face support, but also information support, in particular on online platforms (e.g. websites). Website support needs to be accurate, relevant and updated (Kim and Niehm, 2009). Many customers tend to be money rich and time poor and providing Internet services can require high-tech knowledge, therefore reliable and upto-date information can help support decision making of customers (Hasley and Gregg, 2010; Kim and Niehm, 2009). In short, specific dimensions of an ISP s service quality cannot be accurately addressed by SERVQUAL. E-S-QUAL does not fully address the service quality dimensions of an ISP, although it does consider privacy. E-S-QUAL only assesses the service quality of online services that direct their focus toward online environments and website quality. In the case of ISPs, customers can purchase an Internet service package via online and offline channels. The role of offline services cannot be neglected in an ISP s performance. Moreover, ISPs play a very different role compared with other services delivered via websites, which are the focus of E-S-QUAL. For example, ISPs are more active in controlling privacy and security issues than other types of online service providers. ISPs observe and monitor traffic flowing through their networks and are able to detect suspicious traffic spikes, stop malicious traffic or provide timely warnings to customers (Rowe et al., 2011). In conclusion, by using either SERVQUAL or E-S-QUAL to measure an ISP s service quality, researchers might not sufficiently capture all the service aspects of that ISP. 66

87 In brief, the antecedents to the latent construct of an ISP s service quality dimensions have been identified as network quality, customer service, website support and security (section to provides a complete review of the literature relevant to key concepts underlying the development of the theoretical model) Influence of service quality on the cognitive and affective evaluations of customers This section explores the underlying relationship between overall service quality and customers evaluations of trust, value, satisfaction and commitment. Overall service quality is the excellence of quality of the overall service, customers opinions of service standards and performance in general (Brady et al., 2002). Research into India s mobile phone service providers report that a high level of service quality can improve company competency, which leads to greater customer acquisition and a lower customer turnover rate (Seth et al., 2008). The overall service quality, including reliability and responsiveness, can be considered cognitive evaluations of performance over time, which are anticipated to build affective attachment (Fullerton, 2005). Positive overall service quality influences the commitment a customer has towards a particular service and the associated service provider (Jahanzeb et al., 2011). Once a customer tries the service and, if the initial expectations of the overall service quality are exceeded, they are likely to be satisfied (Parasuraman et al., 1985) and this builds trust with the service provider (Chiou, 2004). In other words, the level of a customer s satisfaction is dependent on the ability of the supplier to meet the customer s norms and expectations (Zeithaml et al., 1996). Trust is evaluated by the perceived reliability of an ISP s billing system, fairness of its contracts and fulfillment of its promised service delivery (Chiou, 2004). Trust is dependent on how receptive a customer is to their service provider s products and offerings, along with its reputation, credibility and expectations (Kim et al., 2008). Trust is closely related to overall service quality. In the ISP industry, trust can be evaluated by exploring how a customer feels about their service provider in terms of the company s honesty, responsibility and professionalism, and if the customer thinks that the ISP understands and cares about them (Chiou, 2004). 67

88 On the other hand, Zeithaml (1988) suggests that perceived value is the customer s overall assessment of the utility of a product based on perceptions of what is received and what is given (p. 14). This view reflects an economic perspective and depicts value as a cognitive construct (Choi et al., 2004). In order to improve customer value, organisations can either add more benefits to their service or reduce the costs associated with service provision (Shirin and Puth, 2011; Tam, 2012; Wang and Wu, 2012). For example, benefits may include extra service attributes, superior customer service, aftersales support (Shirin and Puth, 2011), and enhanced quality of the service (Tam, 2012). Sections to provide a complete review of the literature relevant to key underpinnings underlying the development of the theoretical model, including overall service quality, service quality and customer trust, service quality and customer satisfaction, service quality and customer commitment, service quality and customer value Overview of the dependant constructs It has long been acknowledged that both cognitive and affective evaluations impact customer loyalty (Deng et al., 2010). They include a variety of dimensions representing the different ways customers assess and respond to services provided by their ISP. The final component of this study reflects the impact of customers attitudes towards their ISP, as well as their future intentions. Hence, the dependent constructs refer to the overall impact of the service encounters and projects undertaken during the long-term relationship between the ISP and the customer. Loyalty comprises of attitudinal and behavioural loyalty. Attitudinal loyalty is evaluated by customers thoughts of attachment, positive word-of-mouth and recommendations (Zeithaml et al., 1996). Attitudinal loyalty is determined by exploring whether customers consider themselves loyal patrons of a particular ISP and whether they think that a particular ISP is the best choice for them (Kim and Niehm, 2009). Behavioural loyalty is measured by the number of customers who remain with a particular service provider (Zeithaml et al., 1996). Trust plays an important role in determining customer loyalty. When customers trust the service provider, they show this by continuously using the service and recommending 68

89 the provider to potential customers (Deng et al., 2010). In the Taiwanese ISP context, Chiou (2004) claims that customer trust has a direct positive effect on customer loyalty. As well as trust, satisfaction has a strong impact on customer loyalty, especially in the Hong Kong Internet service context (Cheng et al., 2008). These authors find that customers who experience high levels of satisfaction are most likely to stay with their existing service providers and maintain their service subscriptions. Businesses improve the level of customer satisfaction by upgrading their service quality, which in turn can influence purchase and repurchase intentions in the Internet services market (Seth et al., 2008). Customer commitment has been found to explain customer loyalty behaviour in Pakistan s telecommunications service (Jahanzeb et al., 2011). In Korea as well, emotional commitment was stated to be positively related to attitudinal loyalty (Han et al., 2011). Emotionally committed customers usually intend to make a purchase, and to engage in word-of-mouth recommendation (Han et al., 2011). Moreover, research in the Chinese mobile services, indicates customer value has a direct effect on subscribers attitudinal loyalty (Huan et al., 2005).Similarly, Lien et al. (2011) find a significant link between perceived value and behavioural intentions in Taiwanese online shopping, and Chiou (2004) finds a similar connection in the Taiwanese mobile commerce industry. To summarise, loyalty consists of attitudinal and behavioural loyalty. While attitudinal loyalty is evaluated by customers thoughts of attachment, positive word-of-mouth and recommendations, behavioural loyalty is evaluated by examining whether they decide to remain with a particular ISP going into the future. Customer trust, satisfaction, value and commitment have been identified as antecedents to attitudinal and behavioural loyalty. Overall service quality influences customer trust, satisfaction, value, and commitment. Network quality, customer service and technical support, information quality and website information support, and security and privacy influence overall service quality. A further discussion of customers satisfaction, commitment, trust, and value can be found in sections to section of this thesis. 69

90 3.3 Previously validated models Several empirical studies have been identified using models of loyalty incorporating selected antecedents in B2B service provision contexts. All of these studies used path analysis to test their conceptual models. They informed the development of the conceptual framework for the current study and are presented in chronological order. Venetis and Ghauri (2004) were among the first researchers to undertake empirical research into buyers perception of value, and to combine this with assessments of perceived performance and satisfaction as drivers of repurchase intentions. These authors model relationship commitment as mediating the relationships between all the exogenous variables and future intentions. Their final model incorporated direct, as well as indirect paths, from stuck bonds and service quality to repurchase intentions. Although Venetis and Ghauri (2004) incorporated several affective dimensions in their model, in a B2B relationship one would expect that relationship intentions would also be dependent on some cognitive and objective assessments, as well as subjective evaluations. Durvasula et al. (2004), using a services marketing approach, modelled the antecedents of behavioural and attitudinal loyalty incorporating the dimensions of service quality, value and satisfaction. These authors use two models one incorporating both direct and indirect relationships and the other, a fully mediated model. These were tested in a B2C services context with the final model shown in figure Figure 3.2: Structural model from Durvasula et al. (2004) Caceres and Paparoidamis (2007), drawing on the relationship marketing literature, blend with Grönroos (1984) conceptualisation of service quality, to develop and test a 70

91 model of business loyalty using the clients of advertising agencies. This model, shown in figure 3.3, commences with the elements of service quality as antecedents to a single item measure of global relationship satisfaction. In turn, this mediates the relationship between service quality and affective relationship constructs, particularly trust and commitment, all of which lead directly and indirectly to loyalty. 8 Figure 3.3: Structural model from Caceres and Paparoidamis (2007) In this study, loyalty is representative of future repurchase and recommendation intentions. Caceres and Paparoidamis (2007) methodically deconstruct the elements of the services provided by advertising agents to their clients, while efficiently assessing the subsequent constructs. The key drivers of loyalty in their model are service quality, trust and commitment. This model is similar to Durvasula et al. s (2004) model in that it does not account for evaluative and cognitive constructs. Vlachos and Vrechopoulos (2008) have attempted to develop service quality measurement scales in the mobile phone service industries (Figure 3.4). 71

92 9Figure 3.4: Structural model from Vlachos and Vrechopoulos (2008) The purpose of Vlachos and Vrechopoulos s (2008) paper is to investigate the theoretical and empirical meaningfulness of a composite model of behavioral intentions in a pure mobile Internet services context. Their service quality dimensions include content quality, contextual quality, device quality, connection quality and privacy. The key drivers of loyalty in Vlachos and Vrechopoulos s (2008) model are satisfaction, value and service quality. Nevertheless, Vlachos and Vrechopoulos s (2008) service quality model does not effectively evaluate an ISP s service quality dimensions. Several basic differences exist between Internet services and other telecommunications services. For example, mobile service quality includes value-added services (e.g. SMS, MMS, WAP, GPRS) or mobile devices (Santouridis and Trivellas, 2010), which are not applicable in case of ISPs. As the nature of home Internet services is Internet related, information support on websites is important when assessing an ISP s service quality. However, this might not be significant for other telecommunications services, such as television transmission. Such inconsistencies relating to service quality components of high-tech services deserve further investigation. 3.4 Proposed theoretical model As a result of the extant literature review a theoretical model is developed. The dependent construct of customer loyalty is influenced by several independent and 72

93 mediating constructs, such as overall service quality, customer satisfaction, commitment, trust, and value. These constructs have been operationalised using validated measures from the literature (Figure 3.5). The specific ISP service quality dimensions in this theoretical model consist of the core services offered by an ISP to the customer in the Internet services context. It has long been acknowledged that both cognitive and affective dimensions impact customer loyalty (Deng et al., 2010). The theoretical model includes a variety of such dimensions representing the different ways customers assess and respond to the services of their ISPs. The final component of the model reflects the influence of customers attitudes towards their ISP, as well as their future intentions. The outcome constructs refer to the overall impact of the service encounters and projects undertaken during the long-term relationship between an ISP and its customer. 10Figure 3.5: Proposed theoretical model 3.5 An ISP s service quality dimensions The antecedents to the latent construct of an ISP s service quality dimensions are identified as network quality, customer service, website support and security. Each of the service quality dimensions will be discussed in detail below. 73

94 3.5.1 Network quality In the Internet service industry, network quality refers to the quality of the network and strength of the network signal (Wang et al., 2004). In mobile phone services, network quality is related to the quality of the call, which includes network coverage. Associated problems include dropped calls, static and broken conversations during cellular phone calls (Asaari and Karia, 2003). For an Internet service provider, any break in Internet connection leads to low network quality from the customers perspective. It is necessary for ISPs to focus both on technical quality (i.e. Internet connection speed, download speed, connection reliability) and functional quality (i.e. the effective and rapid solution of technical problems, and employee behaviour) in order to improve service quality (Deng et al., 2010; Kyriazopoulos et al., 2007; Woo and Fock, 1999). Companies that deliver high quality services have a better chance of recruiting customers who are willing to return and to speak positively about the firm s performance to others (Ojo, 2010). Chun and Hahn (2007) and Wang et al. (2004) report that network quality is one of the main drivers of overall service quality in the context of the telecommunications industry. Previous research by Ahn et al. (2006) suggests that network quality and call quality are also the key drivers for customer satisfaction in the mobile communications services context. Similar results were found in the Thai mobile cellular network industry by Leelakulthanit and Hongcharu (2011), who claim that a high performance cellular network should have wide coverage, good sound quality, infrequent dropped calls, and instantaneous connection. Keaveney s (1995) critical incident study of 835 customers switching behaviours in service industries demonstrates that 44% switched their service providers because of core service failures. Asaari and Karia (2003) report that the majority of complaints from cellular customers are due to the quality of calls. Hence, dropped calls, rough handling, static, or generally weak cellular signals could lead to subscriber discontent and switching (Steward, 1996). Customers also like to judge service quality on intrinsic cues, or the characteristics found in the service itself (Schiffman and Kanuk, 2000). Schiffman and Kanuk (2000) explain that intrinsic cues enable customers to make rational or objective product choices. For example, in an ISP context, service quality, such as network quality (Lin 74

95 and Ding, 2009), network coverage (Woo and Fock, 1999), network speed and stability of the Internet connection, are customers main expectations (Kyriazopoulos et al., 2007; Trkman et al., 2008). Trkman et al. (2008) have established that speed, continuous connectivity, network health, and the reliability of the service offered are the most important attributes in telecommunications services. However, there is a difference in the degree of customer reliance on different cues to examine service quality (Schiffman and Kanuk, 2000). For this reason, marketers need to study their target customers insights to identify leading attributes in order to effectively influence customer decisions. For example, an ISP may categorise certain customers as being quality-focused customers. This category could be concerned with the performance associated with network quality (Lin and Ding, 2009); network coverage (Woo and Fock, 1999); and speed and stability of the Internet connection (Kyriazopoulos et al., 2007). Research conducted by Huan et al. (2005) suggests that in the short run, it is difficult for smaller service providers to compete with well-established competitors due to their assumed reduced network coverage. Smaller service providers might want to rely on network quality, rather than network coverage, in order to gain new customers and maintain existing customers in the long run (Huan et al., 2005; Woo and Fock, 1999). Hence, in the telecommunications industries it is important to focus on the quality of services (i.e. network quality) rather than on attributes which seem to dominate competition (Huan et al., 2005). This is illustrated by Wang et al. s (2004) who suggest that in order for service providers to build strong competitive advantage, they need to improve the perceived customer service quality by focusing on tangibility, reliability, responsiveness, assurance, empathy, and network quality. In the telecommunications industry, Wang et al. s (2004), as well as Huan et al. s (2005), research confirm that network quality is one of the most important drivers of overall service quality in the Chinese mobile service provider context (Lai et al., 2009). Similar findings were also found in the Hong Kong (Cheng et al., 2008) and Korean mobile services sector (Kim and Yoon, 2004). A study into Greek Internet services reports that network quality is one of the most important drivers of overall service 75

96 quality (Kyriazopoulos et al., 2007). Therefore, an ISP that has superior network quality is likely to have a higher level of overall service quality from its customers perspective. Hence the following has been hypothesised: H 1 : Network quality is positively related with ISPs overall service quality Customer service and technical support Customer service is an important part of everyday encounters (Oloruntoba and Gray, 2009) and involves interactions between service providers personnel and customers (Asaari and Karia, 2003). Although, customer service does not directly generate revenue for a company, the way a company addresses and handles customer requests, questions, and complaints can be a strong point of differentiation (Zeithaml et al., 2010). In general, superior customer service is an important element of a service provider s value proposition and a fundamental driver of competitive advantage in service industries (Brohman et al., 2009). For this reason, customer service has become an integral strategic objective for many businesses, especially in service contexts (Oloruntoba and Gray, 2009). Hence, in such a growth market as Internet services, ISPs that effectively distinguish themselves on customer service as one of the key strategic dimensions, have a better chance of survival and a better opportunity to be profitable in the long run (Erevelles et al., 2003). There are various conceptualisations of customer service in different areas of marketing (Berry and Parasuraman, 1997; Venetis and Ghauri, 2004): operations management (Roth and van der Velde, 1991); service quality management in business-to-business contexts (Pun, 2002; Woodruff and Gardial, 1996); customer relationship management (Christopher et al., 1994); psychology (Parasuraman et al., 1985); performance measurement (Gunasekaran et al., 2001); competitive advantage and strategic management (Porter, 2008); and total quality management (Pun, 2002). In the retail sector, customer service begins once a customer arrives outside a shop, and continues after the transaction is completed, in the form of after-sales service (Carraher et al., 2010). In the Internet services market, the two major points of interaction between an ISP s personnel and its customers are in shops and through call centres (Leelakulthanit and Hongcharu, 2011). These authors propose that the quality of corporate management practices is partly reflected via these two contact points. In addition, ISPs personnel provide cues to customers regarding the nature of the service (Zeithaml et al., 2010). 76

97 Such cues include their dress, personal appearance, attitudes and behaviour, all of which can influence customers perceptions of the service (Zeithaml et al., 2010). Hence, personnel in call centres are expected to have pleasant voices, polite manners, and pay attention to customers needs and concerns (Carraher et al., 2010). In order to provide better service quality, customer service teams should also be passionate about providing a high level of care to customers (Asaari and Karia, 2003). Salespeople who are found to be unwilling to serve or fail to acknowledge customers usually cause disappointment, which results in a low level of perceived service quality (Carraher et al., 2010). Customers usually take responsiveness of technical and customer service staff into consideration when choosing an ISP (Santouridis and Trivellas, 2010). Erevelles et al. (2003) report that an ISP which differentiates itself from competitors by offering superior customer support has a sustainable strategic advantage in the market place. This view is also supported by Leelakulthanit and Hongcharu (2011). Abdolvand et al. (2006) suggest that businesses should not only focus on network quality, but also pay attention to customer support in order to enhance overall service quality perceptions. Research in the Turkish telecommunications industry confirmed this view by indicating that customer complaints handling is an important factor in determining service quality (Aydin and Özer, 2005). In other words, an ISP with outstanding customer service and technical support is more likely to have higher perceived overall service quality (Asaari and Karia, 2003; Carraher et al., 2010). Therefore, service providers need to proactively invest in improving the quality of customer service, in order to gain advantage over their competitors (Asaari and Karia, 2003). For example, service providers should have detailed guidelines for their employees to address how they should respond to customers who are irate and display undesirable behaviour (Carraher et al., 2010). By equipping employees with innovative technology and the right tools, service providers encourage their staff to take ownership of their jobs and to provide better service for the customers (Asaari and Karia, 2003). On the basis of the above discussion, the following is hypothesised: H 2 : Customer service and technical support are positively related with ISPs overall service quality 77

98 3.5.3 Information quality and website information support The combination of information and communication technology generates massive changes across society by connecting businesses and their customers (Asmussen et al., 2013). Businesses need to provide information that helps customers understand their product offerings and that supports customer decision making. This can include detailed product descriptions and transparent price information (Hasley and Gregg, 2010; Yang et al., 2005). Information quality refers to the accuracy, completeness, presentation and format of the information given by service providers (Elliot et al., 2013). Liu et al. (2009) establish that information adequacy is the most important element of information quality, especially in determining customer satisfaction. In terms of accessibility and reliability of information, customers expect the information provided by the company to be available at all times (Bai et al., 2008; Yang et al., 2005). A well-structured and informative information platform can increase the value of a company s services, and positively affect customers behavioural intentions (Chiu et al., 2005). In fact many businesses rely on the company website as a main communication channel (Lee et al., 2012). They have invested a large amount of time and money in maintaining and improving the perceived quality of their websites for users (Grigoroudis et al., 2008). Websites provide an effective communication and information channel between a business and its customers (Grigoroudis et al., 2008). For some organisations, websites serve as a bank of information for various stakeholders (Kim and Stoel, 2004). For others, websites offer transaction capabilities by providing additional tools to serve customers (Kim and Stoel, 2004). Since websites act as an important point of contact between businesses and customers, it is important to provide the right type and quality of information and interactions in order to satisfy customers (Grigoroudis et al., 2008; Kim and Stoel, 2004). Understanding the important attributes of a website and the information they provide is critical to business success as this can be used to enhance online communication between companies and their customers (Kim and Stoel, 2004; Li et al., 2002; McKinney et al., 2002). Website quality is defined as the ability of a website to provide accurate information and perform the promised service consistently and accurately (Li et al., 2002). Website information quality must be judged according to the customers perception of the 78

99 information provided (McKinney et al., 2002). Information on a company s website should be accurate (Bai et al., 2008; Hasley and Gregg, 2010), timely (Kim and Niehm, 2009; Koivumäki et al., 2008), and relevant to users (Chiu et al., 2005; Hasley and Gregg, 2010). In addition, information on company websites should be supportive (Lin, 2007), reliable (Bai et al., 2008), complete (Hasley and Gregg, 2010; Kim and Stoel, 2004; Lin, 2007), usable (Yang et al., 2005), consistent (Hasley and Gregg, 2010), and adequate (Koivumäki et al., 2008). In other words, information provided on websites should be such that the users believe it to be current and perceive it as meeting their requirements (Bai et al., 2008). Apart from information quality, businesses should have well designed websites in order to be successful (Kim and Niehm, 2009). Website design refers to the extent to which customers perceive the online shopping experience as being user friendly (Lin, 2007). The two main attributes of website designs are ease of use and the content of their information (Dadzie et al., 2005; Kwon et al., 2002). Good websites allow users to navigate easily. The result of this is that the site is perceived to be intelligible, uncomplicated, and accessible by users (Dadzie et al., 2005). It is important for websites to be simple to operate and have effective navigational tools (Kim and Niehm, 2009). Information content refers to information that provides the audience with a better understanding of services or purchase objectives (Resnik and Stern, 1977). In general, websites should deliver a wide variety of information in a format that is easy to understand and follow (Bai et al., 2008; Chiu et al., 2005). A visually appealing and state-of-the-art website can help to form pleasant user feelings, which can influence customers information search behaviour and enhance their perceptions that the website is useful and relevant (Kim and Niehm, 2009). Website quality influences the level to which customers expectations are met (Grigoroudis et al., 2008; Parasuraman et al., 1985). When customers visit a company s website, they hope to obtain exclusive and adequate information (Liu et al., 2009). An easy-to-use website is expected to assist online shoppers in placing orders faster and with less effort than visiting a physical shop (Dadzie et al., 2005). These positive perceptions enhance the possibility of closing a transaction using a website and can influence the amount of money customers are willing to spend (Hasley and Gregg, 79

100 2010). According to Yang et al. s (2005) study, a lack of information on a website makes it difficult for customers to receive the correct message. Businesses need to be aware of the type of information that should to be on their websites and avoid information that distracts customers from decision making (Hasley and Gregg, 2010). Information quality contributes to customers overall attitude towards a company, and has been considered as a key component of service quality (Yang et al., 2005). From a customer s view point, high quality information assists in their decision making and positively contributes to overall evaluation of service quality (Hasley and Gregg, 2010). Moreover, as websites provide an interface between a business and its customers, it is important to provide the right type and quality of information and interactions in order to satisfy customers. The quality of information on websites can be used to improve online communications between organisation and customers (Kim and Stoel, 2004). Research suggests that interface attributes of a website, which include information quality, usability and appeal, play a significant role in how customers perceive a business (Hasley and Gregg, 2010). In fact, positive website environments allow customers to process more information, and consequently, result in positive perceptions of service quality (Kim and Niehm, 2009). On the basis of the foregoing discussion, the following hypothesis has been formulated: H 3 : Information quality and website information support are positively related with ISPs overall service quality Privacy and security Service providers have recognised the importance of customer information and its relationship to their business plan (Resnick and Montania, 2003). Customer information offers competitive advantage to all businesses and plays a crucial role in supporting all activities in a business (Lauer and Deng, 2007). Service providers generally study their customers preferences and behaviours, which later can be used to plan advertising and promotional strategies (Resnick and Montania, 2003). For example, in Internet retailing services, businesses collect three types of data: personal information, purchasing habits and click streams (i.e. the sequence of mouse clicks). By rationalising this information, Internet retailers are able to create a detailed profile of each customer (Resnick and 80

101 Montania, 2003). However, collecting and using customers data comes with a number of privacy and security concerns. Privacy often appears in the marketing literature as customers usually have significant concerns regarding personal data acquisition and the usage of this information (Castañeda and Montoro, 2007; Chang and Chen, 2009). There are privacy concerns surrounding businesses obtaining data about individuals and then using them inappropriately (Roca et al., 2009). These authors demonstrate that there are several warnings and issues with respect to confidentiality of personal information and its illegal usage. There are additional privacy concerns and growing perceived risks when it comes to making purchases online (Nepomuceno et al., 2012). Roca et al. (2009) indicate that customers are unwilling to share their personal information because they are afraid of privacy invasion and misuse of that information. Customers worry about the process of collecting and using data (Roca et al., 2009). Sometimes customers are not well informed as to who acquires their information and how the data is used (Resnick and Montania, 2003). Kim and Lee s (2009) research demonstrates that 70 to 84 per cent of their participants were reluctant to share their personal information owing to concerns about privacy. These authors report that customers are anxious that firms may distribute or use their information in a way that causes harm to them. Therefore, it is important to handle customers private information carefully (Lauer and Deng, 2007). Businesses should have a transparent privacy policy in order to reduce customers privacy concerns (Castañeda and Montoro, 2007; Chang and Chen, 2009; Lauer and Deng, 2007). Without a policy on how customers personal information is handled, there is a risk that information may be used inappropriately (Lauer and Deng, 2007). Many organisations have established company privacy policies based on the code of fair information practices (Castañeda and Montoro, 2007; Chang and Chen, 2009). The code of Fair Information Practices (FIP) is established on five principles introduced by the Health, Education, and Welfare Advisory Committee on Automated Data Systems in 1972 (Lauer and Deng, 2007). FIP is a set of practices to ensure the fair treatment and handling of customers personal information collected by all businesses (Lauer and Deng, 2007). FIP states that any businesses that create, maintain, use, or disseminate records of identifiable personal data must assure the reliability of the data 81

102 for their intended use and must take precautions to prevent misuse of the data (Lauer and Deng, 2007). Privacy policy is often made available on company websites, allowing customers to make informed decisions about using the sites and disclosing their personal information (Resnick and Montania, 2003). On the other hand, security considerations relate to how customers perceive business transactions, including payment methods and instruments for storing and transmitting confidential information (Chang and Chen, 2009). Perceived security concerns involves any threat that forms a case, condition, or incident with potential to cause destruction, exposure, alteration of data, rejection of service, and/or fraud, misuse and mistreatment (Roca et al., 2009). The security of transactions has been a major concern in the online environment (Chang and Chen, 2009; Nepomuceno et al., 2012). High risks associated with transmitting sensitive information, such as credit card numbers, are one of the main deterrents of online shopping (Chang and Chen, 2009). A trustworthy process, therefore, is important in establishing reliability and can significantly affect a customer s attitudes and behaviour (Lauer and Deng, 2007). Previous studies demonstrate that privacy and security are positively related to service quality (Ha and Stoel, 2012; Wolfinbarge and Gilly, 2003). Customers tend to believe that it is safe to purchase services from providers that have a good reputation with regards to their security and privacy protection practice (Roca et al., 2009). Therefore, it is conjectured that security (Yang et al., 2004; Zavareh et al., 2012) and privacy (Zeithaml et al., 2002) are components of overall service quality. As such, a number of researchers have incorporated either security or privacy into their service quality measurement. For instance, Yoo and Donthu (2001) introduce a measurement scale for an Internet shopping site called SITEQUAL, which includes four dimensions: ease of use, aesthetic design, processing speed and security. Parasuraman et al. (1985) developed E-S-QUAL, which aimed to capture core service quality aspects: efficiency, fulfillment, system availability, and privacy. Notwithstanding, both privacy and security are important factors in assessing service quality and are closely related. In the telecommunications industry, security and privacy are usually seen as overlapping (Wolfinbarge and Gilly, 2003). Hence, in a study into the Internet retailing industry, propose a more succinct way to measure these constructs by combining both security 82

103 and privacy into a single scale. This scale is further validated by Vlachos and Vrechopoulos (2008) in the mobile Internet services. This research, therefore, intends to use the combined measure of security and privacy. In this respect, the following hypothesis is formulated: H 4 : Security and privacy are positively related with ISPs overall service quality In summary, improvement in service quality is vital for the success of service-based businesses (Kyriazopoulos et al., 2007; Ojo, 2010; Spiller et al., 2007). As discussed earlier, an ISP s customers are motivated by overall service quality, which emanates from a stable and fast Internet network; responsive and positive customer support team; sufficient and high quality information support; and transparent and strong security and privacy practices that are trusted by customers (Vlachos and Vrechopoulos, 2008). A high level of service quality is a determinant for long-term profitability and competitive advantage, which can help telecommunications businesses to differentiate themselves from their competitors (Deng et al., 2010). In particular, service quality can influence customers cognitive and affective evaluations (i.e. trust, satisfaction, commitment, and value) (Cheng et al., 2008; Dwyer et al., 1987; Ghobadian et al., 1994; Morgan and Hunt, 1994). The following sections will discuss the relationships between an ISP s overall service quality, and customer trust, satisfaction, commitment and value. 3.6 Cognitive and affective evaluations of customers This section examines the underlying relationships between overall service quality and customers cognitive and affective evaluations, namely satisfaction, commitment, trust and value Customer trust Trust is considered the foundation for a long-term relationship (Hong and Cho, 2011). It refers to a customer s perception of a service provider s attributes, including its ability, integrity, and benevolence (Deng et al., 2010). Alshurideh (2010) and Morgan and Hunt (1994) find that trust is positively associated with the extent to which parties involved in a relationship share similar expectations, perceptions, values and whether they exchange timely information to solve disputes. Ou et al. (2011) distinguish between trust and expertise by demonstrating that trust relates to the ability of a service provider to fulfill 83

104 its promise (e.g. delivering desirable performance), while expertise involves a service provider s ability to realise its promises. Numerous studies have endeavoured to relate service quality and trust. Several researchers have attempted to incorporate measures of perceived trust and security into comprehensive measures of overall service quality, especially in e-commerce contexts (Janda et al., 2002; Kaynama and Black, 2000; Liljander et al., 2002). However, a number of studies have also investigated the direct association between quality and trust (Chen et al., 2002; Sultan and Mooraj, 2001). Service quality impacts the level of customer trust towards the service provider and their service (Gounaris and Venetis, 2002). In addition, a high level of service quality has a positive influence on customers word-of-mouth (WOM), which in turn increases customer trust (Sabiote and Roman, 2009). Word-of-mouth is believed to be credible as it is generated without self-interest in promoting a service (Sabiote and Roman, 2009). Since it is usually difficult and risky to evaluate a service, WOM becomes a reliable reference source in decision making and plays an important part in determining trust (Sabiote and Roman, 2009). In contrast, Chen et al. (2002) conclude that service quality is not significantly associated with trust in online pharmacy. Conjecture exists surrounding the link between quality and trust across different contexts. Nevertheless, a majority of studies find that trust is driven by service quality (Chiou, 2004). In fact, service providers strategise to build trust and confidence among customers through service offerings. For example, an ISP may offer a new customer a free one-month trial package to test the quality of its service. By doing so, the customer is more convinced about the firm s reliability (Cronin and Taylor, 1994), and tends to be more confident in the relationship with the service provider (Morgan and Hunt, 1994). In short, trust refers to customers perception of company reputation, credibility and expectation fulfillment (Kim et al., 2008). Therefore, it is closely related to overall service quality. An ISP with better overall service quality is likely to benefit from a higher level of trust among their customers (Gounaris and Venetis, 2002). Based on this discussion the following hypothesis has been developed: H 5 : ISPs overall service quality is positively related to customer trust 84

105 3.6.2 Customer satisfaction In the service management literature, customer satisfaction is found to be an effective post-purchase evaluation of the total experience of a service (Cameran et al., 2010; Deng et al., 2010; Ojo, 2010; Pantouvakis, 2010). Customers are more likely to switch if they are unhappy with their current service provider (Bayraktar et al., 2011). Customer satisfaction is defined as customers feelings of happiness, fulfillment and pleasure towards a service provider and its services through their overall experience with the company (Parasuraman et al., 1985). Satisfaction refers to the ability of the supplier to meet the customer s norms and expectations (Zeithaml et al., 1996). In other words, customers are satisfied if the overall service quality meets or exceeds their initial expectations. Satisfaction is not derived from the service itself but from the customer s personal perceptions and expectations of the service attributes (Cameran et al., 2010). Therefore, different customers might develop different perception toward the same service experience. Researchers in services marketing have invested considerable effort in exploring the relationship between service quality and satisfaction. The predominant view is that service quality is an antecedent of satisfaction. The theoretical rationale is grounded in Bagozzi s (1992) appraisal-emotional response-coping attitudinal framework. Service quality is considered to be a cognition-related construct, whereas satisfaction is primarily operationalised as an affective construct (Bagozzi, 1992). His framework postulates that after assessing a service cognitively, consumers develop emotional responses to the evaluation, which eventually facilitates intentions and behaviours. Supporting this view, Tam (2012) reveals that during the consumption process, customers evaluate their satisfaction with the service based on their expectations and the perceived service quality. Wolfinbarge and Gilly (2003) report that overall service quality evaluations are significantly and positively related to satisfaction in the context of online retailing. Likewise, in an attempt to identify the determinants of channel choice for an online multi-channel firm, Montoya et al. (2003) find that perceived quality influences customer satisfaction. Bayraktar et al. (2011) demonstrate empirical evidence for the relationship between service quality and overall satisfaction in the Turkish mobile 85

106 services. A similar result was found in the Chinese (Deng et al., 2010; Lai et al., 2009) and Hong Kong telecommunications industries (Woo and Fock, 1999). Recent research also reveals that service quality is the main determinant of customer satisfaction, which in turn influences purchase intentions (Cameran et al., 2010; Kyriazopoulos et al., 2007; Ojo, 2010; Sa nchez-herna ndez et al., 2010). In summary, a company with outstanding overall service quality is likely to achieve high level of satisfaction among their customers. Hence, service providers can enhance customer satisfaction through improvements in service quality. Based on the foregoing discussion, the following hypothesis has been developed: H 6 : ISPs overall service quality is positively related to customer satisfaction Customer commitment Customer commitment has been defined as a customer s conviction and enduring desire to maintain a relationship that might produce functional and emotional benefits (Hur et al., 2010). Commitment exists when a party perceives that a relationship is important, and as a result, is willing to put maximum effort into maintaining that relationship in the long term (Morgan and Hunt, 1994). Lin and Wu (2011) consider customer commitment as a customer s persistent attempt to maintain a relationship with a service provider. Commitment signifies a high level of relational bonding and is vital for prosperous long-term relationships (Morgan and Hunt, 1994). Therefore, commitment is critical in relational exchange and can result in important outcomes, such as reduced customer switching intention (Porter et al., 1974) and greater motivation (Farrell and Rusbult, 1981). Previous studies identify three components of customer commitment, namely affective commitment, calculative commitment and normative commitment (Fullerton, 2011; Hur et al., 2010). Affective commitment refers to the degree to which a person is psychologically attached to a service provider on the basis of favorable feelings (Hur et al., 2010). Such affective commitment can lead to strong and reliable relationships on the basis of personal engagement and mutuality (Morgan and Hunt, 1994). Calculative or continuance commitment is developed through a cognitive assessment of gains and losses generated from the termination of the relationship (Hur et al., 2010). Normative commitment is defined as the extent to which a customer feels obligated to be in a relationship with a service provider (Fullerton, 2011). It is developed through 86

107 socialisation, as individuals adopt a set of norms that define appropriate behaviour, such as the suitability or unsuitability of switching service providers (Hur et al., 2010). Jahanzeb et al. (2011) reveal that overall service quality impacts on customer commitment towards a particular brand and service provider. It encompasses reliability and responsiveness as cognitive evaluations of performance over time, contributing to affective attachment (Fullerton, 2005). Research by Morgan and Hunt (1994) reports that overall service quality is a direct antecedent of affective commitment in the retail services context. Moreover, overall service quality represents potential relationship benefits that will be lost if a customer switches to other service providers (Morgan and Hunt, 1994). Thus, service quality also has an impact on continuance commitment (Morgan and Hunt, 1994). This view has been supported by several researchers in various contexts (Fullerton, 2005; Thaichon et al., 2014). For example, service quality is positively related to calculative commitment in the grocery retail services (Cater and Zabkar, 2009). In summary, a high level of perceived service quality leads to greater customer commitment. Based on extant literature the following relationship is hypothesised: H 7 : ISPs overall service quality is positively related to customer commitment Customer value There have been attempts to explore perceived value in different contexts (Parasuraman and Grewal, 2000; Sweeney and Soutar, 2001; Thaichon et al., 2014). Cronin et al. (2000) posit that research on perceived value, together with service quality and satisfaction, have been dominant in the services literature. Although several conceptual models of value have been introduced (Holbrook, 1994; Sweeney and Soutar, 2001), customer value has often been described as an exchange between what customers receive and what customers give in order to purchase a service (Lai et al., 2009; Shirin and Puth, 2011; Tam, 2012). Values can be in the form of quality, quantity, time spent, price, brand name, design, social approval, excitement, experience, knowledge, selfrespect, credibility, and security that customers may obtain from using a service (Bell, 2009). Bolton and Drew (1991), and Whittaker et al. (2007) suggest that value is perceived differently depending on service types, situational factors, past experience with the service and customer characteristics. Hence, the concept of value is likely to be subjective from customer to customer (Zeithaml, 2000). 87

108 Using means-end theory, Zeithaml (1988) determines four scopes of perceived value: value as low price, value as anything that a customer wants in a service, value as the quality acquired for the money spent, and value as what the customer receives for what is given. Scholars distinguish between functional value and symbolic value (Chen and Hu, 2010; Zeithaml, 1988). Functional value involves general evaluations of quality, and value of money (Lai et al., 2009). Functional value signifies the way customers perceive the quality of the provided goods and services, the monetary costs of purchase and the time spent (Zeithaml, 1988). Symbolic value is described as an overall indication of experiential value perceptions in terms of community, sentiments, aesthetics, and reputation (Chen and Hu, 2010). Symbolic value implies customers concern about the opinions of others in the society, perceptions of enjoyment or desire, joy of the visual attraction and the consumption experience (Solomon, 1983). This study only examines functional value as it is directly related to an ISP s service quality and is considered to be important with respect to customers usage intentions and behaviours in the telecommunications sector (Kim, 2012; Vlachos and Vrechopoulos, 2008). According to the traditional perceived quality perceived value model by Kim and Damhorst (2010), perceived quality has a positive relationship with perceived value. This view is supported by research in various industries, such as Internet retailing services (Kim and Damhorst, 2010), telecommunications services (Lai et al., 2009), and heritage tourism (Chen and Chen, 2010). Moreover, in order to enhance customer value, a company can either add more benefits, or reduce the costs associated with the service and the use of the service (Shirin and Puth, 2011; Tam, 2012; Wang and Wu, 2012). Additional benefits could take the form of superior service quality (Tam, 2012), bonus service attributes, and outstanding customer service or after-sales support (Shirin and Puth, 2011). Hence, the following has been hypothesised: H 8 : ISPs overall service quality is positively related to customer value 3.7 Dependant Constructs The final endogenous construct of loyalty comprises attitudinal and behavioural loyalty, which are investigated in light of customer evaluations, namely trust, satisfaction, commitment and value. The concept of loyalty has been researched for almost a century since brand insistence was presented by Copeland in 1923 (Jacoby and Chestnut, 1978). Most of the early research merely explained loyalty in terms of behavioral aspects (i.e., 88

109 repeated purchase). Jacoby and Chestnut (1978) introduced a new conceptualisation of loyalty incorporating both attitudinal and behavioral elements. Dick and Basu (1994) state that consumer loyalty includes both a favourable attitude and repeat purchase, which is empirically evidenced by a study of East (2005). True loyalty can be considered an attitude-based behavior of loyalty, whereas the inertial repurchase with slight or no loyal attitude is referred to as (Kim et al., 2008). Attitudinal loyalty is evaluated by customers inner thoughts of attachment, positive word-of-mouth and recommendations (Zeithaml et al., 1996). Attitudinal loyalty can be determined by exploring whether customers consider themselves to be loyal patrons of a particular ISP, as well as whether they think that a particular ISP is the best choice for them (Kim and Niehm, 2009). In addition, behavioural loyalty refers to customer retention or repurchase behaviour (Zeithaml et al., 1996) Customer trust and loyalty Gundlach and Murphy (1993, p. 41) argue that the variable most universally accepted as a basis of any human interaction or exchange is trust. The development of the relational paradigm has underscored the role of trust in buyer-seller relationships in both B2B (Dwyer et al., 1987; Morgan and Hunt, 1994) and B2C markets (Bennett, 1996; Lau and Lee, 1999). In the study of exchange, trust is typically depicted as the outcome of reflexive concerns on the ability of an actor (e.g., a firm or brand) to fulfil their responsibilities (Chaudhari and Holbrook, 2001; Doney and Cannon, 1997). As such, trust has been considered a fundamental element in relationship establishment, development, and maintenance in a number of exchange contexts (Sirdeshmukh et al., 2002; Verhoef et al., 2002). Trust involves customer confidence in the provider s reliability and integrity (Kim et al., 2008), and can reduce the risk associated with the relationship exchange process (Morgan and Hunt, 1994). Therefore, customers are inclined to be more cooperative with a trustworthy service provider (Morgan and Hunt, 1994). That is, when customers trust the service provider, they are more likely to repurchase and recommend the service to the others (Deng et al., 2010). In contrast, Keller (2003) points out that a low level of trust can destroy the relationship between a service provider and its customers. Hence, trust is a key factor for a company in establishing a relationship with its customers (Kinard and Capella, 2006). 89

110 Trust has been associated with several outcomes in both psychology and marketing literature (Harris and Goode, 2004). Hennig Thurau and Klee (1997) indicate that relational attributes impact on repurchasing decisions. Customer trust, therefore, has a significant and positive impact on both behavioural intentions and actual behaviour (Hsieh and Liao, 2011). In addition, trust towards a firm is a determinant of customers intentions to purchase, while lack of trust discourages customers from engaging in a business relationship (Roca et al., 2009). Trust reduces uncertainty in the relationship exchange, and as a result, increases customers intentions to purchase (Castañeda and Montoro, 2007). This view is supported in sectors as diverse as the Malaysian fast food industry (Ling et al., 2011) and the Pakistani telecommunications industries (Khokhar et al., 2011). In addition, trust can decrease customers switching intentions, as evidenced in Malaysian telecommunications (Amin et al., 2012). Chaudhari and Holbrook (2001) assert that trust is not only significantly related to behavioural loyalty, but also to attitudinal loyalty. Matzler et al. (2008) state that trust is an antecedent of attitudinal loyalty. Chiou and Droge (2006) and Ganesan (1994) support this view by pointing out that customer trust is related to the emotional nature of customer loyalty. In line with this thinking, Sirdeshmukh et al. (2002) find empirical evidence for a strong association between trust and customer loyalty. A similar result is found in mobile commerce (Deng et al., 2010) online services (Hong and Cho, 2011), and telecommunications services (Pirc, 2006). Sultan and Mooraj (2001) propose that trust is crucial in exchange regardless of whether the context is online or offline. However, trust appears to be more prominent in an Internet-related environment, rather than in conventional brick-and-mortar contexts. Reichheld and Schefter (2000, p. 107) propose that to gain the loyalty of customers, you must first gain their trust. That s always been the case, but on the web... it s truer than ever, whereas Grewal et al. (2004) state that customers are required to develop more trust towards Internet firms than their offline counterparts. Trust is particularly important in Internet-related contexts due to the absence of physical contact with the company, as well as the presence of online threats (Harris and Goode, 2004). In particular, payment security and potential fraud have emerged as customers main concerns on the Internet (Wolfinbarge and Gilly, 2003). Lauer and Deng (2007) 90

111 establish that customers tend to repurchase from and re-visit the website they trust. Stewart (2003) proposes a strong relationship between trust and intention to buy, while Lynch et al. (2001) find that trust has a consistent and significant influence on online loyalty across a number of different countries. In summary, the role of trust has been highlighted in the development of relationships and loyalty (Hiscock, 2001; Merrilees and Fry, 2002; Morgan and Hunt, 1994). In light of the above discussion, the following has been hypothesised: H 9a : Customer trust is positively related to attitudinal loyalty H 9b : Customer trust is positively related to behavioural loyalty Customer satisfaction and loyalty Customer satisfaction is a key success factor in service industries (Ojo, 2010). Fornell (1992) postulates that satisfaction is a function of the confirmation or disconfirmation of customer expectations and service performance. Glasman and Albarracín (2006) review the positive association between attitudes and behavior in a study that finds similar paths for the relationship between satisfaction and loyalty intentions. Empirical research concludes that customer satisfaction plays an important role in determining customer loyalty (Bayraktar et al., 2011; Cheng et al., 2008; Deng et al., 2010), and customer repurchase intention (Anderson et al., 1994; Cameran et al., 2010; Flint et al., 2011). Oliver s (1997) quality-satisfaction-behavioural intention relationship framework suggests that perceived quality (appraisal) determines satisfaction evaluation (affective response), playing an important part in the persistence of satisfaction levels and the progress of reactions that imply future behavioural intentions. Customer satisfaction is positively related to company profitability and market share (Flint et al., 2011), because satisfied customers return and tend to make more frequent purchases (Bayraktar et al., 2011). Furthermore, high satisfaction leads to a decrease in the perceived benefits offered by other suppliers, hence reducing the likelihood of customers looking at alternatives (Cameran et al., 2010; Lin and Ding, 2005). Likewise, Jones and Suh (2000) report that satisfaction plays an important part in determining future purchases. Castañeda (2001) confirms that satisfaction is a predictor of customer repurchase intention, empirically supported in the mobile service industries in Thailand (Leelakulthanit and Hongcharu, 2011); China (Huan et al., 2005; Wang et al., 2004); 91

112 Malaysia (Asaari and Karia, 2003); Korea (Kim and Yoon, 2004); Slovenia (Pirc, 2006); and United Kingdom (Ranaweera and Prabhu, 2013). In addition, when customers feel satisfied, they become more devoted, and are more likely to continue doing business with their incumbent provider (Mokhtar et al., 2011). Similar results are found in the US discount retailing context in which satisfaction leads to positive feelings and emotions towards the purchase experience, and consequently enhances customer attitudinal loyalty (Carpenter, 2008; Prause et al., 2011). Satisfied customers are likely to recommend the service, and spread positive word-of-mouth about their service provider (Lai et al., 2009). This view finds support in the Chinese mobile phone services (Qian et al., 2011). To summarise, customer satisfaction can result in higher repurchase rates, positive attitudes and customer retention (Bayraktar et al., 2011; Lai et al., 2009; Ojo, 2010). Therefore, the following has been hypothesised: H 10a : Customer satisfaction is positively related to attitudinal loyalty H 10b : Customer satisfaction is positively related to behavioural loyalty Customer commitment and loyalty Commitment is defined as an essential and necessary condition of customer loyalty (Knox and Walker, 2001). Some studies have suggested the relationship between loyalty and commitment is correlational rather than causal. A number of studies treat commitment as a component of loyalty measurement (Bloemer et al., 1999), rather than a distinctive and antecedent construct. However, Cunningham (1967) was among several earlier scholars who viewed commitment as an antecedent of customer loyalty. According to Fishbein and Ajzen (1975), behavioural intentions are powerful in predicting behaviours. In the area of psychology, commitment is deemed as possessing intentional aspects, supported by Kiesler s (1971, p. 30) definition of commitment, which is the pledging or binding of an individual to behavioral acts. Therefore, commitment is proposed as a direct antecedent of loyalty behaviour (Kim et al., 2008). Customer commitment is the primary determinant of customer loyalty in the relationship marketing literature (Fullerton, 2005). Scholars have conceptualised commitment as a sense of attachment between two parties, which results in a wish to nurture a relationship between them (Fullerton, 2005; Morgan and Hunt, 1994). Customer loyalty and commitment can be distinguished on the basis that commitment 92

113 mainly consists of emotions, beliefs and feelings, whereas loyalty is a combination of attitude and behavior, which is often referred to as repeat purchase and recommendations (Cater and Zabkar, 2009). Strong customer commitment indicates a high level of relationship demonstrated by a strong sense of attachment and compulsion to the business (Lin and Wu, 2011; Pan et al., 2012). The more time and effort invested in a relationship by a customer, the less inclined they are to break up that relationship (Bügel et al., 2010). Consequently, Bügel et al. (2010) reveal that commitment has a significant impact on loyalty, as compared to satisfaction. Jones et al. (2008) establish an association between commitment and customer repurchase intentions, relative attitude, willingness to pay more and willingness to recommend (Jones et al., 2008). Hur et al. (2010) posit that commitment leads to positive word-of-mouth. Cater and Zabkar (2009) and Veloutsou and McAlonan (2012) also note that there is a positive correlation between customer commitment and customer repurchase. Likewise, Verhoef (2003) finds a positive connection between affective commitment and customer loyalty in the financial services industry, which is supported by Qian et al. (2011) and Trinh and Gyarmati (2010) in telecommunications services. In addition, Trinh and Gyarmati (2010) and Cheng et al. (2008) claim that calculative commitment is related to customer retention. Drawing upon the extant research, the following has been hypothesised: H 11a : Customer commitment is positively related to attitudinal loyalty H 11b : Customer commitment is positively related to behavioural loyalty Customer value and loyalty Perceived value associated with a company s offering is critical to the success of any firm (Keeney, 1999; Ruiz et al., 2008). Utility theory provides the theoretical underpinning for the value conceptualisation. Customers obtain bundles of attributes that, as a whole, denote a particular quality of the firm s offering at a specific cost (Lancaster, 1971). Therefore, customers assess value based on the comparison between the utility created by the combination of attributes and the disutility signified by the payment and other costs (Caruana and Ewing, 2010). Zeithaml (2000) points out that perceived value is the customer s overall judgment of a product or service based on perceptions of what is received and what is given. Hence, this research study views 93

114 customer value as a result of customers evaluation of the benefits received versus their perceptions of the costs of obtaining a service. In repeat purchase it is typically assumed that customers will return to a provider that offers greatest value for money (Hansen et al., 2013). This decision, however, requires a detailed cost-benefit calculation to estimate accurate value for all available options (Hansen et al., 2013). However, customers, as cognitive misers, would rather choose shortcuts (Cacioppo et al., 1996), and as a result, replace such precise estimates with a subjective overall perception of which alternative can deliver superior value (Hansen et al., 2013). In other words, customers do not always have the need for cognition and thus are reluctant to be involved in such detailed cognitive analysis (Ariely, 2008). Moreover, a consumer s memory often stores an overall judgment of available alternatives, rather than attribute-specific information regarding every single alternative (Schiffman et al., 2008). Customers are most likely to process an overall value evaluation for each alternative in the consideration set in their memory and make their repurchase decision accordingly (Hansen et al., 2013). Furthermore, Sirdeshmukh et al. (2002) underline that value, an exclusive purchase goal, facilitates a customer s actual behaviour following intentions of loyalty. It is anticipated that consumers will perform certain actions in order to accomplish this goal; this means that they are likely to exhibit loyalty intentions towards a service provider as long as they can derive superior value from the purchase (Chiou, 2004). In fact, customers tend to purchase and repurchase a service that can maximise their benefits (Wang and Wu, 2012). The more customers perceive that a service provider is of high value, the more they become committed and willing to return to that particular service provider (Wallace et al., 2004). Therefore, it can be concluded that higher perceived value leads to more positive attitudes towards the service providers, as well as encouraging customer repurchase (Wallace et al., 2004). Empirical research reports that value is a direct antecedent of loyalty (Lien et al., 2011; Wang and Wu, 2012; Wang et al., 2004). Chiou (2004) finds that value is an important determinant of consumers loyalty intentions towards Taiwanese mobile commerce companies. In line with this thinking, Qian et al. (2011); Shirin and Puth (2011) and 94

115 Wang and Wu (2012) assert that value impacts on customers attitudinal loyalty and repeat purchase in telecommunications services. This view is supported in the Australian (Lee and Murphy, 2008) and Chinese mobile telephony contexts (Huan et al., 2005). Bolton and Drew (1991) also propose that a service with greater benefits is of higher value, thus, strengthening repurchase intention. In other words, customers are likely to stay with a company that offers more benefits than its competitors (Lai et al., 2009; Wang and Wu, 2012). Based on extant literature, the following relationships have been hypothesised: H 12a : Customer value is positively related to attitudinal loyalty H 12b : Customer value is positively related to behavioural loyalty In summary, the exogenous constructs of the theoretical model include influential factors such as network quality, customer service, information support, and security. The endogenous constructs include cognitive and affective determinants such as customers trust, satisfaction, commitment, value, attitudinal loyalty and behavioural loyalty. A summary of the hypotheses developed from the foregoing discussions is thus presented: H 1 : Network quality is positively related with ISPs overall service quality; H 2 : Customer service and technical support are positively related with ISPs overall service quality; H 3 : Information quality and website information support are positively related with ISPs overall service quality; H 4 : Security and privacy are positively related with ISPs overall service quality; H 5 : ISPs overall service quality is positively related to customer trust; H 6 : ISPs overall service quality is positively related to customer satisfaction; H 7 : ISPs overall service quality is positively related to customer commitment; H 8 : ISPs overall service quality is positively related to customer value; H 9a : Customer trust is positively related to attitudinal loyalty; H 9b : Customer trust is positively related to behavioural loyalty; H 10a : Customer satisfaction is positively related to attitudinal loyalty; H 10b : Customer satisfaction is positively related to behavioural loyalty; H 11a : Customer commitment is positively related to attitudinal loyalty; H 11b : Customer commitment is positively related to behavioural loyalty; 95

116 H 12a : Customer value is positively related to attitudinal loyalty; and H 12b : Customer value is positively related to behavioural loyalty. 96

117 3.8 Chapter summary This chapter has presented the theoretical model for the research study as well as the research hypotheses. It commenced with a review of the literature relevant to key concepts underlying the development of the theoretical model (section 3.2). Section 3.2 was subdivided into an overview of an ISP s service quality dimensions, the cognitive and affective evaluations of customers, and discussion about the dependent or outcome constructs. Section 3.3 then outlined the development of the proposed theoretical model. Subsequently, Section 3.4 presented a discussion on an ISP s service quality dimensions, which was subdivided into four sections. These sections included network quality; customer service and technical support; information quality and website information support and privacy and security. The next section (section 3.6) presented the cognitive and affective evaluations of customers. Section 3.6 was subdivided into ISPs overall service quality and customer trust; ISPs overall service quality and customer satisfaction; ISPs overall service quality and customer commitment; and ISPs overall service quality and customer value. Section 3.7 addressed the dependant constructs, and was subdivided into customer trust and customer loyalty, customer satisfaction and customer loyalty, customer commitment and customer loyalty and customer value and customer loyalty. Figure 3.6 provides a roadmap to the structure of the overall thesis. 97

118 Chapter One An introduction to the thesis and overview of the chapters Chapter Two Literature Review Chapter Three Development of the theoretical model and related hypotheses Next chapter Chapter Four Methodology Chapter Five Analysis and Results Chapter Six Discussion, Recommendations and Conclusion 11Figure 3.6: Structure of the overall thesis 98

119 Chapter 4: Methodology 4.1 Chapter overview This chapter discusses the research design including justifications for the use of quantitative methods, online survey, sampling and Structural Equation Modelling (SEM). It commences with a discussion of the research paradigm (section 4.2), survey research (section 4.3) and sampling (section 4.3). Section 4.5 then presents a discussion on the data collection procedures. It is subdivided into the selection of the sample, nonresponse bias, unit of analysis, pre-testing and the pilot study. The next section (section 4.6) presents a justification for the analytical technique. This chapter also discusses the issues of validity and reliability and the steps taken to minimise any related errors (section 4.7). The sampling issues of selection and size are discussed along with response rates. The development of measures and scales used are discussed (section 4.8). Section 4.9 discusses the foundations of the steps involved in Structural Equation Modelling. Ethical considerations are provided in Section The organisation of the discussion in this chapter is shown in Figure

120 4.1 Chapter overview 4.2 The scientific realism paradigm 4.3 The online survey 4.4 Sampling 4.5 Data collection procedures 4.6 Tools for analysis 4.7 Reliability and validity 4.8 Measurement development 4.9 Preparing the data 4.10 Ethical considerations 4.11 Chapter summary 12Figure 4.1: Chapter organisation 100

121 4.2 The scientific realism paradigm The most common paradigms in business research are positivism, realism, interpretivism and pragmatism (Saunders and Lewis, 2011). The positivist paradigm is identified as logical positivism and reflects the philosophical position of the natural scientists (Weaver and Olson, 2006). The positivist philosophy claims that there is one objective reality (Weaver and Olson, 2006). Realism is akin to positivism in that it adopts a scientific approach to the development of knowledge (Saunders and Lewis, 2011). Realists argue that there is a reality relatively independent of the human mind (Saunders and Lewis, 2011). Interpretivism supports the idea that it is important for the researcher to comprehend distinctions between humans in our role as social actors (Saunders and Lewis, 2011). Pragmatism advocates that it is completely possible to work with variations in the researcher s epistemology, ontology and axiology (Saunders and Lewis, 2011). The research question is the most important determinant of the ontology, epistemology, and axiology adopted in the research (Neuman, 1997; Nunnally and Bernstein, 1994). The research questions and hypotheses in this study involve the causal relationship between service quality, and customer affective and cognitive evaluation and customer loyalty. The nature of the research in this study has positivist aspects, as it aims to find a model that can predict human behaviour (Gephart, 1999; Lincoln and Guba, 2000). Positivism has been adopted extensively in the marketing literature (Alshurideh, 2010; Santouridis and Trivellas, 2010; Seth et al., 2008) because of its systemised structures which enable explanation and anticipation of phenomena (Hunt, 1991). The use of scientific categorisation schema for many marketing theories supports positivist paradigms. These schema consist of models for different types of products (convenience, shopping, unsought), decision-making (routine, limited and extensive), and pricing policies (above the market, at the market and below the market) (Hunt, 1991). However, extant research postulates that the field of marketing cannot be completely explained by the original positivist approach, which insists on clear differences between empirically observable concepts and theoretical concepts (Hunt, 1991). In any concept, there always exist some elements of theoretical content (Hunt, 1991). Moreover, researchers usually attempt to understand unobservable constructs, such as perceptions, by using measurable phenomena as measures of a construct, and this clashes with positivism (Hunt, 1991). Therefore, it is argued that scientific realism 101

122 is a more appropriate philosophy of science to guide marketing theory and research (Hunt, 1991). Scientific realism has been developed largely as a reaction to logical positivism. Scientific realism states that truth is the goal for marketing theory and research; science can manage to make sense in the real world, yet not with certainty (Peter, 1992). Scientific realism includes the assumptions of positivism, however, it seeks estimated truth rather than precise truth (Hunt, 1991). Scientific realism declares that the statements in a theory are correct or incorrect (Harre, 1986). However, scientific realism differs from positivism in that it searches for the approximate truth, and as a result, accepts that a pure truth may not exist (Weston, 1992). Scientific realists claim that theoretical constructs can always exist without having any observable referents (Hunt, 1991). For example, the existence of attitudes cannot be denied, although they do not have direct observable referents. This view is specifically related to the use of SEM as an analytical tool. Typically, this technique examines hypotheses between latent constructs which are non-observable (Hunt, 1991). In summary, this research employs the scientific realism paradigm which reflects traits of positivism and has been proven to be appropriate for research in marketing. Fundamentally, the philosophical approach of a researcher can influence the development of research questions and the choice of methodology for that research. This research adopts the quantitative methodology associated with the positivist approach. This will be discussed in more detail in the next section. 4.3 The online survey The literature review has identified nine constructs, which are the proposed antecedents of customer loyalty. The relationships between these constructs are depicted in the conceptual model in Chapter 3. Most of the constructs are unobservable latent variables that have to be operationalised using multi-indicators (Schumacker and Lomax, 1996). In this study, indicators for the constructs are self-reported measures. Hence, surveys were chosen as the primary data collection method as self-reported thoughts and behaviours can be most effectively measured via a survey instrument (Neuman, 1997). 102

123 In general, there are four types of surveys: mail questionnaires, online questionnaires, telephone interviews and face-to-face interviews. The survey for this thesis comprised 69 items which were used to operationalise the constructs of the proposed research model. It would be extremely time consuming to gather data on those items using telephone interviews (Neuman, 1997). Mail and online questionnaires offer considerable ease of implementation and time and cost efficiency as the researchers are not required to be present when conducting surveys (De Vaus, 1995). In addition, mail and online surveys can reduce bias as social desirability and interviewer influence can be avoided, demonstrating their advantage over telephone and face-to-face interviews (De Vaus, 1995; Neuman, 1997). However, mail surveys still incur some costs such as paper printing and postage fees, while online surveys do not entail such expenditure. Online surveys appear to be the most cost effective method among the four survey types mentioned previously, in particular when a large number of respondents need to be reached (i.e. 8,000 in this case). Additionally, online surveys do not delay or obstruct respondents from performing their daily duties. They allow them to complete the survey at their convenience, which is facilitated by ease of access to the Internet. This convenience factor is important in obtaining meaningful answers to a sophisticated questionnaire, as well as resolving low response rates (Baldauf et al., 1999; Wright and Grace, 2011). Moreover, the respondents of this study are customers of an ISP, who essentially have access to the Internet. There are several disadvantages of mail and online questionnaires as compared with other forms of surveys. These include low level of research control and lack of probing and visual observation (Dillman, 1978). In addition, the accuracy of responses depends on the reading skill of the respondents, which can lead to sensitive and complicated questions being ineffective (Dillman, 1978; Neuman, 1997; Wright and Grace, 2011). This study attempts to overcome these disadvantages by undertaking pre-testing, which will be discussed later on. Any identified problems were fixed and additional instructions were provided in the final questionnaire to ensure that respondents correctly understood the questions. Therefore, despite the above mentioned disadvantages, online 103

124 surveys were considered most appropriate for this study owing to their benefits in the context of large scale research. The next section discusses the sampling. 4.4 Sampling This section reviews sample size error, sample selection and the sample size selection Sample size error There are two issues that need to be considered when determining sample size: the type of data analysis and the sampling error (De Vaus, 1995). Firstly, SEM was chosen for the purpose of data analysis. SEM has four criteria that affect sample size: model misspecification, model size, departures from normality, and estimation procedures (Hair et al., 1998). Each of these criteria influences the sample size. A sample size of 200 meets the four criteria and concurs with the general consensus amongst SEM scholars as being the minimum sample size (Hair et al., 1998). This is supported by Nunnally and Bernstein (1994), who claim that while it is difficult to identify the minimum number of responses required for analysis, a good practice to ensure stability in analysis is to use a minimum of 200 subjects (Nunnally and Bernstein, 1994). The second criteria that has an effect on sample size is the sampling error (De Vaus, 1995). The sampling error depends on the level of accuracy and the variation in the population. A sample size of 600 produces a possible sampling error of between 3.5% and 4.0% at 95% confidence level (De Vaus, 1995). This is an acceptable sampling error for homogenous sample groups where either 10% or 90% of the respondents are anticipated to give similar answers (De Vaus, 1995) Sample size Two issues need to be considered in determining sample size: data analysis techniques and sampling error (De Vaus, 1995). This study involves a large number of variables, and the intention was to undertake advanced statistical analyses using Structural Equation Modeling. The minimum sample size concurs with the general consensus among SEM researchers of 200 responses (Hair et al., 1998; Schumacker and Lomax, 1996). In line with this thinking, Nunnally and Bernstein (1994) claim that a good rule of thumb to establish statistical stability is 200 subjects. 104

125 The second factor affecting the choice of sample size is the sampling error (De Vaus, 1995). Sampling error rests on the level of accuracy and the variation in the population (De Vaus, 1995). Using 95% confidence level and a sampling error (confidence interval) of ±2.5, the required sample size is 1537 people. However, as the research also aims to conduct advanced statistical analyses (for example, multiple group analyses via SEM), sample size for the survey is determined to be 2,000. The average response rate for surveys via is approximately 30% (Nulty, 2008). A total of 8000 survey invitations were sent out via . The survey data was then analysed using exploratory factor analysis, confirmatory factor analysis, and structural equation modelling to estimate the proposed and competing models Sample selection Thailand is ranked third in South East Asia in terms of residential Internet usage with an estimated 17,483,000 Internet users in 2009 (CIA, 2013) and over 24 million Internet users in 2012 (IWS, 2013). The number in 2012 represented over one-third of the Thai population. The average annual growth rate of Internet users was 30% over the period from 2000 to 2010 (Srihirun, 2011). The competition among residential Internet service providers in Thailand is intense. Currently there are three major ISPs and sixteen smaller ones across the country. In this highly competitive context, the churn rate of Internet users was approximately 12% in 2009 (True, 2010). This scenario, therefore, poses many challenges to ISPs especially in the area of customer retention. The research hypotheses were tested using a sample of Internet service customers in Thailand. The study employed the customer database of major ISPs, which includes customers from all regions of Thailand and is representative of the Thai population. Simple randomisation was chosen to achieve freedom from human bias and to avoid classification errors (Black, 1999). The respondents were not locked into any contracts and were free to switch to other ISPs. It was a requirement that the participants were over 18 years of age and have used home Internet services. Some of the current customers of the selected ISP may have switched from other competing ISPs. Hence, the switching and retention behavioural aspects of the respondents could be investigated. 105

126 4.5 Data collection procedures This section reviews the data collection process, the efforts to minimise non-response bias, and the anticipated response rates Non-Response bias The average response rate for surveys via is approximately 30% (Nulty, 2008). Non-response is one of a number of respondent issues that can impact the reliability and validity of outcomes (Nunnally and Bernstein, 1994). Non-response bias occurs due to respondents refusing to participate. Those who are not interested in the topic, or have other concerns, are unlikely to respond. Those who think that the survey is too complicated or boring may also turn down the invitation to complete the survey (Nulty, 2008). Based on social exchange theory, Dillman (1978) proposes three strategies for achieving higher response rates. These include rewarding respondents, decreasing the cost for respondents and building trust. It was decided to utilise non-financial incentives to enhance the response rate. The introduction in the was tailored to enhance respondents trust, according to Dillman s (1978) suggestions. The reward values of positive recognition and reliability were demonstrated by using the university s letterhead, signatures and language choice (i.e. English or Thai). Moreover, affiliate with a trustworthy organisation can have a positive impact on participants by establishing trust, which results in higher response rates (Turley, 1999). In this research, trust was built by collaborating with a recognised organisation that has a good reputation in the community. As mentioned earlier, online surveys reduce the costs to the respondents since they offer flexibility and convenience. Moreover, all of the questions were designed to avoid any possibility of humiliation or misunderstanding. To reduce respondent fatigue, all items were measured on five point Likert scales and each construct contained no more than eight items. All scales had been previously validated in extant research with acceptable validity and reliability ratings. 106

127 4.5.2 Unit of analysis The unit of analysis selected for this research was the customer of an Internet service provider. This is consistent with B2C research. The participants were randomly selected from Thai major ISPs database of existing customers (i.e. True Internet, TOT and 3BB). Some of these customers had switched from competing ISPs The focal organisations The three main service providers which account for 95% of the Thai home Internet services are True Internet, TOT and 3BB (True, 2014). As Thai customers have a variety of ISPs to choose from, there is a high churn rate which is essentially what this research intends to investigate. The customers choice of remaining with a particular ISP can be largely explained by their loyalty towards that ISP. The comprehensive customer database from the three major ISPs incorporates diverse customer profiles, including those who have switched from other ISPs or those who wish to change to other service providers. Hence, the sampling frame is representative of the entire population of Thai home Internet service customers. The participants for the survey were randomly selected from the customer database using computer software and this process was supervised by the researchers. The corporations did not have any control in determining who participated in the survey, and did not benefit from influencing or creating bias during the data collection process. The survey responses were stored in the university s Opinio database. The companies did not have access to this data. They were only interested in an independent and academically rigorous process, the findings of which would assist them in developing and designing long-term customer retention strategies The choice of Thailand Thailand is endowed with a wide variety of natural resources, a substantial population and a relatively strong economy. Enhanced investment in education has resulted in knowledge improvement and a larger number of higher educational qualified Thai people. Thailand is fast becoming an information society. This is part of the reason for the considerable development of the telecommunications industry in Thailand. During the period of 2000 to 2010, the average Internet user growth rate per year was 30 per 107

128 cent (Srihirun, 2011). Telecommunications play an important part in Thailand s economy and this is especially true for broadband, which is projected to grow 23 per cent annually (Srihirun, 2011). However limited research exists on customers buyer behaviour of home Internet services in Thailand. This provides the justification for choosing Thailand as the research area. Additionally, the student investigator originally hails from Thailand; hence he is very familiar with respondents cultural norms, expectations and language Pre-Testing In order to overcome the identified limitations of online surveys, pre-testing was carried out. A group of six respondents was asked to complete the questionnaire in front of the researcher. The researcher observed while the respondents were giving their responses to the question. Notes were taken when the researcher noticed respondents hesitating or making mistakes. These hesitations or mistakes signalled that problems and issues exist in the survey questions and layout, and need to be ameliorated. At the end of the session, the six individuals were encouraged to identify ambiguous questions, errors and suggest improvements to the survey. No major problems were identified in the final version of the survey Pilot study A pilot study is a smaller version of a full-scale study and allows for specific pretesting of a particular research instrument (Zikmund, 2000). The pilot study aims to test and examine the questionnaire sequencing, wording, and layout, as well as to evaluate data collection and data analysis procedures. Moreover, it can help to generate familiarity with participants, to anticipate response rates, and to estimate the total time required for questionnaire completion (Veal, 2005). The pilot test allows researchers to assess the precision of data collection, to minimise mistakes and errors from inadequate survey design, and to remove unnecessary interviewing guidelines (Zikmund, 2000). With regards to this study, the pilot test was run with 30 respondents, who were a small sample of respondents in the main study. According to the results from the pilot study, it took approximately 9 to 24 minutes for each respondent to complete the survey. No survey items were found to be problematic, and the wording of the questions was easy 108

129 to comprehend. The participants who took part in the pilot study were not included in the main study Data collection The online survey was made available via the university s Opinio platform. The web link was relayed for the online survey and ed to a representative sample of 8000 customers who were randomly selected from their database. All of the respondents are not locked into time contracts by the ISPs, and are free to switch to an alternate provider or continue their relationship. Care was taken to include a proportionate number of prospective respondents in all the four regions of Thailand. The final estimated sample size was 2000 Thai customers. It was envisaged that 500 completed and valid surveys from each region could be obtained. The survey instrument, which was originally written in English, was translated into Thai by a bilingual researcher. Subsequently, the translated Thai version was back-translated into English. Significant misunderstanding or confusion caused by a cross-cultural transformation was detected through the back-translation process. Discrepancies in the translation were carefully inspected and corrected to ensure that the items reflected the original meaning, and did not contain any social judgments. To confirm the error-free translation, the translated versions were cross-checked by three other bilingual scholars selected from Australian and Thai universities, who were competent in both Thai and English. Responses to the online survey were automatically returned to the student investigator through the Opinio platform. Opinio software enables the production and reporting of a survey and assures the anonymity, confidentiality and privacy of the respondents. In order to achieve accurate results, and in particular, to prevent multiple completions by the same respondent, the default in Opinio was set as follows: not allow multiple submissions, and prevent with cookies and IP-address check. This means that the Opinio software recognised every respondent s IP address (computer ID), and only one completed survey was accepted from a particular computer. The university s Opinio platform was kept live for a period of three months. 109

130 4.6 Tools for analysis The most appropriate tools to analyse quantitative type surveys are multivariate statistical techniques (Hair et al., 1998). These consist of correlation, regression, factor analysis, classification analysis and Structural Equation Modelling (Baldauf et al., 1999). In factor analysis all the variables are simultaneously analysed. They are considered to be inter-related and inter-dependent on each other, thus there are no identified independent and dependent variables (Hair et al., 1998). A drawback of most multivariate statistical techniques is that they only measure one dependent relationship at a time and it is difficult to identify multiple dependent relationships using these techniques (Hair et al., 1998). As SEM is an extension of several multivariate techniques, especially multiple regression and factor analysis, it overcomes this limitation (Hair et al., 1998) and was chosen for this research. Structural Equation Modelling (SEM) is a method for analysing data which is confirmatory in nature and where the variables interrelate simultaneously with each other (Kelloway, 1998). It is commonly used in management research, especially in the area of marketing where investigations of consumer behaviour are performed. SEM was chosen for this research because of the simultaneous interactive nature of the theoretical model, the confirmatory nature of the research questions and hypotheses in the theoretical model, the use of latent constructs in the theoretical model and finally, the use of surveys as the data collection method (Hair et al., 1998). SEM is the most appropriate technique as it allows testing of theoretical models, specifically those which consist of latent constructs (Anderson and Gerbing, 1988). It is also the best choice for analysing the confirmatory nature of the research question and the simultaneous nature of the multiple relationships among the key constructs (Hair et al., 1998). The multi-scale nature of the data and the use of ordinal scales require the use of polychoric correlation matrices of software programs (Hair et al., 1998). Therefore AMOS software was used. The next section examines validity and reliability issues as those are important to ensure consistency and accuracy of the results. 110

131 4.7 Reliability and validity Reliability and validity indicate how consistent and accurate the data is. Results that are low in reliability and validity are most likely to contain errors and can be questioned. Validity tests detect measurement errors arising from combinations of scores from multiple items, while reliability tests minimise measurement errors caused by random or systematic bias (Nunnally and Bernstein, 1994). This section demonstrates how these issues are dealt with through the design and implementation of the online survey used Reliability Reliability refers to the similarity of results measured by independent but comparable items of the same subject, attribute, or concept (Churchill, 1992). Reliability means that the information obtained from indicators (e.g. a measure of service quality) does not differ as an outcome of characteristics of the indicator, instrument (questionnaire), or measurement device per se (Churchill, 1992). Reliability indicates how well errors caused by random or systematic bias during the measurement are managed over time, or how well a scale can reflect the proposed constructs over time. Reliability for multi-item measures can be calculated by correlating each item with other items measuring the same construct, thus creating a reliability coefficient. A coefficient greater than 0.60 confirms reliability among items in a scale (Nunnally and Bernstein, 1994). In order to ensure high reliability in this research, measurement errors that violate the reliability are taken into consideration. Measurement errors that are presented non-randomly are considered systematic. Measurement errors can be grouped into four types: sample, transmittal, response, or analysis error (Green et al., 1988). These are controlled by research design and the measurement process. Sample errors occur when the sample does not correspond to the population (Green et al., 1988). In other words, the sample does not represent the whole population. This has been reduced by randomly selecting respondents from the customer database using computer software. The companies and the researchers did not have any control in determining who would participate in the survey. Transmittal error occurs in relation to the instrument material, the researcher position and the response by the respondent (Green et al., 1988). Transmittal error was minimised by structuring the instruments into 111

132 a reasonable order of questions, detecting the evoked set, and adopting words and phrases that minimise misperception or intimidation (Green et al., 1988). Response error refers to error during the interviewing process. This has been minimised by using an online survey to avoid interviewer bias which is more likely to be present in other methods of data collection (such as telephone interviews and face-to-face interviews). The responses were directly collected from the respondents without being interpreted or modified in any way. Furthermore, errors can occur during the analysis process due to inaccurate data entry and coding, or the use of chosen analysis techniques (Green et al., 1988). In order to avoid any uncertainties for data entry, the survey was pre-coded and majority of answers called for a selection from a series of options, rather than being left open-ended. The SEM was employed as the main analysis technique, because it is appropriate for this research as mentioned earlier, and has been widely used in previous studies in marketing literature Validity Validity refers to how well a particular scale reflects the construct for which it is intended (Green et al., 1988). Validity is an important measure in verifying the accuracy of research. When research is highly valid, the results are considered to be evidenced not by accident or error, but by the means reported in the study. Fundamentally, validity is confirmed when the measure is a flawless representation of the variable the investigator expects to measure (Neuman, 1997). Reliability does not guarantee validity, and vice versa, as consistency (reliability) and accuracy (validity) are two different concepts. Construct validity is considered a prerequisite for the development and testing of theories (Neuman, 1997). Poorly worded questions resulting in different interpretations are a good example of low validity. In addition, inappropriate choice of measuring devices (e.g. using surveys to research sensitive topics) is one of the common reasons for low validity. In general, there are three types of validity: content, construct and criterion validity (Neuman, 1997). The majority of scales consist of a sample of possible items reveal the features of the construct. Content validity relates to how well this sample represents the population of total items of the construct (Nunnally and Bernstein, 1994). Content validity for the 112

133 scales in this research was established through the use of previously validated scales demonstrating key dimensions of the constructs. Criterion validity (predictive or concurrent validity) refers to the ability of the measure to predict, as anticipated, another construct (Hair et al., 1998). Predictive validity offers more objective information than face validity (Faulkner, 1998). Predictive validity relates to the extent to which measures correlate with a construct that is a possible antecedent, requiring longitudinal data (Faulkner, 1998). A measure shows predictive validity when a relationship exists between the antecedents and the measure (Peter, 1992). Construct validity involves the relationships between items in a measurement scale. If the measures are not correlated, it is likely that they are not testing the same construct and thus the generalisability of the results is doubtful (Nunnally and Bernstein, 1994). There are two evaluations of construct validity (McColl-Kennedy and Fetter, 1999): convergent (i.e. measuring the same constructs by applying multiple methods) and discriminant (i.e. assessing the distinctiveness of the measure) (Neuman, 1997). Convergent validity was established via confirmatory factor analysis. The relationships among items were demonstrated by statistically significant estimated parameters (Holmes-Smith, 2000). Discriminant validity refers to whether the measure is unique as compared to other measures (Peter, 1992). Large correlations between latent constructs (greater than 0.80 or 0.90) signify the absence of discriminant validity (Heeler & Ray, 1972). Discriminant validity was examined in this research by calculating squared correlations coeffiecients between each pair of constructs and comparing them with the corresponding average variances extracted (AVE) of each construct. Discriminant validity is confirmed if all squared correlations coefficients are below AVEs (Fornell and Larcker, 1981). To reduce the risk of low validity, the measurement instruments for each construct were derived from pre-existing scales and adapted for a new context (Section 4.8) Constructing reliable and valid online questionnaires To create an instrument that maximises content validity, six issues need to be addressed: the actual scale items, the scale, instrument length, item choice, item analysis, and sample (Nunnally and Bernstein, 1994). To achieve construct validity, researchers must include homogenous items in a construct while assuring heterogeneity in the 113

134 methodology (Nunnally and Bernstein, 1994). Appropriate items should also be chosen and examined. Reliability is confirmed by reducing the systematic bias (using methods explained in the previous section) and the random bias. Minimising random bias can be achieved using clear wording and instructions, consistent data collection procedures and eliminating biased scoring. The items in the scale need to be well written, relevant and specific (Nunnally and Bernstein, 1994). This can be achieved through the use of simple wording, as high complexity tends to not only lower response rates, but also bias the sample towards those with a higher level of education. Explanations are provided throughout the survey to indicate which action is required from the respondents; examples were also used for clarification. The length of the instrument also needs to be considered. Adequate items should be included to assure validity. Although reliability improves as the number of items increases (De Vaus, 1995; Nunnally and Bernstein, 1994), too many items can lead to respondent fatigue. The number of items can be determined based on the scales level of internal consistency and validity. If the scales have been used before, and are deemed high in internal consistency, fewer items are needed (Nunnally and Bernstein, 1994). Additionally, a more heterogeneous population requires fewer items (Nunnally and Bernstein, 1994). Based on the above recommendations, the instruments in this study were carefully designed to ensure validity and reliability, while simultaneously avoiding respondent fatigue. 4.8 Measurement development Likert scales were selected because of their extensive use to measure similar constructs in previous research. In addition, they do not specify the distance between objects, and they combine scores, thus concealing any extreme dissimilarities between the scores (Green et al., 1988; Nunnally and Bernstein, 1994). The Likert scale is an ordinal scale that includes a range of objects ranging from least to most of a specific aspect (Nunnally and Bernstein, 1994). In this research, the Likert scale has strongly disagree, disagree, neither agree nor disagree, agree, strongly agree as the objects. 114

135 Measurement error due to aggregation can be minimised by removing items that do not differentiate well between high and low total scores of an item. This is done by using a single factor congeneric model for each construct. The Likert scale is deemed statistically sound on the basis that it is able to reflect distinct differences of an attribute (Nunnally and Bernstein, 1994). The scale has been derived from existing research to minimise measurement error. However, although being previously validated, these scales have only been administered in English. Therefore, in order to ensure that they are appropriate to use in another language, in this case Thai, these scales were tested by bilingual scholars via the back-translation process. A copy of the final questionnaire is included in Appendix 1. The operationalisation of each construct is explained in following sections Network quality Vlachos and Vrehopoulos s (2008) connection quality scale has been chosen to operationalise this construct. This scale examines the connection quality for mobile phone services, which are very similar to an ISP s network quality. Using a relatively new methodology, this scale originated from Chae et al. (2002), and it possesses strong factor loadings (0.71 to 0.73) and an acceptable Cronbach s alpha (α = 0.76). The purpose of the first item is to identify whether errors occur in the network when using the Internet. Items two and three investigate the speed of online downloading and uploading, as well as the consistency of the network. Construct Network quality 5Table 4.1: Network Quality scale Factor Cronbach s Adapted Measure loading alpha I do not experience any Internet disconnection from this ISP The Internet downloading and uploading speed meet my expectations The Internet speed does not reduce regardless of peak or off-peak hours Notes: Adapted from Vlachos and Vrechopoulos (2008). Item no. in survey

136 4.8.2 Customer service and technical support The customer service and technical support scale was taken from Wolfinbarger and Gilly (2003). This scale was chosen due to its ability to represent both customer service and technical support in the context of Internet services (Table 4.2). In Wolfinbarger and Gilly s (2003) research, eight items were originally tested. However, after a factor analysis, only three items were retained. These items have reasonable factor loadings (0.76 to 0.86) and a Cronbach s alpha of (α = 0.81). Wolfinbarger and Gilly s (2003) scale investigates whether the company is willing and prepared to respond to customers, and whether enquiries are answered promptly. 6Table 4.2: Customer service and technical support scale Factor Cronbach s Item no. Construct Adapted Measure loading alpha in survey Customer service and technical support Customer service personnel are knowledgeable Customer service personnel are willing to respond to my enquiries My technical problems are solved promptly Notes: Adapted from Wolfinbarger and Gilly (2003) Information quality and website information support Four different scales measuring information quality and website support were taken into consideration from Chae et al. (2002); Lin (2007); Kim and Niehm (2009) and Vlachos and Vrehepoulos (2008). After a thoughtful analysis, Kim and Niehm s (2009) information quality scale was chosen to measure information and website support (Table 4.3). This helps to obtain answers relating to information support and website support in Internet services. This scale has strong factor loadings ( ), and a high Cronbach s alpha of α = 0.96 compared to the Chae et al. s (2002) or Lin s (2007) information quality scale. The purpose of the information quality and website support scale is to examine the quality of information that the company offers to its customers. These items explore whether the information is up-to-date, accurate, useful, relevant and adequate for the customers. 116

137 7Table 4.3: Information quality and website information support scale Factor Cronbach s Item no. in Construct Adapted Measure loading alpha survey Information quality and website information support This ISP provides accurate information This ISP is informative The ISP provides updated information This ISP provides sufficient information This ISP provides up-to-date information The information on the ISP s website is relevant to me I can find what I need on the ISP s website This ISP provides relevant information Notes: Adapted from Kim and Niehm (2009) Privacy and security Vlachos and Vrehopoulos s (2008) scale was employed to measure privacy and security (Table 4.4). The original scale developed by Chae et al. (2002) was adapted by Vlachos and Vrehopoulos (2008) and it has strong factor loadings ( ) and a reasonable Cronbach s alpha (i.e. α = 0.87). This scale is relatively new when compared with the other scales and encompasses important aspects of privacy and security in an Internet services context. The items explore whether customers feel that their privacy is protected when doing business with an ISP. In addition, customer perception on security features is evaluated. Construct Privacy and security 8Table 4.4: Privacy and security scale Factor Cronbach s Adapted Measure loading alpha I feel that my personal information is protected at this ISP I feel that my financial information is protected at this ISP I feel that the transactions with this ISP are secured Notes: Adapted from Vlachos and Vrehopoulos (2008). 117 Item no. in survey

138 4.8.5 Customer trust Three different scales measuring perceived trust from Chou (2004), Aydin and Ozer (2005), and Kim and Niehm (2009) were examined. The scale from Kim and Niehm (2009) has very strong factor loadings (0.90 to 0.99) and an ideally high Cronbach s alpha (α = 0.92). However, several items in Kim and Niehm s (2009) trust scale are akin to the measures in the security/privacy scale that was chosen. Therefore, Kim and Niehm s (2009) trust scale was not chosen to ensure the scale discriminant validity. Instead, the scale from Aydin and Ozer (2005) was chosen (Table 4.5) as Aydin and Ozer s (2005) study was conducted in the mobile phone industry which is similar to the context of this study (i.e. residential internet services).this scale covers various aspects of trust and has reasonable Crombach s alpha. The factor loading of each item is not available in the original paper. However, Aydin and Ozer s (2005) analysed the unidimensionality of trust by specifying a measurement model. According to Jöreskog and Sörbom (1993), a goodness of fit index (GFI) of 0.90 or above suggests that the construct is unidimensional. The GFI value of trust was 0.94, satisfying the requirement. Furthermore, Aydin and Ozer s (2005) examined convergent validity of trust by using the Bentler Bonett normed fit index (NFI) (Bentler and Bonett, 1990). NFI value of trust was Therefore, it has been concluded that this scale is suitable for the research. The items in this scale have been designed to investigate customer perceptions of the company s honesty, responsibility and professional manners. The scale also attempts to determine whether customers feel that their service provider appreciates and cares about them. Construct Customer trust 9Table 4.5: Customer trust scale Factor Cronbach s Adapted Measure loading alpha 118 Item no. in survey I trust this ISP N/A I feel that I can rely on this ISP service I trust the ISP s billing system This ISP is reliable because it is mainly concerned with the customer s interests I feel that this ISP will not deceive me in any way Notes: Adapted from Aydin and Ozer (2005). N/A N/A N/A N/A

139 4.8.6 Customer satisfaction There were several scales for satisfaction considered, including Chiou (2004), Vlachos and Vrehepoulos (2008), Wolfinbarger and Gilly (2003), Brady, Cronin and Brand (2002), and Lin (2007). Although majority of them have strong factor loadings and acceptable Cronbach s alpha, measures of customer satisfaction were sourced from the scale of Chiou (2004) (Table 4.6). Chiou s scale (2004) has factor loadings ranging from 0.85 to 0.94 and significantly high Cronbach s alpha (α = 0.92). This scale aims to determine how satisfied customers are with the ISP and their service offerings. Results from the scale items intend to demonstrate whether the customers are happy about their decision to choose this ISP, as well as their service performance. Construct Customer satisfaction Notes: Adapted from Chiou (2004). 10Table 4.6: Customer satisfaction scale Factor Cronbach s Adapted Measure loading alpha I am happy about my decision to choose this ISP I believe that I did the right thing when I chose this ISP Overall, I am satisfied with this ISP Item no. in survey Customer commitment Two different scales relating to commitment from Eisingerich and Rubera (2010) and Hur, Park and Kim (2009) have been taken into consideration. After a thorough analysis, the commitment scale from Eisingerich and Rubera (2010) was chosen (Table 4.7). This scale has significantly strong factor loadings (0.94 to 0.96) and an acceptable Cronbach s alpha (α = 0.87). The items examine customers feelings of attachment and sense of belonging to their service provider. One reverse coded item was included to minimise problems of inattention and acquiescence of the respondents. 119

140 Construct Customer commitment 11Table 4.7: Customer commitment scale Factor Cronbach s Adapted Measure loading alpha Item no. in survey I feel involved with this ISP I am very proud to have this company as my service provider I feel attached to this ISP I will not buy this ISP s services in the future Notes: Adapted from Eisingerich and Rubera (2010) Customer value Four different scales measuring perceived value from Chiou (2004), Hur, Park and Kim (2009), Kim and Niehm (2009), Vlachos and Vrehepoulos (2008) have been examined. After thorough consideration, measures of perceived value scale were sourced from Kim and Niehm s scale (2009) as it has been recently developed and specifically covers aspects of perceived value that the study intended to measure (Table 4.8). This scale has relatively strong factor loadings ( ), and a reasonable Cronbach s alpha (α = 0.87) compared to the others. The measurement items aim to capture how customers perceive functional values of the service offering. Construct Customer values 12Table 4.8: Customer value scale Factor Cronbach s Adapted Measure loading alpha This Internet package is worth my money I would consider this Internet package to be a good buy I feel that I purchased a good Internet package with a reasonable price Notes: Adapted from Kim and Niehm (2009). Item no. in survey Customer loyalty Numerous loyalty scales were considered for use in operationalising this construct, including those developed by Chiou (2004), Aydin and Ozer (2005), Vlachos and Vrehepoulos (2008), Kim and Niehm (2009), and Zeithaml et al. (1996). As discussed earlier, customer loyalty is a bi-dimensional construct consisting of attitudinal and 120

141 behavioural components. Ultimately, the loyalty scale from Kim and Niehm (2009) was selected to measure attitudinal loyalty since it has strong factor loadings (0.71 to 0.95) and considerably high Cronbach s alpha (α 0.93). In addition, this scale is relatively new as compared to Chiou (2004) and Aydin and Ozer (2005). Also, it addresses the inner mechanism of customer loyalty by determining favourable relative attitudes towards the ISP and also considers customers propensity to recommend the service (Table 4.9). To measure the behavioural aspect of loyalty, Zeithaml et al. s (1996) scale was chosen (Table 4.10). This scale is among the most well-known and most cited scales in the loyalty literature. It has reasonable factor loadings (0.74 to 0.94), and is able to capture the outward behavioural manifestations of loyalty. It aims to determine customers propensity to stay and their intention to increase the volume of business with their service provider. Construct Attitudinal loyalty 13Table 4.9: Attitudinal loyalty scale Adapted Measure If I bought/upgraded a new Internet package, I would prefer this ISP I try to use this ISP because it is the best choice for me I consider myself to be a loyal patron of this ISP I would say positive things about this ISP to other people I would recommend this ISP to someone who seeks my advice Notes: Adapted from Kim and Niehm (2009). Factor loading Cronbach s alpha Item no. in survey

142 Construct Behavioural loyalty 14Table 4.10: Behavioural loyalty scale Adapted Measure I would consider this ISP as my first choice to buy services I would do more business with this ISP in the next few years I would do less business with this ISP in the next few years (-) Notes: Adapted from Zeithaml et al. (1996). Factor loading Cronbach s alpha Item no. in survey 0.74 N/A N/A N/A Preparing the data This section provides a brief overview of coding and editing data, as well as the method used to analyse them Coding and editing The survey was designed to ensure data to be coded and edited easily. Each question has numerical values allocated to the response. For example, a standard code used in the study was 99 for non-response or missing data, 1 for yes and 2 for no. Frequencies were run in SPSS to detect data entry errors Structural equation modelling This section discusses the main data analysis technique: Structural Equation Modelling (SEM). SEM was selected as it is the most appropriate method for confirmatory research questions and latent constructs in a quantitative approach (Schumacker and Lomax, 1996; ). There are seven stages in SEM. Each stage is discussed below Stage 1 Developing a theoretical model Structural Equation Modelling is used to evaluate relationships between variables in order to justify changes or deviations in other variables. A particular variable can influence and be influenced by others simultaneously (Hair et al., 1998; Hair, Black, Babin and Anderson, 2009). This notion has been identified in existing theories, as well as in extant research (Hair et al., 1998). A causal relationship between two variables in SEM can only be supported once the following four criteria are met (Hair et al., 1998; 122

143 Hair et al., 2009): significant correlation; temporal antecedence where effects must precede the development of causes; lack of substitute variables; and a theoretical foundation for the relationship (Hair et al., 1998; Hair et al., 2009). A significant concern in building a structural model in SEM is specification error, which occurs when an important independent variable is not included in the model (Kelloway, 1998; Kline, 2013). This type of error can result in a false conclusion about relationships between variables (Schumacker and Lomax, 1996). Dealing with misspecification requires a comprehensive theoretical foundation in specifying a model. Hence, variables need to be researched carefully in the literature in order to reduce the possibility of misspecification error. Moreover, although the number of constructs could vary, it is recommended not to include too many variables (Hair et al., 1998; Hair et al., 2009). Including too many variables may not be a realistic way to obtain data on an extensive scale and measure a large number of variables. A larger number of variables also require a larger sample size, which leads to a higher chance of parameters being significant (Hair et al., 1998; Hair et al., 2009; Streiner, 2013). Furthermore, models including more than 20 variables can be very difficult to comprehend (Hair et al., 1998; Hair et al., 2009) Stage 2 Constructing a path diagram The relationships between the variables can be explained by a path diagram: a graphical illustration method. A path diagram visually depicts the relationships between variables but cannot be used to determine causes (Kelloway, 1998; Schumacker and Lomax, 1996). Path analysis allows the identification of direct and indirect relationships between variables, as well as correlations between criterion and predictor variables (Kelloway, 1998). This is an extension of multiple regression analysis using additive equations; it demonstrates the cumulative effect of explanatory variables on a single predicted variable (Schumacker and Lomax, 1996; Kyndt and Onghena, 2014). However, when compared to path analysis, multiple regression analysis does not infer any relationships between the independent variables. It also does not allow variables to be dependent and independent simultaneously (Schumacker and Lomax, 1996). 123

144 SEM employs path diagrams as a technique of describing existing relationships in the theoretical model (Hair et al., 1998; Hair et al., 2009). They can be used to represent direct and indirect effects (Hair et al., 1998; Hair et al., 2009; Reisinger and Mavondo, 2007). Path diagrams comply with a set of standard conventions: a curved arrow for correlation, a straight arrow for a predictive relationship, and straight arrows from one construct to the second and from the second to the third for a mediated path (Hair et al., 1998; Hair et al., 2009; Reisinger and Mavondo, 2007) Stage 3 Converting the path diagram to a set of structural equations The next step following the construction of a path diagram is to establish measurement models and ultimately a structural model (Hair et al., 1998; Hair et al., 2009). The structural model illustrates each endogenous construct being the predicted variable in a single equation explained by other endogenous or exogenous variables in the model. Every equation has a path coefficient and an associated error/residual (Hair et al., 1998; Hair et al., 2009; Reisinger and Mavondo, 2007). The reliability of indicators was specified by using congeneric single factor models. According to Anderson and Gerbing (1988) and Rowe and Christie (2008) items with standardised residuals greater than 2.54 were removed. In addition, the composite reliability and variance extracted were also calculated. Reliability scores higher than 0.70, together with a variance extracted more than 0.5, demonstrate acceptable reliability (Anderson and Gerbing, 1988). Notwithstanding, it is not feasible to evaluate the reliability of single-item measures. In this case, there are two alternatives: the correction for attenuation formula and factor analysis (Bowling, 2005; Wanous and Reichers, 1996). If one wants to choose a single item from a multi-item scale, one could apply Spearman-Brown s prophecy formula in reverse (Nunnally and Bernstein, 1994). The simultaneous estimation of the measurement and structural models may generate complications in assigning meaning to theoretical constructs. To minimise the risk of interpretational confusion, the measurement model is separated from the structural model (Anderson and Gerbing, 1988). Joreskog (1993) introduces a strategy called model-generating models using Exploratory Factor Analysis (EFA) as the first step in the data analysis. In this approach, the researchers first specify the theoretical model. 124

145 Prior to testing the full model, a series of congeneric single factor models are tested (Cunningham, 2010). Congeneric single factor models can help create meaning by reducing the items to a single construct where construct validity is achieved (Anderson and Gerbing, 1988). Changes are only made to the model if they make substantive sense (Cunningham, 2010). Once constructs have been validated, a full measurement model is examined (Cunningham, 2010). This approach has been used in this research as it is found to be a pragmatic way forward (Cunningham, 2010) Stage 4 Choosing the input matrix This stage embraces the concerns of entering data in a suitable form and selecting an appropriate estimation process. There are two options in formatting the input data: correlation and covariance matrix. SEM does not require raw data to execute the test. If a correlation or covariance matrix exists, either can be used to estimate SEM. This is because the analysis is not related to individual observations, but rather to the configurations of relationships between variables across all respondents (Hair et al., 1998; Hair et al., 2009; Reisinger and Mavondo, 2007). In addition, data cleaning is important before entering them into the SEM software because outliers, non-normal data and missing-value data can significantly mislead the findings. Non-normal data results in the chi-square statistics under-estimating the model fit; hence, a parameter may be concluded as non-significant (a type 2 error). Small sample sizes are more likely to be non-normal, increasing the possibility of type 2 error (Schumacker and Lomax, 1996). In order to decrease non-normality, the Satorra-Bentler chi-square adjustment is advised for small sample sizes (Arbuckle, 1996; Lackner et al. 2006). However, this is not applicable in the current research study. When determining the methods for estimating the proposed model, estimation techniques and computer analysis software should be taken into consideration. The maximum likelihood estimation is the most popular method for normally distributed data. If the data is non-normal (especially in small sample sizes) an asymptotic distribution free estimation before weighted least squares is most appropriate (Hair et al., 1998; Hair et al., 2009; Reisinger and Mavondo, 2007). There are four common resampling techniques: direct estimation, bootstrapping, simulation and jack-knifing 125

146 (Hair et al., 1998; Hair et al., 2009). This research uses bootstrapping as a method of estimating standard errors and confidence intervals of a population parameter in order to deal with multivariate non-normal distribution and to construct hypothesis tests (Shao and Tu, 1995). Due to the well-developed calculation methods, bootstrapping is easy to perform. Moreover, bias correct bootstrapping has also been used to test indirect and direct effects as recommended by Preacher and Kelley (2011) Stage 5 Assessing the identification of the structural model Model identification refers to whether there is a unique solution for the model being tested. If the model is not identified, it is not possible to conduct analysis (Kelloway, 1998). A model is considered under-identified when there is a very large amount of unspecified information for the equations to explain (Holmes-Smith, 2000; Kim, Shin and Grover, 2010). In other words, too many parameters need to be assessed without adequate input data about the constructs. In contrast, models that have more than one possible solution (but one best or optimal solution) for each parameter estimate are said to be over-identified (Loehlin, 1992; Kim, Shin and Grover, 2010). Models in which there is only one solution for each parameter estimate are considered to be justidentified, thus having little theoretical value (Loehlin, 1992). Over-identified models are usually preferred, as these models enable the researcher to undertake statistical hypotheses test, including model fit (Loehlin, 1992). A method used for detecting the estimated quantity of excess parameters is t-rule (Kelloway, 1998). The t-rule reveals that t (t = number of parameters to be estimated) should be no more than (1/2 number of observed variables) x (number of observed variables +1) (Holmes- Smith, 2000). If t is larger than the number of observed variables + 1, the model is deemed to be under-identified. To resolve this issue, the number of estimated parameters should be reduced by deleting paths until the problem is corrected (Holmes- Smith, 2000). This error was not present in the analysis of this data, and thus did not require this correction Stage 6 Evaluating the model fit It is important to measure the fit between the estimated model and the data in the SEM. However, it is essential not to over-fit the model by neglecting the theory. A freed path 126

147 may enhance the model fit but has no theoretical or logical meaning. It is critical to obtain a sufficient number of estimated coefficients to achieve a particular level of fit (Schumacker and Lomax, 1996). Parsimony can be accomplished by eliminating nonsignificant parameters in the model. Model fit indices are categorised into three types: absolute, incremental and parsimonious measures (Iacobucci, 2010; Kelloway, 1998). Absolute fit measures estimate the overall model fit, the incremental measures are related to a null model, and the parsimonious measures are equal to the degrees of freedom (Hair et al., 1998; Hair et al., 2009). Absolute fit measures evaluate how effective the model is in fitting the sample data (Bollen, 1989). In other words, it identifies whether the paths specified in the model represent the observed data. A poor absolute fit shows that the paths identified in the model do not precisely reflect the paths existing in the sample data. This then needs additional or different paths in order to enhance the absolute fit of the model. A common measure of absolute fit is the chi-square statistic, which is reported for each model in this study. However, this fit index can be influenced by extremes in sample size (Kelloway, 1998). It is unlikely to detect significant differences in sample sizes less than 100. Significant results can be found even for minimal differences in sample sizes greater than 1000 (Anderson and Gerbing, 1988). Other measures of absolute fit include Root Mean Square Residual (RMR), Standardised Root Mean Square Residual (SRMR), Root Mean Square Error of Approximation (RMSEA), and the Goodness-of-Fit (GFI) index. The RMR is an average of magnitude of fitted residuals (Hair et al., 1998). However, the RMR does not show which component of the model is not appropriately specified, and this is its major limitation. On the other hand, the SRMR measures the average difference between corresponding elements of the sample and model-implied correlation matrices. As it is related to RMR, RMSEA is another measure which is related to the degrees of freedom (Hair et al., 1998). The GFI index, another popular measure, reveals the degree of variance between the sample covariance matrix and the specified covariance matrix (Bollen, 1989). Notwithstanding, the GFI index can also be affected by large sample sizes (Holmes-Smith, 2000). Considering the limitations of each of these measures, it is 127

148 recommended to combine a number of fit indices to assess absolute fit instead of relying on one particular measure. The second model fit type is incremental measures (Hair et al., 1998; Hooper, Cooghlan and Mullen, 2008). In these measures, the fitted model is compared with a baseline model which is typically a null model (covariance is zero) (Bollen, 1989; Kelloway, 1998). Incremental measures consist of adjusted-goodness-of-fit (AGFI), Tucker Lewis Index (TLI) and the Comparative Fit Index (CFI). AGFI, a common measure, adjusts GFI based on degrees of freedom (similar to the relationship between the RMR and RMSEA). These measures, nevertheless, still have some limitations as they may result in a negative index, provided the degrees of freedom are not high enough. The TLI known as the Non Normed Fit Index (NNFI) approximates the improvement per degree of freedom of the target model over an independent model. Finally the CFI compares non-centrality between the specified and null model, and performs well even when the sample size is small (Holmes-Smith, 2000). The third model fit type is parsimonious fit measures (Hair et al., 1998). Using this measure, the quantity of estimated parameters is compared with degrees of freedom in the sample data (Holmes-Smith, 2000; Hooper et al., 2008). Parsimonius measures consist of the Parsiminous Goodness-of-Fit Index (PGFI), normed chi-square and Akaike Information Criterion (AIC). This measure detects whether GFI has been achieved by simply adding more parameters into the model. No acceptable levels for these measures have been specified because they are contingent on the particular data being tested. However, a researcher can find a satisfactory level by modifying the number of parameters and marking the point where the parsimony measure stops decreasing and starts increasing (Holmes-Smith, 2000). A sample of these model fit indices are reported in this study including chi square, normed chi square, SRMR, GFI, AGFI, RMSEA, TLI and CFI. A summary of the acceptable levels for each has been obtained from Holmes-Smith (2000) and is described in Table A non-significant chi square statistic normally indicates there is no significant difference between the variance/covariance matrix and the model-implied variance/covariance matrix. However, in assessing a model fit one should not rely on 128

149 any single indicator. For example, although a non-significant chi-square suggests a good fit, GFI less than 0.90 could suggest that a better fit can be achieved. 15Table 4.11: Model fit indicators adopted in this study Indicator Abbreviation Preferred level Value of normed chi-square CMIN/df Below 4.0 chi-square P-value χ2 P-value >0.05 >0.05 Goodness-of-Fit Index, Adjusted Goodness-of-Fit GFI AGFI >0.95 >0.95 Index Tucker Lewis Index TLI >0.95, sometimes > 1.00 Comparative Fit Index CFI >0.95, sometimes only >0.90 Root-Mean-Square Error of RMSEA <0.05, sometimes <0.08 Approximation Standardised Root-Mean- Square Residual SRMR < Stage 7 Interpreting and modifying the model Once alterations have been done and parsimonious and good-fit models have been determined, the findings need to be interpreted with regard to theory. The principle relationships in the theoretical model are discussed in chapter six in light of the results, together with a discussion on unexpected findings. A significant issue in interpreting findings is the choice of standardised versus unstandardised solutions (Hair et al., 1998). In this research, the raw data was standardised to allow comparisons between the regression coefficients. In addition, standardised results allow implications to be obtained with regards to the comparative impact of every construct, specifically between those with dissimilar scales. Moreover, the use of standardised solutions enables comparison amongst samples. In the interpretation, another issue found is model re-specification. Eliminating or adding parameters is a method of achieving a better model fit; these modifications are categorised as theoretically and empirically founded (Hair et al., 1998; Lei and Wu, 2007). Theoretical relationships that illustrate the underlying theory must not be changed, however it is acceptable to modify empirical relationships. 129

150 The second step of the two-step model includes assessing the model fit and modifying the model appropriately. This stage also involves testing the hypotheses in the proposed model (Anderson and Gerbing, 1988). Values calculated in unidimensional models are used to fix the lambda and theta values. In addition, the proposed parameters are freed. Based on the level of standardised residuals to enhance the model fit and achieve parsimony, additional parameters can be fixed or freed Ethical considerations The collection of data from human subjects raises some important ethical considerations. The researcher needs to be aware of damage or harm that can happen to participants when undertaking research. The main concerns are: physical or legal harm, deception, informed consent and privacy (Neuman, 1997). Social research does not generally involve physical harm as few interventions occur. However, the researcher needs to identify areas of potential damage, for example, insecure equipment (Neuman, 1997). The use of online surveys in this research eliminated the likelihood of physical harm occurring. Non-respondents, while being encouraged to complete the survey, were not coerced in any way. Deception involves misleading the participants, either for the purpose of the research or the items, which can increase the level of mistrust and may contaminate the results (Neuman, 1997). The cover letter that accompanied the online survey stated the purpose of the research and how the data would be used. The online survey also began with the purpose of the research, the identity of the researcher and invited voluntary participation. Privacy and confidentiality are important issues to participants. Confidentiality was guaranteed for each participant and the identity of each participant was protected. This study obtained ethics clearance from the university s Human Research Ethics Committee (SUHREC) at a meeting held on 13 July 2012 and this is shown in Appendix

151 4.11 Chapter summary This chapter discussed the research design, including justifications for the use of quantitative methods, an online survey data collection method and Structural Equation Modelling (SEM). The chapter began with a discussion of research paradigm, quantitative research and sampling. It then presented a discussion on data collection procedures including sample selection, non-response bias, unit of analysis, pre-testing process and pilot study. The next section presented a justification of the analytical technique. This chapter also discussed the issues of validity and reliability and the steps taken to minimise any related errors. The sampling issues of selection and size were discussed along with response rates. Subsequently the discussion on the development of measures and scales was provided. Finally the stages of Structural Equation Modelling were outlined, and ethical considerations were provided. Figure 4.2 provides a roadmap to the structure of the overall thesis. 131

152 Chapter One An introduction to the thesis and overview of the chapters Chapter Two Literature Review Chapter Three Development of the theoretical model and related hypotheses Chapter Four Methodology Next chapter Chapter Five Analysis and Results Chapter Six Discussion, Recommendations and Conclusion 13Figure 4.2: Structure of the overall thesis 132

153 Chapter 5: Analysis and Results 5.1 Chapter overview Chapter 5 presents the results of the data analysis. This chapter begins with data screening (section 5.2), profile of respondents (section 5.3), details of Internet services used (section 5.4), additional descriptive statistics (section 5.4) and segmentation based on Internet usage patterns (section 5.6). Profiles of the participants are provided, followed by a description of the statistical measures adopted to assess each of the constructs of the conceptual model. The chapter proceeds to assess the validity and reliability of the constructs. This is followed by exploratory factor analysis (section 5.7), confirmatory factor analysis, and building of measurement models (section 5.8) and full Structural Equation Model analysis (section 5.9). Sections 5.10 presents segmentation analysis using invariance testing performed on the basis of Internet usage patterns, age groups and income levels. The organisation of this chapter is shown diagrammatically in Figure

154 5.1 Chapter overview 5.2 Data screening 5.3 Profile of Respondents 5.4 Details of Internet services used 5.5 Additional descriptive statistic 5.6 Segmentation based on Internet usage patterns 5.7 Exploratory analysis and reliability test 5.8 Confirmatory factor analysis and measurement models 5.9 Structural Equation Model (SEM) 5.10 Details of segmentation analysis using invariance testing 5.11 Chapter summary 14Figure 5.1: Chapter organisation 134

155 5.2 Data screening Prior to the analysis, the data was examined using Statistical Package for the Social Sciences (SPSS) Version 20. The accuracy of the data, missing values, outliers and assumptions of multivariate analysis were examined Missing data A total of 3803 online survey responses were collected. Of these 953 were incomplete hence they were discarded, reducing the total number to The data was then screened and cleaned using descriptive analysis. Missing values were detected. Since Structural Equation Modelling (SEM) requires a complete data set (Hair et al., 1998) and the sample size was reasonably large, Complete Case Analysis was chosen to deal with missing data. This technique, also known as the LISTWISE deletion in SPSS, is a simple and direct method of handling missing data in which only cases with valid data are retained. Any cases with missing data are excluded from the whole data set. Due to the large number of responses obtained in this research, the sample size after case deletion was sufficient for the analysis. In all 311 were deleted and the total number of valid responses was reduced to Response time bias Invalid responses can also be identified by the response time, which is the amount of time taken to complete a survey. Responses are considered invalid if particular respondents spend far less time on reading and providing their responses as compared to the sample. The pilot study determined that on average it took approximately 10 to 15 minutes for a respondent to complete the questionnaire. Therefore, it was determined that respondents who spent no more than 7 minutes on the questionnaire were deleted. In doing so, 444 responses were deleted. This brought the total number of valid responses to Outliers In general, outliers affect the normality of the data and are extreme values that noticeably stand out from the rest. There are three types of outliers: univariate, bivariate and multivariate outliers. Univariate outliers are cases that have an unusually high or low value for a single variable. Bivariate and multivariate outliers possess an exclusive 135

156 combination of values across two (i.e. bivariate outliers) or more (i.e. multivariate outliers) variables. This creates a distinctive difference between those observations compared to others (Hair et al., 1998). Frequency analyses were conducted for each variable using SPSS, with a view to screening for outliers. The maximum and minimum statistics were considered and extreme values were revisited and corrected using descriptive analysis. A comparison between means and 5 per cent trimmed means (when 5 per cent of highest and lowest values are deleted) calculated by descriptive statistics reveals issues related to outliers. If the two mean values are similar to each other, it can be assumed that univariate outliers are not a problem in the study (Pallant, 2007). The results of such analysis (Table 5.1) demonstrate that no significant difference between the means and the 5 per cent trimmed means was found. In terms of bivariate outliers, a scatterplot is first used to eyeball the presence of outliers (Filzmoser et al., 2005). Bivariate and multivariate outliers were detected by calculating Mahalanobis Distances, which provide a relative measure of a data point s distance from the centre cluster of remaining cases (Filzmoser et al., 2005). Outliers were found to be problematic for items 55 and 56. Item 55 related to the number people in the household and item 56 related to the number of people using Internet in the household. Logically, the value in item 56 should be equal or less than that in item 55 for the same respondent. Thirty six cases were found problematic and they were deleted from the data set bringing the final number of valid responses to

157 Item no. 16Table 5.1: Means and 5% trimmed means Statement Mean 5% Trimmed Means 1 I do not experience any Internet disconnection from this ISP The Internet downloading and uploading speed meet my expectations 3 The Internet speed does not reduce regardless peak or off-peak hours 4 Customer service personnel are knowledgeable Customer service personnel are willing to respond to my enquiries 6 My technical problems are solved promptly This ISP provides accurate information This ISP is informative The ISP provides updated information This ISP provides sufficient information This ISP provides up-to-date information The information on the ISP s website is relevant to me I can find what I need on the ISP s website This ISP provides relevant information I feel that my personal information is protected at this ISP I feel that my financial information is protected at this ISP I feel that the transactions with this ISP are secured This Internet package is worth my money I would consider this Internet package to be a good buy I feel that I purchase a good Internet package with a reasonable price 21 I am happy about my decision to choose this ISP I believe that I did the right thing when I chose this ISP Overall, I am satisfied with this ISP I trust this ISP I feel that I can rely on this ISP service I trust the ISP s billing system This ISP is reliable because it is mainly concerned with the customer s interests 28 I feel that this ISP will not deceive me in any way I feel involved with this ISP I am very proud to have this company as my service provider I feel attached to this ISP I will not buy this ISP s services in the future If I bought/upgraded a new Internet package, I would prefer this ISP 34 I try to use this ISP because it is the best choice for me I consider myself to be a loyal patron of this ISP I would say positive things about this ISP to other people I would recommend this ISP to someone who seeks my advice I would consider this ISP as my first choice to buy services I would do more business with this ISP in the next few years I would do less business with this ISP in the next few years (-)

158 5.2.4 Assessing normality and reliability It is important to check the normality of data in the early stages of multivariate analysis. This is because it can affect the results and findings of the analysis. Normality can be examined using both statistical and graphical methods (Tabachinick and Fidell, 2001). Normality tests aim to check whether the population from which the data is sampled is normally distributed (Allen and Bennett, 2010). Values of skewness and kurtosis can be used to determine the level of normality (Hair et al., 1998). West et al. (1995) suggest that skewness and kurtosis with absolute values greater than 2 and 7 respectively demonstrate relatively non-normal distributions. Kline (2005) asserts that absolute kurtosis values exceeding 10 show problematic non-normality, and values exceeding 20 indicate a significant departure from normality. The Kolmogorov-Smirnov test and Shapiro-Wilk test are designed to test the normality of distribution. However, these tests are sensitive to large samples. Given that the sample size in this research is large (n = 2059), the significant values of the Kolmogorov-Smirnov test and Shapiro-Wilk tests demonstrated minor deviations from normality. Random sampling was undertaken and reliability analysis was run for every batch of 400 random responses within the 2059 cases. This technique was used to validate the internal consistency of all factors obtained from the EFA. Table 5.2 illustrates the progressive reliability tests done for each batch of responses. The results reveal that the overall internal consistencies of the various batches of responses are acceptable, as the lowest Cronbach s alpha obtained was.726 and the highest.978. This is within the acceptable limits in the early stage of research (Nunnally, 1978). The results of these tests provide evidence that generalisation can be made to the population. 138

159 17Table 5.2: Results of progressive reliability tests Sample size Network Quality Customer Service Information Quality Security and Privacy Customer Satisfaction Customer Commitment Customer Trust Customer Value Attitudinal loyalty Behavioural loyalty Profiles of Respondents Using frequency and descriptive statistics in SPSS, the respondents profiles are analysed and discussed in the following sections Gender In terms of gender distribution, 65.5 per cent (1348 respondents) of the total respondents were male, and 34.5 per cent (711 respondents) were female Age 22.6 per cent (465 respondents) were between 18 and 28 years old, 38.7 per cent (797 respondents) were between 29 and 38 years old, 24.8 per cent (510 respondents) were between 39 and 49 years old and 13.9 per cent (287 respondents) were 50 years old or older Monthly household income The results reveal that only 5 per cent of the total respondents belonged to households with monthly income less than 10,000 baht. Of these, 1.4 per cent (28 households) reported that their household income was less than 5,000 baht per month; and 3.6 per cent (74 respondents) earned between 5,000 and 10,000 baht per month. Of those who earned more than 10,000 baht a month, 26.1 per cent (537 respondents) earned between 10,001 and 30,000 baht per month, 23.5 per cent (484 respondents) earned between 30,001 and 50,000 baht per month, 24.4 per cent (503 respondents) earned between 139

160 50,001 and 100,000 baht per month and 21 per cent (433 respondents) earned more than 100,000 baht per month Level of education 5.7 per cent of the total respondents (117 respondents) held a secondary education qualification or lower; 8.5 per cent (175 respondents) held a 2-year college qualification or associate degree. The majority of respondents (59 per cent or 1215 respondents) held a Bachelor s degree and 26.8 per cent (552 respondents) held a postgraduate degree or higher Main areas of respondents education With regards to the main area of respondents education, 12.3 per cent (2.4 respondents) majored in education, 29.7 per cent (611 respondents) were qualified in business-related disciplines; information technology was the major field of study for 16.6 per cent (342 respondents), 4.5 per cent (93 respondents) specialised in hospitality and 36.9 per cent (759 respondents) selected other areas including law, media, agriculture, and engineering Employment status More than half of the respondents (63.2 per cent or 1302 respondents) were employed in full time positions. 2.3 per cent (48 respondents) worked as part time, casual or seasonal staff per cent (506 respondents) were self-employed and 7.1 per cent (146 respondents) were retired Areas of employment With regards to areas of employment, 0.9 per cent (19 respondents) worked in the agricultural sector, 7.9 per cent (162 respondents) were employed in the education sector, 16.1 per cent (331 respondents) worked in information technology, 34.1 per cent (703 respondents) held positions in business. In addition, hospitality accounted for 13.6 per cent of respondents (280 respondents) and other areas, such as law, media and engineering, made up 27.4 per cent (564 respondents). 140

161 5.3.8 Location In terms of the geographical location, 65.3 per cent (1344 respondents) were from Bangkok s central business district (CBD) and 18.4 per cent (379 respondents) were from Bangkok s suburbs. 1.7 per cent (35 respondents) were from central Thailand, 1.8 per cent (38 respondents) were from eastern Thailand, 4.1 per cent (85 respondents) were from northern Thailand, 3 per cent (61 respondents) were from the north-east, 4.7 per cent (96 respondents) were from the south and 1 per cent (21 respondents) were from the west. 5.4 Details of Internet services used Using frequency and descriptive statistics in SPSS, the details of respondents Internet services used are analysed and discussed in the following sections Switching experience The results show that 61.5 per cent of the total respondents (1267 respondents) had used one or more other Internet service providers in the past. This indicates that 38.5 per cent of respondents (792 respondents) did not have any switching experience Reasons for switching In term of reasons for switching service providers, the most common reason given was poor network performance (43.9 per cent or 556 respondents), followed by low quality customer service (43.1 per cent or 546 respondents). The results show that 37.4 per cent (474 respondents) of the 1267 respondents who switched from other ISPs did so because of lack of promotional packages per cent (314 respondents) switched because they claimed that the previous ISP s service was too expensive and 23.7 per cent (300 respondents) mentioned one or more other reasons for switching, for example poor aftersales service, insecure payment methods and threats of privacy violation Frequency of Internet use The results show that a majority of respondents (73.3 per cent or 1509 respondents) used the Internet twice or more every day per cent (431 respondents) of the total respondents used the Internet once a day. Of those who used the Internet less frequently than once a day, 0.9 per cent (18 respondents) used the Internet once every two days, 141

162 0.4 per cent (8 respondents) used the Internet once every five days and 4.5 per cent (93 respondents) used the Internet once every week or less frequently Internet usage time In terms of weekly time allocation for online activities, nearly half of the respondents (48.5 per cent or 998 respondents) spent more than 30 hours on the Internet each week. 3.7 per cent (76 respondents) spent less than 2 hours on the Internet each week per cent (512 respondents) spent between 2 and 9 hours online per week, 14.9 per cent (307 respondents) spent between 10 and 19 hours on the Internet each week, and 8.1 per cent (166 respondents) spent between 20 and 29 hours on the Internet every week. A recent report reveals that the average time a person spends on the Internet is approximately 20 hours (Nielson, 2013), which is relatively consistent with the finding of this study. However, it is noticeable that almost half of the respondents in this study spent more than 30 hours on the Internet. This suggests that people tend to rely heavily on the Internet, making it a significant aspect of modern life Number of people in a household 5.7 per cent (118 respondents) of the total respondents stated there was only one person in their household per cent (396 respondents) lived in a two-person household, 19.4 per cent (399 respondents) stated that there were three people in their household, 27.1 per cent (557 respondents) said their household comprised four people and 28.6 per cent (396 respondents) reported there were five or more people in their household Number of people using the Internet in a household Although half of the respondents lived in four or five person households (section 5.3.6), a large proportion (35.3 per cent or 727 respondents) reported that just two people used the Internet in their household per cent (250 respondents) said only one person in their household accessed the Internet. For 25.4 per cent of respondents (522 respondents) three people used the Internet in their house, 17.4 per cent (359 respondents) recorded four people in their house using the Internet, while 9.8 per cent (201 respondents selected five or more people). 142

163 5.4.7 Types of Internet connection Broadband Internet was the most popular Internet connection with 59.1 per cent (1217 respondents) using this mode. Wireless USB Internet was the next most popular mode, being used by 24 per cent (494 respondents). Other Internet connection types were significantly less popular than these two modes. Only 5.2 per cent (107 respondents) used a dialup Internet connection, 5 per cent (103 respondents) used a lease line Internet connection and 6.7 per cent (138 respondents) used another type of Internet connection, for example fibre optics Home Internet speed More than two-thirds of respondents (73.7 per cent or 1518 respondents) stated that their Internet speed ranged from 1.6 mbps to 20 mbps which is considered to be medium speed per cent (343 respondents) claimed that their Internet speed was between 512 kbps and 1.5 mbps which is considered to be lower speed. Internet speed for 1.7 per cent (36 respondents) of total respondents was lower than 512 kbps. The category of higher speed, i.e. between 21 mbps to 30 mbps was experienced by 1.7 per cent (36 respondents) and 3.1 per cent (64 respondents) selected Internet speed faster than 30 mbps. 3 per cent (62 respondents) were not sure about their Internet speed Internet expenditure per month Nearly half of all respondents (48.1 per cent or 991 respondents) spent between 601 and 900 baht on Internet services per month per cent (414 respondents) spent slightly less on Internet services (between 301 and 600 baht) monthly. 2.2 per cent of respondents (46 respondents) spent less than 300 baht per month, while of the higher spenders, 12.7 per cent (262 respondents) spent between 901 and 1200 baht per month and 12.2 per cent (251 respondents) spent 1200 baht or more per month for Internet services. 4.6 per cent (95 respondents) were not sure about their expenditure for Internet services. 143

164 5.5 Additional descriptive statistics In order to provide a generic situation of the interrelations between two variables, as well as to detect interactions between them, cross tabulation was employed. Cross tabulation displays a joint frequency distribution of two or more categorical variables (Michael, 2001). The distribution of cases by their values constitutes contingency table analysis and is among the most common analysis methods in survey research (Michael, 2001). In some cases, the analysis can be assisted by the chi square statistics ( 2 ) to determine whether variables are statistically independent or related. If a dependency between variables is confirmed, other indicators, for example Cramer s V and gamma, can be used to measure the strength of association between the variables. Cramer s V values range from 0 (showing no association between the variables) to 1 (indicating complete association) Switching experience versus gender 64.8 per cent of males and 55.3 per cent of female reported having switched ISPs. Chisquare statistics was significant at p = (< 0.05). However, Cramer s V value was 0.093, suggesting a weak relationship (Table 5.3). Therefore, it can be concluded that males and females had similar experiences with regards to switching ISPs. 18Table 5.3: Switching experience versus gender Previous experience with other ISPs Yes No Gender Total Male Female Count % within Previous experience with other ISP 69.0% 31.0% 100.0% % within Gender 64.8% 55.3% 61.5% Count % within Previous experience with other ISP 59.8% 40.2% 100.0% % within Gender 35.2% 44.7% 38.5% Count Total % within Previous experience with other ISP 65.5% 34.5% 100.0% % within Gender 100.0% 100.0% 100.0% Note: Chi-square = , df = 1, p < 0.001, Cramer s V =

165 5.5.2 Switching experience versus age The number of respondents who switched from other ISPs was slightly higher than the number of who had not signed up with any other providers previously. Respondents in the age group had less switching experience as compared to other groups. Although chi-square statistics were significant at p value < 0.05, Cramer s V value was indicating that the relationship between age and switching experience was weak (Table 5.4). Previous experience with other ISPs Yes 19Table 5.4: Switching experience versus age Age group Total and above Count % within Previous experience with other ISP 20.4% 39.5% 24.9% 15.2% 100.0% % within Age group 55.7% 62.7% 61.8% 67.2% 61.5% Count % within Previous experience No with other ISP 26.0% 37.5% 24.6% 11.9% 100.0% % within Age group 44.3% 37.3% 38.2% 32.8% 38.5% 4.6% 38.5% Count Total % within Previous experience with other ISP 22.6% 38.7% 24.8% 13.9% 100.0% % within Age group 100.0% 100.0% 100.0% 100.0% 100.0% Note: Chi-square = , df = 3, p = 0.011, Cramer s V = Switching experience versus household income Table 5.5 depicts the relationship between monthly household income and switching experience. Except for the group with household income under 5,000 baht, it appears that the higher the household income of respondents, the more likely they were to change service providers. For example, approximately 70 per cent of respondents with a household income of 100,000 baht or more had previously purchased from other service providers. Chi square statistics was significant at p < and Cramer s V value was 0.129, demonstrating a moderately weak relationship between these variables. In other words, income may be related to the switching behaviour of an ISP s customers. 145

166 20Table 5.5: Switching experience versus household income Previous experience with other ISPs Yes No Household monthly income Total Under 5,000-10,001-30,001-50,001 - Over 5,000 10,000 30,000 50, , ,000 Count % within Previous experience with 1.4% 3.2% 23.4% 21.6% 26.4% 24.0% 100.0% other ISP % within Household monthly income 64.3% 54.1% 55.3% 56.6% 66.4% 70.2% 61.5% Count % within Previous experience with other ISP 1.3% 4.3% 30.3% 26.5% 21.3% 16.3% 100.0% % within Household monthly income 35.7% 45.9% 44.7% 43.4% 33.6% 29.8% 38.5% Count % within Previous Total experience with 1.4% 3.6% 26.1% 23.5% 24.4% 21.0% 100.0% other ISP % within Household 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% monthly income Note: Chi-square = , df = 5, p < 0.001, Cramer s V = Reasons for switching versus gender Table 5.6 illustrates the reasons for switching amongst males and females. Whereas network performance and customer service were popular reasons among both females and males (47.7 per cent and 45.8 per cent respectively for males and 35.4 per cent and 37.2 per cent respectively for females), less attractive promotion packages were chosen by a larger percentage of females (38.4 per cent). This might suggest that females were more concerned about the attractiveness of promotional offerings. There was a noticeably larger percentage of both males and females who attributed their switching behaviour to other reasons (23.2 per cent and 24.7 percent respectively). 146

167 21Table 5.6: Reasons for switching versus gender Reasons for switching Total Less promotional package Poor Internet service performance High price Poor customer service Other reason Gender Total Male Female Count % within Reasons 68.1% 31.9% % within Gender 37.0% 38.4% % of Total 25.5% 11.9% 37.4% Count % within Reasons 75.0% 25.0% % within Gender 47.7% 35.4% % of Total 32.9% 11.0% 43.9% Count % within Reasons 68.5% 31.5% % within Gender 24.6% 25.2% % of Total 17.0% 7.8% 24.8% Count % within Reasons 73.3% 26.7% % within Gender 45.8% 37.2% % of Total 31.6% 11.5% 43.1% Count % within Reasons 67.7% 32.3% % within Gender 23.2% 24.7% % of Total 16.0% 7.7% 23.7% Count % of Total 69.0% 31.0% 100.0% Reasons for switching versus age No noticeable difference was found among age groups in terms of their reasons for switching. It can be seen that age did not contribute to the respondents motivation for switching and this is illustrated in Table

168 22Table 5.7: Reasons for switching versus age Reasons for switching Total Less promotional package Poor Internet service performance High price Poor customer service Other reason Age group Total and above Count % within Reasons 21.3% 40.5% 23.6% 14.6% % within Age 39.0% 38.4% 35.6% 35.8% % of Total 8.0% 15.2% 8.8% 5.4% 37.4% Count % within Reasons 20.0% 42.8% 23.4% 13.8% % within Age 42.9% 47.6% 41.3% 39.9% % of Total 8.8% 18.8% 10.3% 6.1% 43.9% Count % within Reasons 18.8% 43.3% 26.1% 11.8% % within Age 22.8% 27.2% 26.0% 19.2% % of Total 4.7% 10.7% 6.5% 2.9% 24.8% Count % within Reasons 20.1% 42.5% 24.0% 13.4% % within Age 42.5% 46.4% 41.6% 37.8% % of Total 8.7% 18.3% 10.3% 5.8% 43.1% Count % within Reasons 21.0% 38.0% 25.7% 15.3% % within Age 24.3% 22.8% 24.4% 23.8% % of Total 5.0% 9.0% 6.1% 3.6% 23.7% Count % of Total 20.4% 39.5% 24.9% 15.2% 100.0% Reasons for switching versus household income Table 5.8 depicts the relationship between monthly household income and reasons for switching. A significantly high percentage of respondents who earned less than 5000 baht selected poor Internet service performance as their motivation for switching service providers. In the under 5,000 baht group of respondents, 61.1 per cent selected poor Internet performance as their reason for switching, whereas this figure only ranged from 32.5 per cent to 47.3 percent for other income groups. On the other hand, high price and less attractive promotional packages were the main reasons among the 5000 to 10,000 baht income group (45 per cent and 42.5 per cent respectively). Based on the survey results, it seems that the low-income group was sensitive to the value for money aspect. 148

169 23Table 5.8: Reasons for switching versus household income Reasons for switching Total Less promotional package Poor Internet service performance High price Poor customer service Other reason Household monthly income Total Under 5,000 5,000-10,000 10,001-30,000 30,001-50,000 50, ,000 Count % within Reasons 1.4% 5.0% 27.7% 27.7% 38.3% % within Income 27.8% 45.0% 33.3% 36.1% 41.0% % of Total 0.5% 1.9% 10.3% 10.3% 14.2% 37.2% Count % within Reasons 2.7% 3.2% 27.0% 28.4% 38.7% % within Income 61.1% 32.5% 37.0% 42.3% 47.3% % of Total 1.1% 1.3% 11.4% 12.0% 16.4% 42.4% Count % within Reasons 1.7% 7.1% 32.5% 25.8% 32.9% % within Income 22.2% 42.5% 26.3% 22.6% 23.7% % of Total 0.4% 1.8% 8.1% 6.4% 8.2% 24.9% Count % within Reasons 1.6% 3.5% 31.2% 26.8% 36.8% % within Income 38.9% 37.5% 45.1% 42.0% 47.3% % of Total 0.7% 1.6% 13.9% 11.9% 16.4% 44.5% Count % within Reasons 1.3% 3.9% 32.0% 29.4% 33.3% % within Income 16.7% 22.5% 24.9% 24.8% 23.1% % of Total 0.3% 0.9% 7.7% 7.1% 8.0% 24.0% Count % of Total 1.9% 4.2% 30.8% 28.5% 34.7% % 5.6 Segmentation based on Internet usage patterns Different customers have distinctive needs and require tailored approaches (Mazzoni et al., 2007; Ringle et al., 2013). Based on their usage pattern, ISP customers are generally segmented into heavy, medium and light user categories. On average, an Internet user spends between nine and 20 hours online each week (ACMA, 2012). Heavy users are those who spend more than 29 hours on the Internet every week, while light users are those who use the Internet for less than 9 hours per week (Assael, 2005). A study by the Electronic Transactions Development Agency (ETDA, 2013) reveals that in general, Thai Internet users who spend less than 11 hours per week online account for 35.7 per cent of all users. Those who spend between 11 and 20 hours per week online make up 25.8 per cent per cent of Thai users spend 21 to 41 hours on the Internet weekly and 27.8 per cent spend more than 41 hours weekly online. This study, adapting usage segmentation from previous research, categorises three main groups of Internet users: 149

170 light users who are online less than 9 hours per week, medium users who spend between 9 and 29 hours online each week, and heavy users who are connected to the Internet for more than 29 hours per week. Heavy users made up 48.5% (998 respondents), 23% (474 respondents) considered themselves medium users and 29% (597 respondents) said they were light users. The number of customers in the heavy users segment is expected to increase as Internet connectivity continues to play a growing role in people s day-to-day life General characteristics of Internet usage groups Light users were most likely to be between 29 and 38 years (36.4% of respondents) or 39 and 49 years and working full-time (32.8% of respondents) (Table 5.9). More than 40% of pensioners were also light users (Table 5.10). Possible explanations for these statistics include that full time workers do not have spare time to surf the Internet, and retired people are often not as familiar with technology, especially in an Asian context. In terms of gender, females were more likely to be light users than males (Table 5.11). There were no discernibly unusual patterns among the medium user age groups (Table 5.9) and the distribution of gender was relatively even in this group of users (Table 5.11). Noticeably, more than 50% respondents working in the education sector were identified as medium users (Table 5.12). 24Table 5.9: Age groups among light, medium and heavy users Light user Medium user Heavy user Age group > Count % within Internet usage 15.10% 36.40% 32.80% 15.60% % within Age group 19.10% 26.90% 37.80% 32.10% % of Total 4.30% 10.40% 9.40% 4.50% Count % within Internet usage 19.90% 34.90% 27.30% 18.00% % within Age group 20.20% 20.70% 25.30% 29.60% % of Total 4.60% 8.00% 6.30% 4.10% Count % within Internet usage 28.30% 41.90% 18.80% 11.00% % within Age group 60.60% 52.40% 36.90% 38.30% % of Total 13.70% 20.30% 9.10% 5.30% 150

171 25Table 5.10: Employment status among light, medium and heavy users Full time Part time Selfemployed Unemployed Light Count Retired user % within Internet usage 70.20% 2.00% 18.20% 5.60% 3.90% % within Employment status 31.70% 25.00% 21.10% 22.60% 40.40% % of Total 20.10% 0.60% 5.20% 1.60% 1.10% Medium Count user % within Internet usage 65.30% 1.30% 24.90% 5.30% 3.20% % within Employment status 23.70% 12.50% 23.30% 17.10% 26.30% % of Total 15.00% 0.30% 5.70% 1.20% 0.70% Heavy Count user % within Internet usage 58.10% 3.00% 28.20% 8.80% 1.90% % within Employment status 44.50% 62.50% 55.50% 60.30% 33.30% % of Total 28.20% 1.50% 13.60% 4.30% 0.90% Heavy users were more distinctive when compared to the other two segments. Young Internet users made up the largest percentage of this group; more than 60% were in the age bracket and 50% were in the age bracket (Table 5.9). Therefore, it is not surprising that more than half of the students in this study were classified as heavy users (Table 5.13). This is a logical finding as students tend to be conversant with technology and usually rely on the Internet in their daily lives for example, for study, entertainment and personal networking (Fulps, 2013). 26Table 5.11: Gender among light, medium and heavy users Gender Male Female Light user Count % within Internet usage 60.70% 39.30% % within Gender 26.50% 32.50% % of Total 17.30% 11.20% Medium user Count % within Internet usage 67.00% 33.00% % within Gender 23.50% 21.90% % of Total 15.40% 7.60% Heavy user Count % within Internet usage 67.50% 32.50% % within Gender 50.00% 45.60% % of Total 32.70% 15.70% 151

172 27Table 5.12: Area of employment among light, medium and heavy users Light user Medium user Heavy user Agriculture Educati on IT Business Hospitality Other Count % within Internet usage 0.70% 7.10% 12.80% 34.40% 15.80% 29.30% % within Area of your employment 21.10% 25.90% 22.70% 28.70% 33.20% 30.50% % of Total 0.20% 2.00% 3.60% 9.80% 4.50% 8.40% Count % within Internet usage 1.70% 8.00% 18.20% 33.40% 12.90% 25.80% % within Area of your employment 42.10% 50.60% 26.00% 22.50% 21.80% 21.60% % of Total 0.40% 4.00% 4.20% 7.70% 3.00% 5.90% Count % within Internet usage 0.70% 8.20% 17.00% 34.40% 12.60% 27.10% % within Area of your employment 36.80% 23.50% 51.40% 48.80% 45.00% 47.90% % of Total 0.30% 1.80% 8.30% 16.70% 6.10% 13.10% 28Table 5.13: Percentages of students among light, medium and heavy users Light user Medium user Heavy user Student Yes No Count % within Internet usage 9.5% 90.5% % within Student 23.4% 29.2% % of Total 2.7% 25.8% Count % within Internet usage 10.6% 89.4% % within Student 20.9% 23.2% % of Total 2.4% 20.5% Count % within Internet usage 13.3% 86.7% % within Student 55.6% 47.5% % of Total 6.5% 42.0% Service quality perceptions Four dimensions of an ISP s service quality network quality, customer service, information support and privacy and security were identified and discussed in Chapter 3. In order to gain more insights into the ISP s customers, different Internet usage groups were examined in terms of their perceptions of service quality dimensions. 152

173 In general, respondents were satisfied with their experience of the ISPs customer service, security and privacy practices (Figure 5.2). They were less satisfied with information support and network quality. This pattern repeats itself across light and medium user segments. Noticeably, light users were more generous in their ratings of service quality when compared with the other two groups, while heavy users gave the lowest overall ratings. Furthermore, in contrast to the light and medium users, the heavy user group was more content with the security and privacy provisions of their ISP, followed by customer service. This implies that heavy users had higher expectations of service quality than light and medium users Network quality Customer service Information website Security and privacy Light user Medium user Heavy user All respondents 15Figure 5.2: Evaluating service quality: Mean Statistics Complaints behaviour Heavy users were more aggressive than the other two segments in responding to service failures. When experiencing problems with the services, they did not hesitate to complain to the ISP, as well as to other people in their personal network including friends and family. The intention to complain to other customers was also higher for heavy users than light or medium users. In other words, an ISP which fails to deliver on its promise is more likely to experience negative word of mouth from heavy Internet users. 153

174 The four service quality dimensions were also investigated with respect to their influence on customers intention to complain. Results from multiple regression analyses are shown in Table In general, only network quality and information support had significant effects on intentions to complain. However the effects of all four dimensions of service quality on complaining intention vary across light, medium and heavy users. Only the lack of information support motivated heavy and light users to complain, while poor network performance was the most likely cause for complaints among medium users. Further analyses were carried out to test the effects of service quality dimensions on the criterion variable by eliminating the factors that were not significantly associated with complaining intentions. The beta values for these variables are also presented in Table Table 5.14: Relationships between customers perceptions of service quality, and intention to complain, beta coefficients Intention to complain Intention to complain (after Total Total dropping insignificant factors) Light Medium Heavy Light Medium Heavy users users users users users users NQ -.079*** ** ** -.218*** CS IW -.158*** -.130* ** -.164*** -.180*** -.223** SP Model summaries for complaining intention Model summaries for complaining intention (after dropping insignificant factors) R 2 (total) =.049, f (4, 2054)= R 2 (total) =.047, f (4, 2054)= R 2 (light users) =.036, f (4,583) = R 2 (light users) =.032, f (4,583) = *** R 2 (medium users) =.051, f (4, 468) = 6.3 R 2 (medium users) =.045, f (4, 468) = *** R 2 (heavy users) =.052, f (4, 997) = R 2 (heavy users) =.050, f (4, 997) = Notes: total = coefficient values for all participants, * p <.05, ** p <.01, *** p <.001 Hours spent on Internet per week: light users = use Internet under 9 hrs/week; medium users = use Internet 9 29 /week; heavy users = use Internet more than 29 hrs/week. NQ = network quality; CS = customer service; IW = information support; SP = security and privacy Switching behaviour Drawing on prior knowledge of high churn rates among ISP s customers, it is not surprising that more than 50% of the respondents in this study have used other ISPs in the past. It is noticeable that nearly 80% of heavy users remarked that they had switched service providers previously. This figure was 63.2% and 57.8% for medium and light users respectively. This suggests that customers with higher Internet usage are more likely to switch service providers. Table 5.15 indicates the percentage of heavy users who have had previous experience with other ISPs. 154

175 30Table 5.15: Switching experience Previous experience with other ISPs Heavy user Medium user Light user Yes 79.4% 63.2% 57.8% No 20.6% 36.8% 42.2% In terms of switching behaviour (Table 5.16 and Figure 5.3), more than 50% of heavy users chose poor Internet service performance as the reason for switching service providers. The second most popular reason was unsatisfactory customer service, followed by less attractive promotional package. Only 24.7% of heavy users switched because of the price of the services. It appears that more customers (50.6%) belonging to this group of users focus on the core performance of their Internet service. These users are less concerned with the monetary cost (24.7%) or the type of promotional package offered (38.5%). 41.5% of medium users are likely switch when they perceive that their Internet service is of poor quality and 40.8% would switch when they experience poor customer service. Light users appeared to be equally concerned with customer service (40.3%), poor Internet service performance (39.1) and attractive promotional package (39.1%). It can be concluded that although heavy users associated their switching decision predominantly to Internet performance, all three groups considered poor Internet service performance, poor customer service, and less attractive promotional package as the top three reasons to switch. 31Table 5.16: Reasons to switch Reasons to switch to other ISPs Heavy user Medium user Light user Less promotional package 38.5% 33.8% 38.5% Poor Internet service performance 50.6% 41.5% 39.1% High price 24.7% 24.7% 25.0% Poor customer service 42.7% 40.8% 40.3% Other reason 25.2% 22.1% 22.4% 155

176 Heavy users Medium users Light users 20 0 Less promotional package Poor Internet service performance High price Poor customer service Other reason 16Figure 5.3: Reasons to switch Results for the relationship between switching intentions and the four service quality dimensions are shown in Table For all participants, customers perceptions of network quality, customer service, information support, and security and privacy were negatively related to switching intention. While the core ISP service offering network quality was the predominant reason for the intention to switch, customer service and security and privacy were least influential. As expected, the effects of all four ISP service quality dimensions on customers intention to switch varied among light, medium and heavy users. Network quality played an important role in both medium and heavy users switching intentions, whereas for light users it was inadequate information support. Furthermore, it is noted that customer service was negatively and significantly associated with the switching intentions of heavy users. The findings are confirmed by further analyses in which insignificant factors were deleted. The beta values for these variables are shown in Table

177 32Table 5.17: Relationships between customer perceptions of service quality, and intention to switch, beta coefficients Intention to switch Intention to switch (after dropping Total Total insignificant factors) Light Medium Heavy Light Medium Heavy users users users users users users NQ -.122*** ** -.119** -.122*** -.280** -.159*** CS -.066* * -.066* -.171*** IW -.100** -.168** ** -.294*** SP -.078** ** Model summaries for switching intention Model summaries for switching intention (after dropping insignificant factors) R 2 (Total) =.092, F(4, 2054) = R 2 (Total) =.092, F (4, 2054) = R 2 (Light Users) =.1, F (4,583) = R 2 (Light Users) =.086, F (4,583) = R 2 (Medium Users) =.089, F (4, 468) = R 2 (Medium Users) =.078, F (4, 468) = R 2 (Heavy Users) =.09, F (4, 997) = R 2 (Heavy Users) =.080, F (4, 997) = Notes: total = coefficient values for all participants, * p <.05, ** p <.01, *** p <.001 Hours spent on Internet per week: light users = use Internet under 9 hrs/week; medium users = use Internet 9 29 /week; heavy users = use Internet more than 29 hrs/week. NQ = network quality; CS = customer service; IW = information support; SP = security and privacy Intentions to recommend When it comes to peer recommendation (Table 5.18), light users were most likely to recommend their current service provider to other people (71.4%). Heavy users were the next most likely, with 65.7% likely to recommend their ISP to peers (Table 5.19 and Figure 5.4). Light users would recommend an ISP primarily based on an attractive promotional package (30.2%), followed by good customer service (28.2%) and good Internet service performance (26.7%). In contrast, medium users considered customer service as the most important motivation for a recommendation (33.2%). In addition, good Internet service performance (29.3%) and reasonable promotional packages (25.5%) were also popular motivations for recommendations by medium users. For heavy users, good Internet service performance motivated 41.8% to provide recommendations. In general though, across Internet usage groups, performance, customer service, and attractive promotional packages were the top three reasons why customers would recommended an ISP to others. Price contributed the least towards customer recommendation. 157

178 33Table 5.18: Intention to recommend Will you recommend your current ISP to Heavy user Medium user Light user other people? Yes 65.7% 54.2% 71.4% No 34.3% 45.8% 28.6% 34Table 5.19: Reason to recommend the current ISP to others Why would you recommend your current Heavy user Medium user Light user ISP to other people? More promotional package 26.6% 25.5% 30.2% Good Internet service performance 41.8% 29.3% 26.7% Low price 15.4% 12.7% 19.5% Good customer service 31.3% 33.2% 28.2% Other reason 9.2% 8.1% 11.6% Heavy users Medium users Light users 20 0 More promotional package Good Internet service performance Low price Good customer service Other reason 17Figure 5.4: Reason to recommend the current ISP to others In order to determine whether the four ISP service quality dimensions had any impact on a customer s intention to recommend, multiple regression analyses were carried out and the results are shown in Table The values of R 2 were relatively large suggesting reasonable effect size (Cohen, 1988). Generally, all four service quality dimensions were positively associated with intentions to provide recommendations. However, customer service was not a reason for medium users to recommend the service or the company. Interestingly, information support, and security and privacy 158

179 were the most important determinants of recommendations among light and medium users. Information support and network quality were the two most important dimensions that contributed towards heavy users intentions to recommend. The effects of service quality dimensions on intentions to recommend were further confirmed by eliminating the factors that were not significantly associated with intentions to recommend. The beta values for these variables are presented in Table Table 5.20: Relationships between customer perceptions of service quality, and intention to recommend, beta coefficients Intention to recommend Intention to recommend (after Total Total dropping insignificant factors) Light Medium Heavy Light Medium Heavy users users users users users users NQ.213***.18***.2***.233***.213***.18***.212***.233*** CS.081***.079* **.081***.079*.098** IW.251***.286***.231***.238***.251***.286***.253***.238*** SP.241***.268***.309***.199***.241***.268***.316***.199*** Model summaries for recommend intention Model summaries for recommend intention (after dropping insignificant factors) R 2 (Total) =.432, F(4, 2054) = R 2 (Total) =.432, F (4, 2054) = R 2 (Light Users) =.455, F (4,583) = R 2 (Light Users) =.455, F (4,583) = R 2 (Medium Users) =.462, F (4, 468) = R 2 (Medium Users) =.461, F (4, 468) = R 2 (Heavy Users) =.405, F (4, 997) = R 2 (Heavy Users) =.405, F (4, 997) = Notes: total = coefficient values for all participants, * p <.05, ** p <.01, *** p <.001 Hours spent on Internet per week: light users = use Internet under 9 hrs/week; medium users = use Internet 9 29 /week; heavy users = use Internet more than 29 hrs/week. NQ = network quality; CS = customer service; IW = information support; SP = security and privacy. 5.7 Exploratory analysis and reliability tests Following the steps defined in Chapter 4, factor analysis, validity, and reliability tests were performed prior to the structural modeling to ensure meaningful interpretation of the data (Gay et al., 2006). The measurement items in the current study were sourced from a variety of pre-existing scales and were statistically validated via exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) using separate datasets. This section discusses the results of EFA. Prior to the EFA, Fields (2005) recommends that the intercorrelations between the variables should have an appropriate level of correlation. They should neither be too high nor too low. In addition, the assumptions of normality should be verified. Bartlett s Test of Sphericity and the Kaiser-Meyer-Olkin (KMO) value were examined. The data is only considered suitable for factor analysis if Bartlett s Test of Sphericity statistics is 159

180 significant at p < 0.05 and the Kaiser-Meyer-Olkin (KMO) value is not less than 0.6 (Pallant, 2007). These measures used together are the minimum standards to proceed with exploratory factor analysis. Table 5.21 explains the statistical conditions used to determine the suitability of the data when performing factor analysis. 36Table 5.21: Statistical conditions Statistic Interpretation Critical value Correlation Level of association between variables. Value varies from 0 to 1 and indicates the proportion of common variance Not too small, but < 0.9 Determinant Checks for multicollinearity Must be > 0 Kaiser Meyer Measure of sampling adequacy. Values Olkin vary between 0 and 1 Bartlett s test of sphericity Note: Field (2005). Tests whether the variables are correlated highly enough to undertake factor analysis or whether items are only correlated to themselves, and not with any other items in the matrix 0.5 to 0.7 is mediocre 0.7 to 0.8 good 0.8 to 0.9 great > 0.9 superb p < 0.05 and null hypothesis should be rejected Several constructs in this study are latent variables which were not directly observable. Latent constructs are theoretical or hypothetical constructs measured by multiple observed indicators; they are also known as manifest variables (Reisinger and Mavondo, 2007). EFA is a statistical method which can be utilised to identify interpretable factors and ensure that the indicators of the factors represent the latent construct (Hair et al., 2006). Subsequently, all of the items measuring each construct are subjected to an exploratory factor analysis to ensure dimensionalities of the constructs. Principal axis factoring is a common correlation factor extraction method which seeks to reproduce the intercorrelations among the variables without making assumptions about the underlying distribution (Garson, 2009). This method tends to provide similar results to maximum likelihood estimation (Cunningham, 2008) and principal component analysis (PCA) (Fields, 2005). Varimax rotation can also be used to limit the number of variables with high loadings on every factor (Tabachinick and Fidell, 2001). Thus, it was an appropriate choice for this study. In addition, communality refers to the variance of observed variables accounted for by a common factor, and varies between 0 (i.e. no shared variance) and 1 (i.e. no unique variance) (Fields, 2005). Large 160

181 communality indicates the strong influence of an underlying factor (Fields, 2005). Moreover, only items with factor loadings above 0.4 should be retained (Kline, 1994). If factors share items with high cross loadings, the items are considered complex as they reveal the impact of more than one factor, and could be considered to be discharged (Worthington and Whittaker, 2006). In order to statistically determine the number of factors that should be retained, Kaiser s criterion is the most widely used measure. Kaiser (1960) recommends that all factors with Eigenvalues above 1 should be retained. As such, only one factor with an Eigenvalue of more than 1 is extracted if the measurement items measure a single underlying concept (Manning and Munro, 2007). In contrast, two or more factors with Eigenvalues greater than 1 exist if the measurement items are not homogenous and indicate two or more concepts (Manning and Munro, 2007). The Scree Plot should be considered in which it is recommended to keep all factors above the elbow, or break, because of their significant contribution in explaining variance in the whole data set (Pallant, 2007). An EFA was performed on all items in this study. KMO coefficient was and the result of Bartlett s Test of Sphericity ( 2 (780) = , p < 0.001) was satisfactory. While there was no cross loadings, all of factor loadings ranging from to well exceeded the 0.4 cut off (Kline, 1994). Therefore, the factor analysis produced a ten-factor solution with the items loading on appropriate factors consistent with the theoretically hypothesised relationships. These factors altogether explained 81.08% of the shared variance. The results depicted in Table 5.22 reveal that a single factor solution was obtained for each construct. In other words, each of the ten scales was unidimensional. 161

182 37Table 5.22: Exploratory Factor Analyses (EFA) results Construct Item Factor 1 Factor 2 Network Quality I do not experience any Internet disconnection from this ISP.513 The Internet downloading and uploading speed meet my expectations.636 The Internet speed does not reduce regardless peak or off-peak hours.567 Customer service Customer service personnel are knowledgeable.743 Customer service personnel are willing to respond to my enquiries.739 My technical problems are solved promptly.703 Information quality and website information support Security and Privacy This ISP provides accurate information.537 This ISP is informative.581 The ISP provides updated information.656 This ISP provides sufficient information.666 This ISP provides up-to-date information.590 The information on the ISP s website is relevant to me.693 I can find what I need on the ISP s website.660 This ISP provides relevant information.676 I feel that my personal information is protected at this ISP.693 I feel that my financial information is protected at this ISP.745 I feel that the transactions with this ISP are secured.657 Value This Internet package is worth my money.749 I would consider this Internet package to be a good buy.785 I feel that I purchase a good Internet package with a reasonable price.737 Satisfaction I am happy about my decision to choose this ISP.601 I believe that I did the right thing when I chose this ISP.638 Overall, I am satisfied with this ISP.619 Trust I trust this ISP.626 I feel that I can rely on this ISP service.620 I trust the ISP s billing system.712 This ISP is reliable because it is mainly concerned with the customer s interests.564 I feel that this ISP will not deceive me in any way.694 Commitment I feel involved with this ISP.705 I am very proud to have this company as my service provider.688 I feel attached to this ISP.604 I will not buy this ISP s services in the future.708 Attitudinal loyalty Behavioural loyalty If I bought/upgraded a new Internet package, I would prefer this ISP.664 I try to use this ISP because it is the best choice for me.706 I consider myself to be a loyal patron of this ISP.710 I would say positive things about this ISP to other people.775 I would recommend this ISP to someone who seeks my advice.760 I would consider this ISP as my first choice to buy services.687 I would do more business with this ISP in the next few years.668 I would do less business with this ISP in the next few years.624 Explained variance (%) Note: Total variance: 81.08%; Kaiser-Meyer-Olkin Measure of Sampling Adequacy (KMO) =.976; Bartlett's Test of Sphericity ( 2 (780) = , p < 0.000). 162 Factor 3 Factor 4 Factor 5 Factor 6 Factor 7 Factor 8 Factor 9 Factor 10

183 The next step was to assess the reliability of the scales by examining the Cronbach s alpha coefficients. As such, the reliability of the survey instrument can also be determined. Cronbach s alpha is used to evaluate the internal consistency reliability (Canana et al., 2001). Internal consistency reliability concerns the homogeneity of the measurement items in a scale (DeVellis, 2003). Scales based on classical measurement models are intended to measure a single concept, which is usually associated with Cronbach s coefficient alpha (α) (DeVellis, 2003). Cronbach s alpha values above 0.7 (Cronbach, 1951) are considered acceptable and values above 0.8 indicate good consistency. The reliability test was performed for all the factors generated by the EFA. The overall Cronbach s Alpha ranged from the lowest to the highest 0.964, which indicates good internal consistency of the scales. The results are shown in Table Table 5.23: The reliability test Construct Number of statements Cronbach s alpha Network quality Customer service and technical support Information quality and website information support Security and privacy Customer value Customer satisfaction Customer trust Customer commitment Attitudinal loyalty Behavioural loyalty Total Confirmatory Factor Analysis and measurement models Confirmatory Factor Analysis (CFA) is a technique that requires a priori specification of items to their corresponding latent variables or factors (Cunningham, 2010), and the models generated are usually referred to as measurement models. CFA is used to test whether theoretical relationships between items and their hypothesised factors are supported by the data (Cunningham, 2010; Hair et al., 2006). Ideally, a one-factor model should include three or four statements (Bollen, 1989). As a rule of thumb, three statements are the minimum number of indicators that are required for a measurement model (Kline, 2010). 163

184 CFA is not concerned with deciding the number of underlying factors. Instead, it verifies the factor structure of a set of observed variables, provided that the models in CFA are specified a priori (Cunningham, 2010). CFA allows the researcher to test the relationship between observed variables and their underlying latent constructs (Cunningham, 2010). While EFA is typically a data-driven approach, CFA is theoretically driven (Cunningham, 2010) Maximum Likelihood Maximum Likelihood (ML) is the most commonly applied method of parameter estimation in SEM (Bollen, 1989). ML possesses a number of optimal characteristics forestimation, namely sufficiency (complete information about the parameter of interest comprised in the ML estimator); consistency (true parameter value that generated the data recovered asymptotically, i.e. for data of adequately large samples); efficiency (lowest-possible variance of parameter estimates obtained asymptotically); and parameterisation invariance (same ML solution obtained independently of the parameterisation employed) (Myung, 2003). Provided that the data is drawn from a continuous and multivariate normal distribution, ML performs well with sample sizes over 500 (Tabachnick and Fidell, 2007). In this study, the survey included more than 2000 usable cases, thus ML estimation was suitable for providing estimates for the models parameters as well as for standard errors (Bollen, 1989) Model fit indices Table 5.24 summarises all model fit indicators adopted in this study. 39Table 5.24: Model fit indicators adopted in this study Indicator Abbreviation Preferred level Value of normed chi-square CMIN/df Below 5.0 Chi-square P-value χ2 P-value >0.05 >0.05 Goodness-of-Fit Index, Adjusted Goodness-of-Fit Index GFI AGFI >0.95 >0.95 Tucker Lewis Index TLI >0.95, sometimes > 1.00 Comparative Fit Index CFI >0.95, sometimes only >0.90 Root-Mean-Square Error of RMSEA <0.05, sometimes <0.08 Approximation Standardised Root-Mean-Square Residual SRMR <

185 Chi-square value The model evaluation was assessed by chi-square (χ2) statistics and its associated significance test. If the p-value associated with chi-square is more than 0.05, indicating no significant difference between the sample variance/covariance matrix and the modelimplied variance/covariance matrix, it can be assumed the model is a good fit to the data (Cunningham, 2010). However, chi-square statistics are sensitive to the sample size. To be specific, with a larger sample size the p-value associated with the χ2 is more likely to result in a significant difference between the data and the model (Kline, 2005). Therefore, this study also takes into account other fit indices. One of them is normed chi-square value (χ2/df or CMIN/DF). χ2/df values that are equal to or less than five are indicators of reasonable fit (Kline, 2005). In addition, the Bollen-Stine bootstrap is a post-hoc adjustment which accounts for non-normality and can be used to test the model fit. The bootstrapping technique generates a new chi-square value which is compared against the original chi-square value and an adjusted p-value is simultaneously computed (Cunningham, 2010). If the Bollen-Stine p value is smaller than 0.05, the null hypothesis (i.e. the model is correct) is rejected (Cunningham, 2010). Two thousand bootstrapping samples were generated in this study, producing the chi-square distribution that corrects for departure from multivariate normality Goodness-of-Fit Index and Adjusted Goodness-of-Fit Index Goodness-of-Fit Index (GFI) is an absolute fit index that estimates the proportion of covariance in the sample data matrix explained by the model (Kline, 2005). GFI values range from 0 to 1 with values of 1 indicating perfect model fit (Kline, 2005). Although related to GFI, Adjusted Goodness-of-Fit Index (AGFI) differs from GFI as it adjusts GFI based on degrees of freedom in the specified model (Byrne, 2001). Models with a GFI and AGFI of about 0.95 or over are considered to be a good fit to the data (Cunningham, 2008) Tucker Lewis Index (TLI) The Tucker Lewis Index (TLI), also known as the Non Normed Fit Index (NNFI), estimates the improvement per degree of freedom of the target model over an independent model (Hu and Bentler, 1998). The values of TLI usually range from 0 to 1, with values exceeding 0.95 indicating a good fitting model (Hu and Bentler, 1998). 165

186 However, TLI may exceed a value of 1, especially for over-fitting models (Cunningham, 2010) Comparative Fit Index Comparative Fit Index (CFI) is derived from chi-square statistic and measures how much better the target model is when compared with an independent model (i.e. a model in which all variables are not correlated and only error variances are estimated) (Cunningham, 2010). CFI values range from 0 to 1 with values over 0.90 indicating a good model fit (Hu and Bentler, 1998). A value of 0.95 is preferable (Kline, 2005) and was used as the preferred level in this study. Being relatively independent of sample size, CFI is perhaps the most frequently reported indicator in the literature (Hair et al., 1998; Kline, 2005) Standardised Root-Mean-Square Residual The Standardised Root-Mean-Square Residual (SRMR) measures the average difference between corresponding elements of the sample and model-implied correlation matrices (Cunningham, 2008). SRMR values below 0.05 suggest that the model may be a good fit to the data. Moreover, large values of the SRMR may indicate outliers in the data (Cunningham, 2008) Root-Mean-Square Error of Approximation The Root-Mean-Square Error of Approximation (RMSEA) is a measure of the discrepancy per degree of freedom (Cudeck and Browne, 1992). RMSEA values of 0.05 or below indicate a model of close fit, while values between 0.05 and 0.08 suggest reasonable fit (Cudeck and Browne, 1992). A 90 per cent confidence interval (90% CI) produced for RMSEA in AMOS should range from 0.03 to 0.08 (Hair et al., 2006). RMSEA is strongly recommended by many researchers because the availability of the confidence interval provides significant information about the accuracy of the estimate of fit (Kline, 2005) Congeneric measurement models Joreskog (1993) proposes that an initial step in the analysis of SEMs is to test a series of congeneric models for each of the latent constructs. A one-factor model is the simplest 166

187 form of measurement models demonstrating the regression of a set of observed variables on a single latent variable (Cunningham, 2010). As the latent construct in a congeneric model is unidimensional, the correlated error terms do not exist. The establishment of congeneric models allows the measurement problems to be dealt with prior to the formation of a full SEM, and supports the unidimensionality of the latent variable (Cunningham, 2010). Ten congeneric models corresponding to ten latent constructs were tested and are discussed in this section Network quality CFA was performed with the three statements that measured the network quality construct, as shown in Figure 5.5. The chi square statistics were insignificant. Other descriptive fit statistics, such as AGFI = 0.996, TLI = 0.999, CFI = 1.000, also indicated that the model fitted the data well. Chi-square = 2.061, df = 1, p-value = 0.151, CMIN/DF = 2.061, GFI = 0.999, AGFI = 0.996, TLI = 0.999, CFI = 1.000, RMSEA = 0.023, 90%CI = (0.000, 0.068), SRMR = Figure 5.5: CFA for network quality measures The regression weights for the statements in the network quality measurement model are shown in Table Statement 2 was weighted as unity in order to obtain a solution. The weights for the other two statements were significant at p < levels. 40Table 5.25: Regression weights for network quality measures Estimate S.E. C.R. P Label Statement1 <--- NETWORK_QUALITY *** Statement2 <--- NETWORK_QUALITY Statement3 <--- NETWORK_QUALITY *** Note: *** p <

188 Customer service and technical support The three statements of customer service and technical support construct formed a one factor CFA measurement model, as shown in Figure 5.6. CFA results for the customer service factor show the chi square value was with p-value = The GFI (1.000), AGFI (0.999) and CFI (1.000) reveal that this was a well-fitting model. Chi-square = 0.281, df = 1, p-value = 0.596, CMIN/DF = 0.281, GFI = 1.000, AGFI = 0.999, TLI = 1.001, CFI = 1.000, RMSEA = 0.000, 90%CI = (0.000, 0.047), SRMR = Figure 5.6: CFA for customer service and technical support measures The regression weights for the statements in this measurement model are shown in Table Statement 4 was weighted as unity in order to obtain a solution. The weights for the other two statements were significant at p < level. 41Table 5.26: Regression weights for customer service and technical support measurements Statement4 <--- CUSTOMER_SERVICE Estimate S.E. C.R. P Label Statement5 <--- CUSTOMER_SERVICE *** Statement6 <--- CUSTOMER_SERVICE *** Note: *** p <

189 Information quality and website information support The eight statements of information quality and website information support construct formed a one factor CFA measurement model as shown in Figure 5.7. The chi-square statistic was significant (χ2 (5) = , p-value = 0.000). In addition, GFI, AGFI, TLI, and CFI were below 0.95 with RMSEA > 0.08 and SRMR > This indicates that the model did not fit the data well. The standardised residuals covariance matrix, sample correlations, and modification indices suggested the deletion of Statement 7 The ISP provides accurate information, Statement 9 The ISP provides updated information and Statement 13 I can find what I need on the ISP s website in order to improve the model fit. Chi-square = , df = 20, p-value = 0.000, CMIN/DF = , GFI = 0.861, AGFI = 0.750, TLI = 0.866, CFI = 0.904, RMSEA = 0.172, 90%CI = (0.164, 0.180), SRMR = Figure 5.7: CFA for information quality measures 169

190 After deleting Statements 7, 9 and 13, CFA was re-run and the resultant model was a better fit to the data. The RMSEA value was 0, and the SRMR value was less than The GFI and AGFI indicated that the model represented a good approximation of the data and the CFI suggested a well-fitting model. The results are shown in Figure 5.8. Therefore, it can be concluded that the model is a good fit to the data. Chi-square = 1.817, df = 2, p-value = 0.403, CMIN/DF = 0.909, GFI = 1.000, AGFI= 0.998, TLI = 1.000, CFI=1.000, RMSEA= 0.000, 90%CI = (0.000, 0.042), SRMR = Figure 5.8: A re-specified one factor model for the information quality construct The regression weights for the statements in this measurement model are shown in Table The results indicate that all the factor coefficients were significant at p < levels. 42Table 5.27: Regression weights information quality measurements Estimate S.E. C.R. P Label Statement8 <--- INFORMATION *** Statement10 <--- INFORMATION *** Statement11 <--- INFORMATION *** Statement14 <--- INFORMATION *** Note: *** p <

191 Security and privacy The three statements that measured security and privacy formed a latent construct, as shown in Figure 5.9. The results of CFA show that the chi square statistic was significant, χ2 (1) = , p-value = However, it is well known that chi-square statistics are sensitive to sample size and are more likely to result in a significant difference between the model and the data in larger samples (Cunningham 2010). Therefore, other fit indices were taken into consideration. GFI, AGFI, and CFI were higher than 0.95 with RMSEA < 0.08, 90%CI = (0.430, 0.116) and SRMR < Therefore, it can be determined that the model was a good fit to the data. Table 5.28 shows the regression weights for items in the security measurement model. Chi-square = , df = 1, p-value = 0.000, CMIN/DF = , GFI = 0.996, AGFI = 0.974, TLI = 0.986, CFI = 0.995, RMSEA = 0.076, 90%CI = (0.430, 0.116), SRMR = Figure 5.9: CFA for security and privacy measures According to Table 5.28, statement 17 was weighted as unity in order to obtain a solution. The weights for the other two statements were significant at p < level. 43Table 5.28: Regression weights for security and privacy measurements Estimate S.E. C.R. P Label Statement15 <--- SECURITY *** Statement16 <--- SECURITY *** Statement17 <--- SECURITY Note: *** p <

192 Customer trust The five statements of customer trust formed a one factor CFA measurement model for this construct, as shown in Figure The results show that the chi-square statistic was significant (χ2 (5) = , p-value = 0.000). The other indicators were below the recommended threshold, revealing that the model was a poor fit to the data. The standardised residuals covariance matrix, sample correlations, and modification indices suggested the exclusion of Statement 26 I trust the ISP s billing system and Statement 27 I believe that I can trust this ISP will not try to cheat me in order to improve the model fit, and the final best-fit congeneric measurement model is shown in Figure Chi-square = , df = 5, p-value = 0.000, CMIN/DF = , GFI = 0.813, AGFI = 0.438, TLI = 0.756, CFI = 0.878, RMSEA = 0.316, 90%CI = (0.300, 0.332), SRMR = Figure 5.10: CFA for customer trust measures The removal of Statement 26 and 27 improved the model fit. The chi square value was insignificant. The values of other fit indices were reasonable. All the descriptive fit statistics indicated that model was a good fit to the data. The regression weights for the statements in this measurement model are shown in Table

193 Chi-square = 1.115, df = 1, p-value = 0.291, CMIN/DF = 1.115, GFI = 1.000, AGFI = 0.998, TLI = 1.000, CFI = 1.000, RMSEA = 0.007, 90%CI = (0.000, 0.060), SRMR = Figure 5.11: A re-specified one factor model for customer trust construct The results in the following table show that Statement 25 was weighted as unity in order to obtain a solution. The weights for the other two statements were significant at p <0.001 levels. 44Table 5.29: Regression weights for customer trust measurements Estimate S.E. C.R. P Label Statement24 <--- TRUST *** Statement25 <--- TRUST Statement28 <--- TRUST *** Note: *** p <

194 Customer satisfaction Three statements formed a one-factor model for the customer satisfaction construct, as shown in Figure Chi square statistics (χ2 (1) = 1.418, p-value = 0.234) were not significant, indicating that the model fits the data well. GFI, AGFI, TLI, and CFI exceeded 0.95, with RMSEA = 0.014, 90%CI = (0.000, 0.062), SRMR = The fit indices confirmed that this was a well-fitting model. Chi-square = 1.418, df = 1, p-value = 0.234, CMIN/DF = 1.418, GFI = 1.000, AGFI = 0.997, TLI = 1.000, CFI = 1.000, RMSEA = 0.014, 90%CI = (0.000, 0.062), SRMR = Figure 5.12: CFA for customer satisfaction measures The regression weights for the statements in the customer satisfaction measurement model are shown in Table Statement 22 was weighted as unity in order to obtain a solution. The weights for the other two statements were significant at p < levels. 45Table 5.30: Regression weights for customer satisfaction measurements Estimate S.E. C.R. P Label Statement21 <--- SATISFACTION *** Statement22 <--- SATISFACTION Statement23 <--- SATISFACTION *** Note: *** p <

195 Customer commitment The four statements of the customer commitment construct formed a one factor measurement model, as shown in Figure Although the chi-square statistic was significant (χ2 (2) = , p-value = 0.000) owing to the large sample size, the GFI, AGFI, TLI, and CFI were higher than 0.95, SRMR was less than 0.05, and RMSEA was acceptable at This suggests that the model was a reasonable fit to the data. Chi-square = , df = 2, p-value = 0.000, CMIN/DF = , GFI = 0.994, AGFI = 0.968, TLI = 0.978, CFI = 0.993, RMSEA = 0.079, 90%CI = (0.054, 0.106), SRMR = Figure 5.13: CFA for customer commitment measures The regression weights for the statements in this measurement model are shown in Table This table illustrates that all the factor coefficients were significant at p <0.001 levels. 46Table 5.31: Regression weights for customer commitment measurements Estimate S.E. C.R. P Label Statement29 <--- COMMITMENT *** Statement30 <--- COMMITMENT *** Statement31 <--- COMMITMENT *** Statement32 <--- COMMITMENT *** Note: *** p <

196 Customer value The three statements of customer value formed a one factor CFA measurement model for this construct, as shown in Figure Although the chi-square statistics (χ2 (1) = , p-value = 0.000) was significant, the other indicators (i.e. GFI = 0.994, AGFI = 0.961, TLI = 0.992, CFI = 0.997) showed that this model was a reasonable fit to the data. Chi-square = , df = 1, p-value = 0.000, CMIN/DF = , GFI = 0.994, AGFI = 0.961, TLI = 0.992, CFI = 0.997, RMSEA = 0.000, 90%CI = (0.062, 0.135), SRMR = Figure 5.14: CFA for customer value measures The regression weights for the statements in this measurement model are shown in Table Statement 20 was weighted as unity as its regression weight in order to obtain a solution. The regression weights of the other two statements were significant at p < level. 47Table 5.32: Regression weights for customer value measurements Estimate S.E. C.R. P Label Statement18 <--- VALUE *** Statement19 <--- VALUE Statement20 <--- VALUE *** Note: *** p <

197 Attitudinal loyalty The five items measuring attitudinal loyalty formed a latent construct, as shown in Figure The results indicate that the model was a poor fit to the data. Hence, Statement 33 If I bought/upgraded a new Internet package, I would prefer this ISP and Statement 34 I try to use this ISP because it is the best choice for me, which had high standardised residual covariances, were removed in order to improve the model fit. Chi-square = , df = 5, p-value = 0.000, CMIN/DF = , GFI = 0.947, AGFI = 0.842, TLI = 0.938, CFI = 0.969, RMSEA = 0.155, 90%CI = (0.139, 0.172), SRMR = Figure 5.15: CFA for attitudinal loyalty measures Figure 5.16 shows the measurement model of attitudinal loyalty to be acceptable after the removal of Statement 33 and Statement 34. The results show that the chi-square statistics were not significant at p-value = In addition, each of GFI, AGFI, TLI and CFI had a value of RMSEA value was 0.05 and SRMR value was All the fit indices suggested a well-fitting model. The regression weights for the statements in this measurement model are shown in Table

198 Chi-square = 0.001, df = 1, p-value = 0.974, CMIN/DF = 0.001, GFI = 1.000, AGFI = 1.000, TLI = 1.000, CFI = 1.000, RMSEA = 0.000, 90%CI = (0.000, 0.000), SRMR = Figure 5.16: A re-specified one factor model for attitudinal loyalty construct The results in Table 5.33 indicate that Statement 37 was weighted as unity in order to obtain a solution. Both of the regression weights of the other two statements were significant at p <0.001 levels. 48Table 5.33: Regression weights for attitudinal loyalty measurements Estimate S.E. C.R. P Label Statement35 <--- ATTITUDINAL *** Statement36 <--- ATTITUDINAL *** Statement37 <--- ATTITUDINAL Note: *** p <

199 Behavioural loyalty The three statements that measured the behavioural loyalty construct formed a one factor measurement model, as shown in Figure 5.17 Results of CFA show that the chisquare value was with associated p-value = The other indicators were GFI = 0.999, AGFI = 0.994, CFI = All of them revealed that the model was a good fit to the data. Chi-square = 3.008, df = 1, p-value = 0.083, CMIN/DF = 3.008, GFI = 0.999, AGFI = 0.994, TLI = 0.996, CFI = 0.999, RMSEA = 0.031, 90%CI = (0.000, 0.074), SRMR = Figure 5.17: CFA for behavioural loyalty measures Statement 39 was weighted as unity in order to obtain a solution. The factor coefficients for the other two statements were significant at p < level. The regression weights for the statements in this measurement model are shown in Table Table 5.34: Regression weights for behavioural loyalty measurements Estimate S.E. C.R. P Label Statement38 <--- BEHAVIOURAL *** Statement39 <--- BEHAVIOURAL Statement40 <--- BEHAVIOURAL *** Note: *** p <

200 Summary of the results of CFA A total of eight statements were deleted in order to improve the fit of all the measurement models. Thirty two statements comprising 10 congeneric measurement models were retained. Table 5.35 summarises the reliability scores of the 10 congeneric measurement models. The scores, which range from a high of to a low of 0.732, indicate reasonable reliability. 50Table 5.35: Summary of the reliability of the measurement models Measurement Model Number of statements retained Number of statements deleted Reliability scores Network quality Customer service and technical support Information quality and website information support Security and privacy Customer trust Customer satisfaction Customer commitment Customer value Attitudinal loyalty Behavioural loyalty Total Structural Equation Model (SEM) The SEM technique is applied to test and estimate causal relations between latent variables with a view to seeking support for the theoretical model (Cunningham, 2010). Prior to testing the hypotheses, convergent validity and discriminant validity are examined in the following section Convergent validity Convergent validity refers to the extent to which numerous attempts measuring the same model with different methods are in agreement (Hair et al., 2006). It refers to the level in which items significantly load on one single factor. In CFA, researchers can determine convergent validity within the context of the measurement model (Steenkamp and Van Trijp, 1991). In this study, convergent validity of the constructs was evaluated using CFA. In general, convergent validity is confirmed when all statements have significant factor loadings above 0.30, and they only load on the priori theoretical 180

201 construct. First, the significance and strength of the factor loadings were assessed for each measurement model. It was found that statements in all measurement models had significant regression coefficients exceeding Second, modification indices where the information about how to improve the model fit, could be found were inspected to detect probable cross-loading items (Straub et al., 2004). No cross-loading items were detected Discriminant validity Discriminant validity refers to the extent to which constructs in a model are different from one another (Cunningham, 2010; Hair et al., 2006). A set of variables intended to measure different constructs display discriminant validity when they are not highly inter-correlated (Kline, 2010). The use of pattern and structure coefficients is recommended as a test for discriminant validity and was used in this study to decide whether constructs in measurement models are empirically distinguishable (Thompson, 1997). Discriminant validity was evaluated in three sections of the proposed conceptual model: section one (Figure 5.18) covered the service quality dimensions; section two addressed the cognitive and affective evaluations or customers (Figure 5.19) and section three addresses the outcome of resultant constructs (Figure 5.20) Section 1: The service quality dimensions The results of the service quality dimensions are shown in Table It is a second order model with service quality as the focal construct and four reflective indicators identified as network quality, customer service, information quality and website information support, and security and privacy. The factor pattern and structure coefficients reveal that the four constructs (network quality, customer service, information quality, together with security and privacy) reveal discriminant validity. From Table 5.36, it is apparent that the statements that measured an individual construct had considerably higher factor loadings than the other statements that were not intended for that construct. 181

202 31Figure 5.18: The service quality dimensions of the conceptual model 182

203 51Table 5.36: Factor pattern and structure coefficients for the service quality dimensions Construct Statement Information Customer Network Security Quality Service Quality P S P S P S P S Statement a a a.525 Statement a a a.441 Statement a a a.438 Statement14 0 a a a.553 Statement11 0 a a a.559 Statement10 0 a a a.578 Statement8 0 a a a.558 Statement6 0 a a a.486 Statement5 0 a a a.541 Statement4 0 a a a.541 Statement3 0 a a a Statement2 0 a a a Statement1 0 a a a Note: P = Pattern coefficients. S = Structure coefficients. N = All pattern coefficients are significant at p <.001; a. Parameters fixed at reported level in specifying the model Section 2: Customer cognitive and affective evaluation constructs The results of the discriminant validity are shown in Table Customer cognitive and affective evaluations comprise of four constructs: value, commitment, satisfaction and trust. An examination of factor patterns and structure coefficients shows that the four constructs display discriminant validity. 183

204 32Figure 5.19: Customer cognitive and affective evaluation constructs 184

205 52Table 5.37: Factor pattern and structure coefficients for the customer cognitive and affective evaluations constructs Construct Statement Value Commitment Satisfaction Trust P S P S P S P S Statement a a a.697 Statement a a a.704 Statement a a a.690 Statement32 0 a a a.360 Statement31 0 a a a.634 Statement30 0 a a a.748 Statement29 0 a a a.729 Statement23 0 a a a.868 Statement22 0 a a a.891 Statement21 0 a a a.875 Statement28 0 a a a Statement25 0 a a a Statement24 0 a a a Note: P = Pattern coefficients. S = Structure coefficients. N = All pattern coefficients are significant at p <.001; a. Parameters fixed at reported level in specifying the model Section 3: The outcome or resultant constructs The results of the discriminant validity test are shown in Table There are only two constructs in this section: attitudinal loyalty and behavioural loyalty. Figure 5.37 shows there was a high correlation (0.96) between the two constructs. An inspection of the factor pattern and structure coefficients revealed that there were issues with Statement 40 (Behavioural Behaviour) in terms of discriminant validity, as the structural coefficient of this statement (0.497) was relatively low. This was possibly caused by the statement being reverse-scored, which may have confused respondents. Although this statement was taken from a validated scale, the context in which it was previously used was different. 185

206 33Figure 5.20: Resultant constructs of the conceptual model 53Table 5.38: Factor pattern and structure coefficients for the resultant constructs Construct Behavioural Attitudinal Statement P S P S Statement a.381 Statement a.760 Statement a.853 Statement37 0 a Statement36 0 a Statement35 0 a Note: P = Pattern coefficients. S = Structure coefficients. N = All pattern coefficients are significant at p <.001; a. Parameters fixed at reported level in specifying the model. 186

207 5.9.3 The output of the structural model A structural model is used to determine the direct and indirect effects among latent variables in a model (Bollen, 1989). The structural model was tested and the output of the final stage of analysis is shown in Figure Post-hoc modifications were run with a view to improving the model fit, as well as to producing a more parsimonious model. Chi-square = ; p-value = 0.000, CMIN/DF = 9.775, GFI = 0.872; AGFI = 0.849, TLI = 0.930, CFI = 0.937, RMSEA = 0.065, 90%CI = (0.064, 0.067), SRMR = Figure 5.21: Structural model of customer loyalty 187

208 The hypothesised structural model was tested based on validated measures obtained from CFA and guided by the proposed conceptual framework established in Chapter Three (Figure 5.21). While network quality, customer service, website support, and security are dimensions of overall service quality, overall service quality is an antecedent of customer trust, satisfaction, customer commitment and customer value. Subsequently, customer trust, satisfaction, customer commitment and customer value are directly related to behavioural intentions and attitudinal loyalty. These paths exist with multiple regression relationships corresponding to the hypotheses developed in the proposed conceptual framework (details are shown in Appendix C). The SEM was performed and the results are presented in Figure Chi-square statistics (χ2 = ) were significant at p-value = The other indicators were GFI = 0.872, AGFI = 0.849, TLI = 0.930, CFI = 0.937, RMSEA = 0.065, 90%CI = (0.064, 0.067), SRMR = , all of which revealed that the model was not a good fit to the data. In order to improve the model fit, one statement from information quality and website information support construct (Statement 8) and one from commitment construct (Statement 32) were deleted due to their high standardised residual covariances. The SEM was re-run and the results indicate that the chi-square statistics (χ2 = ) were significant at p-value = 0.000, owing to a large sample size. The other fit indices include GFI = 884; AGFI = 0.862, TLI = 0.941, CFI = 0.947, RMSEA = 0.063, 90%CI = (0.061, 0.065), SRMR = All the above fit indicators suggested a better fitting model; this model was retained as the final best model fit. The results are shown in Figure Latent variables are represented by eclipses, while rectangles represent observable variables. The factor loadings of all statements in the final model ranged from to (Appendix D). 188

209 Chi-square = ; p-value = 0.000, CMIN/DF = 9.118, GFI = 884; AGFI = 0.862, TLI = 0.941, CFI = 0.947, RMSEA = 0.063, 90%CI = (0.061, 0.065), SRMR = Figure 5.22: Re-specified final best fit model of customer loyalty In this study, all latent variables of the measurement models were validated and the respecified models fitted well. The output of the squared multiple correlations (Appendix E) show that the final model explains 91.5 per cent of the variance in attitudinal loyalty and 95.6 per cent of the variance in behavioural loyalty. 189

210 5.9.4 Descriptive statistics and intercorrelations Table 5.39 reports descriptive statistics and correlations among the study variables (five-point scales were used for all statements). 54Table 5.39: Descriptive statistics and correlations among the study variables No. Constructs Mean SD Network quality Customer service Information quality Security Value Satisfaction Trust Commitment Attitudinal Behavioural Note: The diagonal elements are the AVEs (italicised and bolded). The lower-left triangle elements are correlations among the composite measures (unweighted mean of the items for each construct). The upper-right triangle elements are the squared correlations among constructs. All correlations are significant at the 0.01 level (2-tailed). 190

211 5.9.5 Assessing the reliability of the final model Table 5.40 illustrates how the number of statements from the original survey decreased from 40 to 28 in the process of achieving the final best-fitting model. The results indicate that the Cronbach s alphas ranged from the lowest to the highest Ideally, the Cronbach s alpha should be above 0.7. Therefore, it can be concluded that all scales in this study have good internal consistency. Construct 55Table 5.40: Summary of reliability of the final model scales Original number of statements Number of statement deleted in CFA Number of statement deleted in final model Number of statements retained in final model Cronbach s alpha Network quality Customer service Information quality Security Value Satisfaction Trust Commitment Attitudinal Behavioural Discussion of hypothesis tests The testing of hypotheses is discussed in this section and the results are shown in Table Standard errors (SE) and critical ratio (CR) (magnitude > 2 indicates statistical significance at the.05 level) are presented. Statistical significance is determined at p values less than The direction and importance of the relationships is regulated by the magnitude of regression coefficients. 191

212 56Table 5.41: Results of hypotheses testing Hypothesis Std. Coefficient Critical Ratio Hypothesis Support p- value NQ SQ Strong *** CS SQ Strong *** IW SQ Strong *** SP SQ Strong *** SQ TRU Strong *** SQ SAT Strong *** SQ COM Strong *** SQ VAL Strong *** SAT AL Moderate *** TRU AL Moderate *** COM AL Strong *** VAL AL Not supported.414 SAT BL Moderate *** TRU BL Moderate *** COM BL Strong *** VAL BL Weak *** Note: CS = Customer Service; NQ = Network Quality; IW = Information Quality; SP = Security and Privacy; SQ = Service quality; SAT = Satisfaction; COM = Commitment; VAL = Value; TRU = Trust; AL = Attitudinal loyalty; BL = Behavioural loyalty; *** p values are statistical significant at levels. As illustrated in Table 5.42, among the four dimensions, the influence on service quality of information and website support was the greatest, followed by privacy and security. Network quality came next and customer service and technical support had the least influence on perceived service quality. Furthermore, Table 5.43 demonstrates that the effect of service quality was strongest on trust, followed by satisfaction. The service quality s influence on commitment ranked third and the impact of service quality on value was lowest. Moreover, commitment was the most significant determinant of customer attitudinal and behavioural loyalty. Also, it was found that satisfaction created the second strongest impact on customer. Trust was ranked third and value had the least significant impact on customer loyalty. The results are presented in Table Table 5.42: Ranking of service quality dimensions based on their influence on service quality Rank Service quality dimensions Service quality Std. Coefficient 1 Information quality Security and privacy Network quality Customer service

213 58Table 5.43: Ranking of cognitive and affective evaluations based on the influence of service quality Rank Cognitive and affective evaluations Service quality Std. Coefficient 1 Trust Satisfaction Commitment Value Table 5.44: Ranking of cognitive and affective evaluations based on their influence on attitudinal and behavioural loyalty Rank Cognitive and affective Attitudinal Loyalty Behavioural Loyalty evaluations Std. Coefficient Std. Coefficient 1 Commitment Satisfaction Trust Value Network quality H 1 : Network quality is positively related with ISPs overall service quality The results indicate that network quality was strongly related to overall service quality. This proves that customers pay considerable attention to Internet downloading and uploading speed, as well as to the stability of the Internet connection. This finding supports previous research conducted by Cheng et al. (2008), Lai et al. (2009) and Woo and Fock (1999). They find that network quality is one of the most important drivers of overall service quality in the Chinese (Lai et al., 2009) and Hong Kong (Cheng et al., 2008) mobile services markets. The standardised direct effect of overall service quality on network quality was This means that for every one standardised deviation increase in overall service quality, an increase of standardised deviation in network quality would be expected (Cunningham, 2010). Therefore, hypothesis H 1 is accepted. 193

214 Customer service and technical support H 2 : Customer service and technical support are positively related with ISPs overall service quality The results indicate that customer service and technical support was strongly associated with overall service quality. Home Internet services are characterised by technical terms unfamiliar to many users and complex associated problems with hardware and software making customer service necessary. The findings demonstrate that the overall service quality partly depended on how customer service personnel handled and responded to customer enquiries and whether the customer service team and company showed a sincere interest in resolving customer enquiries. Previous research by Abdolvand et al. (2006) report similar findings in mobile service providers. Aydin and Özer (2005) also note that customer complaints handling is important in determining service quality in telecommunications industries. The standardised direct effect of overall service quality on customer service and technical support was 0.692, which can be interpreted that for a one standardised deviation increase in overall service quality, an increase of standardised deviation in customer service and technical support would be expected (Cunningham, 2010). Therefore, hypothesis H 2 is accepted Information quality and website information support H 3 : Information quality and website information support are positively related with ISPs overall service quality Information and website information support was a significant determinant of overall service quality. In fact, this construct was predominantly stronger than the other three antecedents of service quality. This can be explained as Internet users are avid information seekers and are used to browsing websites in the hope of finding information or advice they need. It can be concluded that the accuracy, quality and relevancy of the information that companies provide to customers is important to the overall service quality. The standardised direct effect of service quality on information and website information support was 0.825, which can be interpreted that for a one standardised deviation increase in service quality, an increase of standardised deviation in information and website information support would be expected 194

215 (Cunningham, 2010). Therefore, hypothesis H 3 is accepted Security and privacy H 4 : Security and privacy are positively related with ISPs overall service quality The results reveal a significant direct effect of security and privacy on overall service quality. This finding illustrates that customers are unhappy with the overall service quality if their privacy is not protected or if there is a security breach relating to their online activity. In current times, privacy and security are extremely important issues, especially in the online environment. Previous research by Roca et al. (2009) and Wolfinbarge and Gilly (2003) support these findings. When service providers hold credible reputation in relation to their security practice, consumers tend to trust that doing business with them is safe (Roca et al., 2009). Wolfinbarge and Gilly (2003) state that security of payments and privacy of personal information are positively related to service quality in online retailing. The standardised direct effect of overall service quality on security and privacy was In other words, for a one standardised deviation increase in overall service quality, an increase of standardised deviation in security and privacy would be expected (Cunningham, 2010). Therefore, hypothesis H 4 is accepted Overall service quality and customer trust H 5 : ISPs overall service quality is positively related to customer trust In the second stage of the proposed conceptual framework, the relationship between overall service quality comprising the four dimensions network quality, customer service and technical support, information and website support, and privacy and security and customer affective and cognitive evaluations was evaluated. The results demonstrate that overall service quality was strongly associated with customer trust. Customers tend to trust service providers who can provide a high level of overall service quality. This result is not surprising considering the amount of research conducted in this area that supports this statement (Chiou, 2004; Gounaris and Venetis, 2002). The standardised direct effect of service quality on customer trust was 0.988, which means that one standardised deviation increase in service quality is expected to 195

216 result in an increase of standardised deviation in customer trust (Cunningham, 2010). Therefore, hypothesis H 5 is accepted Overall service quality and customer satisfaction H 6 : ISPs overall service quality is positively related to customer satisfaction The results indicate that service quality had a significant influence on customer satisfaction. When service performance meets or exceeds customer expectation, customers tend to be happy and believe they have made the right choice of service provider. The standardised direct effect of service quality on satisfaction was 0.943, which means that one standardised deviation increase in service quality is expected to result in an increase of standardised deviation in satisfaction (Cunningham, 2010). Therefore, hypothesis H 6 is accepted Overall service quality and customer commitment H 7 : ISPs overall service quality is positively related to customer commitment The results shown in Table 5.41 indicate that the service quality construct had a significant influence on customer commitment (p value was significant). Essentially when customers are delighted with the overall service quality, they are willing to make an effort to continue their relationships with a particular ISP. This finding is similar to that of Al-Hawari (2011) and Fullerton (2005). The standardised direct effect of service quality on commitment was This means that one standardised deviation increase in service quality is expected to result in an increase of standardised deviation in commitment (Cunningham, 2010). Therefore, hypothesis H 7 is accepted Overall service quality and customer value H 8 : ISPs overall service quality is positively related to customer value The results indicate that service quality had a significant impact on value. A service package is considered to be of high value when it provides good overall quality. Previous research by Kim and Damhorst (2010) and Lai et al. (2009) report similar findings. The standardised direct effect of service quality on value was 0.778, which 196

217 means that one standardised deviation increase in service quality is expected to result in an increase of standardised deviation in value (Cunningham, 2010). Therefore, hypothesis H 8 is accepted Customer trust and attitudinal loyalty H 9a : Customer trust is positively related to attitudinal loyalty The results show that customer trust was significantly related to attitudinal loyalty (p value was significant). Customers are more likely to form positive feelings towards a reliable service provider. The standardised direct effect of trust on attitudinal loyalty was 0.253, which means that one standardised deviation increase in trust is expected to result in an increase of standardised deviation in attitudinal loyalty (Cunningham, 2010). Therefore, hypothesis H 9a is accepted Customer trust and behavioural loyalty H 9b : Customer trust is positively related to behavioural loyalty The results confirm that customer trust had a significant influence on behavioural loyalty (p value was significant). Customers are more likely to repurchase from service providers they trust. The standardised direct effect of trust on behavioural loyalty was 0.384, which means that one standardised deviation increase in trust is expected to result in an increase of standardised deviation in behavioural loyalty (Cunningham, 2010). Therefore, hypothesis H 9b is accepted Customer satisfaction and attitudinal loyalty H 10a : Customer satisfaction is positively related to attitudinal loyalty Customer satisfaction was seen as a precursor to customer attitudinal loyalty. This suggests that a customer who is satisfied with the service provider will undoubtedly display favourable attitudinal intentions. Previous research by Gerpott et al. (2001) reported similar findings. The standardised direct effect of customer satisfaction on attitudinal loyalty was 0.362, which means that an increase by one standardised deviation in satisfaction is expected to result in an increase of standardised 197

218 deviation in attitudinal loyalty (Cunningham, 2010). Therefore, hypothesis H 10a is accepted Customer satisfaction and behavioural loyalty H 10b : Customer satisfaction is positively related to behavioural loyalty The results show that customer satisfaction had a significant influence on customer behavioural loyalty. This suggests that satisfied customers are more likely to exhibit repurchase behaviour, supporting previous research by Gerpott et al. (2001) and Cheng et al. (2008). The standardised direct effect of customer satisfaction on behavioural loyalty was 0.571, which means that one standardised deviation increase in satisfaction is expected to result in an increase of standardised deviation in behavioural loyalty (Cunningham, 2010). Therefore, hypothesis H 10b is accepted Customer commitment and attitudinal loyalty H 11a : Customer commitment is positively related to attitudinal loyalty The results indicate that customer commitment had a significant impact on customer behavioural loyalty. When customers have a sense of belonging with a service provider, they tend to feel loyal towards that particular service provider. This supports previous work conducted by the likes of Fullerton (2005). The standardised direct effect of customer commitment on behavioural loyalty was 0.877, which means that an increase by one standardised deviation in commitment is expected to result in an increase of standardised deviation in behavioural loyalty (Cunningham, 2010). Therefore, hypothesis H 11a is accepted Customer commitment and behavioural loyalty H 11b : Customer commitment is positively related to behavioural loyalty Customer commitment was an antecedent of customer behavioural loyalty. A committed customer is more likely to repurchase from their incumbent service provider. This provides evidence for previous work by Fullerton (2005) and Cater and Zabkar (2009). The standardised direct effect of customer commitment on behavioural loyalty 198

219 was 0.880, which means that one standardised deviation increase in commitment is expected to result in an increase of standardised deviation in behavioural loyalty (Cunningham, 2010). Therefore, hypothesis H 11b is accepted Customer value and attitudinal loyalty H 12a : Customer value is positively related to attitudinal loyalty The results (Table 5.41) show that customer value did not significantly influence attitudinal loyalty (p value was not significant). It was negatively related to the attitudinal loyalty of Thai telecommunications consumers. This can be explained by noting that the costs of services, as well as the promotional programs, are very similar among the three main Thai ISPs. Therefore, perceived value may not significantly differ for customers of those ISPs. Also, customer value involves monetary benefits which may be an indicator of price sensitivity. A price conscious customer tends to switch between brands and has low brand inertia (Corstjens and Lal, 2000). Therefore, hypothesis H 12a is rejected Customer value and behavioural loyalty H 12b : Customer value is positively related to behavioural loyalty The results illustrate that customer value had a significant influence on customer behavioural loyalty. High perceived value can facilitate repeat purchase. The standardised direct effect of customer value on behavioural loyalty was 0.097, which means that an increase by one standardised deviation in customer value is expected to result in an increase of standardised deviation in behavioural loyalty (Cunningham, 2010). Therefore, hypothesis H 12b is accepted Overall service quality, and attitudinal and behavioural loyalty The direct effect of overall service quality on attitudinal and behavioural loyalty is not hypothesised. However, the results from bias corrected bootstrapping indicate the significant indirect effect of overall service quality on both attitudinal and behavioural loyalty via customer trust, satisfaction, commitment and value. This demonstrates that improving service quality can result in favourable customer attitudes towards the ISPs and repeat purchase, confirming the importance of service quality. The results are 199

220 reported in Table The standardised indirect effect of overall service quality on behavioural loyalty was 0.865, and the standardised indirect effect of overall service quality on attitudinal loyalty was Table 5.45: Standardised indirect effects of overall service quality on loyalty Attitudinal loyalty Behavioural loyalty 95% CI 95% CI 95% CI 95% CI Std. IE Std. IE (lower) (upper) (lower) (upper) SQ.857*** *** Notes: SQ = overall service quality; Std. IE = Standardised Indirect Effect; The 95% CI is obtained by the bias-corrected bootstrap with 2,000 bootstrap samples; CI (lower) = lower bound of a 95% confidence interval; CI (upper) = upper bound; *** p.001. In short, network quality; customer service and technical support; information quality and website information support; and security and privacy are positively related to overall service quality. Overall service quality is an antecedent of customers trust, satisfaction, commitment and value. In return, customer trust, commitment and satisfaction are positively related to attitudinal and behavioural loyalty. While significantly related to behavioural loyalty, the significant direct effect of value on attitudinal loyalty is not confirmed. Table 5.46 summaries the results of all the hypotheses tests. 61Table 5.46: Summary of the hypotheses tests No. Hypothesis Accepted/ Rejected H 1 Network quality is positively related with ISPs overall service quality Accepted H 2 Customer service and technical support are positively related with ISPs Accepted overall service quality H 3 Information quality and website information are positively related with Accepted ISPs overall service quality H 4 Security and privacy are positively related with ISPs overall service Accepted quality H 5 ISPs overall service quality is positively related to customer trust Accepted H 6 ISPs overall service quality is positively related to customer satisfaction Accepted H 7 ISPs overall service quality is positively related to customer commitment Accepted H 8 ISPs overall service quality is positively related to customer value Accepted H 9a Customer trust is positively related to attitudinal loyalty Accepted H 9b Customer trust is positively related to behavioural loyalty Accepted H 10a Customer satisfaction is positively related to attitudinal loyalty Accepted H 10b Customer satisfaction is positively related to behavioural loyalty Accepted H 11a Customer commitment is positively related to attitudinal loyalty Accepted H 11b Customer commitment is positively related to behavioural loyalty Accepted H 12a Customer value is positively related to attitudinal loyalty Rejected H 12b Customer value is positively related to behavioural loyalty Accepted 200

221 5.10 Details of segmentation analysis using invariance testing Several researchers (Ringle et al., 2013) suggest that studying a single homogenous population in path models is inadequate in to understanding path relationships. This is because customers characteristics and the nature of their demands for services differ (Mazzoni et al., 2007; Ringle et al., 2013). Segmentation is a process of subdividing a heterogeneous market into homogeneous groups of customers who have similar characteristics or who respond to marketing activities in the same way (Ko et al., 2012). Nevertheless, there is very little evidence as to how effective segmentation can be operationalised for ISP customers. In Section 5.6, this study introduced customer segmentation based on usage pattern (heavy, medium and light users), as well as demonstrating several basic differences between customers in different age and income groups. In this section, invariance testing in the final best fit model is employed to extend the segmentation analysis and further provide insights of customers with different level of Internet usage, age and income levels Internet usage pattern Segmentation based on usage is one of the most logical basis of segmentation in similar types of services (Mazzoni et al., 2007; Wedel and Kamakura, 2003). Segmenting markets by consumption patterns is a relatively intuitive step toward understanding customers (Weinstein, 2002). By categorising customers into usage groups, service providers can create suitable marketing strategies for each segment. Furthermore, segmentation by usage in building long-term customer value enhances the profitability of an organisation (McDougall, 2001). Weinstein (2002) concludes that usage analysis can assist in customer retention. To examine the interaction effect, the sample was split into three groups based on their usage level (light users: less than 9 hours a week, medium users: 9-29 hours a week and heavy users more than 29 hours a week) and paths were determined at different levels of the moderating variable. As indicated in Table 5.47, the main research models which include Internet usage show reasonable fit to the data. To determine the differences in structural path for the three groups of users, the structural models were separated for the three subsamples. The moderating effect was tested by constraining the 16 paths (i.e. from four quality dimensions to overall service 201

222 quality; from overall service quality to trust, satisfaction, value, commitment; and from trust, satisfaction, value, commitment to attitudinal and behavioural loyalty) to be equal, using the chi-square difference test for the effect of Internet usage. An unconstrained model that simultaneously fits all three usage groups was run and the paths of interest were fixed to be invariant to all groups in arriving at a constrained model (Cunningham, 2010). In the 16 models, only six models resulted in significant difference in the chisquare test (Table 5.48). Path 62Table 5.47: Regression weights (Internet usage groups) Estimate Light User Medium User Heavy User SQ TRU.889***.863***.959*** SQ SAT.868***.926***.940*** SQ COM.803***.800***.706*** SQ VAL.824***.822***.858*** NQ SQ.666***.764***.700*** CS SQ.518***.547***.550*** IW SQ.606***.608***.711*** SP SQ.630***.731***.690*** SAT AL.332** *** TRU AL.616*** * COM AL.705***.851***.930*** VAL AL SAT BL.659***.239*.633*** TRU BL.823*** ** COM BL.682***.783***.989*** VAL BL **.109*** Goodness of fit indices λ 2 (389) = , CMIN/DF = 3.377, GFI =.858, AGFI =.831, TLI =.941, CFI =.947, RMSEA =.064, 90% CI =.06:.067, SRMR =.0491 λ 2 (389) = , CMIN/DF = 3.054, GFI =.852, AGFI =.823, TLI =.936, CFI =.942, RMSEA =.066, 90% CI =.062:.07, SRMR =.0426 λ 2 (389) = , CMIN/DF = 5.191, GFI =.868, AGFI =.842, TLI =.936, CFI =.942, RMSEA =.065, 90% CI =.062:.068, SRMR =.0457 Notes: CS = Customer Service; NQ = Network Quality; IW = Information Quality; SP = Security and Privacy; COM = Commitment; VAL = Value; TRU = Trust; AL = Attitudinal Loyalty; BL = Behavioural Loyalty; * p.05, ** p.01, *** p

223 63Table 5.48: Chi-square difference test* (Internet usage groups) Model DF CMIN P NQ SQ CS SQ IW SQ PS SQ SQ TRU SQ SAT SQ COM SQ VAL SAT AL SAT BL TRU AL TRU BL COM AL COM BL VAL AL VAL BL Notes: *Assuming model Default model to be correct; CS = Customer Service; NQ = Network Quality; IW = Information Quality; SP = Security and Privacy; COM = Commitment; VAL = Value; TRU = Trust; AL = Attitudinal Loyalty; BL = Behavioural Loyalty. The result suggests that the influence of information quality on service quality was not the same for people from different Internet usage groups (i.e. Δχ 2 (2) = 7.612, p =.022 in the path specifying information quality to service quality). Similar results are also shown in the path specifying satisfaction to behavioural loyalty (i.e. Δχ 2 (2) = 6.627, p =.036). The results suggest that the effects of trust on attitudinal and behavioural loyalty were not the same for people belonging to different Internet usage groups (i.e. Δχ 2 (2) = , p =.004 in the path of trust to attitudinal loyalty; and Δχ 2 (2) = , p =.004 in the path of trust to behavioural loyalty). In addition, the results suggest that the effects of commitment on attitudinal and behavioural loyalty were not the same for people belonging to different Internet usage groups (i.e. Δχ 2 (2) = 9.187, p =.010 in the path of commitment to attitudinal loyalty; and Δχ 2 (2) = 8.712, p =.013 in the path of commitment to behavioural loyalty). These findings demonstrate that the six paths were not invariant among customers from different Internet usage groups, indicating a strong moderating effect of Internet usage. As illustrated in Table 5.47, heavy users were significantly different from the others. The positive impact of information quality on service quality for heavy users was more significant than similar effects found among light and medium users. Also, the positive impact of commitment on attitudinal and behavioural loyalty for heavy users was more 203

224 significant than similar effects found among light and medium users. On the other hand, customer trust did not seem to play a significant role in attitudinal and behavioural loyalty among medium users. Table 5.49 shows that the model explained 62.2 per cent of the variance in value (medium user); 85.2 per cent in commitment (light user); 91.8 per cent towards satisfaction (light user); 98.2 per cent in trust (light user); 93.3 per cent towards behavioural loyalty (light user); and 96.8 per cent in attitudinal loyalty (heavy user). 64Table 5.49: Squared Multiple Correlations (Internet usage groups) Estimate Light User Estimate Medium User Estimate Heavy User VAL COM SAT TRU BL AL SP IW CS NQ Notes: CS = Customer Service; NQ = Network Quality; IW = Information Quality; SP = Security and Privacy; COM = Commitment; VAL = Value; TRU = Trust; AL = Attitudinal Loyalty; BL = Behavioural Loyalty Age Homburg and Giering (2001) consider personal characteristics, such as age, and income to be important factors in studying consumer behaviour. In particular, customers of different ages may possess different attitudes and buying behaviours (Cardoso et al., 2010; Hervé and Mullet, 2009). In line with this thinking, Homburg and Giering (2001) argue that older customers are limited in their information processing ability as compared to younger ones, which results in critical differences in their affective responses and loyalty. Results from a study in member-based services organisations by Daughtrey et al. (2013) confirm that as customers get older, they tend to be more loyal to their service provider and less likely to terminate their memberships. Hence, this study attempts to investigate customers from different age groups. In this process, the sample is split into four groups based on their age (18-28 years, years,

225 years and 50 years or older) and paths are determined at different levels of the moderating variable. As indicated in Table 5.50, the main research models which include age of customers show reasonable fit to the data. To examine the differences in each structural path between the four groups of users, the structural models were separated into four subsamples. The moderating effect was tested by constraining the 16 paths (from four quality dimensions to overall service quality; from overall service quality to trust, satisfaction, value, commitment; and from trust, satisfaction, value, commitment to attitudinal and behavioural loyalty) to be equal, using the chi-square difference test for the effect of age groups. An unconstrained model that simultaneously fits all four age groups was run and the paths of interest were fixed to be invariant in all groups to arrive at a constrained model (Cunningham, 2010). In the 16 models, only four models resulted in significant difference in the chi-square test (Table 5.51). 205

226 65Table 5.50: Regression weights (Age groups) Estimate Estimate Estimate Estimate 50 or more SQ TRU.893***.949***.925***.857*** SQ SAT.914***.913***.941***.866*** SQ COM.709***.751***.823***.796*** SQ VAL.846***.841***.868***.798*** NQ SQ.671***.746***.684***.667*** CS SQ.534***.572***.546***.497*** IW SQ.637***.687***.671***.648*** SP SQ.720***.686***.668***.629*** SAT AL.308***.468***.318**.386* TRU AL ***.416* COM AL.730***.798***.863***.979*** VAL AL SAT BL.576***.620***.405**.800*** TRU BL ***.966** COM BL.682***.799***.913***.923*** VAL BL.126**.111*** Goodness of fit indices λ 2 (389) = , CMIN/DF = 3.185, GFI =.838, AGFI =.807, TLI =.922, CFI =.930, RMSEA =.069, 90% CI =.064:.073, SRMR =.0497 λ 2 (389) = , CMIN/DF = 4.378, GFI =.869, AGFI =.843, TLI =.938, CFI =.945, RMSEA =.065, 90% CI =.062:.068, SRMR =.0484 λ 2 (389) = , CMIN/DF = 3.387, GFI =.840, AGFI =.809, TLI =.935, CFI =.942, RMSEA =.068, 90% CI =.064:.073, SRMR =.0464 λ 2 (389) = , CMIN/DF = 2.276, GFI =.828, AGFI =.795, TLI =.935, CFI =.942, RMSEA =.067, 90% CI =.061:.073, SRMR =.0428 Notes: CS = Customer Service; NQ = Network Quality; IW = Information Quality; SP = Security and Privacy; COM = Commitment; VAL = Value; TRU = Trust; AL = Attitudinal Loyalty; BL = Behavioural Loyalty; * p.05, ** p.01, *** p

227 66Table 5.51: Chi-square difference test* (Age groups) Model DF CMIN P NQ SQ CS SQ IW SQ PS SQ SQ TRU SQ SAT SQ COM SQ VAL SAT AL SAT BL TRU AL TRU BL COM AL COM BL VAL AL VAL BL Notes: *Assuming model Default model to be correct; CS = Customer Service; NQ = Network Quality; IW = Information Quality; SP = Security and Privacy; COM = Commitment; VAL = Value; TRU = Trust; AL = Attitudinal Loyalty; BL = Behavioural Loyalty. This result suggests that the effect of trust on attitudinal loyalty was not the same for people from different age groups (i.e. Δχ2 (3) = , p =.000 in the path of trust to attitudinal loyalty). Similar results are also shown in the path specifying trust to behavioural loyalty (i.e. Δχ2 (3) = , p =.001). In addition, the results suggest that the path specifying commitment to attitudinal and behavioural loyalty were not the same for people from different age groups (i.e. Δχ2 (3) = , p =.000 in the path of commitment to attitudinal loyalty; and Δχ2 (3) = , p =.000 in the path of commitment to behavioural loyalty). These findings demonstrate that the four paths were not invariant among customers from different age groups, indicating a strong moderating effect of age. The results suggest that as customers age, they tend to be more loyal, both attitudinally and behaviourally, to the service providers to which they are highly committed. Noticeably, trust was only significantly related to loyalty among customers who were 39 years old or older. This demonstrates that older customers those older than 39 years of age are more likely than younger customers to develop loyalty based on trust. Table 5.52 shows that the model explained 68.1 per cent of the variance in value (50 or more), 89.2 per cent in commitment (39-49), 91.1 per cent towards satisfaction (50 or 207

228 more), 97.7 per cent in trust (39-49), 94.6 per cent towards behavioural loyalty (39-49), and 98.5 per cent in attitudinal loyalty (50 or more). 67Table 5.52: Squared Multiple Correlations (Age groups) Estimate Estimate Estimate Estimate 50 or more VAL COM SAT TRU BL AL SP IW CS NQ Notes: CS = Customer Service; NQ = Network Quality; IW = Information Quality; SP = Security and Privacy; COM = Commitment; VAL = Value; TRU = Trust; AL = Attitudinal Loyalty; BL = Behavioural Loyalty Income Income is an important demographic characteristic that directs customer behaviour (Homburg and Giering, 2001). This study examined four income levels under 30,000 baht, 30,001-50,000 baht, 50, ,000 baht and over 100,000 baht and determined paths at different levels of the moderating variable. As indicated in Table 5.53, the main research models of income level show reasonable fit to the data. To evaluate the differences in each structural path between the four groups of users, the structural models were separated for the four subsamples. The moderating effect was tested by constraining the 16 paths (from four quality dimensions to overall service quality; from overall service quality to trust, satisfaction, value, commitment; and from trust, satisfaction, value, commitment to attitudinal and behavioural loyalty) to be equal using the chi-square difference test for the effect of income level. An unconstrained model that simultaneously fits all four income groups was run and the paths of interest were fixed to be invariant in all groups to arrive at a constrained model (Cunningham, 2010). In the 16 models, only five models resulted in significant differences in the chisquare tests (Table 5.54). 208

229 68Table 5.53: Regression weights (Income groups) Estimate Under 30,000 Estimate 30,001-50,000 Estimate 50, ,000 Estimate Over 100,000 SQ TRU.844***.907***.979***.921*** SQ SAT.841***.893***.983***.933*** SQ COM.737***.760***.785***.775*** SQ VAL.799***.819***.909***.870*** NQ SQ.614***.696***.754***.754*** CS SQ.534***.496***.551***.592*** IW SQ.668***.626***.652***.707*** SP SQ.675***.670***.704***.672*** SAT AL.305***.593*** *** TRU AL ** * COM AL.909***.962***.933***.921*** VAL AL SAT BL.431***.903***.532***.696*** TRU BL ***.457**.768* COM BL.962***.751***.992***.918*** VAL BL -.141*** -.079* Goodness of fit indices λ 2 (389) = , CMIN/DF = 4.093, GFI =.846, AGFI =.815, TLI =.922, CFI =.930, RMSEA =.07, 90% CI =.066:.073, SRMR =.0519 λ 2 (389) = , CMIN/DF = 3.091, GFI =.847, AGFI =.817, TLI =.934, CFI =.941, RMSEA =.066, 90% CI =.062:.07, SRMR =.0498 λ 2 (389) = , CMIN/DF = 3.298, GFI =.847, AGFI =.817, TLI =.937, CFI =.944, RMSEA =.068, 90% CI =.064:.072, SRMR =.0412 Chi Square difference test Δλ 2 (60) = , p =.165 λ 2 (389) = , CMIN/DF = 3.025, GFI =.841, AGFI =.810, TLI =.933, CFI =.940, RMSEA =.068, 90% CI =.064:.073, SRMR =.0454 Notes: CS = Customer Service; NQ = Network Quality; IW = Information Quality; SP = Security and Privacy; COM = Commitment; VAL = Value; TRU = Trust; AL = Attitudinal Loyalty; BL = Behavioural Loyalty; * p.05, ** p.01, *** p

230 69Table 5.54: Chi-square difference test* (Income groups) Model DF CMIN P NQ SQ CS SQ IW SQ PS SQ SQ TRU SQ SAT SQ COM SQ VAL SAT AL SAT BL TRU AL TRU BL COM AL COM BL VAL AL VAL BL Notes: *Assuming model Default model to be correct; CS = Customer Service; NQ = Network Quality; IW = Information Quality; SP = Security and Privacy; COM = Commitment; VAL = Value; TRU = Trust; AL = Attitudinal Loyalty; BL = Behavioural Loyalty. This result suggests that the effect of service quality on trust was not the same for people with different levels of household income (i.e. Δχ 2 (3) = , p =.012 in the path of service quality to trust). Similar results are also shown in the path specifying service quality to satisfaction (i.e. Δχ 2 (3) = 9.076, p =.028). The results suggest that the effects of satisfaction on attitudinal and behavioural loyalty were not the same for people in different income groups (i.e. Δχ 2 (3) = , p =.017 in the path of satisfaction to attitudinal loyalty and Δχ 2 (3) = 8.851, p =.031 in the path of satisfaction to behavioural loyalty). In addition, the results suggest that the effects of commitment on behavioural loyalty was not the same for people with different income levels (i.e. Δχ 2 (3) = , p =.017 in the path of commitment to attitudinal loyalty). These findings demonstrate that the five paths were not invariant among customers with different household income level, indicating income as a moderating effect. It appears that the effects of service quality on trust and satisfaction were stronger among higher income segments. Noticeably, satisfaction did not have a significant effect on attitudinal loyalty among the upper-middle income group (those earning between 50, ,000 baht). In contrast, the effect of satisfaction on behavioural loyalty was highest among customers in the middle income group (those earning between 30,001-50,000 baht). However, this group also demonstrated the least predictive power for commitment as an 210

231 indicator of behavioural loyalty. Table 5.55 shows that the model explained 69.3 per cent of the variance in value (50, ,000 baht); 84.2 per cent in commitment (50, ,000 baht); 91 per cent towards satisfaction (50, ,000 baht); 95.9 per cent in trust (over 100,000 baht); 92.7 per cent towards behavioural loyalty (over 100,000 baht); and 97.9 per cent in attitudinal loyalty (50, ,000 baht). 70Table 5.55: Squared Multiple Correlations (Income groups) Estimate Under 30,000 Estimate 30,001-50,000 Estimate 50, ,000 Estimate Over 100,000 VAL COM SAT TRU BL AL SP IW CS NQ Notes: CS = Customer Service; NQ = Network Quality; IW = Information Quality; SP = Security and Privacy; COM = Commitment; VAL = Value; TRU = Trust; AL = Attitudinal Loyalty; BL = Behavioural Loyalty. 211

232 5.11 Chapter summary Chapter 5 presented the results of the data analysis. This chapter commenced with data screening (section 5.2), profile of respondents (section 5.3), details of Internet services used (section 5.4), additional descriptive statistic (section 5.5) and segmentation based on Internet usage patterns (section 5.6). Profiles of the participants were depicted, followed by a description of the statistical measures adopted to assess each of the constructs of the conceptual model. The chapter proceeded to assess the validity and reliability of the constructs. This is followed by exploratory factor analysis (section 5.7) confirmatory factor analysis, the building of measurement models (section 5.8) and full Structural Equation Model analysis (section 5.9). Section 5.10 presented segmentation analysis using invariance testing done on the basis of Internet usage patterns, age groups and income levels. Figure 5.23 provides a roadmap to the structure of the overall thesis. 212

233 Chapter One An introduction to the thesis and overview of the chapters Chapter Two Literature Review Chapter Three Development of the theoretical model and related hypotheses Chapter Four Methodology Chapter Five Analysis and Results Next chapter Chapter Six Discussion, Recommendations and Conclusion 36Figure 5.23: Structure of the overall thesis 213

234 Chapter 6: Discussion of Findings, Recommendations and Conclusion 6.1 Chapter overview The results of this research study have been collated in the final chapter. This chapter also presents a detailed discussion of each of the individual dimensions included in the conceptual model. Chapter 6 starts with a discussion of the implications of the best-fit Structural Equation Model (section 6.2). The service quality dimensions are reviewed, and the relationship between service quality, customers cognitive and affective evaluations and customer loyalty in the context of home Internet services is presented. This is followed by a discussion of findings of the segmentation analysis (section 6.3). The overall theoretical and managerial implications, as well as action plans of the research are discussed (section 6.4). The chapter concludes with the limitations of the study, directions for future research (section 6.5) and concluding remarks (section 6.6). The organisation of the discussion in this chapter is shown in Figure

235 6.1 Chapter overview 6.2 Discussion of the findings 6.3 Segmentation analysis 6.4 Contribution of this study 6.5 Limitations of this study and future research directions 6.6 Concluding remarks 6.7 Chapter summary 37Figure 6.1: Chapter organisation 215

236 6.2. Discussion of the findings The following section provides the discussion of findings relating to the research questions. The main objective of this research was to investigate the antecedents of customer retention and customer loyalty of Internet service providers in Thailand. In addition, this exploratory study intended to identify the dimensions of overall service quality of Internet service providers, as well as the relationships between these dimensions and service quality. The overall aim was to develop and empirically test a model presenting the role of relevant service quality dimensions in influencing customer loyalty in the home Internet services market. A three stage conceptual framework was proposed to achieve this aim. The three stages were termed: (1) the specific service quality dimensions of an ISP; (2) cognitive and affective evaluations of customers and (3) outcomes or resultants (Figure 6.2). The specific service quality dimensions in this conceptual framework consist of the core services offered by ISPs to their customers in the Internet services market. It has been demonstrated that both cognitive and affective evaluations influence customer loyalty. These evaluations represent the various ways customers assess and respond to the services of their ISPs after experiencing their service quality dimensions. The final component of the model illustrates the influence of customers attitudes towards their ISP, as well as their future intentions. Hence, the outcome or resultants relate to the overall impact of the service encounters on the development of long-term relationships between an ISP and its customer. There were three research questions, which were sequentially associated with the three stages of the conceptual framework. Service quality dimensions of an ISP Cognitive and affective evaluations Outcomes or Resultants 38Figure 6.2: Initial conceptual framework for the study 216

237 An ISP s service quality dimensions play a vital role in the development of customer loyalty. After an extensive literature review, service quality dimensions were identified as being network quality, customer service and technical support, information quality and website information support, and security and privacy Discussion of findings relating to stage 1 of the conceptual framework: service quality dimensions As discussed earlier, SERVQUAL and E-S-QUAL focus on service providers that operate via the Internet platform (Vlachos and Vrechopoulos, 2008), but not on those actually providing the Internet connection and platform activities. Several studies using SERVQUAL AND E-S-QUAL have been done in the telecommunications industry, especially in the mobile telephony market (He and Li, 2010). However, many basic differences exist between Internet services and other telecommunications services. This results in a need to identify the unique characteristics of an ISP s service quality. Extant research shows that network quality reflects the core performance of a telecommunications service. Moreover, customer service represents the empathy and responsiveness dimensions of the SERVQUAL scale, and is an important dimension of an ISP s service quality. Internet services are based on high-tech information technology in both online and offline communications modes. Existing research provides evidence of the contribution of information quality and website support in enhancing customers perceptions of an ISP s service quality. Additionally, owing to concerns regarding cybercrimes and privacy breaches, ISPs are expected to demonstrate their capability and motivation in protecting their customers privacy and security, which is another measure of service quality. The results demonstrate that overall service quality is determined by network quality, customer service and technical support, information quality and website information support, and security and privacy. However, these dimensions have varying influences on the overall perception of service quality of an ISP. Among the four dimensions, the influence on service quality of information and website support (i.e. β = 0.825) is the greatest, followed by privacy and security (i.e. β = 0.782). Network quality (i.e. β = 0.743) comes next and customer service and technical support (i.e. β = 0.692) have the least impact on perceived service quality. The predominant role of information quality 217

238 and website support and privacy and security in determining overall service quality confirms that ISP customers place greater emphasis on these dimensions compared with core service performance, namely network quality and customer service, which are standard across various service providers. A detailed discussion is provided for each individual dimension ordered by their importance to service quality Information quality and website information support The results of this study demonstrate strong support for the hypothesis that information quality and website information support are associated with service quality. Despite scant evidence on the relationship between website information support and an ISP s service quality, the current study confirms that this dimension significantly contributes to customers perceptions of an ISP s overall service quality. In fact, this dimension ranked highest in influencing overall service quality. It is likely that this is because Internet users are avid information seekers and are experienced browsing websites and searching for knowledge. At the inception of this research study, information quality and website information support was operationalised using eight items. However, during the analysis, it became clear that some of those items were not relevant. It was initially assumed that customers would most likely consider information about an ISP using material provided on the company s websites, or by other industry or commercial organisations linked to that ISP, or through advertising and print material, such as brochures, and flyers. However in the analysis, these particular items were eliminated to improve the fit of the measurement model. The final measurement model for information quality and website support included items relating to the sufficiency, relevance and currency of information. This illustrates that a customer s need for information quality is very specific, and quality is preferable over quantity. ISPs need to tailor their information support to suit a variety of customer needs and provide timely updates to their customers. The results indicate that information quality and website information support have a significant relationship with service quality. Businesses need to provide information that helps customers understand their products offerings and to support customer decision 218

239 making (Hasley and Gregg, 2010). Such information includes detailed product description, transparent price information, company information, professional advice, research reports, contact information, and hyperlinks to relevant websites (Yang et al., 2005). According to Yang et al. (2005), insufficient information hinders the process of conveying the right message and actual situation to the customers. On the other hand, too much information can cause customers to become confused and overwhelmed. Therefore, ISPs need to be vigilant about the kind of information they display on their websites and they should avoid displaying information which distracts customers from their decision making (Hasley and Gregg, 2010). In short, it can be concluded that the accuracy, quality and relevancy of the information ISPs provide to customers is paramount in influencing overall service quality Security and privacy The results of this study show that security and privacy have a significant relationship with overall service quality. This dimension has the second strongest influence on service quality; it is surpassed only by information quality and website support. This indicates customers are unhappy when their privacy is not protected or when there is a security breach relating to their transactions. Online privacy and security continue to be regarded as extremely important issues, particularly when it comes to incidents of online fraud and privacy breaches. More than $500 million was reportedly lost due to cybercrime in 2012 (IC3, 2012). Additionally, unprotected online data can lead to the illegal dissemination of customers demographic information (Steel, 2013). It must be noted that Thailand led the world in phishing in 2012 (Aaron and Rasmussen, 2013) and was in the top 30 countries with the highest piracy rate (Nationmaster, 2012). Previous research indicates that security and privacy are positively related to service quality, especially in e-commerce contexts (Ha and Stoel, 2012; Wolfinbarge and Gilly, 2003). In line with this thinking, Roca, García and Vega (2009) state that consumers tend to believe that it is safe to purchase from service providers who have credible reputation in relation to security practices. In this study, customer privacy was represented through the collection and use of personal and financial information by ISPs for administrative and billing purposes. Having access to such important personal details means that customers rely on their 219

240 service providers to manage and protect that data. As both personal and financial information are closely related, leaks from an ISP s database could lead to untoward incidents, such as stolen identity, online fraud or scams. As such, it is important to ensure confidentiality during the process of collecting, handling, storing and transmitting customers private information. Furthermore, as this dimension it has the highest factor loading, a secure network and secure transactions between customers and service providers directly relates to customer perceptions of an ISP s security. Customers are aware that their home Internet connection does not always provide a high level of security for online transactions. Because of this, ISPs who are able to alert their customers to the dangers of malicious traffic and provide cleaner networks tend to be highly evaluated (Streamshield, 2004) Network quality The results of this study demonstrate that network quality is positively associated with service quality. This dimension has frequently been reported in the service quality literature (He and Li, 2010). Network quality, which is only ranked third in terms of contribution to overall service quality perception, appears to have become largely reliable in home Internet services. It is, therefore, no longer a differentiator between ISPs, nor a useful measure of competitive advantage. However, ISPs need to ensure optimal standards of their network quality to allow them to remain competitive. The development of technology and the accessibility of a wide range of specialist websites have provided customers with the ability to log their connectivity using speed tests and problem diagnosis tools. The significant influence of network quality on overall service quality proves that customers pay considerable attention to the mechanics of Internet downloading and uploading speed. This item had the highest loading among items measuring network quality. Following closely was the item relating to consistency of the network speed. ISP customers expect their provider to deliver a satisfactory service measured by the quality of the connection, in particular agreed downloading and uploading speed during peak and off-peak periods. Although it is less important, a stable connection is still a concern for customers. Temporary Internet disconnections, similar to call drop-outs among mobile phone service providers, causes inconvenience for customers. This 220

241 finding supports previous works conducted by the likes of Cheng et al. (2008), Lai et al. (2009) and Vlachos and Vrechopoulos (2008). Network quality is deemed to be one of the most important drivers of overall service quality in the Chinese (Lai et al., 2009), Hong Kong (Cheng et al., 2008) and Greek (Vlachos and Vrechopoulos, 2008) mobile service provides Customer service and technical support Despite being the least important dimension of service quality, customer service and technical support is significantly and positively related to overall service quality. The Internet services industry is characterised by technical jargon and often requires specialist knowledge to address the problems that arise. These characteristics make customer service vital. The study s findings demonstrate that overall service quality depends on how well customer service personnel handle and respond to customer enquiries. It is important that the customer services team and the company show a sincere interest in solving customer enquiries. The quality of customer service is also evaluated based on the knowledge and expertise of staff providing it. Customers demand that their enquiries are dealt with promptly and professionally by skillful and knowledgeable staff. Previous research by Abdolvand et al. (2006) report similar findings. These authors suggest that businesses should pay attention to customer support in order to improve overall service quality perception. The quality of technical support also refers to how quickly technical problems are solved. Time is an important factor for an ISP s customers. The Internet increasingly plays a critical role in many people s daily lives, hence disruptions to Internet services can result in significant inconvenience for customers. When customers face problems with high-tech Internet services, they often seek help and support from technical and customer service staff (Thaichon et al., 2014). For this reason, customer service teams are under constant pressure to perform reliably, dependably and according to set protocols in order to meet their productivity goals (Lounsbury et al., 2012) and deliver quality customer service (Rod and Ashill, 2013). A study of the Turkish telecommunications industry demonstrates that efficiently handling customer complaints enhances perceptions of overall service quality (Aydin and Özer, 2005). 221

242 Four dimensions network quality, customer service and technical support, information quality and website support and privacy and security exert varying levels of influence on overall ISP s service quality. This proves that service quality is a multi-dimensional construct. Perceived service quality reflects customer experience in service encounters and may result in both cognitive and emotional responses, thereby forming the basis for the relationship between customers and their service providers (Edvardsson et al., 2005). The next section discusses the influence of service quality on relationship factors, or more specifically, customers cognitive and affective evaluations including satisfaction, trust, commitment, and value Discussion of findings relating to stage 2 of the conceptual framework: customers cognitive and affective evaluations In the second stage of the proposed conceptual framework, the relationships between overall service quality comprising the four dimensions (network quality, customer service and technical support, information and website support, and privacy and security) and customer affective and cognitive evaluations were evaluated. This research endeavours to understand the relationships between overall service quality and customer satisfaction, value, trust and commitment towards Internet service providers in Thailand. In general, all of the relationships are found to be positive and significant. Clearly, overall service quality shapes a multitude of customer perceptions toward service providers. More specifically, the effect of service quality is found to be strongest on trust (i.e. β = 0.988), followed by satisfaction (i.e. β = 0.943). The service quality s influence on commitment (i.e. β = 0.887) ranks third and the impact of service quality on value (i.e. β = 0.778) ranks the lowest. Details of these cognitive and affective evaluations are discussed on the basis of the significance of the influence of service quality on them ISP s overall service quality and customer trust In home Internet services, Chiou (2004) suggests that customer trust can be evaluated by exploring how customers feel about their service provider with regards to the company s honesty, responsibility, professional manners and degree of understanding and care. Initially in this research, trust was operationalised using five items. However, the items relating to the billing system and attention to customer interests, were found to 222

243 be problematic. Reliability in billing systems may have been confusing as it appears to be closely related to the construct of privacy and security. Furthermore, customers interests in billings form part of the quality of customer service. Hence, removing these items resulted in better scale reliability. Consistent with Morgan and Hunt (1994) and Sirdeshmukh et al. (2002), in this study, trust is measured by how highly a customer rates an ISP s reliability and honesty. Moreover, trust also reflects the expectations of customers that the company can be relied upon to fulfil its promises. The results of this study reveal that among the four customer affective and cognitive evaluations, service quality is highly related to customer trust. Customers feel that they can trust and rely an ISP which delivers high overall service quality. This result is not surprising considering the amount of research conducted in this area which predominantly concurs that service quality is significantly associated with customer trust (Chiou, 2004; Gounaris and Venetis, 2002). Gounaris and Venetis (2002) demonstrate that the quality of the service provided influences the level of customers trust in the service provider. Network performance, customer support and information quality represent a company s promise and reputation. Moreover, service quality also comprises reliability and integrity, attributes which constitute the foundation of trust (Doney and Cannon, 1997). In line with this thinking, in the context of online marketing of Thai gemstones, Alessandro et al. (2012) report that a company s privacy and security practices have a significant influence on customer trust. Previous research indicates that features such as procedural fairness contributes to impersonal trust; as such, when the companies act in a fair manner with a view to addressing customer concerns, their customers are more likely to consider the companies to be reliable and trustworthy (Culnan and Armstrong, 1999). Similarly Hoffman et al. (1999) suggest that informing customers clearly about the process of information collection and usage can decrease the risk associated with purchase. The level of trust an individual develops toward the other party in a transaction is a function of the level of risk involved in the situation (Koller, 1988). Hence, it can be proposed that privacy risk is associated with trust, and improving customers perception of privacy, which is a part of overall service quality, can lead to greater trust. Likewise, favourable perceptions of an ISP s level of security can give 223

244 customers a sense of safety and reliability, which then leads to trust, towards the company ISPs overall service quality and customer satisfaction Results of this study demonstrate that service quality is positively associated with customer satisfaction. Service quality is first experienced by the customer at an encounter level, and over time, customers build an overall evaluation of the quality of services. Satisfaction is mainly an affective response to the service provider, which takes into account the service provided and relationships and outcomes produced, as judged by the company s performance and efforts. Oliver (1993) and Rust et al. (1995) argue that satisfaction with individual aspects of services can be incorporated as part of an overall judgment of satisfaction. In this study, satisfaction as a global judgment is defined by customers feeling delighted with the service provider. When customers believe they have chosen the right service provider, this manifests itself through a feeling of happiness and satisfaction. The level of a customer s satisfaction depends on the ability of the supplier to meet the customer s norms and expectations (Zeithaml et al., 1996). In other words, if the initial expectations of the overall service quality are exceeded, a customer would be satisfied. On the basis of the findings, the level of customer satisfaction can be enhanced if ISPs introduce steps to improve the core service performance or network quality; the human element of service delivery or customer service and technical support; the information element or information quality and website support and provide clear and transparent privacy and security policies. Researchers generally agree that favorable service quality perceptions lead to improved satisfaction (Parasuraman et al., 1985). Extant research by Deng et al. (2010) and Woo and Fock (1999) report similar findings. In the telecommunications industry, service quality is found to have a strong relationship with customer satisfaction in the Chinese (Deng et al., 2010) and Hong Kong contexts (Woo and Fock, 1999). In the Australian higher educational context, service quality has a positive influence on customer satisfaction (Sultan and Wong, 2012). 224

245 ISPs overall service quality and customer commitment The importance of relationship commitment as an indicator of relationship quality has been highlighted by many authors (Roberts et al., 2003). Essentially, customer commitment is a psychological attribute that links customers to a company. Commitment has been discussed in various studies in several ways. Caceres and Paparoidamis (2007) integrate measures relating to connection and pride, whereas La et al. (2005) identifies two dimensions of commitment, namely attitudinal disposition and relationship orientation. In this study, commitment is conceptualised as a relational involvement, a sense of belonging and a desire to maintain the relationship, thereby implying emotional attachment. Service quality could be considered as one of several evaluative benefits that customers pursue in their relationship with service providers to help create commitment (Gruen et al., 2000). Feelings of attachment are likely to come about as a result of overall service quality judgment (Gruen et al., 2000). Moreover, overall service quality represents potential relationship benefits that will be lost if a customer switches to other service providers (Morgan and Hunt, 1994). On the basis of these findings, in the context of an ISP s services, overall service quality is an immediate antecedent of commitment, thereby leading to service provider-customer relationships. Service quality has a significant influence on customer commitment, as explained in the works of Al-Hawari (2011) and Fullerton (2005). In other words, when customers are happy with the service quality being provided, they are inclined to stay with their service provider because of what that provider has done for them (Cater and Zabkar, 2009). Hence, a company that can deliver a high level of overall service quality will likely have higher customer commitment. Essentially when customers are delighted with the overall service quality, they are willing to dedicate themselves to continuing their relationships with a particular ISP. Fullerton (2005) reports that service quality is positively related to customer commitment in the Canadian telecommunications services. This is supported by Morgan and Hunt (1994), who find that overall service quality is a direct antecedent of commitment in the retail services context. Likewise, a study conducted in retail banking in the United Arab Emirates confirms that service quality has a direct and positive influence on customer commitment (Al-Hawari, 2011). 225

246 ISPs overall service quality and customer value Despite many attempts to conceptualise the role of value in the process of customer decision making, previous research has revealed several aspects to the concept of value. The precursors and outcomes of customer perceived value vary across studies (Dodds et al., 1991). Moreover, only few studies have simultaneously examined perceived value in conjunction with variables that have been proved to be significant in evaluating consumer service experience, such as service quality (Oh, 1999). As mentioned earlier, in this study, value is characterised by monetary and functional value. A service is considered to be of high value when customers gains outweigh their losses. In other words, customers consider whether the service they are receiving is worth the costs incurred. Likewise, Dodds et al. (1991) propose that perceived value is a tradeoff between perceived quality and perceived sacrifice. In general, the total perceived value of a service originates from two sources: value that comes from the service performance itself and value that comes from the quality of the service act (Oh, 1999). This study posits that perceived quality contributes towards the majority of the service s total perceived value, mirroring the study of Jayanti and Ghosh (1996) who find perceived quality as a direct function of perceived value. The results of this study reveal that service quality is a determinant of customer value. This suggests that when customers perceive the overall service quality to be high they consequently perceive their Internet services package as being good value for money. Previous research by Kim and Damhorst (2010) and Lai et al. (2009) report similar findings. This is because higher service quality is associated with greater benefits received by customers, which also means higher value. Kim and Damhorst (2010) state that service quality positively influences perceived value in the Internet retail setting. Lai et al. (2009) state that apart from monetary costs, value is determined by service quality in the telecommunications market. In an e-commerce setting, researchers state that perceived service quality is positively related to perceived value (Chen and Dubinsky, 2003). In the hospitality industry, Hu et al. (2009) confirm that service quality exerts a positive impact on perceived value for hotel guests in Mauritius. Supporting this view, Cronin et al. (2000) confirm that service quality affects perceived value. 226

247 The results demonstrate that overall service quality is an important factor in the development of relationship between service providers and customers. However, the ultimate goal of a relationship is customer loyalty, which manifests itself via favorable attitudes, advocating positive behaviour, and repeat purchase. The discussion on the effects of service quality, trust, satisfaction, commitment and value on both attitudinal and behavioural loyalty is provided in the next section Discussion of findings relating to stage 3 of the conceptual framework: customers cognitive and affective evaluations on customer loyalty When investigating how customers feelings of trust, satisfaction, commitment and value are associated with attitudinal and behavioural loyalty, it was found that these feelings are positively associated with both types of loyalty. Commitment is the most significant antecedent of customer attitudinal and behavioural loyalty (i.e. attitudinal loyalty (dependent variable): β = 0.877; behavioural loyalty (dependent variable): β = 0.880). This is not surprising as commitment and loyalty are linked and commitment has long been considered to be an inevitable precursor to loyalty (Caceres, & Paparoidamis, 2007). Oliver (1999, p. 34) defines loyalty as: a deeply held commitment to rebuy or repatronise a preferred product or service consistently in the future, thereby causing repetitive same-brand or same brand-set purchasing, despite situational influences and marketing efforts having the potential to cause switching behaviour. It was found that satisfaction has the second strongest influence on customer loyalty (i.e. attitudinal loyalty (dependent variable): β = 0.362; behavioural loyalty (dependent variable): β = 0.571). Trust (i.e. attitudinal loyalty (dependent variable): β = 0.253; behavioural loyalty (dependent variable): β = 0.384) is ranked third and value (i.e. attitudinal loyalty (dependent variable): β = ; behavioural loyalty (dependent variable): β = 0.097) has the least significant impact on customer loyalty. Noticeably, the direct effect of value on attitudinal loyalty is also not significant. The predominance of affective evaluation on customer loyalty indicates the significance of affection over cognition in customer decision making. On the other hand, the indirect effect of overall service quality on customer loyalty can be confirmed, demonstrating the importance of service performance. 227

248 The subsequent section details the influence of the cognitive and affective evaluations of customers, and overall service quality on customer attitudinal and behavioural loyalty within the Thai home Internet services market. A discussion is provided for each individual construct ordered by their importance to customer loyalty Customer commitment and customer loyalty Most marketing researchers agree that commitment is an attachment between two parties that results in a desire to retain a relationship (Moorman et al., 1992; Morgan and Hunt, 1994). Beyond the favourable or unfavourable evaluation of a service provider, commitment helps strengthen customer behaviour over time, regardless of the situation (Scholl, 1981). Commitment is therefore a critical component of long-term loyalty (Morgan and Hunt, 1994). Commitment is based on a preference for a particular company. It is acted upon by continual use of a service and resistance to competitors counterattack (Zeithaml et al., 1996). Under a relationship marketing paradigm, customer commitment, rather than service quality, is the major driver of consumer loyalty towards the service provider (Morgan and Hunt, 1994). It has been recognised that relationships can be based upon several forms of commitment (Harrison-Walker, 2001; Kumar et al., 1995), for example affective and continuance commitment. As mentioned earlier, in this study commitment is conceptualised as being primarily emotion-based. Therefore, it is not surprising that commitment is confirmed to have a significant effect on attitudinal loyalty. Attitudinal loyalty is measured by customers favourable attitudes, word of mouth and recommendations. Customers who have a strong sense of belonging to a service provider will likely feel loyal towards that company. Committed customers also exhibit advocacy behaviour and become a voluntary promoter of the company. This supports previous work conducted by the likes of Fullerton (2005). Furthermore, the results of this study provide support for the hypothesis regarding the relationship between customer commitment and behavioural loyalty. A committed customer is more likely to repurchase from a service provider, echoing the findings of Fullerton (2005) and Cater and Zabkar (2009). As such, even if it is more difficult to 228

249 purchase from the service provider, customers will still consider the service provider as the number one choice and ignore alternatives. The more time and effort customers have invested in the relationship with their service provider, the less inclined they are to terminate that relationship (Bügel et al., 2010). Hence, customers with a high level of commitment are more likely to stay with their incumbent service providers and unwilling to seek alternatives (Bügel et al., 2010; Cater and Zabkar, 2009). In fact, extant research demonstrates that there is a positive relationship between customer commitment and customer loyalty. For example, Cater and Zabkar (2009) report that affective and calculative commitment positively affects customers intention to continue a relationship with their service provider in Central and Eastern Europe in the services sector. Verhoef (2003) finds a positive connection between affective commitment and customer loyalty in the financial services industry and Fullerton (2005) states that calculative commitment and affective commitment positively influence behavioural intentions (Cater and Zabkar, 2009; Fullerton, 2005) Customer satisfaction and customer loyalty Customer satisfaction is the second greatest influence on customers loyalty to ISPs. This is due to both the complex nature of the high-tech services industry and the tendency for relationships in the telecommunications industry to be lasting. Satisfaction is a crucial aspect of relationships and is necessary for their continuity (Crosby et al., 1990). The results of this study reveal that customer satisfaction has a positive impact on both types of loyalty: attitudinal and behavioural. Customer satisfaction is a precursor to customer attitudinal loyalty, suggesting that a satisfied customer displays favourable attitudinal intention towards their service provider. Satisfaction can also result in positive word of mouth and recommendations for the business, echoing the findings of Anderson and Sullivan (1993). Research by Gerpott et al. (2001) reports similar findings. Research in the American Customer Satisfaction Index provides additional empirical evidence for loyalty responses as the key outcome of consumer satisfaction (Fornell et al., 1996). The influence of customer satisfaction on behavioural loyalty demonstrates that customers who are happy with the service tend to exhibit an intention to repurchase. 229

250 The results indicate that satisfied customers are more likely to consider their incumbent service provider as the primary choice when a purchasing decision is required. This supports previous research by Gerpott et al. (2001) and Cheng et al. (2008). It is worth noting that extant research finds that customer satisfaction is a determinant of customer loyalty in the Hong Kong Internet services market (Cheng et al., 2008). These authors suggest that customers who experience a high level of satisfaction are most likely to stay with their existing service provider and maintain their service subscription. Similarly, a recent study in the Malaysian mobile services context concludes that when customers feel satisfied, they tend to become more loyal and desire a long-term relationship with their service provider (Mokhtar et al., 2011). Khokhar et al. (2011) assert that that there is a positive relationship between customer satisfaction and customer loyalty in Pakistan s telecommunications industry Customer trust and customer loyalty Trust and commitment are the most important factors in the relationship marketing paradigm (Morgan and Hunt, 1994). The results in the current study demonstrate that customer trust is significantly related to attitudinal loyalty. This finding is supported by Oliver (1999), who states that trust appears inherent to true attitudinal loyalty. Customers tend to form positive feelings towards a service provider they trust. Likewise, they spread favourable word of mouth and recommend others to purchase from a service provider they rely on. Since trust is defined as confidence in the reliability and honesty of the partner in a business exchange relationship, it is an essential element for long-term orientation leveraging shifting customer attention to future circumstances and continuity (Doney and Cannon, 1997; Ganesan, 1994). Similarly, the results confirm that customer trust has a significant influence on behavioural loyalty. In other words, customers are more likely to repurchase from service providers whom they trust. In addition, a reliable service provider has more chances of outperforming its competitors owing to favour customers perceptions. Previous research also demonstrates that trust plays an important role in determining customer loyalty. In the telecommunications market, Chiou and Droge (2006) reveal that customer trust is related to the emotional nature of consumer loyalty and it 230

251 determines long-term orientation of the relationship. Likewise, several researchers suggest that trust positively influences customer attitude and behavioural intention in the mobile commerce context (Deng et al., 2010). Pirc (2006) claims that customer trust has a positive and direct effect on loyalty in the Slovenia mobile phone industry. Similar results are also reported in the Taiwanese ISP industry (Chiou, 2004), and in the mobile phone market of the United Kingdom (Ranaweera and Prabhu, 2013). Therefore, an ISP which is trustworthy has a higher level of attitudinal and behavioural loyalty among its consumers Customer value and customer loyalty The results reveal that customer value is not significantly associated with attitudinal loyalty, despite the fact that perceived value has previously been considered a powerful determinant of customer loyalty in the services context (Qian et al., 2011). Value was negatively related to the attitudinal loyalty of the Thai Internet service consumers. This can be attributed to the fact that the costs of services, as well as the promotional programs, are very similar among the three main Thai ISPs. Therefore, perceived value may not be significantly different among customers of those ISPs. In this study, perceived value is also conceptualised as the functional and financial value relating to quality performance and monetary benefits, which may be an indicator of price sensitivity. A price conscious customer tends to switch between brands and has minimal brand inertia, as well as a low sense of loyalty (Corstjens and Lal, 2000). Moreover, services offered at a comparatively low price might be associated with a negative image and a lesser brand value. In other words, a perception of high value does not guarantee positive attitudes towards the service providers nor does it encourage advocacy behaviour. Nevertheless, the results illustrate that customer value still has a significant influence on customer behavioural loyalty. This means high perceived value can facilitate repeat purchase. In addition, customers still favourably consider high-value services. This finding echoes the results reported by Lee and Murphy (2008). These authors report that in the Australian mobile service context, value is a stronger antecedent of loyalty as compared to service quality and switching costs (Lee and Murphy, 2008). Similarly, in the Thai mobile services context, value has a positive influence on customer retention 231

252 (Leelakulthanit and Hongcharu, 2011). In the Chinese mobile data services, customer value directly influences repurchase intentions (Qian et al., 2011; Wang et al., 2004). Likewise, Lien et al. (2011) find a significant link between perceived value and behavioural intentions in the Taiwanese online shopping context. Similar findings are also reported in the Taiwanese mobile commerce context (Chiou, 2004) Overall service quality and customer loyalty The results confirm the indirect effect of service quality on customer loyalty. Overall service quality impacts customers affective and cognitive responses, which are antecedents of customer loyalty. In other words, the indirect effect is channelled via customers trust, satisfaction, commitment and value. A high level of service quality generates positive attitudes, motivates positive word of mouth and recommendations, and encourages customers repeat purchase. These findings support previous research by Prentice (2013). Overall service quality influences customers perceptions of the service provider and its offerings. Superior service quality often results in positive attitudes from customers towards the business, can make customers feel loyal to the company (Prentice, 2013) and can encourage customers to spread favourable word-ofmouth to others (Sabiote and Roman, 2009). Moreover, it is likely that customers will stay with a company that delivers high quality services and in the process, they develop positive feelings and thoughts. Given the importance of service quality, improvements in service quality are vital for the success of service-based businesses. 6.3 Segmentation analysis Ringle et al. (2013) and Mazzoni et al. (2007) suggest that studying a single homogenous population is insufficient to understand relationships because customer characteristics and the nature of demand for services differ. Nevertheless, there is scant evidence as to how effective segmentation is operationalised for an ISP s customers. Using simple analysis techniques, this study introduced customer segmentation based on usage pattern (heavy, medium and light users), as well as it demonstrated several basic differences between customers in different age and income groups. In addition, invariance testing in the final best fit model was employed to extend the segmentation analysis and provide further insights into customers with different levels of Internet usage. The following sections discuss the results of the segmentation analysis. 232

253 6.3.1 Internet usage Segmentation based on usage is one of the most logical bases of segmentation when considering similar types of services (Mazzoni et al., 2007; Wedel and Kamakura, 2003). Segmenting markets by consumption pattern is a relatively intuitive step toward understanding customer characteristics (Weinstein, 2002). By categorising customers into usage groups, service providers can create suitable marketing strategies for each segment. The ISP s customers were categorised based on their Internet usage: light users who used the Internet for fewer than 9 hours a week, medium users who used the Internet for 9-29 hours a week and heavy users who used the Internet for more than 29 hours a week Basic segmentation analysis The results demonstrate that ISP s customers cohorts are not a homogeneous group. For instance, heavy users tend to have higher standards when evaluating service quality aspects and they comprise the highest percentage of switchers. Heavy users are most concerned with the security and privacy practices of their ISPs, whereas information support receives the highest rating from the other two user groups. Heavy users complain more often when experiencing service failures as compared to other user groups. When heavy users have problems with their service they complain to the ISP s employees, external agencies and people in their personal network, such as friends and families. On the other hand, light users tend to recommend their service provider to other people more than the other segments, which can be an advantage for ISPs in recruiting new customers. The effects of an ISP s service quality dimensions (network quality, customer service, information support, and privacy and security) on customers intention to complain, switch, and recommend among the three Internet user groups deserves the attention of an ISP s management. These findings can be utilised to design marketing strategies targeting specific market segments. For example, in order to decrease the churn rate, companies need to ensure good network quality for medium and heavy users, whereas emphasis on information support should be given to light users. Furthermore, while all customers are inclined to give recommendations on any well performed service quality aspects (except for customer service among medium users), they are more likely to 233

254 complain when they are unhappy with network quality (among light users) or are receiving inadequate information support (among heavy users). The findings of this study demonstrate that all three user groups consider Internet service performance, customer service and promotional packages as the top three motivations for switching and for recommending their current service providers. However, each of the group attaches different levels of importance to those three areas. For example, while heavy users consider a failure of Internet service performance the primary reason for changing service providers, poor Internet performance and customer service are invariably selected by medium users and light users as their rationale for switching. Similarly, while Internet service performance is the prime motivation for recommendation by heavy users, it is not surprising that customer service is the most popular reason for recommendation by medium users. Light users recommend a particular ISP mainly because of its promotional package Detailed segmentation analysis The invariance tests for different Internet usage groups reveal some interesting findings. As Internet usage increases, the influence of information support on the overall perception of service quality becomes stronger. The positive impact of information on service quality by heavy users was considerably more significant than the same influence found for light and medium users. As such, while network quality remains the dominant dimension of service quality among light and medium users, information quality and website support is the strongest determinant of an ISP s service quality for heavy users. Following the same pattern, the positive impact of commitment on attitudinal and behavioural loyalty is more significant among users with higher Internet usage. On the other hand, customer trust is not significantly related to attitudinal and behavioural loyalty of medium users. In contrast, the effect of trust on both types of loyalty is significantly higher for light users. This suggests that light users are more likely to be loyal to an ISP which they trust as compared to other groups of users Age By categorising customers into age groups, service providers can create effective marketing strategies for each segment. This study splits the sample into four groups 234

255 based on their age: years, years, years and 50 years or over Basic segmentation analysis With regards to switching experience, respondents in the age group had less switching experience compared to the other groups. The highest churn rate was among those aged over 50. However, the percentage of switchers across other age groups varies only slightly. In addition, no noticeable differences were found among age groups in terms of the reasons for their switching. Therefore, it is likely that age did not contribute to the respondents motivation for switching Detailed segmentation analysis The invariance tests for different age groups reveal some significant insights. Age shows a positive moderating effect on the relationship between customer trust and attitudinal and behavioural loyalty. Trust was a strong predictor for loyalty among older customers and was less important or even insignificant for younger age groups. Trust alone appeared to be inadequate for the younger age groups to determine their loyalty. Specifically, the results show that respondents under 39 years of age did not consider trust as a determinant of their decision to stay with a service provider or their feeling of liking or being associated with a particular ISP. Additionally, the moderating effects of age on the link between commitment and attitudinal and behavioural loyalty strengthen as age increases. Older customers appear to behave more rationally. They express intentions to associate with the ISP and engage in repeat purchase only if they feel a sense of belonging to the service provider Income By categorising customers into different income levels, service providers can create suitable marketing strategies for each segment Basic segmentation analysis There were five groups with different levels of monthly household income: less than 5,000 baht; 5,000 to 10,000 baht; 10,001 to 30,000 baht; 30,001 to 50,000 baht; 50,001 to 100,000 baht; and over 100,000 baht per month. The results demonstrate that there is a relationship between income and switching experience. Apart from the group with 235

256 household income of less than 5,000 baht, it appears that higher household income is associated with a greater likelihood to switch providers. With regard to reasons for switching, customers with higher income are more likely to choose poor Internet service performance as the main reason to switch. Customer service was also chosen as a reason to switch by a large percentage of higher income groups, as compared to the lower income ones. In addition, high price and fewer promotional packages are popular reasons for switching among groups with the lowest levels of household income. This suggests that low income groups are relatively concerned about value for money compared to higher income groups, who pay more attention to service performance and customer support Detailed segmentation analysis This study examined respondents belonging to four groups of income levels. These levels were low income, or under 30,000 baht; lower-middle income or 30,001-50,000 baht; upper-middle income, or 50, ,000 baht; and high income, or over 100,000 baht. The findings demonstrate the differences among customers with different household income levels and indicate a moderating effect of income. Customers with higher income are more likely to develop satisfaction and trust based on service quality. Noticeably the effect of satisfaction on attitudinal loyalty is not significant among the upper-middle income group, revealing that satisfaction should not be the only goal of a company seeking customer loyalty within this segment. Regardless of their income, customer commitment is a more reliable indicator of loyalty. The findings suggest that the effect of commitment on loyalty was considered significant across all income segments. This is likely due to the nexus of commitment and loyalty, hence achieving commitment is a crucial step in realising loyalty. 6.4 Contribution of this study The findings of this research study have both academic and practical implications. First, they enhance the knowledge and measurement of service quality in the home Internet services industry, which is not adequately addressed by instruments such as SERVQUAL and E-S-QUAL. Second, a robust model of customer loyalty is developed in the context of home Internet services which contributes to the current loyalty literature. Additionally, it provides empirical evidence for the relationships between 236

257 service quality, various cognitive and affective evaluation constructs and loyalty in high-tech home Internet services. Furthermore, this study reveals that various groups of customers, segmented based on usage patterns, age and income, perceived the relationships between variables in the conceptual model differently, confirming the importance of customer segmentation. In terms of practical contributions, the findings of this research study provide managerial implications which can be considered by ISPs intending to improve their overall service quality and retain existing customers. The current study also endeavours to outline necessary steps in achieving customer loyalty by improving a company s performance and understanding customers perceptions by using segmentation. This is critical, especially for the home Internet services market which is experiencing a high churn rate among its customers Academic contribution Primary research question: What are the specific service quality dimensions and attributes which influence the overall service quality of an ISP? As mentioned previously, high-tech telecommunications service quality cannot be effectively measured by SERVQUAL or E-S-QUAL because these scales fail to cover specific issues relevant to this particular context (He and Li, 2010). Hence, theoretically rooted in the service quality literature, this study proposes and tests a measurement scale for an ISP s service quality. This approach confirms the multidimensional aspect of service quality existing in the services literature (Grönroos, 1998; Parasuraman et al., 1988). Four specific service quality dimensions, identified as network quality, customer service and technical support, information quality and website information support, and security and privacy, were utilised to capture customer perceptions of an ISP s service quality. This customized scale has been created based on a thorough literature review which not only considers other researchers recommendations (Aydin and Özer, 2005; Lai et al., 2009; Roca et al., 2009), but also relates unique characteristics of an ISP s services. The four dimensions of service quality demonstrate a valuable initial step in an effort to capture the fundamentals characteristics of this important service. Noticeably, the inclusion of the unique dimensions information quality and website support, and privacy and security and their contribution to the perception of overall service quality highlights the distinctive characteristics of an ISP s service. This demonstrates that an 237

258 ISP s service quality needs to be measured using a specialised scale in order to fully capture all service dimensions. Hence, the contribution of this research study is original because it is the first of its kind to investigate the dimensions of an ISP s service quality and their effects on customer loyalty in high-tech services. It contributes to the body of knowledge relating to service quality and consumer loyalty in the home Internet services market. Secondary research question 1: What are the clear and unambiguous relationships between service quality and customers cognitive and affective evaluations in the home Internet services market? The present model is an empirical attempt aimed at capturing the key part of a business transaction. Its objective is to demonstrate causality between service quality and its outcomes. Relationships are critical to all aspects of life, especially businesses. There are many advantages to developing and sustaining relationships between ISPs and their customers. This study has identified a set of customers affective and cognitive evaluations, namely trust, satisfaction, commitment and value, initiated by their perceptions of overall service quality. Increasing customers perceptions of service quality can result in higher levels of customer trust, customer satisfaction, commitment, and finally perceived value. This finding demonstrates the benefits of investing in service quality with a view to developing long-term business-customer relationships. This knowledge enables service providers to formulate appropriate marketing strategies by focusing on the key dimensions of service quality which would work in their favour in achieving competitive advantage and long-term sustainability. Moreover, there has been little research investigating the relationships between various constructs that are antecedents to customer loyalty and overall service quality within the home Internet services market. This study is an empirical attempt to provide insights into how service quality contributes to the way customers affectively and cognitively evaluate services in the residential ISP market. 238

259 Secondary research question 2: What are the effects of customers cognitive and affective evaluations on customer loyalty in the home Internet services market? The telecommunications industry is under pressure to retain existing customers and acquire new customers (Spiller et al., 2007). Despite the importance of the Thai telecommunications market and current issues related to this industry, surprisingly few research studies have been performed on Thai home Internet services. Therefore, it is apparent that the Thai market is an under researched context. Given the high usage and increasing churn rates in this sector, it is timely to investigate the antecedents to customer loyalty and customer retention within the Thai Internet services market. This study investigates the development of loyalty with service quality as an initiator and provides empirical evidence for the relationships between various constructs service quality dimensions, overall service quality, customer cognitive and affective evaluations, and customer loyalty in a unique research context. Moreover, cognitive and affective evaluations seem to moderate the direct influence of service quality on loyalty, linking service performance with relational aspects. Essentially, the role of trust, satisfaction and commitment development is of great importance because they are directly related to the formation of a relational atmosphere for strong relationship quality leading to both types of loyalty. The weak effects of value on loyalty could be an indicator of a slightly less significant role of purely cognitive assessment in determining loyalty. In general, the model helps outline different elements affecting relationship loyalty, starting from network quality, customer service, information support and privacy and security which constitute overall service quality, and continuing onto trust, satisfaction, commitment, and value, to ultimately establish loyalty. The results confirm that service quality manifests its influence on loyalty through customer trust, satisfaction, commitment and value Practical contributions and managerial implications The following section provides practical contributions from this study which address the following three areas: the telecommunications industry, ISP service quality and customer retention. 239

260 Telecommunications industry The telecommunications industry has witnessed high switching rates. Many companies seek to build a customer base using price signals and price competition. This strategy has flaws, as attracting customers via pricing mechanisms can lead to a vulnerable and imbalanced market where smaller companies with low levels of capital cannot survive against larger corporations (Buil et al., 2013). Such competition can also diminish a firm s performance and service infrastructure. This study seeks to circumvent this problem by investigating customer loyalty determinants with a view to improving customer retention and preventing future switching behaviour. The results from this study illustrate that service quality has a significant impact on customer trust, satisfaction, commitment and value, all of which lead to customer loyalty. Moreover, by understanding the importance of the service quality dimension, the industry can focus on promoting key areas of service that need development. For example, where a business has limited resources to allocate to service quality improvements, management could concentrate on areas of service quality where ratings are consistently low across customer segments. Management can also assess which areas of service quality are important relative to others and focus on those as a means of retaining customers. By encouraging quality-based competition and a customer-focused approach, the telecommunications industry would be able to achieve sustainable development in the long run ISPs service quality This research study evaluates the service quality dimensions relating to the high-tech Internet service industry. The results emphasise the critical role of service quality and its dimensions to this industry. They also highlight the need for ISPs to dedicate resources in improving service quality dimensions, especially information quality and website support which are the main focus of overall service quality by an ISP s customers. Nowadays customers have access to information and tools in order to make effective and quick comparisons between services and competing ISPs. Hence it is crucial that companies invest in improving service quality in order to enhance positive cognitive and affective responses of their customers and eventually increase their loyalty. 240

261 Marketers would be able to focus on the key dimensions of an ISP s service quality, which in turn would lead to long-term profitability. This particular finding on an ISP s service quality dimensions would be beneficial to high-tech service providers in other countries which have similar profiles to Thailand, for example Indonesia, Vietnam, and India (Jahanzeb et al., 2011). By enhancing service quality, firms can influence customers behavioural and attitudinal loyalty, which are critical for an ISP s success. Moreover, it is vital for ISPs to obtain accurate information regarding their service quality as perceived by their customers. This information would assist them to formulate appropriate marketing strategies that could work towards achieving competitive advantage and long-term sustainability (Vlachos and Vrechopoulos, 2008) Customer retention The results illustrate that trust, satisfaction, commitment and value are significantly related to behavioural loyalty. The effects of trust, satisfaction, and commitment on attitudinal loyalty are also confirmed. By using a set of customer cognitive and affective evaluations, this study confirms the importance of perceived service quality in customer loyalty development. It suggests that, apart from improving service quality, companies can enhance customer loyalty through customer affective and cognitive evaluations of trust, satisfaction, commitment and value. However, marketers need to be cautious in formulating retention strategies. By decreasing the price of services or undertaking promotions they might increase perceived functional value, thereby encouraging repurchase behaviour. However, increased value for money does not necessarily lead to stronger attitudinal loyalty and it risks opening the door to competitors (Oliver, 1999). The segmentation analysis reveals that customers with different profiles tend to differ in their psychology and buyer behaviour. For example, heavy Internet users are more likely to exhibit greater switching behaviour, as compared to light and medium users. Network quality and customer service determines the switching behaviour of heavy users, whereas information support is the only motivation for switchers among light users. Therefore, the findings of this study can be a foundation for service providers in the home ISP sector to selectively serve the needs of different segments of customers and to create strategies to retain them. Moreover, by making customers more central in 241

262 the company s operations, these strategies can potentially reduce the expenses associated with acquiring new customers (Spiller et al., 2007) Action plans for the decision makers: improving service quality As mentioned earlier, a high level of overall service quality increases customers affective and cognitive evaluations of trust, satisfaction, commitment and value. This finding suggests that ISPs need to be proactive in improving their core services, such as Internet signal consistency and stability, quick response customer support, high level of information support quality and a high level of security and privacy, all of which contribute to overall service quality Service quality dimensions There are several recommendations which arise out of this study. First, the findings suggest that it is critical for ISPs to maintain and improve network quality. Customers generally pay considerable attention to Internet downloading and uploading speed, as well as to the stability of the Internet connection. Therefore management should work to guarantee network consistency and reliability, and satisfactory Internet speed. ISPs could employ specialists to check and maintain Internet network quality on a regular basis. Moreover, they should conduct research projects with a view to improving their Internet network quality. For example, ISPs need to enhance the efficiency of the network through improvements in the network architecture by engaging bandwidth brokers for differentiated services networks, traffic engineering and network engineering and interconnections. Also, developing fibre-optic networks enables Internet service providers to deliver faster connections to their users. Additionally, the company might want to be a first mover in network development, which would likely result in favourable perceptions towards the company, especial in this fast moving hightech sector. ISPs should concentrate on customer service and technical support. Customer service and technical support personnel should be readily available. Outlets and call centres are the two major points of interactions between the ISPs personnel and customers (Leelakulthanit and Hongcharu, 2011). These authors claim that these two contact points reflect the quality of corporate management practices. An ISP s employees are 242

263 expected to have pleasant voices, professional appearance and exhibit polite manners, as well as dedicate their attention to customers needs and concerns. Staff with appropriate skills and competencies must demonstrate sincere interest in dealing with problems or issues when handling customers' enquiries and complaints. Moreover, ISPs should provide their employees with detailed guidelines that outline measures to be taken when dealing with difficult customers in order to ensure consistency. In addition, by offering the right technology, instructions, tools, and training, a company empowers their employees to take ownership of their work, which consequently improves the overall quality of the services provided. Findings from this study highlight the need for management to establish easy access to information channels, especially to their websites, in order to provide information desired by customers. Company management should keep in mind that these channels are means of information provision and communications and should have user-friendly interfaces and reliable functions. ISPs need to provide accurate, informative, up-to-date, relevant and clear information to customers. Examples of basic information ISPs could incorporate into their communication material targeting customers include detailed service descriptions, transparent price information, company information, professional advice, technical or trouble shooting guides, research reports, contact information and hyperlinks to relevant websites (Yang et al., 2005). Additionally, ISPs should create good rapport with customers through s or mail delivering information which is useful and of interest to customers. Finally, a transparent privacy policy and secure transaction procedures can provide ISPs with strong points of differentiation. An ISP s server contains the account information of many users, which, if unsecured, might put customers personal data at risk (Rowe et al., 2011). Therefore, ISPs should protect customers from data leaks and clearly inform customers on how their personal information is handled, used and stored. ISPs can also enhance the security of their network. They are able to observe and monitor traffic flow through their networks, detect suspicious traffic spikes and either stop malicious traffic or give timely warnings to customers (Rowe et al., 2011). A study in 2004 reports that 66% of consumers would switch to competing ISPs if they offered more secured Internet services (Streamshield, 2004). Therefore, it can be concluded that protecting 243

264 both a customer s private details and the network at large can be beneficial to perceptions of an ISP s integrity Customer segmentation This study suggests that ISPs should delve deeper into customer segmentation in order to effectively and efficiently understand the market and develop appropriate marketing strategies for different groups of customers. In particular, they should distinguish heavy Internet users from other segments and offer different service packages for these segments. This will enhance service offerings and maximise the use of organisational resources. For example, information packages could be designed to suit customers with different levels of Internet usage and knowledge, such as novices, intermediate and expert. Communications material should be tailored to suit each of the identified segment on the basis of their usage and knowledge characteristics. Similarly, customer care could be varied by offering customers alternatives, for instance, face-to-face consultation and online support. Follow-up calls and s are also a good idea to improve customer experience with an ISP s services. These would eventually result in significant long-term profitability. Information support was the most significant service quality dimension among heavy Internet users. Therefore, it is advisable for ISPs to focus on the design of the information support platform in both online and offline environments to enhance perceived service quality for heavy users. It is suggested that heavy Internet users tend to be less comfortable communicating and establishing relationships offline (Thayer and Ray, 2006). Therefore, as well as face-to-face consultation, online support could be an appropriate add-on for heavy users. While being convenient, online interactions may be also favourable for this group of Internet users as they tend to spend considerable time on the Internet and are most likely to prefer Internet-related activities to non-internet ones (Assael, 2005). Network quality is the most significant determinant of overall service quality among light and medium users, followed by security and privacy. Hence, in addition to clarifying the transparency and reliability of the company s security and privacy practices, an ISP could also promote high network quality via their communications 244

265 with light and medium users, for example in their brochures and flyers. Ensuring satisfactory network quality helps to prevent heavy and medium users from switching providers or complaining. In addition, with regards to information support, ISPs could provide information packages tailored to the needs and wants of customers for these Internet usage groups. Light users are usually novices in the online environment and therefore are likely to prefer simple information over complicated descriptions. Hence, any communications material intended for this segment should be specialised to target beginners. Medium users are expected to have better knowledge of the Internet than light users, albeit they are less knowledgeable than heavy users. Therefore, an intermediate level of information is considered appropriate for this segment. ISPs should also keep in mind that customers in different age and income groups have differing attitudes and behaviour. For example, a reliable reputation alone cannot motivate younger customers (those under 39 years) to develop loyalty towards an ISP. Furthermore, satisfied customers who belong to the upper-middle income group (those earning between 50, ,000 baht) do not necessarily exhibit attitudinal loyalty. As a result, the common goal of achieving satisfaction might not be effective when targeting this group of customers. This, therefore, signals that ISPs should focus on giving due importance to customer segmentation in order to effectively and efficiently target their preferred market. In other words, marketers of ISPs should take customers' personal information into account when examining their attitudes and behaviours. Such information can assist ISPs better segment customers and create suitable marketing strategies. Also, although the relationships between variables vary across the different segments, ISPs should endeavor to prioritise delivering optimal services to these segments. It is necessary to avoid perceived prejudice by customers, which may affect the perception of the firm s service quality, their affective and cognitive evaluations and, subsequently their loyalty. In general it is critical for ISPs to make customers central in their daily operations as a means of achieving a fair competitive advantage. 6.5 Limitations of this study and future research directions There are several limitations of this study. Firstly, the proposed conceptual model in this study was empirically tested using perceptions of a major ISP s customers in Thailand. It is probable that the results of the study would be different if based on perceptions of 245

266 customers in other service industries or customers in other countries. In the interest of generalisation, future research should be conducted to test this model in other contexts in countries, such as Vietnam, Cambodia and Burma. Second, this study did not examine the interactions between the affective and cognitive evaluation constructs of customer trust, satisfaction, commitment, and value. An investigation on a wide range of mediation effects on various relationships could generate more insights. Third, the model in this study only includes satisfaction, value, trust and commitment as customers cognitive and affective responses. An investigation into other dimensions, such as switching barriers, might be useful. Furthermore, this study segmented ISP customers based on Internet usage, income level and age groups. It may be desirable to segment customers based on other characteristics, such as location or lifestyle, in order to understand fully the importance of segmentation in the ISP market. Finally, a longitudinal study of ISP customers might be worthwhile to capture the changing patterns of customers psychological and behavioural attitudes over time. 6.6 Concluding remarks The current research develops an understanding about consumer buyer behaviour of home Internet services in Thailand and creates an ideal model to capture service quality and customer loyalty development. This study is original in that it is the first of its kind to investigate the dimensions of an ISP s service quality, the effects of service quality on customer affective and cognitive evaluations and customer loyalty in high-tech services. It provides valuable insights into consumer retention and customer loyalty in the Thai home Internet services. Beneficiaries of this study include various stakeholders, such as an ISP s consumers, ISPs themselves, the government, and other commercial interests. This study provides both theoretical and practical implications. Overall service quality is widely considered one of the key factors in determining customer attitudinal loyalty and customer retention. This study has developed a model to evaluate and measure the service quality of ISPs. The finding that an ISP s four service quality dimensions influences its overall service quality supports the notion that service quality is a multidimensional construct. The service quality dimensions are identified as network quality, customer service and technical support, information quality and website information 246

267 support and security and privacy. Practically, ISPs need to consider other factors apart from network quality, for example information and website support, which emerged as the strongest influencer of overall service quality. This study has identified the differences in service quality dimensions relating to an ISP, which results in a better understanding of service quality. Furthermore, it investigates the effects of service quality on affective and cognitive evaluations of customers and their loyalty towards high-tech service providers. By enhancing service quality, firms can influence customer trust, satisfaction, commitment and value, which in turn influences loyalty and encourages repurchase. This knowledge enables service providers to formulate appropriate marketing strategies by focusing on the key dimensions of service quality in order to achieve competitive advantage and long term sustainability. The results also demonstrate that various groups of customers, who were segmented based on Internet usage patterns, household income and age, exhibit different perceptions, attitudes and behaviour, confirming the importance of customer segmentation. In other words, Internet customers are not homogeneous. Managers of ISPs should be aware that, although the concept of service quality is multidimensional, not all dimensions contribute equally to perceptions of overall service quality. The dimensions of service quality have varying influences on service quality across different groups of customers. Understanding relevant customer characteristics is critical to retaining customers. Hence, ISPs should address the issues that are appropriate for specific customer segments in order to maximise the use of its resources and become more market oriented. Practical implications that can be drawn from this research will form a foundation for service providers in the home Internet services sector to develop new retention strategies. An action plan has been suggested. By increasing customer retention rates, the company can potentially reduce the need to recruit new customers and reduce its expenditure. By using the findings of this study, the ISP industry will be able to deal with current issues relating to customer switching in the home Internet services market in Thailand and in other countries. In general, this study contributes towards the service quality and loyalty literature, and provides managerial implications with regards to the sustainable development of ISPs in Thailand and elsewhere. The findings of this study 247

268 would benefit ISPs in countries with similar market characteristics to Thailand, for example Malaysia, Vietnam, the Philippines and India. In summary, the proposed research develops an understanding of consumer buyer behaviour in the Thai home Internet services and creates an ideal model to enhance customer loyalty. It provides valuable insights into consumer retention and loyalty in the home Internet services market. This research contributes a new body of knowledge with regards to the future potential of the ISPs in Thailand and for other ISPs in developing countries. Beneficiaries of this study include various stakeholders in Thailand, such as consumers of ISPs, ISPs themselves, the government, and other commercial interests. Practical implications that can be drawn from this research will form a foundation for service providers in the home ISP to develop new retention strategies. By being more customer-oriented in company operations, these strategies can possibly not only increase customer retention rates, but also relieve the financial pressure of acquiring new customers. 248

269 6.7 Chapter summary This thesis has been developed based on existing knowledge from the areas of marketing and psychology, including services marketing and service quality literature, with a view to providing both theoretical contributions and practical implications. The topic of customer loyalty is of interest as it is essential for all businesses wishing to retain existing customers. This thesis aims to address three objectives, namely: (1) Identify the specific service quality dimensions and attributes which influence the overall service quality of an ISP; (2) Identify the clear and unambiguous relationships between service quality and customers cognitive and affective evaluations in the home Internet services market; (3) Identify the effects of customers cognitive and affective evaluations on customer loyalty in the home Internet services market. In this chapter the results from the data analysis were incorporated and a detailed discussion on the relationships between various constructs was provided. In addition, theoretical and managerial contributions were also presented. This chapter started with a discussion of the results from the data analysis (section 6.2). The model of ISP service quality was reviewed and the relationship between the drivers of customer loyalty and service quality in the context of home Internet services was discussed. This was followed by a discussion of findings related to the segmentation analysis (section 6.3). Subsequently, the overall theoretical and managerial implications of the research were discussed (section 6.4). The chapter concluded with the limitations of the study and directions for future research (section 6.5) and concluding remarks (section 6.6). 249

270 References Aaron, G. and Rasmussen, R. (2013), "Global Phishing Survey: Trends and Domain Name Use in 2H2012", APWG, available at: (accessed 1st August 2013). Abdolvand, N., Charkari, M. and Mohammadi, R. (2006), "Technical model for improving customer loyalty with m-commerce: mobile service providers", International Journal on Computer Science and Information Systems, Vol. 1 No. 2, pp Abdullah, F. (2006), "Measuring service quality in higher education: HEdPERF versus SERVPERF", Marketing Intelligence & Planning, Vol. 24 No. 1, pp ACMA (2012), "Internet use and social networking by young people", Australian Communications and Media Authority, available at: lib310665/no1_internet_use_social_networking_young_people.pdf. (accessed 1/04/2014. Ahn, J., Han, S. and Lee, Y. (2006), "Customer churn analysis: Churn determinants and mediation effects of partial defection in the Korean mobile telecommunications service industry", Telecommunications Policy, Vol. 30, pp Al-Hawari, M. (2011), "Automated service quality as a predictor of customers commitment A practical study within the UAE retail banking context", Asia Pacific Journal of Marketing and Logistics, Vol. 23 No. 3, pp Alessandro, S., Girardi, A. and Tiangsoongnern, L. (2012), "Perceived risk and trust as antecedents of online purchasing behavior in the USA gemstone industry", Asia Pacific Journal of Marketing and Logistics, Vol. 24 No. 3, pp Allen, P. J. and Bennett, K. (2010), PASW statistics by SPSS: a practical guide : version 18.0, Cengage Learning, South Melbourne, Vic. Alshurideh, M. (2010), "Customer Service Retention-A Behavioural Perspective of the UK Mobile Market", paper presented to Doctoral dissertation, Durham University. 250

271 Álvarez, L. S., Casielles, R. V. and Martín, A. M. D. (2010), "Analysis of the role of complaint management in the context of relationship marketing", Journal of Marketing Management, Vol. 27 No. 1/2, pp Amin, S. M., Ahmad, U. N. U. and Hui, L. S. (2012), "Factors Contributing to Customer Loyalty Towards Telecommunication Service Provider", Procedia- Social and Behavioral Sciences, Vol. 40, pp Anderson, E., Fornell, C. and Lehmann, R. (1994), "Customer satisfaction, market share, and profitability: Findings from Sweden", Journal of Marketing, Vol. 58 No. 1, pp Anderson, E. W. and Sullivan, M. W. (1993), "The antecedents and consequences of customer satisfaction for firms", Marketing Science, Vol. 12 No. 2, pp Anderson, J. and Gerbing, W. (1988), "Structural equation modelling in practice: A review and recommended two stage approach", Psychological Bulletin, Vol. 27 No. 1, pp Andronikidis, A. (2009), "Linking dimensions of perceived service quality to actual purchase behaviour", EuroMed Journal of Business, Vol. 4 No. 1, pp Arbuckle, J. L. (1996), AMOS users' guide. Version 3.6, Smallwaters, Chicago. Ariely, D. (2008), Predictably Irrational: The Hidden Forces That Shape Our Decisions, HarperCollins Publishers, London. Asaari, M. and Karia, N. (2003), "Business Strategy: Customer Satisfaction Among Cellular Providers in Malaysia", paper presented to The European Applied Business Research Conference, Venice, Italy. Asmussen, B., Harridge-March, S., Occhiocupo, N. and Farquhar, J. (2013), "The multilayered nature of the internet-based democratization of brand management", Journal of Business Research, Vol. 66, pp Assael, H. (2005), "A demographic and psychographic profile of heavy internet users and users by type of internet usage ", Journal of advertising research, Vol. 45 No. 1, pp Asubonteng, P., McCleary, K. J. and Swan, J. E. (1996), "SERVQUAL revisited: a critical review of service quality", Journal of Services Marketing, Vol. 10 No. 6, pp

272 Aydin, S. and Özer, G. (2005), "The analysis of antecedents of customer loyalty in the Turkish mobile telecommunication market", European Journal of Marketing, Vol. 39 No. 7/8, pp Bagozzi, R. P. (1981), "Attitudes, intentions, and behavior: A test of some key hypotheses", Journal of Personality and Social Psychology, Vol. 41 No. 4, pp Bagozzi, R. P. (1992), "The self-regulation of attitudes, intentions, and behaviour", Social Psychology Quarterly, Vol. 55 No. 2, pp Bai, B., Law, R. and Wen, I. (2008), "The impact of website quality on customer satisfaction and purchase intentions: Evidence from Chinese online visitors", International Journal of Hospitality Management, Vol. 27 No. 3, pp Baldauf, A., Reisinger, H. and Moncrief, W. C. (1999), "Examining motivations to refuse in industrial mail surveys", Journal of the Market Research Society, Vol. 41 No. 3, pp Baldinger, A. L. and Rubinson, J. (1996), "Brand loyalty: the link between attitude and behavior", Journal of advertising research, Vol. 36, pp Bateson, J. E. (1995), Managing Services Marketing, Dryden Press, Forth Worth, Texas. Bayraktar, E., Tatoglu, E., Turkyilmaz, A., Delen, D. and Zaim, S. (2011), "Measuring the efficiency of customer satisfaction and loyalty for mobile phone brands with DEA", Expert Systems with Applications, Vol. 39, pp Bell, R. (2009), An Exploratory Research on Perceived Value in Professional Business Services, Australian Catholic University, Australia. Bentler, P. M. and Bonett, D. G. (1990), "Comparative fit indices in structural models", Psychological Bulletin, Vol. 107 No. 2, pp Bennett, R. (1996), "Relationship Formation and Governance in Consumer Markets: Transactional Analysis versus the Behavioural Approach", Journal of Marketing Management, Vol. 12 No. 5, pp Bennett, R. (2001), "Back to the future: Lessons for the future of brand loyalty reserch from the past. In: M.Gardiner", paper presented to MRSA Conference 2001, Brisbane. Berry, L. L. (2000), "Cultivating service brand equity", Journal of the Academy of Marketing Science, Vol. 28 No. 1, pp

273 Berry, L. L. and Parasuraman, A. (1997), "Listening to the customer-the concept of a service-quality information system", Sloan management review, Vol. 38 No. 3, pp Berry, L. L., Zeithaml, V. A. and Parasuraman, A. (1985), "Quality counts in services, too", Business horizons, Vol. 28 No. 3, pp Biczók, G., Kardos, S. and Trinh, T. A. (2010), "Pricing Internet access in the presence of user loyalty", NETNOMICS: Economic Research and Electronic Networking, Vol. 11 No. 2, pp Black, T. R. (1999), Doing quantitative research in the social sciences: An integrated approach to research design, measurement, and statistics, SAGE Publications, Thousand Oaks, CA. Blery, E., Batistatos, N., Papastratou, E., Perifanos, I., Remoundaki, G. and Retsina, M. (2009), "Service quality and customer retention in mobile telephony", Journal of Targeting, Measurement and Analysis for Marketing, Vol. 17 No. 1, pp Bloemer, J., Ruyter, K. and Wetzels, M. (1999), "Linking perceived service quality and service loyalty: a multi-dimensional perspective", European Journal of Marketing, Vol. 33 No. 11/12, pp Bodey, K. and Grace, D. (2006), "Segmenting service complainers and noncomplainers on the basis of consumer characteristics", Journal of Services Marketing, Vol. 20 No. 3, pp Bollen, K. A. (1989), Structural Equations with Latent Variables, Wiley, New York. Bolton, R. and Drew, J. (1991), "A longitudinal analysis of the impact of service changes on customer attitudes", Journal of Marketing, Vol. 55, pp Boulding, W., Kalra, A., Staelin, R. and Zeithaml, V. (1993), "A dynamic process model of service quality: form expectations to behavioral intentions", Journal of Marketing Research, Vol. 30, pp Bowling, A. (2005), "Just one question: If one question works, why ask several?", Journal of epidemiology and community health, Vol. 59 No. 5, pp Bowen, T. and Chen, L. (2001), "The relationshio between customer loyalty and customer satisfaction", International Journal of Contemporary Hospitality Management, Vol. 13 No. 4, pp Bowen, T. and Shoemaker, S. (1998), "Loyalty: A strategic commitment", Cornell Hotel and Restaurant Administration Quarterly, Vol. 39, pp

274 Brady, M. and Cronin, J. (2001), "Customer Orientation Effects on Customer Service Perceptions and Outcome Behaviors", Journal of Service Research, Vol. 3 No. 3, pp Brady, M., Cronin, J. and Brand, R. (2002), "Performance-only measurement of service quality: a replication and extension", Journal of Business Research, Vol. 55, pp Brohman, K., Piccoli, G., Martin, P., Zulkernine, F., Parasuraman, A. and Watson, A. (2009), "A Design Theory Approach to Building Strategic Network-Based Customer Service Systems", Decision Sciences, Vol. 40 No. 3, pp Brown, T. J., Churchill, G. A. and Peter, J. P. (1993), "Improving the measurement of service quality", Journal of Retailing, Vol. 69 No. 1, pp Bügel, M., Buunk, A. and Verhoef, P. (2010), "A comparison of customer commitment in five sectors using the psychological investment model", Journal of Relationship Marketing, Vol. 9, pp Buil, I., de Chernatony, L. and Martínez, E. (2013), "Examining the role of advertising and sales promotions in brand equity creation", Journal of Business Research, Vol. 66, pp Butt, M. M. and de Run, E. C. (2009), "Modeling customer satisfaction in cellular phone services", Jurnal Kemanusiaan bil, Vol. 13. Byrne, B. M. (2001), Structural equation modeling with AMOS: Basic con- cepts, applications and programming, Erlbaum, Mahwah, NJ. Caceres, R. and Paparoidamis, N. G. (2007), "Service quality, relationship satisfaction, trust, commitment and business-to-business loyalty", European Journal of Marketing, Vol. 41 No. 7/8, pp Cacioppo, J. T., Petty, R. E., Feinstein, J. A. and Jarvis, W. B. G. (1996), "Dispositional differences in cognitive motivation: The life and times of individuals varying in need for cognition", Psychological Bulletin, Vol. 119, pp Cameran, M., Moizer, P. and Pettinicchio, A. (2010), "Customer Satisfaction, Corporate Image, and Service Quality in Professional Services", The Service Industries Journal, Vol. 30 No. 3, pp Canana, R. T., Delahaye, B. L. and Sekaran, U. (2001), Applied business research: Qualitative and quantitative methods, John Wiley & Sons, Brisbane: Australia. 254

275 Cardoso, P. R., Costa, H. S. and Novais, L. A. (2010), "Fashion consumer profiles in the Portuguese market: involvement, innovativeness, self-expression and impulsiveness as segmentation criteria", International Journal of Consumer Studies, Vol. 34, pp Carlson, J. and O Cass, A. (2011), "Developing a framework for understanding e- service quality, its antecedents, consequences, and mediators", Managing Service Quality, Vol. 21 No. 3, pp Carman, J. M. (1990), "Consumer perceptions of service quality: an assessment of the SERVQUAL dimensions", Journal of Retailing, Vol. 66 No. 1, pp Carpenter, J. M. (2008), "Consumer shopping value, satisfaction and loyalty in discount retailing", Journal of Retailing and Consumer Services, Vol. 15 No. 5, pp Carraher, S., Welsh, D. and Cash, R. (2010), "An examination of Chinese entrepreneurs and intrapreneurs: High tech or high touch customer service? ", paper presented to The 2010 Small Business Institute (SBI) Conference, Albuquerque, NM. Carrillat, F. A., Jaramillo, F. and Mulki, J. P. (2007), "The validity of the SERVQUAL and SERVPERF scales: a meta-analytic view of 17 years of research across five continents", International Journal of Service Industry Management, Vol. 18 No. 5, pp Caruana, A. and Ewing, M. T. (2010), "How corporate reputation, quality, and value influence online loyalty", Journal of Business Research, Vol. 63, pp Castañeda, J. (2001), "Relationship between customer satisfaction and loyalty on the internet", Journal of Business and Psychology, Vol. 26 No. 3, pp Castañeda, J. and Montoro, F. (2007), "The effect of Internet general privacy concern on customer behaviour", Electronic Commerce Research, Vol. 7, pp Cater, B. and Zabkar, V. (2009), "Antecedents and consequences of commitment in marketing research services: the client s perspective", Industrial Marketing Management, Vol. 38 No. 7, pp Chae, M., Kim, J., Kim, H. and Ryu, H. (2002), "Information quality for mobile internet services: a theoretical model with empirical validation", Electronic Markets, Vol. 12 No. 1, pp Chang, H. and Chen, S. (2009), "Consumer perception of interface quality, security, and loyalty in electronic commerce", Information & Management, Vol. 46 No. 7, pp

276 Chaudhari, A. and Holbrook, B. (2001), "The chain of effects from brand trust and brand affect to brand performance: The role of brand loyalty", Journal of Marketing, Vol. 65 No. 2, pp Chen, C. and Chen, F. S. (2010), "Experience quality, perceived value, satisfaction and behavioral intentions for heritage tourists", Tourism Management, Vol. 31 No. 1, pp Chen, C. and Green, D. (2009), "Marketing communications and Branding: Competitive Hypermarket Strategies", International Journal of Management and Marketing Research, Vol. 2 No. 1, pp Chen, L. D., Gillenson, M. L. and Sherrell, D. L. (2002), "Enticing online consumers: an extended technology acceptance perspective", Information & Management, Vol. 39 No. 8, pp Chen, T. and Hu, H. H. (2010), "The effect of relational benefits on perceived value in relation to customer loyalty: An empirical study in the Australian coffee outlets industry", International Journal of Hospitality Management, Vol. 39 No. 3, pp Chen, Z. and Dubinsky, A. (2003), "A conceptual model of perceived customer value in e-commerce: A preliminary investigation", Psychology & Marketing, Vol. 20 No. 1, pp Cheng, T., Lai, L. and Yeung, A. (2008), "The Driving Forces of Customer Loyalty: A Study of Internet Service Providers in Hong Kong", International Journal of E- Business Research, Vol. 4 No. 4, pp Chiou, S. (2004), "The antecedents of consumers loyalty toward Internet Service Providers", Information & Management, Vol. 41, pp Chiou, S. and Droge, C. (2006), "Service quality, trust, specific asset investment, and expertise: direct and indirect effects in a satisfaction-loyalty framework", Journal of the Academy of Marketing Science, Vol. 34 No. 4, pp Chiu, H. C., Hsieh, Y. C. and Kao, C. Y. (2005), "Website quality and customer s behavioural intention: An exploratory study of the role of information asymmetry, " Total Quality Management and Business Excellence, Vol. 16 No. 2, pp Choi, K. S., Cho, W. H., Lee, S., Lee, H. and Kim, C. (2004), "The relationships among quality, value, satisfaction and behavioral intention in health care provider choice: 256

277 A South Korean study", Journal of Business Research, Vol. 57 No. 8, pp Christodoulides, G. and Michaelidou, N. (2010), "Shopping motives as antecedents of e-satisfaction and e-loyalty", Journal of Marketing Management, Vol. 27 No. 1/2, pp Christopher, M., Payne, A. and Ballantyne, D. (1994), Relationship marketing-bringing quality, customer service and marketing together, Butterworth-Heinemann Ltd., Oxford. Chun, S. and Hahn, M. (2007), "Network externality and future usage of Internet services", Internet Research, Vol. 17 No. 2, pp Churchill, N. C. (1992), Research issues in entrepreneurship in D.L. Sexton & J.D. Kasarda eds The State of the Art of Entrepreneurship, PWS-Kent Publishing, Boston, Massachusetts. CIA (2013), "The world factbook", Central Intelligence Agency, available at: (accessed 1st August 2013). Clow, K. E., Fischer, A. K. and O'Bryan, D. (1995), "Patient expectations of dental services. Image affects expectations, and expectations affect perceived service quality", Journal of Health Care Marketing, Vol. 15 No. 3, pp. 23. Cohen, J. (1988), Statistical power analysis for the behavioral sciences, Lawrence Erlbaum Associates, Hillsdale: New Jersey. Cooper, A. R., Cooper, M. B. and Duhan, D. F. (1989), "Measurement Instrument Development Using Two Competing Concepts of Customer Satisfaction", Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior, Vol. 2, pp Corstjens, M. and Lal, R. (2000), "Building store loyalty through store brands", Journal of Marketing Research, Vol. 37 No. 3, pp Cronbach, L. J. (1951), "Coefficient alpha and the internal structure of tests", Psychometrika, Vol. 22 No. 3, pp Cronin, J. J., Brady, K. and Hult, M. (2000), "Assessing the effects of quality, value, and customer satisfaction on consumer behavioural intentions in service environments", Journal of Retailing, Vol. 76 No. 2, pp

278 Cronin, J. J. and Taylor, S. A. (1992), "Measuring service quality: a reexamination and extension", The Journal of Marketing, pp Cronin, J. J. and Taylor, S. A. (1994), "SERVPERF versus SERVQUAL: reconciling performance-based and perceptions-minus-expectations measurement of service quality", Journal of Marketing, Vol. 58 No. 1, pp Crosby, A., Evans, R. and Cowles, D. (1990), "Relationship quality in services selling: An interpersonal influence perspective", Journal of Marketing, Vol. 54, pp Crosby, P. B. (1979), Quality is free: The art of marketing quality certain, New American Library, New York. Cudeck, R. and Browne, M. W. (1992), "Constructing a covariance matrix that yields a specified minimizer and a specified minimum discrepancy function value", Psychometrika, Vol. 57 No. 3, pp Cui, Y., Trent, E. S., Sullivan, P. M. and Matiru, G. N. (2003), "Cause-related marketing: How generation Y responds", International Journal of Retail & Distribution Management, Vol. 31 No. 6, pp Culnan, M. J. and Armstrong, P. K. (1999), "Information privacy concerns, procedural fairness, and impersonal trust: An empirical investigation", Organization Science, Vol. 10 No. 1, pp Cunningham, E. (2008), A practical guide to Structural Equation Modelling using AMOS, Statsline, Melbourne. Cunningham, E. (2010), A practical guide to structural equation modelling using Amos, 1 edn, Statsline, Melbourne. Cunningham, M. (1956), "Brand loyalty - what where how much?", Harvard Business Review, Vol. 34, pp Cunningham, M. (1967), Perceived Risk and Brand Loyalty in Risk Taking and Information Handling in Consumer Behavior,, Harvar University Press, Cambridge, MA. Czepiel, J. A., Solomon, M. R., Surprenant, C. F. and Gutman, E. G. (1985), "Service encounters: an overview", The service encounter, pp Dadzie, K. Q., Chelariu, C. and Winston, E. (2005), "Customer service in the internetenabled logistics supply chain: website design antecedents and loyalty effects", Journal of Business Logistics, Vol. 26 No. 1, pp

279 Daughtrey, C. L., Vowles, N. S. and Black, G. S. (2013), "The Effects of Membership and Demographics on Consumer Satisfaction and Loyalty in Service Organizations", Services Marketing Quarterly, Vol. 34 No. 4, pp Day, G. S. (1969), "A two-dimensional concept of brand loyalty", Journal of advertising research, Vol. 9, pp De Vaus, D. A. (1995), "Surveys in Social Research", Contemporary Social Research Series, Vol. 11. Dekimpe, M. G., Steenkamp, J. B. E., Mellens, M. and Vanden Abeele, P. (1997), "Decline and variability in brand loyalty", International Journal of Research in Marketing, Vol. 14 No. 5, pp Deng, Z., Lu, Y., Wei, K. and Zhang, J. (2010), "Understanding customer satisfaction and loyalty: An empirical study of mobile instant messages in China", International Journal of Information Management, Vol. 30, pp DeVellis, R. F. (2003), Scale development: Theory and applications, 2 edn, Sage Publications, Inc., California. Díaz, E., Martín-Consuegra, D. and Esteban, Á. (2011), "Consumer loyalty in mobile telephony", African Journal of Business Management, Vol. 5 No. 33, pp Dick, S. and Basu, K. (1994), "Customer loyalty: Toward an integrated conceptual framework", Journal of the Academy of Marketing Science, Vol. 22 No. 2, pp Dillman, D. A. (1978), Mail and telephone surveys, vol. 3, Wiley, New York. Dodds, W. B., Monroe, K. B. and Grewal, D. (1991), "Effect of price, brand and store information on buyers product evaluations", Journal of Marketing Research, Vol. 28 No. 3, pp Doney, M. and Cannon, P. (1997), "An examination of the nature of trust in the buyerseller relationship", Journal of Marketing, Vol. 61, pp Dowling, G. (2002), "In B2C Markets, Often Less Is More", California Management Review, Vol. 44 No. 3, pp. 87. Durvasula, S., Lysonski, S., Mehta, S. C. and Tang, B. P. (2004), "Forging relationships with services: The antecedents that have an impact on behavioural outcomes in the life insurance industry", Journal of Financial Services Marketing, Vol. 8 No. 4, pp

280 Dwivedi, Y. K., Papazafeiropoulou, A., Brinkman, W. P. and Lal, B. (2010), "Examining the influence of service quality and secondary influence on the behavioural intention to change internet service provider", Information Systems Frontiers, Vol. 12 No. 2, pp Dwyer, R., Schurr, H. and Oh, S. (1987), "Developing buyer-seller relationships", Journal of Marketing, Vol. 51, pp East, R. (1997), Consumer behaviour: Advances and applications in marketing Prentice Hall, London. East, R. (2000), Fact, Fiction and Fallacy in Retention Marketing, Kingston Business School, United Kingdom, London. East, R., Gendall, P., Hammond, K. and Lomax, W. (2005), "Consumer loyalty: singular, additive or interactive?", Australasian Marketing Journal, Vol. 13 No. 2, pp Edvardsson, B., Gustafsson, A. and Roos, I. (2005), "Service portraits in service research: a critical review", International Journal of Service Industry Management, Vol. 16 No. 1, pp Eisingerich, A. and Rubera, G. (2010), "Drivers of Brand Commitment: A Cross- National Investigation", Journal of International Marketing, Vol. 18 No. 2, pp El-Manstrly, D. and Harrison, T. (2013), "A critical examination of service loyalty measures", Journal of Marketing Management, / X Elliot, S., Li, G. and Choi, C. (2013), "Understanding service quality in a virtual travel community environment", Journal of Business Research, Vol. 66, pp Erevelles, S., Srinivasan, S. and Rangel, S. (2003), "Consumer Satisfaction of Internet Service Providers: A Process Based Approach", Journal of Information Technology and Management, Vol. 4 No. 2, pp Eshghi, A., Haughton, D. and Topi, H. (2007), "Determinants of customer loyalty in the wireless telecommunications industry", Telecommunications Policy, Vol. 31 No. 2, pp ETDA (2013), "Thailand Internet User Profile 2013", Electronic Transactions Development Agency, available at: 13.pdf 15/05/2014). 260

281 Evanschitzky, H., Iyer, G. R., Plassmann, H., Niessing, J. and Meffert, H. (2006), "The relative strength of affective commitment in securing loyalty in service relationships", Journal of Business Research, Vol. 59 No. 12, pp Farrell, D. and Rusbult, C. E. (1981), "Exchange variables as predictors of job satisfaction, job commitment, and turnover: The impact of rewards, costs, alternatives, and investments", Organizational Behavior and Human Performance, Vol. 27, pp Faulkner, M. E. (1998), Business Orientation Scale Development and Psychometric Assessment, School of Marketing, University of South Australia, Adelaide. Feeny, A. and Vongpatanasin, T. (1996), "Retailing in Thailand", International Journal of Retail & Distribution Management, Vol. 24 No. 8, pp Fields, A. (2005), Discovering statistics using SPSS, Sage Publications, Beverly Hills. Filzmoser, P., Garrett, R. G. and Reimann, C. (2005), "Multivariate outlier detection in exploration geochemistry", Computers & Geosciences, Vol. 31 No. 5, pp Fishbein, M. and Ajzen, I. (1975), Belief, Attitude, Intention, and Behavior: An Introduction to Theory and Research, Addison-Wesley, MA. Fisk, R. P., Brown, S. W. and Bitner, M. J. (1993), "Tracking the evolution of the services marketing literature", Journal of Retailing, Vol. 69 No. 1, pp Flint, J., Blocker, C. and Boutin, P. (2011), "Customer value anticipation, customer satisfaction and loyalty: An empirical examination", Industrial Marketing Management, Vol. 40, pp Fornell, C. (1992), "A national customer satisfaction barometer: The Swedish experience", Journal of Marketing, Vol. 56, pp Fornell, C., Johnson, D., Anderson, W., Cha, J. and Bryant, E. (1996), "The American customer satisfaction index: Nature, purpose, and findings", Journal of Marketing, Vol. 60 No. 4, pp Fornell, C. and Larcker, D. F. (1981), "Evaluating structural equation models with unobservable variables and measurement error", Journal of Marketing Research, Vol. 18 No. 1, pp Frow, P., Payne, A., Wilkinson, I. F. and Young, L. (2011), "Customer management and CRM: addressing the dark side", Journal of Services Marketing, Vol. 25 No. 2, pp

282 Fullerton, G. (2005), "How commitment both enables and undermines marketing relationships", European Journal of Marketing, Vol. 39 No. 11, pp Fullerton, G. (2011), "Creating advocates: the roles of satisfaction, trust and commitment", Journal of Retailing and Consumer Services, Vol. 18, pp Fulps, L. (2013), "College students heavy Internet use shares symptoms of addiction", Missouri S&T, available at: 3/02/2014). Furrer, O. and Sollberger, P. (2007), "The dynamics and evolution of the service marketing literature: ", Service Business, Vol. 1 No. 2, pp Fusilier, M. R. and Simpson, P. M. (1955), "AIDS patients perceptions of nursing care quality", Journal of Health Care Marketing, Vol. 15 No. 1, pp Ganesan, S. (1994), "Determinants of long-term orientation in buyer-seller relationships", Journal of Marketing, Vol. 58 No. 1, pp Ganguli, S. and Roy, S. K. (2010), "Service quality dimensions of hybrid services", Managing Service Quality, Vol. 20 No. 5, pp Garson, G. D. (2009), "Factor Analysis", NCSU, available at: 10 April 2010). Gay, L., Mills, G. and Airasian, P. (2006), Educational research: Competencies for analysis and application 8edn, Prentice Hall, New York. Gephart, R. (1999), "Paradigms and research methods", Research Methods Forum, available at: November 13, 2005). Gerpott, T. J., Rams, W. and Schindler, A. (2001), "Customer retention, loyalty and satisfaction in the German mobile cellular telecommunications market", Telecommunications Policy, Vol. 25 No. 4, pp Ghobadian, A., Speller, S. and Jones, M. (1994), "Service Quality: Concepts and Models", International Journal of Quality & Reliability Management, Vol. 1 No. 9, pp Glasman, L. R. and Albarracín, D. (2006), "Forming attitudes that predict future behavior: a meta-analysis of the attitude-behavior relation", Psychological Bulletin, Vol. 132 No. 5, pp Gounaris, S. and Venetis, K. (2002), "Trust in industrial service relationships: behavioral consequences, antecedents and the moderating effect of the duration of the relationship", Journal of Services Marketing, Vol. 16 No. 7, pp

283 Green, P. E., Tull, D. S. and Albaum, G. (1988), Research for marketing decisions, Prentice-Hall, Englewood Cliffs, NJ. Grewal, D., Levy, M. and Lehmann, D. R. (2004), "Retail Branding and Customer Loyalty: An Overview", Journal of Retailing, Vol. 80 No. 4, pp. ix-xii. Grigoroudis, E., Litos, C., Moustakis, V., Politis, Y. and Tsironis, L. (2008), "The assessment of user-perceived web quality: application of a satisfaction benchmarking approach", European Journal of Operational Research, Vol. 187 No. 3, pp Grönroos, C. (1984), "A service quality model and its marketing implications", European Journal of Marketing, Vol. 18 No. 4, pp Grönroos, C. (1994), "From scientific management to service management: a management perspective for the age of service competition", International Journal of Service Industry Management, Vol. 5 No. 1, pp Grönroos, C. (1998), "Marketing services: the case of a missing product", Journal of business & industrial marketing, Vol. 13 No. 4/5, pp Grönroos, C. (2000), "Creating a relationship dialogue: communication, interaction and value", The marketing review, Vol. 1 No. 1, pp Grossmanova, M., Vojtkova, M. and Kita, P. (2009), "Role of Brand Loyalty in Purchase of Sports Shoes.", Vezetestudomany, Vol. 6, pp Gruen, T. W., Summers, J. O. and Acito, F. (2000), "Relationship marketing activities, commitment, and membership behaviors in professional associations", The Journal of Marketing, pp Guardian (2010), "Country profile: Thailand", Guardian, available at: (accessed 1st August 2013). Guest, L. (1944), "A study of brand loyalty", Journal of Applied Psychology, Vol. 28 No. 1, pp Gummesson, E. (1994), "Making relationship marketing operational", International Journal of Service Industry Management, Vol. 5 No. 5, pp Gummesson, E. (1999), "Total relationship marketing: experimenting with a synthesis of research frontiers", Australasian Marketing Journal, Vol. 7 No. 1, pp Gummesson, E. (2007), "Exit services marketing-enter service marketing", Journal of Customer Behaviour, Vol. 6 No. 2, pp

284 Gummesson, E., Lusch, R. F. and Vargo, S. L. (2010), "Transitioning from service management to service-dominant logic: Observations and recommendations", International Journal of Quality and Service Sciences, Vol. 2 No. 1, pp Gunasekaran, A., Patel, C. and Tirtiroglu, E. (2001), "Performance measures and metrics in a supply chain environment", International Journal of Operations & Production Management, Vol. 21 No. 1/2, pp Gundlach, G. T. and Murphy, P. E. (1993), "Ethical and legal foundations of relational marketing exchanges", The Journal of Marketing, Vol. 57 No. 4, pp Ha, S. and Stoel, L. (2012), "Online apparel retailing: roles of e-shopping quality and experiential e-shopping motives", Journal of Service Management, Vol. 23 No. 2, pp Hair, J., Anderson, R., Tatham, R. and Black, C. (1998), Multivariate Data Analysis, 5 edn, Prentice Hall, Upper Saddle River: NJ. Hair, J. F., Black, W. C., Babin, B. J. and Anderson, R. E. (2009), Multivariate data analysis, Prentice Hall, Upper Saddle River, NJ. Hair, J., Tatham, R. L., Anderson, R. E. and Black, W. (2006), Multivariate data analysis, 6 edn, Pearson Prentice Hall, Upper Saddle River, NJ. Han, H., Kim, Y. and Kim, E. (2011), "Cognitive, affective, conative, and action loyalty: Testing the impact of inertia", International Journal of Hospitality Management, Vol. 30 No. 4, pp Han, S., Lu, S. and Leung, S. (2012), "Segmentation of telecom customers based on customer lifetime value by decision tree model", Expert Systems with Applications, Vol. 39, pp Han, X., Kwortnik, R. J. and Wang, C. (2008), "Service Loyalty. An Integrative Model and Examination across Service Contexts", Journal of Service Research, Vol. 11 No. 1, pp Hansen, H., Samuelsen, B. M. and Sallis, J. E. (2013), "The moderating effects of need for cognition on drivers of customer loyalty", European Journal of Marketing, Vol. 47 No. 8, pp Harre, R. (1986), Varieties of Realism, Basil Blackwell Ltd., Oxford. Harris, L. C. and Goode, M. M. (2004), "The four levels of loyalty and the pivotal role of trust: a study of online service dynamics", Journal of Retailing, Vol. 80 No. 2, pp

285 Harrison-Walker, L. J. (2001), "The measurement of word-of-mouth communication and an investigation of service quality and customer commitment as potential antecedents", Journal of Service Research, Vol. 4 No. 4, pp Hasley, J. and Gregg, D. (2010), "An Exploratory Study of Website Information Content", Journal of Theoretical and Applied Electronic Commerce Research, Vol. 5 No. 3, pp He, H. and Li, Y. (2010), "Key service drivers for high-tech service brand equity: The mediating role of overall service quality and perceived value", Journal of Marketing Management, Vol. 27 No. 1/2, pp Healy, M. (2007), "School choice, brand loyalty and civic loyalty", Journal of Philosophy of Education, Vol. 41 No. 4, pp Hennig Thurau, T. and Klee, A. (1997), "The impact of customer satisfaction and relationship quality on customer retention: A critical reassessment and model development", Psychology & Marketing, Vol. 14 No. 8, pp Hervé, C. and Mullet, E. (2009), "Age and factors influencing consumer behaviour", International Journal of Consumer Studies, Vol. 33, pp Hiscock, J. (2001), "Most trusted brands", Marketing, Vol. March, 1st, pp Hoffman, D. L., Novak, T. P. and Peralta, M. (1999), "Building Con Trust Online", Communications of the ACM, Vol. 42 No. 4, pp Holbrook, M. (1994), The Nature of Customer Value: An Axiology of Services in the Consumption Context in Service Quality: New Directions in Theory and Practice, SAGE Publications, Thousand Oaks. Holmes-Smith, P. (2000), "Intoduction to Stractural Equation Modelling Using AMOS 4.0 and LISREL 8.30, School Research", Evaluation and Meaurement Services, pp. Canberra. Homburg, C. and Giering, A. (2001), "Personal characteristics as moderators of the relationship between customer satisfaction and loyalty- An empirical analysis. ", Psychology & Marketing, Vol. 18, pp Hong, P. and Cho, H. (2011), "The impact of consumer trust on attitudinal loyalty and purchase intentions in B2C e-marketplaces: Intermediary trust vs. seller trust", International Journal of Information Management, Vol. 31 No. 5, pp

286 Hong, Y. and John, J. (2010), "Role of customer orientation in an integrative model of brand loyalty in services", The Service Industries Journal, Vol. 30 No. 7, pp Hooper, D., Cooghlan, J. and Mullen, M. R. (2008), "Structural equation modeling: Guidelines for determining model fit", Electronic Journal of Business Research Methods, Vol. 6 No. 1, pp Hsieh, J. Y. and Liao, P. W. (2011), "Antecedents and moderators of online shopping behavior in undergraduate students", Social Behavior and Personality: an international journal, Vol. 39 No. 9, pp Hu, H., Kandampully, J. and Juwaheer, T. (2009), "Relationships and Impacts of Service Quality, Perceived Value, Customer Satisfaction and Image: An empirical study", Service Industries Journal, Vol. 29 No. 2, pp Hu, L. T. and Bentler, P. M. (1998), "Fit indices in covariance structure modeling: Sensitivity to underparameterized model misspecification", Psychological methods, Vol. 3 No. 4, pp Huan, Z., Xu, X. and Li, H. (2005), "The impact of service quality, satisfaction, value and switching barrier on customer loyalty in Chinese mobile telecommunication industry", paper presented to ICSSSM 2005 International Conference, vol. 1, pp Hunt, S. D. (1991), "Positivism and paradigm dominance in consumer research: toward critical pluralism and rapprochement", Journal of Consumer Research, pp Hunt, S. D., Arnett, D. B. and Madhavaram, S. (2006), "The explanatory foundations of relationship marketing theory", Journal of business & industrial marketing, Vol. 21 No. 2, pp Hur, W., Park, J. and Kim, M. (2010), "The role of commitment on the customer benefits-loyalty relationship in mobile service industry", Service Industries Journal, Vol. 30 No. 14, pp Iacobucci, D. (2010), "Structural equations modeling: Fit indices, sample size, and advanced topics", Journal of Consumer Psychology, Vol. 20 No. 1, pp IC3 (2012), "Internet crime report 2012", Internet crime complain center, available at: 266

287 IWS (2013), "Asia Marketing Research, Internet Usage, Population Statistics and Facebook Information", Internet World Stats, available at: (accessed 4th November 2013). Jacoby, J. (1971), "Personality and innovation proneness", Journal of Marketing Research, Vol. 8 No. 2, pp Jacoby, J. and Chestnut, R. W. (1978), Brand Loyalty: Measurement and Management, John Wiley & Sons, New York. Jahanzeb, S., Fatima, T. and Khan, M. (2011), "An empirical analysis of customer loyalty in Pakistan's telecommunication industry", Journal of Database Marketing & Customer Strategy Management, Vol. 18 No. 1, pp Jahromi, A., Sepehri, M. M., Teimourpour, B. and Choobdar, S. (2010), "Modeling customer churn in a non-contractual setting: the case of telecommunications service providers", Journal of Strategic Marketing, Vol. 18 No. 7, pp Jaiswal, K. and Niraj, R. (2011), "Examining mediating role of attitudinal loyalty and nonlinear effects in satisfaction-behavioral intentions relationship", Journal of Services Marketing, Vol. 25 No. 1, pp Janda, S., Trocchia, P. J. and Gwinner, K. P. (2002), "Consumer perceptions of Internet retail service quality", International Journal of Service Industry Management, Vol. 13 No. 5, pp Jandaghi, G., Mehranjani, N., Seresht, N. and Mokhles, A. (2011), "Studying The Effect Of Brand Loyalty On Customer Service In Kerman Asia Insurance Company", Australian Journal of Business and Management Research, Vol. 1 No. 6, pp Jayanti, R. K. and Ghosh, A. K. (1996), "Service value determination: An integrative perspective", Journal of Hospitality and Leisure Marketing, Vol. 34 No. 4, pp Johnson, W. C. and Sirikit, A. (2002), "Service quality in the Thai telecommunication industry: a tool for achieving a sustainable competitive advantage", Management Decision, Vol. 40 No. 7, pp Jones, A. and Suh, J. (2000), "Transaction-specific satisfaction and overall satisfaction: An empirical analysis", Journal of Services Marketing, Vol. 14 No. 2, pp Jones, T. and Taylor, S. F. (2007), "The conceptual domain of service loyalty: how many dimensions?", Journal of Services Marketing, Vol. 21 No. 1, pp

288 Jones, T., Taylor, S. F. and Bansal, H. S. (2008), "Commitment to a friend, a service provider, or a service company-are they distinctions worth making?", Journal of the Academy of Marketing Science, Vol. 36 No. 4, pp Joreskog, K. G. (1993), Testing structural equation models, Sage, Newbury Park, CA. Jöreskog, K. G. and Sörbom, D. (1993), LISREL 8: A Guide to the Program and Applications, Scientific Software International, Chicago, IL. Judd, R. C. (1964), "The case for redefining services", Journal of Marketing, Vol. 28 No. 1, pp Kabiraj, S. and Shanmugan, J. (2010), "Development of a conceptual framework for brand loyalty: A Euro-Mediterranean perspective", Journal of Brand Management, Vol. 18, pp Kaiser, H. F. (1960), "The application of electronic computers to factor analysis", Educational and Psychological Measurement, Vol. 20, pp Kaynama, S. A. and Black, C. I. (2000), "A proposal to assess the service quality of online travel agencies: An exploratory study", Journal of Professional Services Marketing, Vol. 21 No. 1, pp Keaveney, S. M. (1995), "Customer switching behavior in service industries: an exploratory study", The Journal of Marketing, pp Keeney, R. L. (1999), "The value of Internet commerce to the customer", Management Science, Vol. 45 No. 4, pp Keller, L. (2003), "Brand Synthesis: The Multidimensionality of Brand Knowledge", Journal of Consumer Research, Vol. 29 No. 4, pp Kelloway, E. K. (1998), Using LISREL for Structural Equation Modeling: A Researcher s Guide, Sage, Thousand, Oaks, CA. Kenny, R. and Dennis, A. (2013), "Consumer lockin for fixed broadband", Communications Chamber, available at: 22nd July 2014). Khokhar, S., Hussain, F., Qureshi, T., Anjum, I., Samran, A. and Arshad, R. (2011), "Only customer satisfaction and customer loyalty is not enough: A study of Pakistan s telecom sector", African Journal of Business Management, Vol. 5 No. 24, pp Kiesler, C. A. (1971), The Psychology of Commitment: Experiments Linking Behavior to Belief, Academic Press, New York. 268

289 Kim, B. (2012), "The diffusion of mobile data services and applications: Exploring the role of habit and its antecedents", Telecommunications Policy, Vol. 36 No. 1, pp Kim, H. and Niehm, L. S. (2009), "The impact of website quality on information quality, value, and loyalty intentions in apparel retailing", Journal of Interactive Marketing, Vol. 23 No. 3, pp Kim, H. and Yoon, C. (2004), "Determinants of subscriber churn and customer loyalty in the Korean mobile telephony market", Telecommunications Policy, Vol. 28 No. 9, pp Kim, J. and Damhorst, M. (2010), "Effects of Level of Internet Retailer s Service Quality on Perceived Apparel Quality, Perceived Service Quality, Perceived Value, Satisfaction, and Behavioral Intentions Toward an Internet Retailer", Clothing and Textiles Research Journal, Vol. 28, pp Kim, J., Morris, D. and Swait, J. (2008), "Antecedents of True Brand Loyalty", Journal of advertising research, Vol. 37 No. 2, pp Kim, G., Shin, B. and Grover, V. (2010), "Investigating Two Contradictory Views of Formative Measurement in Information Systems Research". MIS Quarterly, Vol. 34 No. 2, pp Kim, M. K., Park, M. C. and Jeong, D. H. (2004), "The effects of customer satisfaction and switching barrier on customer loyalty in Korean mobile telecommunication services", Telecommunications Policy, Vol. 28 No. 2, pp Kim, S. and Stoel, L. (2004), "Apparel retailers: web site quality dimensions and satisfaction", Journal of Retailing and Consumer Services, Vol. 11, pp Kim, Y. J. and Lee, W. N. (2009), "Overcoming consumer skepticism in cause-related marketing: The effects of corporate social responsibility and donation size claim objectivity", Journal of Promotion Management, Vol. 15 No. 4, pp Kinard, B. R. and Capella, M. L. (2006), "Relationship marketing: the influence of consumer involvement on perceived service benefits", Journal of Services Marketing, Vol. 20 No. 6, pp Kline, P. (1994), An Easy Guide to Factor Analysis, Routledge, London. Kline, R. B. (2005), Principles and practice of structural equation modeling, Guilford Press, New York. 269

290 Kline, R. B. (2010), Principles and practice of structural equation modeling, 3 edn, Guilford Press, New York. Kline, R. B. (2013), "Assessing statistical aspects of test fairness with structural equation modelling", Educational Research and Evaluation, Vol. 19 No. 2/3, pp Knox, S. and Walker, D. (2001), "Measuring and managing brand loyalty", Journal of Strategic Marketing, Vol. 9, pp Ko, E., Taylor, C. R., Sung, H., Lee, J., Wagner, U., Navarro, D. M. and Wang, F. (2012), "Global marketing segmentation usefulness in the sportswear industry", Journal of Business Research, Vol. 65, pp Koivumäki, T., Ristola, A. and Kesti, M. (2008), "The perceptions towards mobile services: an empirical analysis of the role of use facilitators", Personal and Ubiquitous Computing, Vol. 12 No. 1, pp Koller, M. (1988), "Risk as a Determinant of Trust", Basic and Applied Social Psychology, Vol. 9 No. 4, pp Kotler, P. (2003), Marketing insights from A to Z: 80 concepts every manager needs to know, John Wiley & Sons, Hoboken, N.J. Kumar, N., Scheer, L. K. and Steenkamp, J. B. E. (1995), "The effects of perceived interdependence on dealer attitudes", Journal of Marketing Research, pp Kuo, Y., Hu, T. and Yang, S. (2013), "Effects of inertia and satisfaction in female online shoppers on repeat-purchase intention. The moderating roles of word-ofmouth and alternative attraction", Managing Service Quality, Vol. 23 No. 3, pp Kurt, S. D. and Atrek, B. (2012), "The classification and importance of E-S-Qual quality attributes: an evaluation of online shoppers", Managing Service Quality, Vol. 22 No. 6, pp Kwon, O. B., Kim, C. R. and Lee, E. J. (2002), "Impact of website information design factors on consumer ratings of web-based auction sites", Behaviour & Information Technology, Vol. 21 No. 6, pp Kyriazopoulos, P., Grigoroudis, E., Siskos, Y., Spyridakos, A. and Yannacopoulos, D. (2007), "Tracking changes of e-customer preferences using multicriteria analysis", Managing Service Quality, Vol. 17 No. 5, pp

291 Kyndt, E. and Onghena, P. (2014), The integration of work and learning: Tackling the complexity with structural equation modelling, Springer, Netherlands. La, K. V., Walker, R. and Kam, B. (2005), "Customer loyalty in e-retailing: an exploratory study", International Journal of Internet Marketing and Advertising, Vol. 2 No. 3, pp Lackner, J. M., Lou Coad, M., Mertz, H. R., Wack, D. S., Katz, L. A., Krasner, S. S., Firth, R., Mahl, T. C. and Lockwood, A. H. (2006), "Cognitive therapy for irritable bowel syndrome is associated with reduced limbic activity, GI symptoms, and anxiety", Behaviour research and therapy, Vol. 44 No. 5, pp Lai, F., Griffin, M. and Babin, B. (2009), "How quality, value, image, and satisfaction create loyalty at a Chinese telecom", Journal of Business Research, Vol. 62, pp Lambe, C., Speakman, R. and Hunt, S. D. (2002), "Alliance competence, resources, and alliance success: Conceptualisation, measurement, and initial test", Academy of Marketing Science Journal, Vol. 30 No. 2, pp Lancaster, K. (1971), Consumer demand: A new approach, Columbia University Press, New York. Lau, G. T. and Lee, S. H. (1999), "Consumers trust in a brand and the link to brand loyalty", Journal of Market-Focused Management, Vol. 4 No. 4, pp Lauer, T. and Deng, X. (2007), "Building online trust through privacy practices", International Journal of Information Security, Vol. 6, pp Leahy, R. (2011), "Relationships in fast moving consumer goods markets: The consumers' perspective", European Journal of Marketing, Vol. 45 No. 4, pp Lee, J., Lee, J. and Feick, L. (2001), "The impact of switching costs on the customer satisfaction-loyalty link: mobile phone service in France", Journal of Services Marketing, Vol. 15 No. 1, pp Lee, R. and Murphy, J. (2008), "The Moderating Influence of Enjoyment on Customer Loyalty", Australasian Marketing Journal, Vol. 16 No. 2, pp Lee, Y., Park, J., Chung, N. and Blakeney, A. (2012), "A unified perspective on the factors influencing usage intention toward mobile financial services", Journal of Business Research, Vol. 65, pp

292 Leelakulthanit, O. and Hongcharu, B. (2011), "Factors that Impact Customer Satisfaction: Evidence from the Thailand Mobile Cellular Network Industry", International Journal of Management and Marketing Research, Vol. 4 No. 2, pp Lei, P. W. and Wu, Q. (2007), "Introduction to structural equation modeling: Issues and practical considerations", Educational Measurement: Issues and Practice, Vol. 26 No. 3, pp Lekagul, A. (2002), Toward Preservation of the Traditional Marketplace: A Preference Study of Traditional and Modern Shopping Environments in Bangkok, Thailand, Virginia Polytechnic Institute and State University, Bangkok, Thailand. Li, Y. N., Tan, K. C. and Xie, M. (2002), "Measuring web-based service quality", Total Quality Management, Vol. 13 No. 5, pp Lien, C., Wen, M. and Wu, C. (2011), "Investigating the Relationships among E- Service Quality, Perceived Value, Satisfaction, and Behavioral Intentions in Taiwanese Online Shopping", Asia Pacific Management Review, Vol. 16 No. 3, pp Liljander, V., van Riel, A. C. R. and Pura, M. (2002), Customer Satisfaction with E- Services: The Case of an On-Line Recruitment Portal, Gabler, Wiesbaden. Lim, H., Widdows, R. and Park, J. (2006), "M-loyalty: winning strategies for mobile carriers", Journal of Consumer Marketing, Vol. 23 No. 4, pp Lin, C. P. and Ding, C. G. (2005), "Opening the black box: assessing the mediating mechanism of relationship quality and the moderating effects of prior experience in ISP service", International Journal of Service Industry Management, Vol. 16 No. 1, pp Lin, C. P. and Ding, C. G. (2009), "Comparing the main effects and moderating effects of education among three models in IT service: a quantitative approach", Quality & Quantity, Vol. 43 No. 4, pp Lin, H. F. (2007), "The impact of website quality dimensions on customer satisfaction in the B2C e-commerce context", Total Quality Management and Business Excellence, Vol. 18 No. 4, pp Lin, J. and Wu, C. (2011), "The role of expected future use in relationship-based service retention", Managing Service Quality, Vol. 21 No. 5, pp

293 Lincoln, Y. S. and Guba, E. G. (2000), Paradigmatic controversies, contradictions, and emerging confluences, The handbook of qualitative research, Sage, London. Ling, K. C., Mun, Y. W. and Ling, H. M. (2011), "Exploring factors that influence customer loyalty among Generation Y for the fast food industry in Malaysia", African Journal of Business Management, Vol. 5 No. 12, pp Liu, C. T., Du, T. C. and Tsai, H. H. (2009), "A study of the service quality of general portals", Information & Management, Vol. 46 No. 1, pp Llusar, J. C. B. and Zornoza, C. C. (2000), "Validity and reliability in perceived quality measurement models: an empirical investigation in Spanish ceramic companies", International Journal of Quality & Reliability Management, Vol. 17 No. 8, pp Loehlin, J. C. (1992), Latent variable models, Lawrence Erlbaum Publishers, Hillsdale, NJ. Lounsbury, J. W., Foster, N., Carmody, P. C., Kim, J. Y., Gibson, L. W. and Drost, A. W. (2012), "Key personality traits and career satisfaction of customer service workers", Managing Service Quality, Vol. 22 No. 5, pp Lovelock, C. H. (1983), "Classifying services to gain strategic marketing insights", Journal of Marketing, Vol. 47 No. 3, pp Lovelock, C. H. (1991), "The customer experience", Services Marketing Quarterly, Vol. 2, pp Lovelock, C. H., Wirtz, J., Keh, H. T. and Lu, X. (2002), Services marketing in Asia: managing people, technology, and strategy, Prentice Hall. Lovelock, C. H. and Gummesson, E. (2004), Whither services marketing? In search of a new paradigm and fresh perspectives, Journal of Service Research, Vol. 7 No. 1, pp Lynch, P. D., Kent, R. J. and Srinivasan, S. S. (2001), "The Global Internet Shopper: Evidence from Shopping Tasks in Twelve Countries", Journal of advertising research, Vol. 4 No. 2, pp Madden, G., Savage, S. and Coble-Neal, G. (1999), "Subscriber Churn in the Australian ISP Market", Information Economics and Policy, Vol. 11, pp Magin, S., Algesheimer, R., Huber, F. and Herrmann, A. (2003), "The impact of brand personality and customer satisfaction on customer's loyalty: theoretical approach 273

294 and findings of a causal analytical study in the sector of internet service providers", Electronic Markets, Vol. 13 No. 4, pp Manning, M. L. and Munro, D. (2007), The survey researcher s SPSS cookbook, 2 edn, Pearson Education, Australia, Frenchs Forest, NSW. Mao, J. (2010), "Customer Brand Loyalty", International Journal of Business and Management, Vol. 5 No. 7, pp Mathur, A. (2011), "An Empirical Evaluation of the Impact of Services Quality on Customer Loyalty at Customer Care Centre-A Case Study of Customer Service in Telecom Sector at Jodhpur", International Journal of Management and Tourism, Vol. 19 No. 1, pp Matzler, K., Grabner-Kräuter, S. and Bidmon, S. (2008), "Risk aversion and brand loyalty: the mediating role of brand trust and brand affect", Journal of Product & Brand Management, Vol. 17 No. 3, pp Mazzoni, C., Castaldi, L. and Addeo, F. (2007), "Consumer behavior in the Italian mobile telecommunication market", Telecommunications Policy, Vol. 31 No. 10/11, pp McAlexander, J. H., Kaldenberg, D. O. and Koenig, H. F. (1994), "Service quality measurement", Journal of Health Care Marketing, Vol. 14 No. 3, pp McColl-Kennedy, J. R. and Fetter, R. E. (1999), "Dimensions of consumer search behavior in services", Journal of Services Marketing, Vol. 13 No. 3, pp McDougall, G. (2001), "Customer retention strategies: When do they pay off?", Services Marketing Quarterly, Vol. 22 No. 1, pp McKinney, V., Yoon, K. and Zahedi, F. M. (2002), "The measurement of web-customer satisfaction: An expectation and disconfirmation approach", Information Systems Research, Vol. 13, pp Mehta, S. C., Lalwani, A. K. and Han, S. L. (2000), "Service quality in retailing: relative efficiency of alternative measurement scales for different product-service environments", International Journal of Retail & Distribution Management, Vol. 28 No. 2, pp Merrilees, B. and Fry, L. (2002), "Corporate branding: A framework for e-retailers", Corporate Reputation Review, Vol. 5 No. 2, pp Michael, R. S. (2001), Crosstabulation & Chi square, Indiana University, Bloomington, IN. 274

295 Mokhtar, S., Maiyaki, A. and Mohd Noor, N. (2011), "The Relationship Between Service Quality and Satisfaction on Customer Loyalty in Malaysian Mobile Communication Industry", School of Doctoral Studies (European Union) Journal, pp Montoya, W., Mitzi, M., Glenn, B. V. and Dhruv, G. (2003), "Determinants of Online Channel Use and Overall Satisfaction with a Relational Multichannel Service Provider", Journal of the Academy of Marketing Science, Vol. 31 No. 4, pp Moorman, C., Zaltman, G. and Deshpande, R. (1992), "Relationships between providers and users of market research: the dynamics of trust within and between organisations", Journal of Marketing Research, Vol. 24 No. 1, pp Morgan, M. and Hunt, D. (1994), "The Commitment-Trust Theory of Relationship Marketing", Journal of Marketing, Vol. 58 No. 3, pp Myung, I. J. (2003), "Tutorial on maximum likelihood estimation", Journal of mathematical Psychology, Vol. 47 No. 1, pp Nam, J., Ekinci, Y. and Whyatt, G. (2011), "Brand equity, brand loyalty and consumer satisfaction", Annals of Tourism Research, Vol. 38 No. 3, pp Nationmaster (2012), "Software piracy rate (most recent) by country", Nation Master, available at: (accessed 1st August 2013). NBTC (2014), "Information about Telecommunications Industry", The National Broadcasting and Telecommunications Commission, available at: (accessed 20/05/2014). NECTEC (2014), "Internet User and Statistics in Thailand", National Electronics and Computer Technology Centre, available at: (accessed 1st August 2014). Nejati, M. and Nejati, M. (2008), "Service quality at University of Tehran central library", Library Management, Vol. 29 No. 6/7, pp Nemati, A., Khan, K. and Iftikhar, M. (2010), "Impact of innovation on customer satisfaction and brand loyalty: a study of mobile phones users in Pakistan", European Journal of Social Sciences, Vol. 16 No. 2, pp

296 Nepomuceno, M. V., Laroche, M., Richard, M. O. and Eggert, A. (2012), "Relationship between intangibility and perceived risk: moderating effect of privacy, system security and general security concerns", Journal of Consumer Marketing, Vol. 29 No. 3, pp Neuman, W. (1997), Social research methods: Qualitative and quantitative approaches, 3rd edn, Allyn & Bacon, Needham Heights, MA. Nielson (2013), "Australian Connected Consumers Report", Nielson, available at: 02/03/2014). Nulty, D. (2008), "The adequacy of response rates to online and paper surveys: what can be done?", Assessment & Evaluation in Higher Education, Vol. 33 No. 3, pp Nunnally, J. (1978), Psychometric Theory, Mcgraw Hill, New York. Nunnally, J. and Bernstein, I. H. (1994), Psychometric theory, 3 edn, McGraw-Hill, New York. Oh, H. (1999), "Service quality, customer satisfaction, and customer value: A holistic perspective", International Journal of Hospitality Management, Vol. 18 No. 1, pp Ojo, O. (2010), "The Relationship between Service Quality and Customer Satisfaction in the Telecommunication Industry: Evidence from Nigeria. ", Broad Research in Accounting, Negotiation, and Distribution, Vol. 1 No. 1, pp Oliver, L. (1980), "A cognitive model of the antecedents and consequences of satisfaction decisions", Journal of Marketing Research, Vol. 17 No. 4, pp Oliver, L. (1993), "Cognitive, affective, and attribute bases of the satisfaction response", Journal of Consumer Research, Vol. 20 No. 1, pp Oliver, L. (1997), Satisfaction: A behavioural perspective on the consumer, McGraw- Hill, Singapore. Oliver, L. (1999), "Whence consumer loyalty", Journal of Marketing, Vol. 63 No. 1, pp Oloruntoba, R. and Gray, R. (2009), "Customer service in emergency relief chains", International Journal of Physical Distribution & Logistics Management, Vol. 39 No. 6, pp

297 Ou, W., Shih, C., Chen, C. and Wang, K. (2011), "Relationships among customer loyalty programs, service quality, relationship quality and loyalty: An empirical study", Chinese Management Studies, Vol. 5 No. 2, pp Pallant, J. (2007), SPSS: A Step by Step guide to Data Analysis Using SPSS for Windows (Version 15), 3 edn, Allen & Unwin, Crows Nest, NSW. Pan, Y., Sheng, S. and Xie, F. (2012), "Antecedents of Customer Loyalty: An Empirical Synthesis and Reexamination", Journal of Retailing & Consumer Services, Vol. 19 No. 1, pp Pantouvakis, A. (2010), "The relative importance of service features in explaining customer satisfaction: A comparison of measurement models", Managing Service Quality, Vol. 20 No. 4, pp Parasuraman, A. (2004), "Assessing and improving service performance for maximum impact: Insights from a two-decade-long research journey", Performance Measurement and Metrics, Vol. 5 No. 2, pp Parasuraman, A., Berry, L. L. and Zeithaml, V. A. (1991), "Perceived service quality as a customer based performance measure: An empirical examination of organizational barriers using an extended service quality model", Human Resource Management, Vol. 30 No. 3, pp Parasuraman, A. and Grewal, D. (2000), "The Impact of Technology on the Quality- Value-Loyalty Chain: A Research Agenda", Journal of the Academy of Marketing Science, Vol. 28 No. 1, pp Parasuraman, A., Zeithaml, V. and Berry, L. L. (1993), "Research note: more on improving service quality measurement", Journal of Retailing, Vol. 69 No. 1, pp Parasuraman, A., Zeithaml, V. A. and Berry, L. L. (1985), "A Conceptual Model of Service Quality and Its Implications for Future Research?", Journal of Marketing Management, Vol. 49 No. 1, pp Parasuraman, A., Zeithaml, V. A. and Berry, L. L. (1988), "Servqual: A multiple-item scale for measuring consumer perceptions of service quality", Journal of Retailing and Consumer Services, Vol. 64 No. 1, pp Parasuraman, A., Zeithaml, V. A. and Berry, L. L. (1998), "SERVQUAL: A multipleitem scale for measuring consumer perceptions of service quality", Journal of Retailing, Vol. 64 No. 1, pp

298 Patterson, P. G. and Smith, T. (2001), "Modeling relationship strength across service types in a south-east Asian context", International Journal of Service Industry Management, Vol. 12 No. 2, pp Peter, J. P. (1992), "Realism or Relativism for Marketing Theory and Research", Journal of Marketing, Vol. 56 No. 2, pp Pirc, M. (2006), "Mobile Service and Phone as Consumption System - the Impact on Customer Switching. Proceedings of Helsinki Mobility Roundtable.", Sprouts: Working Papers on Information Systems, Vol. 6 No. 33. Porter, L. W., Steers, R. M., Mowday, R. T. and Boulian, P. V. (1974), "Organizational commitment, job satisfaction, and turnover among psychiatric technicians", Journal of Applled Psychology, Vol. 59, pp Porter, M. E. (2008), Competitive advantage: Creating and sustaining superior performance, Simon and Schuster. Prause, G., Mendez, M. and Garcia-Agreda, S. (2011), "Attitudinal loyalty and trust in entrepreneurship: building new relationships", International Entrepreneurship and Management Journal, /s y. Preacher, K. J. and Kelley, K. (2011), "Effect size measures for mediation models: quantitative strategies for communicating indirect effects", Psychological methods, Vol. 16 No. 2, pp Prentice, C. (2013), "Who stays, who walks, and why in high-intensity service contexts", Journal of Business Research, Pun, K. F. (2002), "Development of an integrated total quality management and performance measurement system for self-assessment: A method", Total Quality Management, Vol. 13 No. 6, pp Qian, S., Peiji, S. and Quanfu, Y. (2011), "An Integrated Analysis Framework for Customer Value, Customer Satisfactory, Switching barriers, Repurchase Intention and Attitudinal Loyalty: Evidences from China Mobile Data Services", Management Science and Engineering, Vol. 5 No. 3, pp Qin, H., Prybutok, V. R. and Zhao, Q. (2010), "Perceived service quality in fast-food restaurants: empirical evidence from China", International Journal of Quality & Reliability Management, Vol. 27 No. 4, pp

299 Rafiq, M., Fulford, H. and Lu, X. (2013), "Building customer loyalty in online retailing: The role of relationship quality", Journal of Marketing Management, Vol. 29 No. 3/4, pp Rafiq, M., Lu, X. and Fulford, H. (2012), "Measuring Internet retail service quality using E-S-QUAL", Journal of Marketing Management, Vol. 28 No. 9/10, pp Rahman, K. and Areni, S. C. (2010), "Product-Line Sub-branding versus Company as the Brand in Services", Marketing Review, Vol. 10 No. 1, pp Ranaweera, C. and Prabhu, J. (2013), "The influence of satisfaction, trust and switching barriers on customer retention in a continuous purchasing setting", International Journal of Service Industry Management, Vol. 14 No. 1, pp Rathmell, M. J. (1966), "What is Meant by Services?", Journal of Marketing, Vol. 30, pp Reichheld, F. and Sasser, E. (1990), "Zero defects: Quality comes to services", Harvard Business Review, Vol. 68 No. 5, pp Reichheld, F. F. and Schefter, P. (2000), "E-Loyalty", Harvard Business Review, Vol. 78 No. 4, pp Reisinger, Y. and Mavondo, F. (2007), "Structural Equation Modeling: Critical Issues and New Developments", Journal of Travel and Tourism Marketing, Vol. 21 No. 4, pp Resnick, M. and Montania, R. (2003), "Perceptions of customer service, information privacy, and product quality from semiotic design features in an online web store", International journal of human-computer interaction, Vol. 16 No. 2, pp Resnik, A. J. and Stern, B. L. (1977), "An analysis of information content in television advertising", Journal of Marketing, Vol. 41, pp Ringham, L. I., Johnson, L. W. and Morton, C. P. (1994), "Customer satisfaction and loyalty for continuous consumer services", Australian Journal of Market Research, Vol. 2 No. 2, pp Ringle, C. M., Sarstedt, M., Schlittgen, R. and Taylor, C. R. (2013), "PLS path modeling and evolutionary segmentation", Journal of Business Research, Vol. 66, pp

300 Roberts, K., Varki, S. and Brodie, R. (2003), "Measuring the quality of relationships in consumer services: an empirical study", European Journal of Marketing, Vol. 37 No. 1/2, pp Roca, J., García, J. and Vega, J. (2009), "The importance of perceived trust, security and privacy in online trading systems", Information Management & Computer Security, Vol. 17 No. 2, pp Rod, M. and Ashill, N. J. (2013), "The impact of call centre stressors on inbound and outbound call-centre agent burnout", Managing Service Quality, Vol. 23 No. 3, pp Roth, A. V. and van der Velde, M. (1991), "Operations as marketing: a competitive service strategy", Journal of Operations Management, Vol. 10 No. 3, pp Rothschild, M. L. and Gaidis, W. C. (1981), "Behavioral Learning Theory: Its Relevance to Marketing and Promotions", Journal of Marketing, Vol. 45, pp Rowe, B., Wood, D., Reeves, D. and Braun, F. (2011), "The Role of Internet Service Providers in Cyber Security", Institute for Homeland Security Solutions, available at: Brief_Rowe.pdf 20/05/2014). Rowe, P. A. and Christie, M. J. (2008), "Civic entrepreneurship in Australia: Opening the black box of tacit knowledge in local government top management teams", International Journal of Public Sector Management, Vol. 21 No. 5, pp Ruiz, D. M., Gremler, D. D., Washburn, J. H. and Carrion, G. C. (2008), "Service value revisited: specifying a higher-order, formative measure", Journal of Business Research, Vol. 61 No. 12, pp Rust, T. and Chung, T. S. (2006), "Marketing Models of Service and Relationships", Marketing Science, Vol. 25 No. 6, pp Rust, T. and Oliver, R. L. (1994), Service quality: New directions in theory and practice, Sage Publications, California. Rust, T., Zahorik, J. and Keiningham, L. (1995), "Return on quality (ROQ): Making service quality financially accountable", Journal of Marketing, Vol. 59 No. 2, pp

301 Sabiote, E. and Roman, S. (2009), "The Influence of Social Regard on the Customer- Service Firm Relationship: The Moderating Role of Length of Relationship", Journal of Business and Psychology, Vol. 24 No. 4, pp Sachdev, S. B. and Verma, H. V. (2004), "Relative importance of service quality dimensions: a multisectoral study", Journal of Services Research. Sa nchez-herna ndez, R. M., Marti nez-tur, V., Peiro, J. M. and Moliner, C. (2010), "Linking functional and relational service quality to customer satisfaction and loyalty: differences between men and women", Psychological Reports, Vol. 106 No. 2, pp Santouridis, I. and Trivellas, P. (2010), "Investigating the impact of service quality and customer satisfaction on customer loyalty in mobile telephony in Greece", The Total Quality Management Journal, Vol. 22 No. 3, pp Sasser, W. E., Olsen, R. P. and Wyckoff, D. D. (1978), Management of Service Operations, Allyn & Bacon, Boston, MA. Saunders, M. N. K. and Lewis, P. (2011), Doing Research in Business and Management, Pearson, Harlow, United Kingdom. Schiffman, G., Bednall, D., O'Cass, A., Paladino, A., Ward, S. and Kanuk, L. (2008), Consumer Behaviour, 4th Edition edn, Pearson Education, Australia. Schiffman, G. and Kanuk, L. (1997), Consumer Behaviour, A Simon & Schuster Company, Upper Saddel River, New Jersey. Schiffman, G. and Kanuk, L. L. (2000), Consumer Behavior, 7th edn, Prentice Hall, Wisconsin. Schumacker, R. E. and Lomax, R. G. (1996), A Beginner s Guide to Structural Equation Modeling, Lawrence Erlbaum Associates, Mahwah, New Jersey. Seth, A., Momaya, K. and Gupta, M. (2008), "Managing the Customer Perceived Service Quality for Cellular Mobile Telephony: An Empirical Investigation", VIKALPA, Vol. 33, pp Shamdasani, P., Mukherjee, A. and Malhotra, N. (2008), "Antecedents and consequences of service quality in consumer evaluation of self-service internet technologies", The Service Industries Journal, Vol. 28 No. 1, pp Shang, R., Chen, Y. and Liao, H. (2006), "The value of participation in virtual consumer communities on brand loyalty", Internet Research, Vol. 16 No. 4, pp

302 Shani, D. and Chalasani, S. (1992), "Exploiting Niches Using Relationship Marketing", Journal of Consumer Marketing, Vol. 9 No. 3, pp Shao, J. and Tu, D. (1995), The Jackknife an d Boot strap, Springer, New York. Sheth, J. (1994), "A Normative Model of Retaining Customer Satisfaction", Gamma News Journal, Vol. July-August, pp Sheth, J. and Parvatiyar, A. (2002), "Evolving Relationship Marketing into a Discipline", Journal of Relationship Marketing, Vol. 1 No. 1, pp Shirin, A. and Puth, G. (2011), "Customer satisfaction, brand trust and variety seeking as determinants of brand loyalty", African Journal of Business Management, Vol. 5, pp Shostack, G. L. (1977), "Breaking free from product marketing", Journal of Marketing, Vol. 41 No. 2, pp Shostack, G. L. (1984), "Designing Services that Deliver", Harvard Business Review, Vol. 62 No. 1, pp Silayoi, P. and Speece, M. (2007), "The importance of packaging attributes: a conjoint analyis approach", European Journal of Marketing, Vol. 41 No. 11/12, pp Simsim, M. T. (2011), "Internet usage and user preferences in Saudi Arabia", Journal of King Saud University-Engineering Sciences, Vol. 23 No. 2, pp Sirdeshmukh, D., Singh, J. and Sabol, B. (2002), "Consumer Trust, Value, and Loyalty in Relational Exchange", Journal of Marketing, Vol. 66, pp Solomon, M. R. (1983), "The Role of Products as Social Stimuli: A Symbolic Interactionism Perspective", Journal of Consumer Research, Vol. 10 No. 3, pp Solomon, M. R., Surprenant, C., Czepiel, J. A. and Gutman, E. G. (1985), "A role theory perspective on dyadic interactions: the service encounter", Journal of Marketing, Vol. 49 No. Winter, pp Songpradit, C. (2005), "The marketing factors affecting consumers' purchasing decision of direct-sale cosmetics in Pranakornsri Ayuttaya province", Management Science Review, Vol. 8 No. 8, pp Spiller, J., Vlasic, A. and Yetton, P. (2007), "Post-adoption behavior of users of Internet Service Providers", Information & Management, Vol. 44 No. 1, pp

303 Spreng, A. and Mackoy, D. (1996), "An empirical examination of a model of perceived service quality and satisfaction", Journal of Retailing, Vol. 72 No. 1, pp Srihirun, P. (2011), "Telecommunications in Thailand", Office of National Telecommunication Commission, available at: /CMS/Events/2011/ITU-ADB/FinalWorkshop/ Thailand-S2.pdf. Steel, E. (2013), "Companies scramble for consumer data", Financial Times, available at: feab7de.html#axzz2W5lJhSoC (accessed 1st August 2013). Steenkamp, J. B. E. and Van Trijp, H. (1991), "The use of LISREL in validating marketing constructs", International Journal of Research in Marketing, Vol. 8 No. 4, pp Stewart, K. J. (2003), "Trust transfer on the World Wide Web", Organization Science, Vol. 14 No. 1, pp Streiner, D. L. (2013), Building a Better Model: An Introduction to Structural Equation Modelling. A Guide for the Statistically Perplexed: Selected Readings for Clinical Researchers, University of Toronto Press, Toronto. Straub, D., Boudreau, M. C. and Gefen, D. (2004), "Validation guidelines for IS positivist research", The Communications of the Association for Information Systems, Vol. 13 No. 1, pp. 63. Streamshield (2004), "Consumers prepared to pay extra for clean and safe Internet service", StreamShield Networks, available at: &Itemid=130 20/05/2014). Sukato, N. and Elsey, B. (2009), "A model of male consumer behaviour in buying skin care products in Thailand", ABAC Journal, Vol. 29 No. 1, pp Sultan, F. and Mooraj, H. A. (2001), "Design a trust-based e-business strategy trust", Marketing Management, Vol. 10 No. 4, pp Sung, Y. and Kim, J. (2010), "Effects of Brand Personality on Brand Trust and Brand Affect", Psychology & Marketing, Vol. 27 No. 7, pp Swartz, T. A., Bowen, D. E. and Brown, S. W. (1992), Advances in Services Marketing and Management,, JAI Press, Greenwich, CT. Sweeney, J. C. and Soutar, G. N. (2001), "Customer perceived value: The development of a multiple item scale", Journal of Retailing, Vol. 77 No. 2, pp

304 Tabachinick, B. G. and Fidell, L. S. (2001), Using multivariate statistics, Allyn and Bacon, Boston. Tabachnick, B. G. and Fidell, L. S. (2007), Using Multivariate Statistics, 5 edn, Allyn and Bacon, New York. Tam, J. (2012), "Linking Perceived Service Quality to Relational Outcomes in a Chinese Context", Journal of International Consumer Marketing, Vol. 24 No. 1, pp Tamosiuniene, R. and Jasilioniene, R. (2007), "Customer relationship management as business strategy appliance: Theoretical and practical dimensions", Journal of Business Economics and Management, Vol. 8 No. 1, pp Teas, K. R. (1993), "Expectations, performance evaluation and consumers perceptions of quality", Journal of Marketing, Vol. 57 No. 4, pp Teo, T. S. H. and Tan, M. (1998), "An Empirical Study of Adopters and Non-Adopters of the Internet in Singapore", Information and Management, Vol. 34, pp Thaichon, P., Lobo, A. and Mitsis, A. (2012), "Investigating the antecedents to loyalty of Internet service providers in Thailand: Developing a conceptual model", paper presented to Australian and New Zealand Marketing Academy Conference, Adelaide, Australia. Thaichon, P., Lobo, A. and Mitsis, A. (2014), "An Empirical Model of Home Internet Services Quality in Thailand", Asia Pacific Journal of Marketing and Logistics, Vol. 26 No. 2, pp Thaichon, P., Quach, T. N. and Lobo, A. (2013), "Marketing communications: Factors influencing brand loyalty of internet service provider", paper presented to Australian and New Zealand Marketing Academy Conference, Auckland, New Zealand. Thananuraksakul, S. 2007, Factors Affecting Online Shopping Behaviour: A Study of Thai Consumers. Thayer, S. E. and Ray, S. (2006), "Online communication preferences across age, gender, and duration of internet use", CyberPsychology & Behavior, Vol. 9, pp Thompson, B. (1997), "The importance of structural coefficients in structural equation modeling confirmatory factor analysis", Educational and Psychological Measurement, Vol. 57 No. 1, pp

305 TNSO (2010), "Key Statistics of Thailand 2008", Thai National Statistical Office, available at: (accessed 1 July 2010). Tohidinia, Z. and Haghighi, M. (2011), "Predictors and outcomes of relationship quality: a guide for customer-oriented strategies", Business Strategy Series, Vol. 12 No. 5, pp TOT (2013), "Our Company", TOT Public Company Limited, available at: 9&Itemid=172&lang=en (accessed 1 July 2013). Trinh, T. A. and Gyarmati, L. (2010), "How to price Internet access for disloyal users under uncertainty", annals of telecommunications-annales des télécommunications, Vol. 65 No. 3-4, pp Tripathi, S. N. and Siddiqui, M. H. (2010), "An empirical study of tourist of preferences using conjoint analysis", Int. Journal of Business Science and Applied Management, Vol. 5 No. 2, pp Trkman, P., Jerman-Blažič, B. and Turk, T. (2008), "Factors of broadband development and the design of a strategic policy framework", Telecommunications Policy, Vol. 32 No. 2, pp True (2010), "Insights of Thailand Internet Market", True Internet, available at: _ThailandInternetMarket_APF2010.pdf (accessed 1 July 2010). True (2014), "About True", True Corporation, available at: (accessed 1 July 2014). Tucker, W. T. (1964), "The Development of Brand Loyalty", Journal of Marketing Research, Vol. 1 No. 3, pp Ulaga, W. and Eggert, A. (2006), "Relationship value and relationship quality: Broadening the nomological network of business-to-business relationships", European Journal of Marketing, Vol. 40 No. 3/4, pp Vargo, S. L. and Lusch, R. F. (2004), "Evolving to a New Dominant Logic for Marketing", Journal of Marketing, Vol. 68 No. 1, pp Veal, A. J. (2005), Business Research Methods: A Managerial Approach, 2 edn, Addison nd Wesley, Sydney. 285

306 Veloutsou, C. and McAlonan, A. (2012), "Loyalty and or disloyalty to a search engine: the case of young Millennials", Journal of Consumer Marketing, Vol. 29 No. 2, pp Venetis, K. A. and Ghauri, P. N. (2004), "Service quality and customer retention: building long-term relationships", European Journal of Marketing, Vol. 38 No. 11/12, pp Verhoef, C. (2003), "Understanding the Effect of Customer Relationship Management Efforts on Customer Retention and Customer Share Development", Journal of Marketing, Vol. 67 No. 4, pp Verhoef, C., Franses, P. and Hoekstra, J. C. (2002), "The Effect of Relational Constructs on Customer Referrals and Number of Services Purchased: Does Age of the Relationship Matter?", Journal of the Academy of Marketing Science, Vol. 30 No. Summer, pp Vlachos, A. and Vrechopoulos, P. (2008), "Determinants of behavioral intentions in the mobile internet services market", Journal of Services Marketing, Vol. 22 No. 4, pp Wahab, S., Zahari, A., Momani, K. and Nor, N. (2011), "The influence of perceived privacy on customer loyalty in mobile phone services: An Empirical Research in Jordan", International Journal of Computer Science Issues, Vol. 8 No. 2. Walbridge, S. W. and Delene, L. M. (1993), "Measuring physician attitudes of service quality", Journal of Health Care Marketing, Vol. 13 No. 4, pp Wallace, D., Giese, J. and Johnson, J. (2004), "Customer retailer loyalty in the context of multiple channel strategies", Journal of Retailing, Vol. 80 No. 4, pp Wang, C. and Wu, L. (2012), "Customer loyalty and the role of relationship length", Managing Service Quality, Vol. 22 No. 1, pp Wang, Y., Lo, H. and Yang, Y. (2004), "An integrated framework for service quality, customer value, satisfaction: evidence from China s telecommunication industry", Information Systems Frontiers, Vol. 6 No. 4, pp Wanous, J. P. and Reichers, A. E. (1996), "Estimating the reliability of a single-item measure", Psychological Reports, Vol. 78 No. 2, pp Weaver, K. and Olson, J. K. (2006), "Understanding paradigms used for nursing research", Journal of Advanced Nursing, Vol. 53 No. 4, pp

307 Webster, C. (2000), "Is spousal decision making a culturally situated phenomenon?", Psychology & Marketing, Vol. 17 No. 12, pp Wedel, M. and Kamakura, W. (2003), Market segmentation. Conceptual and methodological foundations, Kluwer Academic Publishers, Norwell, MA. Weinstein, A. (2002), "Customer retention: a usage segmentation and customer value approach", Journal of Targeting, Measurement and Analysis for Marketing, Vol. 10 No. 3, pp West, S. G., Finch, J. F. and Curran, P. J. (1995), Structural equation models with nonnormal variables, Problems and remedies, Sage, Newbury Park, CA. Weston, T. (1992), "Approximate Truth and Scientific Realism", Philosophy of Science, Vol. 59 No. 1, pp Whittaker, G., Ledden, L. and Kalafatis, S. (2007), "A re-examination of the relationship between value, satisfaction and intention in business services", Journal of Services Marketing, Vol. 21 No. 5, pp Wolfinbarge, M. and Gilly, C. (2003), "etailq: dimensionalizing, measuring and predicting etail quality", Journal of Retailing, Vol. 79 No. 3, pp Woo, S. and Fock, Y. (1999), "Customer Satisfaction in the Hong Kong Mobile Phone Industry", The Service Industries Journal, Vol. 19 No. 3, pp Woodruff, R. B. and Gardial, S. F. (1996), Know Your Customer: New Approaches to Understanding Customer Value and Satisfaction, Blackwell Publications, Cambridge, Mass. WorldBank (2014), "Thailand and the World Bank", Word Bank, available at: dextn (accessed 1 August 2014). Worthington, R. L. and Whittaker, T. A. (2006), "Scale development research: A content analysis and recommendations for best practices", The Counseling Psychologist, Vol. 34 No. 6, pp Worthington, S., Russell-Bennett, R. and Härtel, C. (2010), "A tri-dimensional approach for auditing brand loyalty", Journal of Brand Management, Vol. 17 No. 4, pp Wright, O. and Grace, A. (2011), "Trust and commitment within franchise systems: an Australian and New Zealand perspective", Asia Pacific Journal of Marketing and Logistics, Vol. 23 No. 4, pp

308 Xevelonakis, E. (2004), "Developing retention strategies based on customer profitability in telecommunications: An empirical study", Database Marketing & Customer Strategy Management, Vol. 12 No. 3, pp Yang, Z., Bi, Z. and Zhou, N. (2005), "The double jeopardy phenomenon and the mediating effect of brand penetration between advertising and brand loyalty", Journal of advertising research, Vol. 45 No. 2, pp Yang, Z., Jun, M. and Peterson, R. (2004), "Measuring costumer perceived online service quality: Scale development and managerial implications", International Journal of Operations & Production Management, Vol. 24 No. 11/12, pp Yoo, B. and Donthu, N. (2001), "Developing and validating a multidimensional consumer-based brand equity scale", Journal of Business Research, Vol. 52 No. 1, pp Zavareh, F., Ariff, M., Jusoha, A., Zakuan, N., Bahari, A. and Ashourian, M. 2012, "E- Service Quality Dimensions and Their Effects on E- Customer Satisfaction in Internet Banking Services", paper presented to The 2012 International Conference on Asia Pacific Business Innovation and Technology Management, Procedia - Social and Behavioral Sciences, vol. 40, pp Zeithaml, V. A. (1988), "Consumer Perceptions of Price, Quality, and Value: A Meansend Model and Synthesis of Evidence", Journal of Marketing, Vol. 52 No. 3, pp Zeithaml, V. A. (2000), "Service quality, profitability and the economic worth of customers: What we know and what we need to learn", Journal of the Academy of Marketing Science, Vol. 28 No. 1, pp Zeithaml, V.A., Berry, L.L. and Parasuraman, A. (1993), "The nature and determinants of customer expectations of service", Journal of the Academy of Marketing Science, Vol. 21 No. 1, pp Zeithaml, V. A., Berry, L. L. and Parasuraman, A. (1996), "The Behavioral Consequences of Service Quality", Journal of Marketing, Vol. 60 No. 2, pp Zeithaml, V. A., Bitner, M. and Gremler, D. (2010), Services Marketing Strategy, John Wiley & Sons, Chichester, UK. 288

309 Zeithaml, V. A., Parasuraman, A. and Berry, L. L. (1985), "Problems and strategies in services marketing", Journal of Marketing, pp Zeithaml, V. A., Parasuraman, A. and Berry, L. L. (1990), Delivering service quality: Balancing customer perceptions and expectations, The Free Pres, New York. Zeithaml, V. A., Parasuraman, A. and Malhotra, A. (2002), "Service quality delivery through web sites: A critical review of extant knowledge", Journal of the Academy of Marketing Science, Vol. 30 No. 4, pp Zhao, Y. L. and Benedetto, C. A. D. (2013), "Designing service quality to survive: Empirical evidence from Chinese new ventures", Journal of Business Research, Vol. 66, pp Zikmund, W. G. (2000), Business Research Methods, 6th edn, Harcourt, USA. Zinkhan, G. M. (2002), "Promoting services via the Internet: new opportunities and challenges", Journal of Services Marketing, Vol. 16 No. 5, pp

310 Appendix 1: Final questionnaire Swinburne University of Technology PROJECT CONSENT INFORMATION STATEMENT Project title: A study investigating the determinants of customer retention and brand loyalty in the home Internet services market of Thailand Investigators: Associate Professor Antonio Lobo; Paramaporn Thaichon (Park); Dr Ann Mitsis Thai National Statistical Office (TNSO 2010) and National Electronics and Computer Technology Centre of Thailand (NECTEC 2010) indicate that 10% of the home Internet customers on average switched service providers each year between 2003 to 2008 in Thailand. In 2009, the figure was 12% (True, 2010). Therefore we need to understand determinants that influence brand loyalty in order to retain customers in telecommunications market, especially in Thailand. The study Investigating the determinants of customer retention and brand loyalty in the home Internet services market of Thailand is conducted by Mr Park Thaichon, a PhD student at Swinburne University of Technology, Australia. His research examines the factors associated with customer retention and brand loyalty of the Internet Service Providers (ISP) in Thailand. The information you provide will help him to develop a better understanding of the factors that influence Internet service providers acquisition and retention of customers. We will send you a weblink for the online survey. It is expected that you would take approximately 15 minutes to complete this survey. Completion of this survey is taken as your Informed Consent to participate in this research. The involvement in the project is voluntary and you can withdraw at any time. Data security is ensured. All data collected and analysed will be stored in a password protected computer, and locked in a filing cabinet, in accordance with Swinburne University s Policy on the Conduct of Research. 290

311 If you have any questions or want more information about this survey, please do not hesitate to contact Mr Park Thaichon. Mr Park Thaichon Associate Professor Antonio Lobo PhD student Principal Coordinating Supervisor Faculty of Business and Enterprise Faculty of Business and Enterprise Swinburne University of Technology Swinburne University of Technology PO Box 218 Hawthorn PO Box 218 Hawthorn VIC 3122 Australia VIC 3122 Australia Mobile: Phone This project has been approved by or on behalf of Swinburne s Human Research Ethics Committee (SUHREC) in line with the National Statement on Ethical Conduct in Research Involving Humans. If you have any concerns or complaints about the conduct of this project, you can contact: Research Ethics Officer, Swinburne Research (H68), Swinburne University of Technology, PO Box 218, Hawthorn, VIC Tel (03) or or resethics@swin.edu.au Park Thaichon Associate Professor Antonio Lobo 291

312 Swinburne University of Technology CONSENT FORM FOR PARTICIPANTS Project title: A study investigating the determinants of customer retention and brand loyalty in the home Internet services market of Thailand Investigators: Associate Professor Antonio Lobo; Paramaporn Thaichon (Park); Dr Ann Mitsis 1. I consent to participate in the project named above. I have been provided a copy of the project consent information statement to which this consent form relates and any questions I have asked have been answered to my satisfaction. 2. In relation to this project, by completing the survey: I agree to complete questions asking me about my opinion as a home Internet service user I consent to allow the researcher to collect, analyse and further use data obtained I agree to make myself available for further information if required 3. I acknowledge that: (a) my participation is voluntary and that I am free to withdraw from the project at any time without explanation; (b) the Swinburne project is for the purpose of research and not for profit; (c) any identifiable information about me which is gathered in the course of and as the result of my participating in this project will be (i) collected and retained for the purpose of this project and (ii) accessed and analysed by the researcher(s) for the purpose of conducting this project; (d) my anonymity is preserved and I will not be identified in publications or otherwise without my express written consent. By clicking Next Page I agree to participate in this project. 292

313 Swinburne University of Technology แบบฟอร มข อม ลงานว จ ย แบบสอบถามน เป นการทาว จ ยเพ อการศ กษาใน ห วข อ การส ารวจป จจ ยท ม ผลต อการร กษาล กค า และความผ กพ นในการให บร การ ทางอ นเตอร เน ต ในประเทศไทย ผ ว จ ย: Associate Professor Antonio Lobo; Paramaporn Thaichon (Park); Dr Ann Mitsis สาน กงานสถ ต แห งชาต (TNSO 2010) และศ นย เทคโนโลย อ เล กทรอน กส และคอมพ วเตอร แห งชาต (NECTEC 2010) ได ระบ ว า ล กค าท ใช อ นเตอร เน ตคร วเร อนประมาณ 10% เปล ยนผ ให บร การท กป ระหว างป 2003 ป 2008 และในป 2009 ม อ ตราการเปล ยนแปลงเฉล ย 12% (True 2010) ด งน นการทาว จ ยน จ งเป นการเป ดโอกาสให ผ บร การ อ นเตอร เน ตในประเทศไทยเข าใจถ งป จจ ยท ม ผลกระทบต อการใช บร การ และสามารถเข าใจถ งการ ร กษาล กค าใน ตลาดอ นเตอร เน ตสาหร บคร วเร อน ซ งจะส งผลต อการเพ มหร อร กษาส วนแบ งตลาดโดยอาศ ยฐาน ของล กค าท ม ความผ กพ น แบบสอบถามน เป นการทาว จ ยเพ อการศ กษาในห วข อ การส ารวจป จจ ยท ม ผลต อการร กษาล กค า และความผ ก พ น ในการให บร การทางอ นเตอร เน ตในประเทศไทย โดย นาย ปรมาภรณ ไทยชน น กศ กษาปร ญญาเอก Swinburne University of Technology ประเทศ ออสเตรเล ย งานว จ ยฉบ บน ม งสารวจหาป จจ ย ท ม ส วนเก ยวข องก บการ ให บร การล กค า และความพ งพอใจของล กค าในการใช บร การอ นเตอร เน ตในประเทศไทย โดยข อม ลท ได ร บจากท าน จะนามาว เคราะห เพ อหาป จจ ยท เก ยวข องก บผ ให บร การทางอ นเตอร เน ตและการ ร กษาล กค า โดยท านจะได ร บ weblink สาหร บแบบสอนถาม online ซ งจะใช เวลาประมาณ 15 นาท ในการตอบแบบสอบถาม โดยท านต อง ประสงค และย นยอมก บ แบบฟอร มย นยอม หากท านไม ประสงค จะตอบคาถาม ท านสามารถหย ด ตอบคาถามได ท นท ข อม ลท งหมดจะถ กเก บไว ใน คอมพ วเตอร ฐานข อม ลท Swinburne University และตามนโยบายของการดาเน นการว จ ย Swinburne University 293

314 หากท านม ข อสงส ยหร อต องการข อม ลเพ มเต มในการตอบแบบสอบถามฉบ บน ท านสามารถต ดต อกระผม นาย ปรมาภรณ ไทยชน ได ตามท อย ด านล าง Mr Park Thaichon Associate Professor Antonio Lobo PhD student Principal Coordinating Supervisor Faculty of Business and Enterprise Faculty of Business and Enterprise Swinburne University of Technology Swinburne University of Technology PO Box 218 Hawthorn PO Box 218 Hawthorn VIC 3122 Australia VIC 3122 Australia Mobile: Phone การทาว จ ยเพ อการศ กษาคร งน ได ร บอน ม ต จาก คณะกรรมการจร ยธรรมของ Swinburne (SUHREC) ว าได สอดคล องก บข อกาหนดทางจร ยธรรมในการว จ ยท เก ยวข องก บทร พยากรบ คคล หากท านม ข อสงส ย เก ยวก บ ข อกาหนดจร ยธรรมของการทาว จ ยน ท านสามารถต ดต อได ท Research Ethics Officer, Swinburne Research (H68), Swinburne University of Technology, PO Box 218, Hawthorn, VIC Tel (03) or or resethics@swin.edu.au Park Thaichon Associate Professor Antonio Lobo 294

315 Swinburne University of Technology แบบฟอร มย นยอม แบบสอบถามน เป นการทาว จ ยเพ อการศ กษาในห ว ข อ การส ารวจป จจ ยท ม ผลต อการร กษาล กค า และความผ กพ นในการให บร การ ทาง อ นเตอร เน ตในประเทศไทย ผ ว จ ย: Associate Professor Antonio Lobo; Paramaporn Thaichon (Park); Dr Ann Mitsis 1. ท านประสงค และย นยอมท จะตอบแบบสอบถามน ท านได อ านและทาความเข าใจก บแบบฟอร มข อม ลงานว จ ย น โดยระเอ ยด 2. หากท านประสงค ตอบแบบสอบถาม โปรดให การย นยอม ตามละเอ ยดด งน : ท านย นด จะตอบท กคาถามเพ อการว จ ย ท ม ส วนเก ยวข องก บการให บร การล กค า และความพ งพอใจของ ล กค าในการใช บร การอ นเตอร เน ต ท านย นยอมให เก บข อม ล ว เคราะข อม ล และใช ข อม ลน ในอนาคต ถ าจาเป น ท านย นยอมให ข อม ลเพ มอ กในอนาคต 3. เพ อโปรดทราบ ตามละเอ ยดด งน : (a) ท านเต มใจท จะตอบท กคาถามด วยความสม ครใจ หากท านไม ประสงค จะตอบคาถามต อไป ท านสามารถ หย ด ตอบคาถามได ท นท (b) การสารวจน เป น ส วนหน งของการว จ ยทางการศ กษาซ งไม ม ความเก ยวข องใดๆก บผลประโยชน ทางทาง ธ รก จ (c) ผลสารวจของแบบสอบถามฉบ บน เป นส วนหน งของการทาว จ ยทางการศ กษา (d) แบบสอบถามน ไม ประสงค ระบ ผ ตอบคาถาม และคาตอบจะไม ม การเป ดเผย โปรดเล อก Next Page ท านประสงค ท จะให การย นยอม ตามละเอ ยดข างต นน 295