Research objectives - To research an overview of e-banking service quality, customer satisfaction, customer loyalty, switching cost, customer trust and the relationship among variables.
Trang 1INTRODUCTION
1 Necessity of the topic
The development of information technology in banking has remarkably
changed business strategies of commercial banks Instead of supplying traditional
banking services, commercial banks has been taking advantage of information
technology, especially internet, to provide customers with modern banking services
(hereinafter referred to as e-banking)
E-banking services bringvariousbenefits to banks such as decrease in
number of banking staff and branches, and increase in number of transactions as
well, thereby, banking cost will slump whereas profit will rocket For customers,
e-banking services help them to rapidly and conveniently make transactions, and
their travelling time and transaction cost can be saved as a result
Thanks to a numberof benefits brought by e-banking services, commercial
banks in Vietnam have been proactively investing in electronic infrastructure
system to compete in this sector, however, they have not met higher and higher
demand of customers and failed to satisfy customers as well as get loyal
customers
On the other hands, to well compete in e-banking sector, banks should
understand criteria used by customers to assess e-banking service quality and
then make solutions to improve their service quality and customer satisfaction
In order to do this, it is vital to develop a research model to determine the
relationship among assessment criteria fore-banking service quality, customer
satisfaction and customer loyalty affected by other intermediate factors
(switching cost, customer trust, etc.) as well as to make orientation for
competitive strategies of the banks
Because of the above reasons, the topic named “Research on the
relationship among e-banking service quality, customer satisfaction and
customer loyalty in Vietnam” is chosen as research topic of the thesis
1.2 Research objectives
- To research an overview of e-banking service quality, customer
satisfaction, customer loyalty, switching cost, customer trust and the relationship
among variables
- To develop a research model on the relationship among e-banking service
quality, customer satisfaction and customer loyalty affected by intermediate
factors including switching cost and customer trust in e-banking sector in
Vietnam
- To test a hypothesis model to determine the relationship among e-banking
service quality, customer satisfaction and customer loyalty in commercial banks
in Vietnam
- To propose some solutions for commercial banks in Vietnam to improve
banking service quality based on the research result on the relationship among
e-banking service quality, customer satisfaction and customer loyalty
1.3 Research questions
Based on research objectives, research questions include:
(1) Is there a relationship between overall banking service quality and e-banking service quality?
(2) Is there a relationship between overall e-banking service quality and electronic information system quality?
(3) Is there a relationship between overall e-banking service quality and banking service product quality?
(4) Is there a relationship between customer satisfaction and overall e-banking service?
(5) Is there a relationship between customer loyalty and customer satisfaction in e-banking service sector?
(6) Does switching cost affect the relationship between customer satisfaction and customer loyalty in e-banking service sector?
(7) Does customer trust affect the relationship between customer satisfaction and customer loyalty in e-banking service sector?
1.3 Research subjects and scope
- Research subjects: the relationship among e-banking service quality, customer satisfaction and customer loyalty Impacts of switching cost and customer trust on this relationship
- Research scope: in commercial banks of Vietnam, including state-owned commercial banks, private joint stock commercial banks and foreign commercial banks in Vietnam Research data is collected from customers of commercial banks in Vietnam
1.4 Research methodology
• Research methodology:
- Research methodology in this thesis is the combination of dialectical materialism and historical materialism
• Research methods:
- Statistical and comparative methods;
- Synthetic and analytical methods;
- Regressive and correlative methods
1.5 Thesis content
Besides introduction, conclusion, list of reference documents, list of figures and appendix, the thesis is divided into five chapters:
Chapter 1: Overview of the research issues Chapter 2: Rationale
Chapter 3: Research methods Chapter 4: Research results Chapter 5: Some conclusions, proposals and further research directions
1.6 New contributions of the thesis
• Academic and theoretical contributions
Based on a general research of over 170 related documents in e-banking
Trang 2service quality sector, the thesis has made new academic contributions as
follows:
(1) Development of a new and comprehensive structural model to explain the
relationship amongservice quality, customer satisfaction and customer loyalty in
e-banking sector in Vietnam The previous works mainly focused on pair relationship
(2) Extended research of two intermediate variables including bank
switching cost and customer trust to consider their impacts on the relationship
between customer satisfaction and customer loyalty
(3) The thesis states a positive and close relationship between overall
e-banking service quality and customer satisfaction and between customer
satisfaction and customer loyalty However, unlike previous works, in the
context of e-banking in Vietnam, intermediate factors such as switching cost and
customer trust are stated to have no impacts
• New proposals drawn from research results
(4) Descriptive statistics result shows that e-banking service users in
Vietnam are young and highly qualified (higher education level) with good
income Thus, marketing strategies of banks should aim at such potential
customers
(5) The research result shows that overall e-banking service quality has a
positive correlation with online information system quality and banking service
product quality and there is no basis to conclude that there is a correlation
between overall e-banking service quality and online customer service quality
Thus, it is proposed that in order to improve overall e-banking service quality,
banks should focus on strongly investing enhancing information system quality
and developing new products to meet customers’ demands
(6) Customer satisfaction has a very strong positive correlation with overall
e-banking service quality and customer loyalty has a very strong positive with
customer satisfaction as well Thus, to get loyal customers, marketing strategies
of banks should aim at customer satisfaction by enhancing service quality
(7) Intermediate variables including switching cost and customer trust have
no impacts on the relationship between customer satisfaction and customer
loyalty in e-banking sector in Vietnam
(8) Based on the research result, the thesis makes some proposals on
business strategies to commercial banks in e-banking sector
CHAPTER 1 OVERVIEW OF THE RESEARCH ISSUES 1.1 Situation of researches in the world
So far, there have been some research works in the world on customer service
quality, customer satisfaction and customer loyalty Such works focus on
developing an assessment model of customer service quality, the relationship
among service quality, customer satisfaction and customer loyalty, developing
assessment criteria and measurement scale as well as using econometric methods to
test the relationship among variables
Gronroos (1984,1990)introduced a service quality model to research
customer feelings based on three basic factors including distinction of functional quality (serving manners such as behaviors and acts of staff to customers) and technical quality (serving media such as computer system, technical and technological solutions); image played the most important part
in service supply enterprises;overall feeling about quality was a mathematic function of customers’ assessment of service quality and difference between such assessment and their service expectations The author generalized as follows:
Service quality = Technical quality + Functional quality
Parasuraman et al (1985, 1988) introduced a service quality measurement
model (called as SERVQUAL) SERVQUAL model was developed based on the concept that service quality was the comparison between expected value and experienced value of customers According to the author,
Service quality = Experienced value - Expected value
Cronin & Taylor (1994) proposed a SERVEPERF model to assess service
quality based on experimental research result in advertising industry in Australia and stated that this model was better than SERVQUAL model of Parasuraman However, this is still in argument because such model result may vary depending
on research contexts in research
Besides, there are some research works of other authors mentioning impacts of service quality on customer satisfaction and customer loyalty Some works are in industries other than banking such as mobile information, advertising, tourism, etc.(Oliver, R.L 1993; Zeithaml, V.A., A Parasuraman &
A Malhotra, 2001; Shankar, V., Smith, K., &Rangaswamy, A., 2003; Ting, D.H, 2004)
Some researches are conducted in traditional banking context where direct contact between customers and banking staff is preferred (such asBloemer, J., Ruyter, K.,& Peeters, P, 1998; Caruana, A 2002; Jamal, A & K Naser, 2002; Baumann, C.S Burton & G Elliott, 2005; Ehigie, B.O, 2006; Calik and Balta, 2006)
Moreover, a few researches are conducted in e-banking context where the first distribution and contact channel are not via direct human contact (Jun & Cai, 2001; Broderick,A.J & S Vachirapornpuk, 2002; Rotchanakitumnuai, S &
M Speece (2003); Jun, M., Z Yang & D.S Kim, 2004; Veloutsou, C., S Daskou & A Daskou, 2004; Jaruwachirathanakul, B & D Fink (2005); Siu, N.Y.M & J.C.W Mou, 2005; Pikkarainen, K., T Pikkarainen., H Karjaluoto &
S Pahnila, 2006; Maenpaa, 2006)
1.2 Situation of researches in Vietnam
So far in Vietnam, there have been some research works on service quality, customer satisfaction and customer loyalty as well as the relationship among variables Such researches have chosen assessment model, determined assessment criteria of service quality, customer satisfaction and loyalty; used
Trang 3econometric methods to test the relationship among variables, typically research
of Nguyen Thi Mai Trang (2006) on service quality, customer satisfaction and
customer loyalty in supermarkets in Ho Chi Minh City; research of Nguyen Huy
Phong and Pham Ngoc Thuy (2007) on comparison of SERVQUAL
(Parasuraman, 1985) and SERVPERF (Cronin and Taylor, 1997) in retail
supermarkets of Vietnam; research of Nguyen Thanh Cong and Pham Ngoc
Thuy (2007) on factors affecting customer loyalty to cell phone brands; research
of Nguyen Thi Phuong Cham (2008) on comparison of SERVQUAL and
Gronroos on service quality and customer satisfaction to select the best model
for Vietnam; research of Nguyen Thi Mai Trang and Tran Xuan Thu Huong
(2010) on library service quality and comparison of functional/technical quality
model and SERVQUAL; research of Nguyen Duy Thanh and Cao Hao Thi
(2011) on e-banking acceptance and use model in Vietnam; research of Le The
Gioi and PhD Le Van Huy (2012) on the relationship among service quality,
customer satisfaction and customer loyalty in banking sector; research of Le Thi
Tuyet Trinh (2012) on customer satisfaction with Viettel mobile
telecommunication services in Binh Dinh; research of Pham Van Tuan (2014)
on impacts of customer satisfaction on customer loyalty and repeated purchasing
behavior of customers in garment industry, etc
1.3 Research gaps and research issues
After studying research works in the world and in Vietnam, there are some
research gaps to be researched as follows:
- There has been no thesis comprehensively researching the relationship
among service quality, customer satisfaction and customer loyalty in e-banking
sector The previous researches in Vietnam only mention the relationship
between service quality and customer satisfaction or between service quality and
customer loyalty Such researches are not in banking sector (mobile information,
tourism, etc.) or in traditional banking sector other than e-banking sector (retail
banking service) Moreover, researches in e-banking sector focus on factors
affecting the model adoption instead of the relationship among service quality,
customer satisfaction and customer loyalty There are some researches on the
relationship amongservice quality, customer satisfaction and customer loyalty in
traditional banking sector but they fail to consider impacts of intermediate
factors such as switching cost and customer trust on the relationship among
variables
- Previous researches have not developed an appropriate theoretical model
to explain the relationship among e-banking service quality, customer
satisfaction and customer loyalty in Vietnam (with impacts of intermediate
factors)
- Analysis methods used in previous researches are limited and the most
commonly used method islinear regression to test relationships and some researches
use Structural Equation Model (SEM) but they fail to consider impacts of
intermediate variables on relationships due to limited tools
Because of such research gaps, the topic named “Research on the relationship among e-banking service quality, customer satisfaction and customer loyalty in Vietnam” is chosen as the research topic of the thesis
CHAPTER 2 RATIONALE 2.1 Electronic commerce
2.1.1 Definition of electronic commerce
There are various definitions of electronic commerce (e-commerce) According to World Trade Organization (WTO), “e-commerce includes production, advertisement, sales, distribution and payment of goods purchased and sold on the Internet but with tangible deliverables, including delivered goods and digital information via Internet.”
According to Electronic Commerce Steering Group of Asia-Pacific Economic Cooperation (APEC), "e-commerce relates to commercial transaction
of goods and services conducted among groups (individuals) electronically via Internet-based systems.”
According to European Commission, “e-commerce can be generally defined as purchase, sales and exchange of goods or services among enterprises, families, individuals and private organizations by electronic transactions via Internet or intermediate computer networks (online communication information) This term includes order and translation via computer network, however, final payment and transport of goods and services may be performed online or manual.”
Zwass (1996) defined e-commerce as business information sharing, relationship maintenance and transaction making via Internet-based devices According to Turban et al (2000), e-commerce isa process of purchase, sale, transport and exchange of products, services and information via computer network including Internet Jelassi and Enders (2004) defined e-commerce as the settlement of transaction problems, product sales and other online services Based on the above definitions, we can seethat e-commerce definitions are used in both broad and narrow meaning, including Internet use for electronic communication and information exchange of products and services However, e-commerce also includes business transactions related to order and payment via Internet (Lin, 2003)
2.2 E-banking service
2.2.1 Definition of e-banking service
The State Bank of Vietnam defines e-banking service as modern and multi-functional banking products and services rapidly distributed to wholesale and retail customers (online, 24 hours per day and seven days per week, anywhere atanytime) via distribution channels (Internet and other terminal access equipment such as computers, ATM, POS, landline telephone, cell phones, etc.) Thus, it can be understood that e-banking services are banking services
Trang 4supplied via electronic media and telecommunication network
2.2.2 Benefits of e-banking services
For banks: E-banking services create a new distribution channel for
banking services
For customers:Using e-banking services with electronic media connected
to telecommunication network enable customers to make their transactions
anywhere and anytime
For society:E-banking services create a brand-new operation method and
contribute to promoting economic activities, trade, services and tourism as well as
facilitate the expansion of economic and commercial cooperation with the region and
the world
2.2.3 Distribution channels of e-banking services
Nowadays, popular distribution channels of e-banking services include:
Internet banking
Home banking
Phone banking
Mobile banking/SMS banking
Call center
2.2.4 Electronic payment instruments
To make electronic transactions, various payment instruments are used
such as:
Digital payment
Digital cheque
Stored value smart card
2.3 E-banking service quality
2.3.1 Service quality
There are various definitions of service quality by researchers such as
definitions of European scholars (Gronroos 1982, 1984; Lehtinen & Lehtinen,
1982) and viewpoints of American scholars (Parasuraman et al., 1985, 1988)
According to Gronroos (1982, 1984), service quality must include three
aspects including technology quality (technical) and performance quality
(functional) and impacts of company image
Service quality = Technical quality + Functional quality
According to Parasuraman et al (1985), service quality is the comparison
between expected value and experienced value of customers and the comparison
can be shown in the following formula:
Service quality = Experienced value - Expected value
According to Zeithaml et al (2002), e-service quality may be considered as
a website to make purchase, sale and receipt of products and services more
effective and convenient According to this definition, service is comprehensive
and includes pre-electronic services and post-electronic services
2.3.2 E-banking service quality
Despite the importance of assessment criteria of e-banking service quality, there is only a few documents includingand studying typical characteristics of e-banking service quality Such characteristics are mentioned in researches of Jun&Cai(2001); Polatoglu&Ekin(2001); Flavian, Tores&Guinaliu(2004); Jayawardhena (2004); Yang and Fang (2004); Bauer &Hammerschmidt(2005); Siu&Mou(2005);Maenpaa(2006); Pikkarainen et al (2006), etc
Within research scope of this thesis, the author uses the definition of
overall e-banking service quality (Jun&Cai, 2001; Yang et al., 2004) This
definition generally assesses e-banking service quality based on three aspects
including online customer service quality measured in availability, reliability, satisfaction and understanding (Jun&Cai, 2001); online information system quality (Jun & Cai, 2001) measured in ease of use, accuracy, confidentiality, contents and appearance; banking service product quality (the number of
services, additional services, free utilities, necessary functions for customers, service characteristics that customers need) (Jun&Cai, 2001, Yang et al., 2004)
2.4 Customer satisfaction
Satisfaction is a customer reaction when their demands are met (Oliver, 1993) It is also a customer reaction when there is a difference between their expectations and experience after using products and services (Tse & Wilton, 1988) Satisfaction is a personal feeling such as pleasure or disappointment resulting from comparing perceived performance(or outcome)from product in relation to his or her expectations (Kotler, 2000) Customer satisfaction is a psychological process of assessing perceived outcome in comparison with their expectations (Egan, 2004)
According to Parasuraman et al (2005), there are some differences between service quality and customer satisfaction and the main difference is causal issue According to Zeithalm (2011) [147], customer satisfaction is affected by many factors including service quality, product quality, situation, price and personal factors
In general, “Customer satisfaction is an emotional state that their demands and expectations of product and service values/benefits are met at a lower, equal or higher level than their expectations and itresults in customer loyalty and repeated product and service purchase”
2.5 Customer loyalty
There are three main approaches to research customer loyalty including behavioral approach, attitudinal approach and general approach The combinationapproach includes both behavior and attitude variables to make a separate definition of customer loyalty
“Behavioral loyalty” is mainly shown by the number of purchased goods, purchase frequency and brand switching (Allen and Meyer, 1990; Oliver, 1993)
“Attitudinal loyalty” is the combination of purchase favorites and priority
of customers to determineloyalty level (Egan, 2004) Loyal customers with“attitudinal loyalty” rarely accept negative information on brand in
Trang 5comparison with other customers (Ahluwali et al., 1999), have a few motives to
search for other alternative services even they are extremely disappointed (Dick
and Basu, 2001) and tend to rapidly and positively advertise good features of
service via oral communication as well as use additional services and accept
reasonable price (Gremler and Brown, 1999)
Thus, loyalty is a definition which can be approached and shown in
various ways However, in this thesis, the author chooses the
combinationapproach which includesboth behavioral and attitudinal loyalty
Accordingly, loyal customers tend to keep on using products and services
(repeated purchase), commit to prioritizing selection of products and services
and introduce products and services to other people (oral communication)
2.6 Research model
2.6.1 Research model
Because some researches have tested the relationship among
decisivefactors of e-baking service quality, overall e-banking service quality,
customer satisfaction, trust, switching cost and loyalty, in order to handle
research gaps, this research proposes some hypotheses, in which, the above
definitions are integrated in the context of Vietnamese banking system A
research model of the relationship among the above definitions is developed as
follows:
Figure 2.1: Research model
2.6.2 Research hypotheses
Based on a general research of documents on e-service quality, customer satisfaction and customer loyalty, some hypotheses are developed in the context
of Vietnamese banking system:
Hypothesis 1 (H1): Online customer service quality and overall e-banking service quality have a positive correlation
Hypothesis 2 (H2): Online information system quality and overall e-banking service quality have a positive correlation
Hypothesis 3 (H3): Banking service product quality and overall e-banking service quality have a positive correlation
Hypothesis 4 (H4): Overall e-banking service quality and customer satisfaction have a positive correlation
Hypothesis 5 (H5): Customer satisfaction and customer loyalty have a positive correlation
Hypothesis 6 (H6): Switching cost has an impact on the relationship between customer satisfaction and customer loyalty
Hypothesis 7 (H7): Customer trust has an impact on the relationship between customer satisfaction and customer loyalty
CHAPTER 3 RESEARCH METHODS
3.1 Research design
3.1.1 Research steps
The research is performed via the following specific steps:
Step 1: Generally research and develop model and system of research
hypotheses
Step 2: Develop questionnaire for data collection
Step 3: Run a measurement model to assess reliability and validity of
measurement scale In order to assess reliability and validity of measurement
Online customer service quality
Information system quality
Banking service product quality
Customer satisfaction
Overall e-banking service quality
Customer loyalty Customer trust
Switching cost
H1+
H2+
H6+
H3+
H4+
H7+
H5+
Trang 6scale, the thesis uses Cronbach’s Alpha coefficients, composite reliability and
average variance extracted (AVE)
Step 4: RunStructural Equation Model (SEM) to test hypotheses in
theoretical model In order to test theoretical model, the author develops three
alternative structural models and selects the best model based on model running
result
Step 5: Summarize research results based on model test result
Step 6: Conclude and propose solutions
3.1.2 Research population and samples
In order to collect research data, the thesis uses questionnaires
Questionnaires are sent to personal customers of commercial banks in large
cities of Vietnam, both in hard and soft copies The number of issued
questionnaire is 550 and the number of collected questionnaire is 511 (the
number of observations is 10 times more than the number of variables)
3.1.3 Measurement scale
3.1.3.1 Measurement scale development
The thesis uses 7-point Likert scale, “1= strongly disagree, 2= disagree, 3=
slightly disagree, 4= neither agree nor disagree, 5= slightly agree, 6= agree, 7=
strongly agree” Based on the research objectives after general research, there
are eight latent variables put into the research model including Online customer
service quality, Online information system quality, Banking service product
quality, Overall e-banking service quality, Customer satisfaction, Customer
loyalty, Switching cost and Customer trust
* Online customer service quality includes four components such as
tangible, reliability, responsivenessand understanding (Yan et al., 2004)
Tangibleis measured bythree observed variables, reliability is measured bythree
observed variables, responsivenessis measured by two observed variables and
empathyis measured by four observed variables
Tangible
T_1 Banks’ websites supply customers with a lot of valuede-banking information
T_2 Searching for e-banking information on banks’ websites is simple
T_3 Banks’ websites are attractive
Reliability
R_1
Banks always fulfill their responsibilities in e-banking transactions with
customers
R_2 E-banking transactions are always made properly
R_3
In case of problems in e-banking transactions of customer, banks strictly
handle such problems
Responsiveness
Res_
1
Bank staff supply customers with information of exact time that e-banking
services requested by customers are performed
Res_ Bank staff help customers rapidly complete e-banking transactions
2
Understanding
E_1
Bank staff consider customer benefits as a top priority in e-banking transactions
E_2
Bank staff understand specific demands of customers in e-banking transactions
E_3
Bank staff pay attention to issues related to customers in e-banking transactions
E_4 Customer support hotline of banks is very useful
* Online information system quality includes three components such as
ease of use (access), accuracy and safety Ease of use is measured byfive observed variables, accuracy is measured bythree observed variables and safety
is measured by four observed variables
Ease of access
EU_1 Information arrangement on banks’ websites helps customers simply search for information
EU_2 Access to e-banking accounts is simple EU_3 Using banks’ websites in e-banking transactions requires much effort EU_4 Making an e-banking transaction via banks’ websites is simple EU_5 I do not have to spend much time searching for e-banking service information via banks’ websites
Accuracy
A_1 Online transactions of customers are always made properly A_2 E-banking service information on banks’ website is accurate A_3 Online transactions are made properly
Safety
S_1 I believe that Banks do not use personal information of customers for wrong purposes
S_2 I feel safe in e-banking transactions via banks’ websites S_3 Sensitive information of customers in e-banking transactions via banks’ websites is secured
S_4 Exposure of risks related to online transactions via banks’ websites is low
* Banking service product quality is measured by five observed variables
BSP_1 Banks supply e-banking services that customers want BSP_2 Banks supply most of online service functions that customers need BSP_3 All demands on online services of customers are in categories of banks BSP_4 Banks supply various online service packages
BSP_5 Many free online services are supplied by banks
* Overall e-banking service quality is measured by two observed
Trang 7variables
O_1 Generally, e-service quality of banks is good
O_2 Generally, banks meet customers’ expectations of components of good
e-banking service quality
* Customer satisfaction is measured by four observed variables
CS_1 Generally, I am satisfied with e-banking transactions with Banks
CS_2 Generally, I am satisfied with Internet-based transactions of Banks
CS_3 Generally, I am satisfied with products and services supplied by Banks
CS_4 Generally, I am satisfied with the bank that I am making transactions
with
* Customer loyalty is measured by three observed variables
L_1 I will keep on making e-banking transactions via the bank that I am
making transactions with
L_2 The bank that I am making transactions with always is my first choice in
e-banking transactions
L_3 I will introduce the bank that I am making transactions with to my
friends and relatives
* Switching costis measured by five observed variables
SC_1 It takes me a lot of time to make e-banking transactions in another bank
SC_2 Making e-banking transactions in another bank will be more expensive
SC_3 It is difficult to familiar with new procedures on e-banking transactions
in other banks
SC_4 It will be uncomfortable to make e-banking transactions in other banks
SC_5 I have invested much (effort, money, time, etc.) into building relationship with Banks
* Customer trust is measured by three observed variables
Tr_1 I trust in e-banking services supplied by Banks
Tr_2 Accounts for my e-banking transactions in Banks are extremely safe
Tr_3 I believe that heads of banks always pay attention to customer benefits
E-banking service quality and electronic information system quality
variables are comprised of many components and each component is measured
by observed variables, thus, the author uses summated scale, in which, each
component variable is the average of observed variables Such average
calculation has no impacts on representation of variables (Bagozzy &
Heatheation, 1994)
3.1.3.2 Reliability of measurement scale
In order to consider reliability of measurement scale, Cronbach’s Alpha
coefficient will be applied This method helps the analyser to eliminate
inappropriate variables and limit garbage itemsduring research and assess of reliability of measurement scale by Cronbach’s Alpha coefficient
Table 3.1: Summary of Cronbach’s Alpha coefficient of variables
Source: Calculation from survey data
Thus, after calculating Cronbach’s Alpha coefficient of variables in the model, it can be shown that all variables have Cronbach’s Alpha coefficient of more than 0.8 soall measurement scales are reliable
3.2 Structural Equation Model (SEM)
3.2.1 Overview of SEM
SEM (Structural Equation Model) is one of the most complex and flexible techniques used to analyze the complicated relationship in causal model SEM is
an expansion of the generalized linear model (GLM) which allows researchers
to test a group of regression equations at the same time
3.2.2 Statistical steps in SEM Step 1: Test reliability of measurement scale
Use Cronbach’s Alpha coefficient (Hair et al., 1998; Segar, 1997) Estimate regression coefficient and tvalue
Confirmatory factor analysis (CFA): Use measurement scale to eliminate variables with low potential factor loading Apply CFA to each sub-model before test the general model
Perform Square Multiple Correlation (SMC)to each external latent concept (CFA result the above measurement scale), similar to coefficient R2 in linear regression, SMC is a variance to explain each latent concept (Bollen, 1989)
Step 2: Goodness of fit of the model population
Actual goodness of fit of the general model is assessed by the following criteria:
Chi-Square (χ2) test shows the general goodness of fit of the whole model at pv=0.05 (Joserkog & Sorbom, 1989) This rarely occurs in fact because
χ2is extremely sensitive to large sample size and test strength, thus, only χ2/df is used to assess in fact
Chi-Square to df ratio (χ2/df) is also used to measure goodness of fit of the whole model more detailedly Some authors suggest 1< χ2/df<3 (Hair et al., 1998) whereas other authors suggest that χ2/df should be as small as possible (Segar, Grover, 1993) and suppose χ2/df<3:1 (Chin& Todd, 1995) Besides, some practical
Trang 8researches have divided into two cases: χ2/df<5 (N>200); or χ2/df<3 (N<200) and in
the latter case, the model is considered to have goodness of fit(Kettingger và Lee,
1995)
If other related indexes such as GFI, AGFI, CFI, NFI, etc., are more than
0.9, the model is considered to have goodness of fit If such indexes equal1, the
model will be perfect (Segar, Grover, 1993), (Chin & Todd, 1995)
GFI measures absolute goodness of fit (without adjusting degree of
freedom) of structural model and measurement model with data set
AGFI adjusts GFI value fordegrees of freedom in the model
RMR assesses residual variance of observed variables andtheir
correlation with that of other observed variables The larger RMR is, the higher
residual variance is and this shows a model with goodness of fit
RMSEA is an important criteria to determine goodness of fit of the
model with population If required RMSEA and RMR are less than 0.05, the
model will have goodness of fit If they are less than 0.08, the model will be
acceptable (Taylor, Sharland, Cronin and Bullard, 1993)
NFI measures the difference of normal distribution of χ2 between
independent model (single factor and coefficients ofzero) with variance
measurement and multifactor model NFI = (χ2null - χ2proposed)/ χ2null = (χ2Mo- χ2Mn)/
χ2Mo
Mo: Original model; Mn: Fit model Proposed NFI is more than 0.9 (Hair
et al., 1998) & (Chin &Todd,1995)
Probability level is more than 0.05 and the model is considered to have
goodness of fit (Rupp & Segal, 1989) This means that Ho (good model
hypothesis) cannot be rejected and there is no model better than current model
Step 3: Modification Indices (MI)
Modification Indices are indices estimating variation of χ2 corresponding
to each case of adding a possible relationship (corresponding to a degree of
freedom) In case ∆ χ2 is more than 3.84 (corresponding to a degree of
freedom), a relationship may be suggested to increase goodness of fit of the
model (Hair et al., 1998) However, researchers should be careful because the
relationship added to the model is only considered if it supports the theory
and researchers should not try their best to improve indices to make the
model get more goodness of fit (Hair et al., 1998)
3.2.5 Data handling and analysis tools in SEM
In this research, the author uses two software including SPSS 18 to handle
data and MPLUS 6.1 to run SEM because this is a software running SEM
simply and effectively and it can be easily applied to intermediate variables
CHAPTER 4 RESEARCH RESULTS
4.2 Descriptive statistics
Research samples are collected from customers of commercial banks in Vietnam There are 511 customers asked via questionnaires in hard copies and soft copies Research samples have some basic statistical characteristics as follows: for gender, male customers account for 36% whereas female customers account for 64%; for age, customers at the age of 20-45 years old account for 93.8%, thus, most service users are young and middle-aged people; for income, customers having income of more than 5 million VND account for 62.6%; for qualification, customers graduating from colleges and universities and at higher education level account for 97.3%, thus, most customers are qualified and can use modern services
Banks’ website access frequency of customers is extremely low Only 11.7% of asked customers accesses banks’ websites once per day This shows that banks’ websites have low attractiveness Most customers have used e-banking services for a short time (less than five years) and only 7.6% of customers uses e-banking services for more than five years.This is reasonable because e-banking services have been deployed in Vietnam for about 10 years and most customers have just used such services For e-banking service products used by customers, the survey shows that customers in Vietnam using e-banking services for account information check, money transfer, personal payment, bill payment and ATM services account for 86%, 70%, 57%, 29% and 19% respectively There are only a few customers using e-banking services for securities trading transaction, credit cards, bill payment, etc
4.3 Model testing result
4.3.1 Measurement model
Trang 9Table 4.15: Statistical criteria of measurement model
Latent variables Observed
variables
Factor loading Heteroscedasticity
of measurement
Average variance extracted (ρ v )
Cronbach’s alpha coefficient (α)
Composite reliability (ρ c )
Online customer
service quality
(F1)
Online
information
system quality
(F2)
Banking service
product quality
(F3)
Overall e-banking
service quality
(F4)
Customer
satisfaction (F5)
Customer loyalty
(F6)
Switching cost
(F7)
Customer trust
(F8)
Tr_2 0.878 0.229 Tr_3 0.687 0.528
Source: Calculation from survey data 4.3.2 Structural Equation Model (SEM)
4.3.2.1 Fit model selection
After running measurement model, the result shows that measurements
scales have sufficient reliability to run SEM with the aim at testing relationship
hypotheses Based on research hypotheses and models developed in overview
section, in order to test model hypotheses, the author develops three SEMs in
variables addition method:
Model 1 (Basic model) includes six latent variables including online
customer service quality (F1), online information system quality (F2), banking
service product quality (F3), overall e-banking service quality (F4), customer satisfaction (F5) and customer loyalty (F6) Model 1 is used to test hypotheses H1, H2, H3, H4 and H5
Model 2 adds switching cost variable (F6) other than six latent variables in
model 1 to test one more hypothesis:
H6: Switching cost has an impact on the relationship between customer satisfaction and customer loyalty
Model 3 adds two latent variables including switching cost (F6) and
customer trust (F7) other than six latent variables in model 1 to test two hypotheses:
H6: Switching cost has an impact on the relationship between customer satisfaction and customer loyalty
H7: Customer trust has an impact on the relationship between customer satisfaction and customer loyalty
After running three structural models above, results will be summarized
and the fittest model will be selected as well
Test result of model 1
Table 4.16: Summary of goodness of fit of model 1
Adjacent values in Logarithm
unit
Value H0
Value H1
-13228.985
-12919.946
Values of AIC, BIC and adjusted
BIC
Coefficient AIC Coefficient BIC
Coefficient adjusted BIC
26599.971 26900.753
26675.389
Chi-square when assessing
goodness of fit of model
Value Degree of freedom
Value p
618.078
181
0.0000 Coefficient RMSEA Estimation
Reliability 90%
Coefficient RMSEA
0.069 0.063 0.075
0.000
Coefficient CFI and TLI Coefficient CFI
Coefficient TLI
0.948
0.940
Chi-square when assessing goodness of fit of base model
developed by software
Value Degree of freedom Value p
8689.220
210
0.0000
Source: Calculation from survey data
After running model 1, it shows that this model has degree of freedom df = 181> 0
Coefficient RAMSEA is 0.069<0.08 and this shows goodness of fit of the model (Diamantopoulos and Siguaw, 2007)
Coefficient SRMR is 0.037<0.05 and this shows goodness of fit of the model (Taylor, Sharland, Cronin and Bullard, 1993)
Coefficient CFI is 0.948>0.9; coefficient TLI is 0.940>0.9 and this shows goodness of fit of the model (Segar & Grover, 1993); (Chin & Todd, 1995)
Trang 10Squared chi to the number of degree of freedom ratio (χ2/df) is 3.4
(618.078/181)<5 and this shows goodness of fit of the model because sample
size is 511>200(Kettingger & Lee,1995)
Results from running two alternative models including model 2 and model
3 are summarized in the following table (Table 4.17) (in comparison with model
1)
Summary table of information on goodness of fit and correlation of three
models shows that model 1 is the best one Model 1 has AIC = 26599.971, BIC
= 26900.753, ABIC = 26675.389 whereas model 2 has AIC = 34351.168, BIC =
34732.44, ABIC = 34446.768 and model 3 has AIC = 38650.728, BIC =
39091.311, ABIC = 38761.199 Thus, model 1 is the best one because it has the
smallest AIC, BIC and ABIC, followed by model 2 and model 3
Moreover, in considering correlation coefficient (γ), correlation
coefficient of model 1 has the higheststatistical significance (equivalent to that
of model 2 and more than that of model 3)
Table 4.17: Goodness of fit of models and Correlation coefficient
Information on goodness of
fit
Model 1 Model 2 Model 3
Adjacent value in Logarithm
unit
Value H0
Value H1
-13228.985
-12919.946
-17085.584
1.503
-19221.364
1.492
Coefficient AIC
Coefficient BIC
Coefficient adjusted BIC
26599.971 26900.753
26675.389
34351.168 34732.441
34446.768
38650.728 39091.311
38761.199
Note: F1: Online customer service quality; F2: Online information system
quality; F3: Banking service product quality, F4: Overall e-banking service
quality; F5: Customer satisfaction; F6: Customer loyalty γ is the relation
coefficient among latent variables
* Significant at value p < 0.05
** Significant at value p < 0.01
Source: Calculation from survey data
In model 2 and model 3, hypotheses H6 and H7 have no statistical
significance In other words, there is no basis to conclude that switching cost
(F6) and customer trust (f7) variables have impacts on the relationship between customer satisfaction and customer loyalty
Thus, it can be concluded that model 1 is the fittest one to sample data and model 1 will be selected in this research
4.3.2.2 Summary of hypothesis test result
Hypothesis test results are summarized and reported based on the best model (model 1) and summarized in Table 4.18 below:
Table 4.18: Summary of hypothesis testing result on model 1 Hypothesis Variable
relationship
Correlation coefficient (γ) Value p
Hypothesis acceptance
Note: F1: Online customer service quality; F2: Online information system
quality; F3: Banking service product quality, F4: Overall e-banking service quality; F5: Customer satisfaction; F6: Customer loyalty
* Significant at value p < 0.05 ** Significant at value p < 0.01
Source: Calculation from survey data
After running model, it shows that four out of five hypotheses of model 1 have relationships with statistical significance as follows
Hypothesis H1: Online customer service quality and overall e-banking service quality have a positive correlation
Model testing result shows that γ1 is equal to -0.030 and smaller than 0 whereas value p is equal to 0.852 and larger than 0.05, thus, there is no basis for this hypothesis acceptance In other words, it cannot be concluded that online customer service quality and overall e-banking service quality have a positive correlation This result is different from result inresearches of Jun & Cai (2001) [81]; Han & Baek (2004) [68]; Nguyen Thi Mai Trang & Tran Xuan Thu Huong (2010) [171]; Le Thi Tuyet Trinh (2012) [163]
Hypothesis H2: Online information system quality and overall e-banking service quality have a positive correlation
Model testing result shows that γ2is equal to 0.537>0 whereas value p =
0.004<0.01, thus, this hypothesis can be accepted In other words, it can be concluded that online information system quality and overall e-banking service quality have a positive correlation This conclusion is generally similar to that of previous researches of Jun&Cai (2001)[81]; Broderick & Vachirapornpuk (2002)[27]; Han &Baek (2004)[68]; Flavian, Tores & Guinaliu (2004)[57]; Yang et al (2004)[150]; Siu &Mou (2005)[134]; Maenpaa (2006)[98]; Nguyen Thi Phuong Cham (2008)[166], etc