Determinants of customer satisfaction

Global Economy and Finance Journal
Vol. 7. No. 1. March 2014. Pp. 63 – 82

Determinants of Customer Satisfaction on Retail Banks in
New Zealand: An Empirical Analysis Using Structural
Equation Modeling
Moha Asri Abdullah1, Noor Hazilah A. Manaf2, Muhammad-Bashir Owolabi
Yusuf3, Kamrul Ahsan4 and S. M. Ferdous Azam5
Customer retention is very crucial to the continuous survival of retail
banking anywhere in the world, most especially when the deregulation
of the sector has provided the customers with different choices to
satisfy their financial needs. This has made many banks to pursue
different strategies that will increase their customer satisfaction through
enhanced service quality. This study examined the determinants of
retail bank customer satisfaction in New Zealand through the survey of
their perception about the banks service quality. The five dimensions of
service quality were initially analysed in relation to customer
satisfaction using the structural equation modeling technique but three
were eventually used. The three factors specified to determine
customer satisfaction in retail banking were found to be both practically
and statistically significant. The implication is that the core, the
enabling and the relational aspect of service quality must be taken care
of by the banks to satisfy their customers in order to retain their loyalty.

Keywords: Customer Satisfaction, Service Quality, Structural Equation Modeling,
Reliability, Assurance, Enabling, Retail Banking.

1. Introduction
Improving service quality has being the primary goal of service industries for the past
five decades, most especially when studies have linked customer satisfaction with good
service quality. This is true, particularly, in retail banking where there is little or no

differentiation of the products offered. The alternative means of retaining-expanding the
customer base is to enhance the quality of services provided to sustain customer
satisfaction. Maintaining customer satisfaction is very crucial to retail bank continuous
existence since no bank can remain in business without loyal customers. Researchers
have enumerated the benefits of customer loyalty as a result of their satisfaction in the
quality of services obtained from their service providers. These include increased profit,
1

Professor of Economics, Faculty of Economics and Management Sciences, International Islamic
University, , +0361964649
2
Associate Professor, Faculty of Economics and Management Sciences, International Islamic University
Malaysia, +0361964756
3
Post Doctoral Research Fellow, Department of Economics, Kulliyyah of Economics and Management
Sciences, International Islamic University, , +60186656956
4
Senior Lecturer, Victoria University of Melbourne, City Flinders Campus Room CF10.29, 300 Flinders
Street, Melbourne VIC 3000, , +613 99191174
5
PhD Research Fellow, Department of Business Administration, Kulliyyah of Economics and Management
Sciences, International Islamic University, , +60166831785


Abdullah, Manaf, Yusuf, Ahsan & Azam
reduction in service cost, better understanding of financial affairs and needs of their
clients and the opportunity to cross-sell the old and new products (Levesque and
McDougall, 1996). Some other benefits are positive words of mouth, readiness to pay
charged price and inclination to see one‟s bank as a “relationship” bank (Arbore and
Busacca, 2009).

Thus, it becomes the duty of retail bank managers to devote their strategies to activities
that will surpass their customers‟ service expectations. Parasuraman et al (1995)
identified ten dimensions of service quality which was later refined to five - Reliability,
Responsiveness, Assurance, Empathy and Tangibles - an assertion that has been
strongly debated in the literature (Levesque and McDougall, 1996). However, there
seems to be a unanimous opinion on the fact that the dimensions of service quality
depend on the service setting and environment with empirical and theoretical evidence
pointing to two of these as the overriding dimensions- the core and the relational
factors- to be universal. The banking sector in New Zealand comprises of both
indigenously owned and foreign incorporated banks that obtained license to operate in
the country. This creates a competitive environment which requires a lot of initiatives in
order to survive. Since it is difficult to differentiate the services provided, the probable
option is to improve customer satisfaction through enhancement of the quality of service
they provide. This study, therefore, tests these dimensions of service quality in New
Zealand with a view to establish those that determine customer satisfaction in retail
banking. Survey data were obtained from 115 customers of six different banks in New
Zealand, and the data collected were analysed using the structural equation modeling
technique. This study confirmed the five dimensions of service quality in New Zealand
out of which three determined the retail bank customer satisfaction. The output of this
work is expected to assist the policy makers to know which of the aspects of service
quality should be given the highest priority.
The rest of this paper is organized with next section (literature review) focusing on the
customer satisfaction in retail banking. Emphasising on service quality (SERVQUAL)
model, the paper reviews the most cited articles on the topic published in academic
journals to analyze the determinants of customer satisfaction as well as constructing
several hypotheses. After the literature review section, materials and methods section
discusses the study area, questionnaire development, sample selection, instrument and
scaling of measurement, data analysis and hypotheses testing etc. After that, the results
and discussion section illustrates demographic characteristics of the respondents,
measurement model, analysis of the structural model, revised structural model, results

of the hypotheses testing and the discussions of the findings. Finally, the Conclusion
provides some policy implications in the real world perspective as well as suggestions
for further studies to achieve more accurate understandings of retail banking services in
New Zealand.

2. Literature Review
The importance of customer satisfaction both practically and theoretically for firms‟
continuous survival cannot be over emphasized (Naser, 2003; Zalatar, 2012). The idea

64


Abdullah, Manaf, Yusuf, Ahsan & Azam
of Customer‟s satisfaction refers to fulfillment of customer‟s expectation (Vesel and
Zabkar, 2009). This is a perception a customer has after using a particular product or
service (Naser, 2003), which antecedents may stem from either emotion or cognition
(Yu and Dean, 2001; Vesel and Zabkar, 2009). Past studies have highlighted the
intangibility of service. Unlike products, service can only be experienced and not seen in
real life, which the assessment is ex-post. As such “it has been argued that intangibility
is the single most important difference between products and services” (Santos, 2002,
p. 1).Thus a service produced, whether good or bad will have to be definitely
experienced by a customer (Jamal & Naser, 2003). Therefore, it becomes paramount to
every organization to monitor the quality of service they provided.
Service quality is an important primary concern of every service organization. This is
because it is a prerequisite for service company both for its survival and to gain
competitive advantage over and above its rival (Zalatar, 2012). Service quality refers to
the difference that exists between customers‟ service expectation and what he actually
received in a particular transaction. The dimensions of service quality were first
conceptualized by Parasuraman et al (1985). They identified five different aspects
employed by customers to assess the quality of service they receive. These are:

Reliability, Responsiveness, Assurance, Empathy and Tangibles. To effectively quantify
these service quality dimensions, Parasuraman et al (1988) developed a 22-item
questionnaire, known as „SERVQUAL‟ instrument, to assess customer‟s expectation
and service performance through these dimensions. Since then, many models and
instruments to quantify service quality have been developed. The models and
instruments have been widely employed in studies conducted on service quality in
different service industries (Zalatar, 2012).
Looking at some of the studies on the determinants of customer satisfaction in retail
banking, Levesque and McDougall (1996) studied the determinants of customer
satisfaction in retail bank in Canada. Data was obtained from a survey of 325 church
goers. They used 17 items to measure service quality and service features on a 7-point
Likert scale, ranging from 1, strongly disagree to 7, strongly agree. All the explanatory
variables which include the service quality dimension proposed by Parasuraman et al
(1985) except bank location were found to be significant determinants of customer
satisfaction in retail banking in Canada. Arbore and Busacca (2009) conducted an
extensive study on the determinants of customer satisfaction in retail banks by obtaining
data from a well-known retail bank in Italy. Using a survey data from 5000 customers,
and a revised methodology that deviate from the traditional approach, they were able to
confirm non-linear and asymmetry relationship among the characteristics of
performances and customers‟ overall satisfaction. In essence, their finding shows
disparity between the results obtained using the tradition and revised methodology.
Jamal and Naser (2003) examined the determinants of customer satisfaction in retail
banks in Pakistan. Using a survey of 300 questionnaires that was randomly distributed
to the customers of women bank in Pakistan, they were able to show strong relationship
between various dimensions of service quality and customer satisfaction. However, the
relationship between tangible and customer satisfaction was not supported in their

65

Abdullah, Manaf, Yusuf, Ahsan & Azam
study. Alhemoud (2010) studied the determinants of customer satisfaction in Kuwait
retail banks, using 605 randomly distributed questionnaire to both citizen and noncitizen resident of Kuwait. He found that customers are generally satisfied with the
service quality provided by Kuwaiti banks. However, the ANOVA result of the data
revealed differences in the aspects of service quality that satisfy the Kuwaiti and their
non-Kuwaiti counterpart. While the Kuwaitis are thrilled with the enabling features of
banking services, it is the reliability dimension that pleases the non-indigenes. The
aspect of service quality that relates to competitiveness measured by interest and the
likes was least valued by the respondents as reported in the study.
Ehigie (2006) studies how customer expectation of service quality and satisfaction
predict the loyalty to their banks in Nigeria. The study which employed mixed methods
combined both focus group discussion (18 participants) and in-depth interview (24
respondents) to develop a measurement scale which was used to survey 247
respondents to obtain its data. Using hierarchical regression, the study revealed that
both service quality and satisfaction are significant determinants of loyalty in retail bank
with customer satisfaction contributing the more. Addo and Kwarteng (2012) assess the
determinants of customer satisfaction and the level of acceptability of services provided
by private banks in Ghana, using the service quality dimensions. They surveyed 140
respondents to take their perception about the five dimensions of service quality as
regards their banks. They analysed the data using descriptive statistics, factor analysis
and correlation. Their results indicate that all the five dimensions of service quality are
significant predictors of customer satisfaction in retail banks in Ghana. In addition, the
result showed that responsiveness and assurance are the most valued service qualities
with highest loadings. Finally they confirm direct link between customer satisfaction and
loyalty.
However, it should be noted that using „SERVQUAL‟ to measure service quality to
determine customer satisfaction has received its own share of criticism in the literature.
Specifically, it has been criticized on a number of grounds. The first is the actual number
of dimensions of service quality. Ten dimensions were original proposed which was
later refined to five. Even these five tend to vary with context and environment. The

second criticism is the stability of the instruments from one context to another which
may warrant adaptation, addition and/or deletion of the items. The third criticism is the
psychometric problem that may arise as a result of the calculation of the difference of
score (expectation minus perception of service quality) which may result in customers
overstating their prior expectation as result of their previous bad experience with the
organizations (Buttle, 1996). Nevertheless these criticisms, „SERVQUAL‟ has found
wide usage among researchers since its development. This study will not be an
exception as it used the five dimensions of service quality to establish the determinants
of customer satisfaction in retail banking in New Zealand. To the best of the authors‟
knowledge, no such study has been conducted in New Zealand before. The five
dimensions are as follow:
Reliability: Ability to perform the promised service dependably and accurately.

66


Abdullah, Manaf, Yusuf, Ahsan & Azam
Responsiveness: Willingness to help customers and provide prompt service.
Assurance: Knowledge and courtesy of employees and their ability to inspire trust and
confidence.
Empathy: Caring, individualized attention the firm provides its customers.
Tangibles: Appearance of physical facilities, equipment, personnel, and communication
materials.
2.1 Hypotheses of the Study
Following the past studies that have used service quality dimensions to predict the
determinants of customer satisfaction, this study tested five hypotheses that relate to
the link between the five service quality dimensions and customer satisfaction for New
Zealand retail banking, using the 22 SERQUAL measurement instruments. These
hypotheses are:
H1: There is significant positive relationship between tangible and customer satisfaction

H2: There is significant positive relationship between reliability and customer satisfaction
H3: There is significant positive relationship between responsiveness and customer
satisfaction
H4: There is significant positive relationship between assurance and customer
satisfaction
H5: There is significant positive relationship between empathy and customer satisfaction

67


Abdullah, Manaf, Yusuf, Ahsan & Azam
Figure 1: Proposed Model of the Determinants of Customer Satisfaction in Retail
Banking in New Zealand

3. Materials and Methods
3.1 Study Area
This study was carried out in New Zealand. New Zealand is an Island country that is
situated in the south western Pacific Ocean. Geographically, it is made up of two main
land masses- the north and south island together with several smaller islands. New
Zealand is located roughly 1,500km to the east of Australia transversely to the Tasman
Sea and approximately 1,000km to the countries in the south of Pacific Island- New
Caledonia, Fiji and Tonga. Due to its isolation, it falls under one of the few last islands to
be inhabited by man. New Zealand economy is based on primary products which serve
as the bases of its export that comes from agriculture. The country has efficient
agricultural system and is the world leading exporter of a number of agricultural
products such as dairy products, meat, fish, wool, fruits, vegetables and forest products.
New Zealand also has a good reserve of natural gas and its leading manufacturing
sector includes metal fabrication, food processing, paper products and wood. The
banking sector in New Zealand is regulated by its central bank known as the Reserve
Bank of New Zealand which manages the country‟s monetary policy, maintains price

stability, promotes efficient financial system and supplies the its currency. The banks
operating in New Zealand are registered under the reserve bank. At present, there are
22 registered banks in New Zealand out of which 10 operate in New Zealand as
branches of oversea incorporated banks.

68


Abdullah, Manaf, Yusuf, Ahsan & Azam
3.2 Questionnaire Development
The questionnaire used for this study contains three parts. The first part was on the
demographic characteristics of the respondents. This part surveys the general
characteristics of the respondents, which include age, gender, academic qualification,
marital status, type of banks etc. The second part contains questions on service quality
of the banks. This aspect adopted the measurement scale of service quality developed
by Parasuraman et al (1994). This scale consists of 22 items that measure various
dimensions of service quality. The respondents were asked to give their perception of
their banks service qualities based on a 5-point Likert-scale. These numbers represent:
1 –strongly disagree, 2 –disagree, 3 –neutral, 4 -agreed and 5 –strongly agree. This
was to allow the respondents some degree of flexibility when responding to the
questions. The final part of the questionnaire was on the overall satisfaction of
customers based on the services provided by their banks. This part was adopted from
Levesque and McDougall (1996), and was also based on a 5 –point Likert scale. In this
part, the meanings of the numbers are: 1 –very satisfied, 2 –satisfied, 3 – neutral, 4 –
dissatisfied and 5 – very dissatisfied. This was printed and administered to the
respondents.
3.3 Sample
This study uses purposive sampling method to select its respondents. Participants for
this study were individuals who operate current or saving account in any of the banks,
age 18 years and above, in New Zealand. Purposive sampling, a convenience sampling

method, is a non-probability sample that satisfies certain criteria (Cooper and Schindler,
2001).In all, 300 questionnaires were distributed to different respondents from which
120 was returned, making the effective response rate to be 40%. However, 5 of these
questionnaires were excluded from further analysis because of non-conformity to the
requirement (criteria) to be used as samples and excessive missing data. These are
questionnaires which parts are missing completely at random (MCAR).Following the
suggestion by Hair et al (2010), any solution to rectify missing data could be used.
Nevertheless, given the fact that the missing information was so great as to render the
questionnaire un-usable, we preferred to remove the responses in these questionnaires
from our subsequent analysis. Therefore, the final sample size was 115 respondents.
Given the minimum sample size requirement of five per indicator, this number is
deemed to be adequate. However, it should be noted that the maximum likelihood
method employed by SEM software requires a large sample for consistent output.
3.4 Instrument and Scaling of Measurement
As mentioned above, this study employs questionnaire items from the existing literature.
Because our study entails using an already developed instrument in a completely new
environment, we conducted exploratory factor analysis (EFA) on the data collected to
make sure that the items loaded well on their designated constructs with a very high
reliability. The EFA was conducted factor by factor since we are using a standardized
measurement scale in other to remove poorly loaded indicator(s) from each of the
constructs before carrying out the reliability test and to be sure the items on each
69


Abdullah, Manaf, Yusuf, Ahsan & Azam
construct are measuring the same thing (Hair et al., 2010). The initial questionnaire
items for the latent variables of the service quality in the model consisted of 22 manifest
indicators from which one was eventually removed. The remaining 21 indicators were
adopted and used for this study. This tally with Muthen (2001) who recommends
conducting EFA in a confirmatory factor analysis (CFA) framework to confirm more

concrete nature of the structural model.
We employed principal component analysis (PCA) of the factor extraction technique,
using verimax rotation option to obtain factors of maximum variance with Eigen value of
1 and above from a data set with few orthogonal components. This is appropriate for
variable reduction prior to performing CFA. All the items have factor loadings of more
than 0.7 on their construct (Table 1).
Table 1: Result of the EFA on the dimensions of service quality and customer
satisfaction
Construct

Factor
Communality Crombach’s
loadings
Alpha
Tangible
0.879
KMO: 0.827
Latest technology
0.843
0.711
Bartlett‟s Test:
Attractive office
0.891
0.794
χ2 (231.41; 6) = Neat
appearing 0.865
0.748
000
employees
Attractive materials

0.824
0.697
Variance
Extracted
Reliability=
0.910
73.3
KMO: 0.885
Sincere
interest
in 0.865
0.749
Bartlett‟s Test:
customer
2
χ (381.39; 10) = Performing service right 0.897
0.805
000
Providing
promised 0.908
0.825
service
Variance
Accuracy of bank record 0.832
0.692
Extracted =
Informing
customer 0.793
0.629
74.00

about service
Responsiveness
KMO: 0.736
Bartlett‟s Test:
χ2 (229.72; 3) =
000

Indicators

0.908
Giving prompt service
0.904
Willingness
to
help
customers
0.943
Increased
customer‟s
confidence
0.912

0.817
0.888
0.831

Variance
Extracted =
84.57

70


Abdullah, Manaf, Yusuf, Ahsan & Azam
Assurance
KMO: 0.849
Customer feel safe with
Bartlett‟s Test:
the bank
2
χ (259.78; 10) = Courteous employees
000
Employee
knowledgeable
Employee
give
Variance
customer attention
Extracted =
Employee understand
65.36
customer

0.866
0.753

0.567

0.849

0.720

0.766

0.587

0.865

0.748

0.803

0.645

Enabling
KMO: 0.724
Many ATM at different 0.890
Bartlett‟s Test:
locations
2
χ (139.41; 55) = Many branches at diff 0.869
000
locations
Different product and 0.858
Variance
service mix
Extracted =
76.14
Source: Authors‟ computation

0.842
0.792
0.755
0.736

The measure of sampling adequacy and suitability of the latent constructs for EFA
indicated by their KMOs (all above 0.5) and Bartlett‟s tests (all highly significant)
together with the factor loading of each of the indicators is presented in the Table 1
above. The factor loadings represent each of the latent variable‟s level of construct
validity. Hair et al (2010) observe that the entire factor loading should be more than
0.50, which means 25 percent of the total variance is accounted for by the factor. The
implication of this is that the loading of each of the indicator should be 0.70 and above
for it to account for 50 percent of the variance of the construct it measures. However,
this value should not be more than 0.9 which also indicates singularity. The presence of
variables with loadings above 0.9 is an indication of singularity in the model. All the total
variance extracted were more than 60 percent, the value recommended in the literature.
3.5 Data Analysis and Hypotheses Testing
We started our data analysis with the descriptive statistics of the demographic part to
summarize patterns in the responses of cases in the sample. Means, frequency
distribution tables etc, were used to bring out the salient information of the data.
Structural Equation Modeling Technique (SEM) was used to test the hypotheses of this
study. Structural equation modeling is a second generation regression techniques that
is superior to other first generation regression analysis such as multivariate regression

71


Abdullah, Manaf, Yusuf, Ahsan & Azam
technique. SEM superiority has been especially noted in its ability to handle a large
number of dependent and independent variables simultaneously. SEM is distinct from

other multivariate techniques for its usefulness when an endogenous variable becomes
an exogenous variable in the same analysis. We chose SEM for this study for two basic
reasons. The first reason is the presence of multiple observed variables because of the
latent nature of our constructs for better understanding in the area of scientific inquiry
(Schumacker& Lomax, 2004). The other ground is the ability of SEM, unlike other
methods, to be able to combine both observed and unobserved variables together in
one shot (Byrnes, 2001). SEM techniques, therefore, have become the toast of
researchers in conforming theoretical models to utilizing a quantitative approach.

4. Results and Discussion
4.1 Demographic Characteristics of the Respondents
The sex distribution of the respondents shows that 62 percent are male while the
remaining 38 percent are female. The age of the participants shows that most of them
(51 percent)are between the age range of 26 and 40 years, the remaining 49 percent is
divided between those who are 25 years and less (37 percent) and those above 40
years of age (12 percent). The distribution of the participants based on marital status
are roughly equal with married 49 percent, single 46 percent and the remaining 5
percent falls to others. The respondents are highly educated with overwhelming
majority, 89.6 percent, having attended college or university, 7.8 percent of the
participants did not go beyond high school while the remaining 2.6 percent have other
qualification. The participants fall into different occupation group with private sector
taking the lead, 29.6 percent, followed by public sector and student only, 22.6 percent
each. The students that engage in part time work are 16.5 percent while less than 1
percent of the respondents are unemployed. Finally, we explore the respondents‟
number of years with their banks. The data show that 57 percent of the respondents
have 5 or more years of experience with their current bank while the remaining 43
percent have less than 5 years of experience with their current banks.

72

Abdullah, Manaf, Yusuf, Ahsan & Azam
Table 2: Demographic Distribution of Respondents
Demographic Characteristics
Frequency (N)
Percentage (%)
Gender
Males
71
61.7
Females
44
38.3
Age
<20
13
11.3
21-25
29
25.2
26-30
25
21.7
31-35
17
14.8
36-40
9
7.8
41-45

9
7.8
46-50
7
6.1
>50
6
5.2
Marital Status
Single
52
45.6
Married
56
49.1
Others
6
5.3

Academic Qualification
High school
College/University
Others
Occupation
Public
Private
Self
Student
Student-Part time
Housewife

Unemployed
Others
Experience with current bank (Yr)
<5
5-10
>10

9
103
3

7.8
89.6
2.6

26
34
3
26
19
2
1
4

22.6
29.6
2.6
22.6
16.5
1.7

0.9
3.5

49
46
20

42.6
40.0
17.4

73


Abdullah, Manaf, Yusuf, Ahsan & Azam
4.2 Measurement Model
The first step in data analysis using the structural equation modeling technique is to
conduct confirmatory factor analysis to validate the constructs of the model (Hair et al,
2010). A confirmatory factor analysis was conducted on the data collected from
respondents through Structural Equation Modeling in AMOS (Version 18), using
Maximum Likelihood (ML) estimation (Byrne, 2010). The measurement model of the five
dimensions of service quality revealed that the overall data model fit was χ2 (160) =
278.717, p = .000. The significance of p statistics of the model is an indication of misfit
between the covariance matrix of the observed data and the implied covariance matrix
of the model. However, scholars have observed that chi-square is sensitive to sample
size; therefore they recommended the use of the fit indices. Following the suggestions
of Byrne (2001, 2010) and Hair et al. (2010), which propose the use of at least one
absolute fit index and one incremental fit index in addition to the chi-square and its
associated degree of freedom, we selected the normed chi-square (CMIN/DF), the
Comparative Fit Index (CFI) and the Root Mean Square Error of Approximation

(RMSEA) to evaluate our model.
The fit indices, CFI of 0.929 (above the threshold of 0.9 and above), CMIN/DF of 2.398
(within the recommended ≤ 3 cut- off point) and RMSEA of 0.08 (within the
recommended ≤ .08) were found to be appropriate (Byrne, 2001, 2010; Hair et al,
2010). All the loading values of observed variables of the model are also above .60 (well
above the recommended cut-off value of 0.5), showing that they are all statistically
significant. Thus it can be concluded that our model fit the collected data well.
However, it should be noted that some of the co-variances among the variables are
above 0.9, a sign of an existence of singularity. This prompted us to assess the
convergent and divergent validities of the model. As could be seen in table 3, while our
model exhibit convergent validity, the same could not be said of the divergent
(discriminant) validity. Another somewhat unusual observation in the measurement
model is the presence of negative sign in some of the co-variances, which may indicate
that the presence of a factor exert a negative influence on the other. Notwithstanding,
we still progressed with the analysis to the examination of structural model.
Table 3: Result of the Confirmatory Factor Analysis
Model
χ2/df
CFI
RMSEA
Cut-off Point
≤5
>0.9
<0.1
CFA (Measurement
1.74
0.929
.081
Model
CFI: Comparative fit index, RMSEA: Root mean square error of approximation.

74


Abdullah, Manaf, Yusuf, Ahsan & Azam
Table 4: Result of Convergent and Discriminant Validity
Factors
CR
AVE
MSV
ASV
Tangible
0.88
0.65
0.77
0.38
Reliability
0.92
0.68
0.96
0.45
Responsiveness
0.91
0.77
0.96
0.44
Assurance
0.87
0.57
0.37

0.13
Enabling
0.84
0.64
0.37
0.10
CR: Critical ratio, AVE: Average variance extracted, MSV: Maximum shared square
variance, ASV: Average shared square variance.
Source: Authors‟ computation
The Critical Ratio (CR) is more than 0.7 and the Average Variance Explained (AVE), a
condition for convergent validity. The meaning of this is that each item loaded on the
expected construct. However, each construct could not be said to be distinct from each
other (divergent validity) as only one of the two conditions that establishes this is met by
our model (AVE > ASV but not MSV).
Figure 2: Measurement Model of the Determinants of Customer Satisfaction in
Retail Banking in New Zealand

75


Abdullah, Manaf, Yusuf, Ahsan & Azam
4.3 Analysis of the Structural Model
The model hypothesized in Figure 1 was analysed using the same criteria as stated
above: the chi-square test, the comparative fit index (CFI) and the root mean square
error of approximation (RMSEA). Additionally, the path coefficients were examined for
statistical significance at p < .05; and practical significance at path loading of ≥ .20. First
we examined the factor loadings. All the items are well loaded on their factors and none
of them was below 0.5, the cut-off point. Then we turn to the fit indices to see how well
our data fit the model. As shown in Figure 3, the chi-square is significant, χ2 (270) =
766.690, p =0.000, normed chi-square, 2.84 (within the acceptable value of <5), CFI,

.785 (below the acceptable value of >0.9) and RMSEA of .127 (well above the
recommended value of 0.08). This shows that our model does not fit the data collected,
thus the need to find a model that fits the data.
Figure 3: Structural Model of the Determinants of Customer Satisfaction in Retail
Banking in New Zealand

76


Abdullah, Manaf, Yusuf, Ahsan & Azam
4.4 Revised Structural Model
In revising the model, we dropped the construct, tangible, that did not contribute to the
variance explained of the customer satisfaction in retail banking in New Zealand and
merged reliability and responsiveness that are highly correlated (Figure 4). Our
examination of the revised model shows that the factor loadings and fit indices are all in
order. All the item loadings are above 0.5 and all the fit indices fall within the acceptable
range. This was established with a Normed chi-square (CMIN) value of 1.871, which is
well below the cut-off value of 5 often indicated as the benchmark in SEM literature. The
CFI also yielded an impressive index of 0.915, also the RMSEA value of 0.087 is below
the 0.1 cut-off point. All these show a good fit of the model.
Figure 4: Revised Structural Model of the Determinants of Customer Satisfaction
in Retail Banking in New Zealand

77


Abdullah, Manaf, Yusuf, Ahsan & Azam
4.5 Results of the Hypotheses Testing
This study was initially set up to test five hypotheses that were represented by their
corresponding path as presented in figure 3 with their respective path loading values.

However, after revising the model to fit the data, two of the five hypotheses were
eventually dropped. These are those related to the path coefficients from the construct
tangible to customer satisfaction that was removed and the one from responsiveness to
customer satisfaction that was merged with reliability. All the three remaining
hypotheses were supported by the data with t-value significant at 5 percent. These are
the path coefficients from reliability, assurance and enabling to customer satisfaction
respectively. The three hypotheses exhibited positive relationships towards customer
satisfaction in retail bank in New Zealand (Table 5).
Table 5: Result of the structural Model
Hypothesis
Causal Path
Estimate
pstatistics
H2
Reliability
Sat
0.498
.001
H4
Assurance
Sat
0.238
.048
H5
Enabling
Sat
0.227
.048
Source: Authors‟ computation

Decision
Supported
Supported
Supported

H2 posited positive relationship between reliability and customer satisfaction in retail
bank in New Zealand. This hypothesis is supported as in table 4 with the significant of
path coefficient both practically and statistically with correct sign that indicate positive
relationship. The more the customers perceived their banks as being reliable based on
the provision of the core elements of banking operation, the more they are satisfied with
the banks. This claim is supported at the 95 percent confidence level with path loading
of 0.498 and t-score of 3.54. The second hypothesis out of the three hypotheses that
made our final model is H4 which hypothesized positive effect of assurance factor on
customer satisfaction. This hypothesis is also supported both practically and
significantly, at 95 confidence level with path loading of 0.238 and t-score of 1.980. The
last hypothesis in our final model is H5. This hypothesis tests the positive influence of
enabling factor on customer satisfaction in retail banks in New Zealand. Like the first
two hypotheses, this is also significant both practically and statistically at 95 percent
confidence level with path coefficient of 0.23 and t-score of 1.977. Finally, this model
explained 20 percent of the variance of the determinants of customer satisfaction in
retail bank in New Zealand. All these indicate good fit of our final model to the data
collected.
4.6 Discussions of the Findings
The first step in structural equation model analysis is to examine the measurement
model of the constructs for confirmation of the factor loadings. As presented in figure 3,
all the measurement items loaded very well on their constructs with all factor loadings
above 0.5, the minimum threshold recommended in literature and t-scores more than
1.96, which shows that they all reach 5 percent level of statistical significant.

78

Abdullah, Manaf, Yusuf, Ahsan & Azam
An assessment of items measuring the construct, tangible, indicates visually attractive
front office facility as the indicator with highest loading. This is reasonable since this is
the first thing that a potential customer will notice and it is a rough indication of what
should be expected. Beautiful environment is a bait to lure potential customers to go in
and see what exactly is going on within the premises of bank before they can
experience all other aspects of service quality they valued. Next on the loading is the
neat appearance of the staff, this is followed by the presence of latest service
technology and the use of attractive materials for the service. This is a logical sequence,
from front attraction that draws the customer into the bank to meeting good looking staff
that are using latest technology and serving their client with beautiful materials. Looking
at the factor, reliability that measures the core services provided by the banks, the
indicators with highest loading is the one that relates to the bank providing the service at
the promised time. The main reason why people use bank facilities is to obtain their
services. Getting this service at the time promised is very important. Thus, it is not
surprising that this indicator has the highest factor loading on reliability in the
determinant of customer satisfaction in New Zealand. This was followed by the
employee showing sincere interest in solving customer problem and performing the
service right the first time. The item with the least loading on this factor is the one on
employee informing the customer the exact services that will be performed. Assessing
the factor, responsiveness, it is discovered that employee willingness to help the
customer is the most important factor on this construct. This was followed by employee
performance that boosts customer‟s confidence and giving prompt services
respectively. However, it should be noted that these three constructs are highly
correlated and were eventually merged in our final analysis of the structural model. This
buttresses the argument in literature that some of the constructs of service quality are
overlapping (Buttle, 1996).
On the factor that measures the assurance, the most important thing to the customer

here is having personal attention from the employees. This was followed by the
employees being consistently courteous. The indicator with least loading on this
construct is that on employee having appropriate knowledge to answer customer
questions. This is rather strange, but a close look at the order shows a reasonable logic.
Of what importance is a disrespectful staff with all their knowledge if it will not be useful
to the clients? A courteous staff can ask others what he/she doesn‟t know and still be
able to present it to the customer in the best way possible, after all nobody is an island
of knowledge. Finally, turning to the last construct, enabling, our assessments show that
the customers really value bank having ATM machines at different location. This boils
down to convenience and having access to ones funds easily at ones convenience. This
is even more valued than the banks having branches at different locations or providing
different product or service mixed. However, it should be noted that the respondents
valued many bank branches and different product and service mix equally.
The final three hypotheses of this study were statistically significant. All the path
coefficients (estimates) – from reliability, assurance and enabling to customer
satisfaction – are significant at 95 percent confidence level. They are also practically
significant (≥.2) and exhibit the correct signs (+ sign). The implication of this is that the

79


Abdullah, Manaf, Yusuf, Ahsan & Azam
constructs, reliability, assurance and enabling are all predictors of customer satisfaction
in retail banks in New Zealand. The more effort exercised to enhance these factors, the
more satisfied the customers would be and the good for the banks. These three factors
jointly predicted 20 percent of the variance of customer satisfaction explained by our
model. Looking at the strength of the path coefficient, the core construct, reliability,
contributed most to the variance explained with the path coefficient of 0.498. This
showed that though the customers valued various service quality dimensions put in
place by the banks, their first and major concern is getting the primary services which

the banks were established to do. These findings are in consonant with the previous
empirical works on the determinants of customer satisfaction in retail bank in other parts
of the world. It conforms to Levesque and McDougall (1996) that found reliability,
assurance and enabling factors among the constructs that were significant in their
study. This work also corroborates Jamal and Naser (2003) that found highly significant
correlation between customer satisfaction with both core and relational factors. Finally,
the study has also learnt credence to Alhemoud (2010) that found availability of ATMs in
several locations and easy to use ATMs as some of the factors that determine customer
satisfaction in Kuwait.

5. Conclusion
The main objective of this work is to assess the service quality dimensions that
determine customer satisfaction in retail banking in New Zealand. Our model initially
consists of five factors: tangible, reliability, responsiveness, assurance and enabling,
leading to five hypotheses. However it was discovered that some of the constructs were
highly correlated which led to review of the model that eventually reduced the
hypotheses to three. From the findings, all the indicators used to measure the
constructs are all statistically significant based on the survey responses. Thus, we
conclude that the indicators are good measure of the constructs. All the three
hypotheses in the final model were statistically significant at 5 percent.
Essentially, this study has done three things. The first was the profiling of the socioeconomic characteristics of the respondents, using descriptive statistics. Our findings
show that males are the majority of the respondents, they are of average age of
between 26 and 40 years, highly educated with various types of occupations. Most of
them (57%) have more than 5 years experience with their current banks. The second
achievement of this study is the establishment of measurement model for customer
satisfaction in retail bank. All the items included in the measurement scale with the
exception of one were good measures of their corresponding constructs. Finally, the
study tested the relationship among the three dimensions of service quality and
customer satisfaction. All the three factors, reliability, assurance and enabling are
significant predictors of customer satisfaction in retail banking in New Zealand.

However, the finding of this study should be taken with caution and more studies are
needed to validate the finding especially with large samples.

80


Abdullah, Manaf, Yusuf, Ahsan & Azam
5.1 Policy Implications and Suggestion for Further Studies
This study has a number of implications for the retail bank industry in New Zealand. It
has not only identified the dimensions of service quality that contribute to customer
satisfaction, but also how important each of these indicators is to their dimensions
according to the customers‟ perceptions. Therefore, any policy to improve customer
satisfaction in retail banking in New Zealand will know what to target and according to
what priority, either from the government or the managers of these industries. However,
it should be noted that more studies need to be done to validate this finding, most
especially with a larger sample size. Further studies that will assess this finding across
the demographic variables and bank ownership are also recommended for the good of
retail bank industries in New Zealand.

References
Addo A and Kwarteng, K 2012, „Customers satisfaction in retail banking services: a
study of selected private banks in Ghana‟, International Journal of Social Science
Tomorrow, Vol.1, No. 6, pp. 1-9
Alhemoud, AN 2010, „Banking in Kuwait: a customer satisfaction case study‟,
International Business Journal, Vol. 20, No. 4, pp. 333-342.
Arbore, A and Busacca, B 2009, „Customer satisfaction and dissatisfaction in retail
banking: exploring the asymmetric impact of attribute performance‟,Journal of
Retailing and Consumer Services, Vol.16, pp. 271–280.
Benjamin, O E 2006, „Correlates of customer loyalty to their bank: a case study in
Nigeria‟, International Journal of Bank Marketing, Vol. 24, No. 7, pp.494 – 508

Buttle, F 1996, „SERVQUAL: review, critique, research agenda‟. European Journal of
Marketing, Vol.30 No.1, pp. 8-32.
Byrne, BM2010,Structural equation modelling with AMOS; basic concepts, applications
and programming.New York: Taylor and Francis.
Hair, JF, Black, W, Babin, BJ, Anderson, RE, and Tatham, RL2010,Multivariate data
analaysis.New Jersey: Pierson Hall.
Jamal, A and Naser, K 2003, „Factors influencing customer satisfaction in the retail
banking sector in Pakistan‟.International Journal of Commerce and Manage
ment,Vol. 13, No. 2, pp. 29–53.
Santos, J 2002, „From intangibility to tangibility on service quality perceptions: a
comparison study between consumers and service providers in four service
industries‟, Managing Service Quality, Vol. 12 No. 5, pp.292 – 302
Levesque,T and McDougall,GHG 1996, „Determinants of customer satisfaction in retail
banking‟, International Journal of Bank Marketing, Vol. 14 No. 7, pp.12 – 20.
Parasuraman, A, Zeithaml, VA and Berry, LL 1985, „A conceptual model of service
quality and implications for future research‟.Journal of Marketing,Vol. 49, pp. 41–
50.
Parasuraman, A, Zeithaml, VA and Berry, LL 1988, „SERVQUAL: a multi-item scale for
measuring consumer perceptions of service quality‟,Journal of Retailing,Vol. 64,
No. 1, pp. 12–40.

81


Abdullah, Manaf, Yusuf, Ahsan & Azam
Santos, J 2002, „From Intangibility to Tangibility on Service Quality Perceptions: A
Comparison Study between Consumers and Service Providers in Four Service
Industries‟, Managing Service Quality, Vol.12, No. 5, pp. 292-302.
Schindler, DR and Cooper, PS 2001, Business Research Methods, Eight Edition, New
York, NY.McGraw-Hill.

Schumacker, RE and Lomax,RG 2004, A Beginner’s Guide to Structural Equation
Modeling, 2nd Edition. Lawrence Erlbaum Associates.
Vesel, PandZabkar, V 2009, „Managing customer loyalty through the mediating role of
satisfaction in the DIY retail loyalty program‟,Journal of retailing and customer
services, Vol. 16, pp. 396-406.
Yu, Y and Dean, A 2001, „The contribution of emotional satisfaction to consumer
loyalty‟, International Journal of Service Industry Management,Vol. 12,No. 3, pp.
234– 250.
Zalatar, WF 2012, „Quantifying Customers‟ Gender Effects on Service Quality
Perceptions of Philippine Commercial Banks, International Conference on Asia
Business Innovation and Technology Management‟, Procedia, Social and
Behaviour Sciences,Vol. 57, pp. 268-274.

82