Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890

Vol. 32 Iss. 1 (2024)

MEASURING CUSTOMER LOYALTY IN AI-ENHANCED DIGITAL BANKING -

A MULTI-DIMENSIONAL SCALE DEVELOPMENT AND VALIDATION

Mohammed Arif Hussain

Research Scholar, GITAM School of Business, GITAM University, Hyderabad, Telangana,
India
Dr. Sudha Vemaraju
Associate Professor, GITAM School of Business, GITAM University, Hyderabad, Telangana,
India
ABSTRACT
The motive of this study is the development of a multiple-dimensional/item scale for evaluating
customer loyalty to AI-powered digital banking. This research employs the modified
technology acceptance model, which accounts for Artificial Intelligence (AI) technology
factors, service quality dimensions, and customer satisfaction to establish a framework for
assessing customer loyalty towards AI-powered digital banking services. For this purpose,
400 digital banking customers in India are the survey's target audience, and the purposive
sampling method is used for the empirical investigations.
The refined scale is determined by confirmatory factor analysis. The developed measurement
scale for assessing customer loyalty to AI-powered digital banking comprises of twelve
dimensions and its thirty-one items. The twelve dimensions are: perceived usefulness, perceived
ease of use, perceived risk, perceived trust, perceived benefit, reliability, assurance,
customization, empathy, responsiveness, customer satisfaction, and customer loyalty.
The findings show that AI technology factors and aspects of service quality positively influence
customer satisfaction and loyalty. The results of the study resemble that the scores of Cronbach's
alpha, ranges between 0.72 and 0.9, and composite reliability, ranges within 0.73 and 0.96,
validates the good construct reliability and high internal consistency. The TLI and CFI values in
this study are 0.84 and 0.876, respectively that is close to threshold limits, indicates a good model
fit.
The study adopts altered TAM factors (PU, PEoU, PR, PT, and PB) and modified Service Quality
dimensions (RACER) to quantify customer loyalty to digital banking services in relation to AI.
By applying this measurement scale Managers and Policymakers may find ways to increase
banking customer satisfaction, foster customer trust, and nurture a devoted customer base depend
upon their awareness of the critical AI technology factors, service quality dimensions, and
customer perception toward customer loyalty.
KEYWORDS: Digital Banking, AI technology factors, Service quality dimensions, Customer
loyalty, Measurement instruments, Confirmatory factor analysis
I. INTRODUCTION
The use of information technology as a tool to gain a competitive edge has been growing quickly
in recent years. Digital services have undergone a significant transformation due to the rise in
popularity of the internet, technological advancements, and evolving customer needs (N.
Halvadia & S. Patel, 2009). Information technology and digital platforms are becoming

13 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

increasingly vital for banks to deliver financial services to customers efficiently. Banks have set
up their web portals to offer digital services and benefit from digital banking's limitless time,
space, cost savings, and increased customer base (Chien-Ta Bruce Ho and Wen-Chuan Lin,
2009). Using AI technology and high-quality services is crucial for banks to remain competitive.
Consequently, it is critical to leverage effective technology and high-quality service to set oneself
apart from other service providers.
The main attribute of digital banking remains its service quality, even though its form differs
from that of traditional services. In light of these developments, service quality is a critical
concern in digital banking. One of the concepts in service marketing studied the most is service
quality. Many studies have created metrics to evaluate traditional banks' services; however,
relatively few studies have developed metrics to assess digital banking services' services (Jun
and Cai, 2001). Furthermore, there needs to be more research that examines the AI technology,
service quality characteristics, and user perceptions of individual banks in the context of digital
banking. Researchers and managers are increasingly interested in measuring service quality in
digital banking services. In addition to developing a scale with multiple-item for assessing digital
banking service quality, the study aims to investigate the fundamental aspects of digital banking
service quality. The TAM and service quality model are modified for this research. This approach
is more appropriate than earlier research that relied on the conventional service quality model.
The remaining components of the study are organized accordingly: the review of literature is
presented in the next segment, followed by an empirical study's methodology and results; the
discussion, research limitations, and future scope are shown in the final section.
II. LITERATURE REVIEW
Artificial Intelligence (AI)
AI is prescribed as "the theory and development of computer systems able to perform tasks, such
as speech recognition, visual perception, decision making, and language translation that normally
require human intelligence". Artificial intelligence (AI) blends machines that emulate human
behavior and cognitive processes with algorithms that resemble human intelligence. Since people
cannot adequately analyze, comprehend, and make decisions from such large databases, artificial
intelligence (AI) has been integrated into several corporate processes.
AI in the Digital Banking Environment
Applying methodologies based on the combination of information and technology can increase
operational efficiency, resource optimization, and profitability growth for banking firms. This
became possible because of the fundamental impact of AI in the sector. In particular, risk
management, cost savings, automation, and credit evaluation (credit score) have all been
impacted by AI in the banking and financial sectors. Chatbots—intelligence messaging robots
that offer personalized financial advice and instant solutions—allow businesses to introduce new
products and services more quickly. This leads to quicker transactions and decisions. A bank that
uses AI employs a variety of AI techniques. The methods provide various approaches to problem-
solving. Banks can leverage the potential of AI techniques by comprehending and applying these

14 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

methods. Neural networks, natural language processing, and support machine learning are banks'
most widely used methods.
Service quality
Academic research on service quality has increased significantly in the last few decades.
According to Parasuraman et al. (1985), services are distinct due to three factors: intangibility,
inseparability and heterogeneity. These characteristics make quality of services an elusive and
abstract concept. Service quality has been defined in numerous studies.
According to Parasuraman et al. (1988), service quality is the distinction between the perceived
performance of the company's offerings and the expectations of its customers.
Asubonteng et al. (1996) defined service quality as the difference between the expectations of
clients and the actual services rendered. According to Yasin et al. (2004), service quality is
considered an organizational asset and a significant factor in determining business sales and
profitability.
Digital service quality
An electronic service is a digitally delivered or interactive online access. E-services are
interactive information services, according to Ghosh et al. (2004). Web services delivered
digitally are known as e-services, according to Zeithaml et al. (2000). The significance of e-
service is growing in terms of both e-commerce success and failure and offering customers an
enhanced interactive information flow experience (Yang et al., 2001; Santos, 2003). Traditional
service delivery and online service delivery are very dissimilar. The e-service provider has the
ability to collect and evaluate customer data, which it then utilizes to personalize the services that
the business offers to the client (Rowley, 2006).
Service quality in digital banking
According to Jayawardhena and Foley (2000), rival banks have few opportunities to differentiate
themselves from one another as digital banking gains popularity. According to Ranganathan and
Ganapathy (2002), banks can attain competitive differentiation by providing superior quality
services compared to their competitors. Attracting and keeping customers may be primarily
determined by the quality of service provided, given the lack of physical or geographic
restrictions associated with digital banking (Liao & Cheung, 2002). Improvising the quality of
services is only possible if it is first measurable. According to Parasuraman et al. (1988), there is
ample evidence that efficient service quality measurement can be highly beneficial for both
resource allocation and customer segmentation. Researchers and managers are increasingly
interested in measuring service quality in digital banking.
OBJECTIVES OF THE STUDY:
 To identify altered dimensions for t h e a c c e p t a n c e o f A I t e c h n o l o g y a n d service
quality of digital banking services.
 To study scale development framework that could be used to measure the customer
loyaltyof AI powered digital banking.
 To develop model for measuring customer loyaltyof AI powered digital banking.

15 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

III. RESEARCH METHODOLOGY

Phase I: Scale Development and Instrument

Step 1: Item scale development

This study established a measuring instrument primarily based on the dimensions of the
Parasuraman et al. (1988) service quality scale and the technology acceptance model (TAM) by
Davis et al. (2002), with a few modified dimensions in each model. The Appendix contains the
questionnaire items below. The items have been developed using responses from users of digital
banking as well as the literature review. Initially, 36 items comprised the questionnaire, designed
using two independent variables—AI technology factors and service quality aspects—and two
dependent variables—customer satisfaction and loyalty in the digital banking interfaces.

There are five components in the altered TAM model. The following is an overview of these
factors.

Perceived Usefulness (PU)

It is the rate at which an individual assumes that digital technology has the desire to improve
banking performance. Therefore, using AI-based banking services positively affects users'
satisfaction if they reckon that the services are more valuable (Davis et al., 1989).
Perceived Ease of Use (PEoU)
Customer satisfaction and loyalty are the ultimate results of a person's belief that a particular
system will boost achievement and favorably influence users' attitudes toward technological
devices (Davis et al., 1989).
Perceived Risk (PR)
The degree of uncertainty and unpredictability that customers experience when purchasing a
good or service is known as their perceived risk (Yang et al., (2015). Strategies should be
developed with the understanding that customers are aware of the risks involved. It has been
found that the feedback from users and the technology's actual capabilities are the same. The
more risk a customer perceives, the more likely they will have a negative experience using
banking services.
Perceived Trust (PT)
It is employed to ascertain the degree of confidence associated with any financial transaction,
which is directly related to the satisfaction of the individual. According to Jane Upton (2013), it
also aids in maintaining the transactional relationship between the bank and the end user. As a
result, the proposed study has traced perceived trust as a critical component for assessing
customer loyalty in digital banking.
Perceived Benefit (PB)
The primary advantage of digital banking is that it allows consumers to feel more at ease. Online
customers have gained numerous advantages over traditional banking methods, such as lower
costs, more convenient banking, time savings, and comprehensive services (Peha & Khamitov,
2004). Therefore, customers are more likely to complete online transactions and experience
satisfaction and loyalty the more benefits they perceive.

16 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

There are five dimensions in the modified Service Quality model (RACER). Here is a discussion
of these dimensions.
Reliability (R)
It has been proposed that accurate service delivery and proper technical operation of a self-service
technology constitute reliability (Weijters et al., 2005). Numerous documented interactions have
recognized dependability as a crucial factor in assessing the quality of services (Bagozzi, 1990).
Assurance (A)
A noteworthy recommendation in an intangible service means that the related customers were
effectively instructed about the consumable services, reinforcing the positive outcome of their
service encounters. Parasuraman et al. (1991) have reported that it also appropriately enhances
user trust and reduces transactional risks.
Customization (C)
According to Kaplan and Haenlein (2007), customization represents a build-to-order technique
that provides a service that satisfies the end user's needs. Customization can assist businesses in
creating close relationships with their customers and increase customer retention or loyalty
(Lovelock & Wirtz, 2004).
Empathy (E)
According to Parasuraman et al. (2002), the fundamental concept of empathy is the expression
of the belief that a customer is special and unique. The SERVQUAL model outlines the
quantitative research that identifies service quality dimensions, such as credibility, security, and
accessibility, and how these factors impact customer satisfaction through measuring empathy.
Responsiveness (R)
According to Sheng and Liu (2010), responsiveness refers to how quickly bank employees
respond to customers. According to research, responsiveness significantly influences customer
satisfaction, loyalty, and the quality of digital banking services (Suleman et al., 2012).
Customer Satisfaction
Oliver (1997) & L. Margherio, (1998) characterized satisfaction as a customer's successful
response. Customer sentiment toward previous services and experiences is evaluated by the post-
service activity measured index, satisfaction. Experienced services that assess customer
satisfaction have a significant impact on their decisions.
Customer Loyalty
According to Oliver (R.L., 1999), it has been associated with the profound dedication connected
to re-patronizing or repeatedly purchasing the desired services consistently in the future. This
initiates the habit of buying the same set of brands repeatedly despite marketing efforts and
external factors that may influence behavior.
This study adhered to the Menor and Roth (2007) scale development framework. This research
primarily used the TAM and service quality scales put forth by other academics exploring digital
banking and quality of online platform services to develop a measuring tool for A.I. technology
factors and digital banking service quality dimensions. Following the outcome of the literature
review, a 36-item questionnaire was designed, with 12 dimensions that covered the acceptance
of A.I. technology factors, elements of service quality, satisfaction, and loyalty of customer in
digital banking.
The measurement scales' face and content validity were evaluated through a questionnaire
pretest. A panel of judges—occasionally experts—who read or examine a measuring method and

17 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

determine whether they believe it measures what its name implies can assess a measure's face
validity. The operational definition of every dimension and recommendations from specialists in
the field of services marketing were used for the content analysis. In the continuous-scale
agreement task, judges used a Likert scale with Strongly Disagree=1 and Strongly Agree=5 to
assess how well each item matched an assumed construct (Hardesty & Bearden, 2004). Seven
academicians reviewed the original questionnaire, which was then updated. Each of them offered
insightful criticism. A few items were added, deleted, or revised.
Subsequently, the participants reviewed and provided responses to the provided questionnaire.
Several adjustments were made in response to the pilot test's feedback. 31 items and 12
dimensions comprised the final questionnaire, which represented the acceptance of AI
technology, the quality of digital banking services, customer satisfaction, and loyalty.

Measurement Scale
Table 1: Scale items of AI technology factors, service quality dimensions, customer satisfaction and
loyalty.
Statement
Item Statement
Construct Source Code
Reliability Parasuraman et RL1
(RL) al. (1988);
My bank performs AI …. right the
Zeithaml &
first time
Bitner (2000);
Yang et al. (2004)
I believe my bank provides AI ….. RL2
promise to do so
I see accuracy ………. AI RL3
technology
Assurance Parasuraman et ASS1
(ASS) al. (1988); My bank provides clear …… digital
Zeithaml & banking services
Bitner (2005)
AI technology enables me …… more ASS2
quickly
I think my bank is committed ….. ASS3
banking services
Customizatio Parasuraman, CUS1
My bank greets customers
n Zeithaml and
……services
(CUS) Berry (1988)
Bank communicates ……banking CUS2
transactions
AI technology fulfills …… CUS3
customers

18 | P a g e
Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

Empathy(EM) Parasuraman et EM1

al. (1988); Bank staff serves well
Zeithaml & ………services
Bitner (2000)
My bank pays individual ….services EM2
queries
Employees of my bank …their EM3
customers
Responsivene Parasuraman et RE1
ss al. (1988);
(RE) Zeithaml & I get in time information ….. banking
Bitner (2002; services
Yang et al.(2004)

My bank's AI…… speed is high RE2

Overall the cooperation ….its RE3
customers
Perceived Davis et al AI technology is very much useful PU1
Usefulness (1989); …. Transactions
(PU) Yuan et al. (2016) AI technology improves… PU2
performance
AI enabled digital banking services PU3
….anytime
Perceived Davis et al PEU1
I find it simple to learn ….banking
Ease of Use (1989);
services
(PEU) Yuan et al. (2016)
Interaction with Robots/Chatbots PEU2
….enjoyable
It is convenient to get AI technology PEU3
…. to do
Perceived Yang et al (2015); I hope my transaction information PR1
Risk (PR) Yuan et al. (2016) …. using AI technology
I believe AI technology…..my PR2
personal information
The risk associated ….. the PR3
transactions is low
Perceived Owusu Kwateng I rely on AI technology…. of PT1
Trust (PT) et al (2019) financial transactions
AI technology ….. secure all the time PT2
I have confidence in the privacy of AI PT3
…..banking services
Perceived Owusu et al AI technology banking services … PB1
Benefit (PB) (2020) lot of comforts

19 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

The time spent is worthwhile …. AI PB2

technology-enabled digital banking
AI technology allows me …. banking PB3
services
Customer Sikdar et al I am generally convinced with AI CS1
Satisfaction (2015) technology….
(CS) I am happy with …. banking services CS2
Overall I am fully satisfied …my CS3
bank
Customer Alonso-Dos- CL1
I prefer AI technology…..other
Loyalty (CL) Santos et al
banking services
(2020)
I will recommend AI technology- CL2
enabled digital banking ……
I intend to continue …..offered by CL3
my bank

Step 2: Data collection

Sampling Design
The selection of digital banking as a sample industry was based on several factors, including the
industry's high service intensity, the complexity of its services, its rapid growth, and the
heightened sensitivity of its clientele to the quality of banking services. Respondents to this study
were Union Bank, State Bank of India, ICICI Bank, and HDFC Bank customers who used digital
banking services. These banks were chosen based on their market capitalization. Since this
study's target population is unknown, and a sampling frame is unavailable, non-probability
sampling was employed. In this case, the researcher decided that the most significant number of
AI-powered digital banking customers of chosen banks could be accessible at branches of chosen
banks, a sort of convenience sampling known as judgmental or purposeful sampling was used to
select study participants.
A verification question about whether the respondent was using the AI-powered digital banking
services of the chosen Indian public or private sector banks was included at the questionnaire's
start to ensure the instrument met its goal. Additionally, respondents were advised to focus on
the digital banking services that they utilize most regularly. A purposive sampling was employed
to gather data from 400 digital banking users to evaluate customer loyalty to AI-powered digital
banking services.

Phase II: Scale refinement and purification

IV. DATA AND EMPIRICAL RESULTS
The measurement model was validated using a confirmatory factor analysis (CFA). SMARTPLS
and SPSS were used in this study to analyze the data. In order to enhance the model fit, some
residual covariance of the indicators representing various constructs was released, and some
indicators were loaded on constructs they were not meant to represent. These adjustments were

20 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

made to the modified AI technology factors and digital banking service quality model. The
questionnaire contained 36 items initially, and the final scale's retained items numbered 31. The
appendix contains information about the retained items.
The Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy (less than 0.5 is unacceptable)
yields an index ranging from zero to one when all other variables predict every variable exactly
and without error. This study data's KMO index was 0.897, regarded as "meritorious" (Hair et
al., 1998).

KMO and Bartlett's Test

Table 2: Sampling Adequacy Measurement

KMO and Bartlett's Test
Kaiser-Meyer-Olkin Measure of
0.897
Sampling Adequacy
Approx. Chi-Square 22103.455
Bartlett's Test of
df 621
Sphericity
Sig. <.001

Assessment of Reliability
The standardized Cronbach's alpha coefficient was used to assess reliability (Cronbach, 1951).
As stated by Hair et al. (2007), Cronbach's alpha value of 0.6 is the lower threshold of acceptance.
For every distinct construct of AI technology factors and service quality, Cronbach's alpha was
calculated after reexamining each dimension and removing items by the SMARTPLS
recommended criteria. Table 3 initially displays Cronbach's alpha scores for each item on the AI
technology factors and service quality scale, as well as customer satisfaction and loyalty. All
items, except PB, CS, and CL, have Cronbach's alpha scores close to or lower than 0.6. Table 4
indicates that after removing the PB3, CS1, CS2, CL1, and CL2 items from the questionnaire,
for all the remaining items the final Cronbach's alpha coefficients ranges from 0.72 to 0.90 (refer
to Table 5), indicating good internal consistency and construct reliability within each construct.
Furthermore, the 31 items' combined scale reliability is 0.90. According to Parasuraman et al.
(1991), the combined scale's high alpha value suggests that the scale's convergent validity and
reliability were both satisfied. All of the scales used in this study had strong internal reliability,
according to Cronbach's alpha scores displayed in Table 5.
Measurement Model
Table 3: FL, CA, AVE and CR scores for all items of AI technology factors, service quality
dimensions, customer satisfaction
and customer loyalty.
Dimensions Statements FL CA AVE CR
Code

Reliability RL1
0.871

21 | P a g e
Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

RL2 0.817 0.733 0.84

0.882
RL3
0.810
Assurance ASS1
0.811
0.759 0.665 0.759
ASS2
0.832
ASS3
0.804
Customization CUS1
0.807
0.775 0.689 0.779
CUS2
0.838
CUS3
0.844
Empathy EM1
0.754
0.724 0.642 0.738
EM2
0.843
EM3
0.803
Responsiveness RE1
0.855
0.808 0.722 0.82
RE2
0.853
RE3
0.842
Perceived Usefulness PU1
0.791
0.833 0.76 0.844
PU2
0.92
PU3
0.883
Perceived Ease of Use PEU1
0.833
0.75 0.658 0.745
PEU2
0.807
PEU3
0.792
Perceived Risk PR1
0.916
0.903 0.839 0.906
PR2
0.895
PR3
0.935
Perceived Trust PT1
0.909
0.857 0.789 0.864
PT2
0.837
PT3
0.9
Perceived Benefit PB1
0.16
0.495 0.349 0.862
PB2
0.197

22 | P a g e
Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

PB3
0.989
Customer Satisfaction CS1
0.937
0.548 0.589 0.862
CS2
0.938
CS3
0.082
Customer Loyalty CL1
0.915
0.386 0.562 0.789
CL2
0.907
CL3
-0.153
*FL=Factor Loading, CA=Cronbach’s alpha, AVE=Average variance exracted,
CR=Composite reliability

Table 4: Refined scale items of AI technology factors, service quality dimensions, customer
satisfaction and customer loyalty.
Statement
Final Item Statement
Construct Source Code
Reliability Parasuraman et RL1
(RL) al. (1988),
My bank performs AI …. right the
Zeithaml &
first time
Bitner (2000),
Yang et al. (2004)
I believe my bank provides AI ….. RL2
promise to do so
I see accuracy ………. AI RL3
technology
Assurance Parasuraman et ASS1
(ASS) al. (1988), My bank provides clear …… digital
Zeithaml & banking services
Bitner (2005)
AI technology enables me …… more ASS2
quickly
I think my bank is committed ….. ASS3
banking services
Customizatio Parasuraman, CUS1
My bank greets customers
n Zeithaml and
……services
(CUS) Berry (1988)
Bank communicates ……banking CUS2
transactions
AI technology fulfills …… CUS3
customers
Empathy(EM) Parasuraman et Bank staff serves well EM1
al. (1988), ………services

23 | P a g e
Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

Zeithaml &
Bitner (2000)
My bank pays individual ….services EM2
queries
Employees of my bank …their EM3
customers
Responsivene Parasuraman et RE1
ss al. (1988),
(RE) Zeithaml &
I get in time information ….. banking
Bitner
services
(2002),Yang et
al.(2004)

My bank's AI…… speed is high RE2

Overall the cooperation ….its RE3
customers
Perceived Davis et al (1989) AI technology is very much useful PU1
Usefulness …. Transactions
(PU) Yuan et al. (2016) AI technology improves… PU2
performance
AI enabled digital banking services PU3
….anytime
Perceived Davis et al (1989) I find it simple to learn ….banking PEU1
Ease of Use Yuan et al. (2016) services
(PEU) Interaction with Robots/Chatbots PEU2
….enjoyable
It is convenient to get AI technology PEU3
…. to do
Perceived Yang et al (2015) I hope my transaction information PR1
Risk (PR) Yuan et al. (2016) …. using AI technology
I believe AI technology…..my PR2
personal information
The risk associated ….. the PR3
transactions is low
Perceived Owusu Kwateng I rely on AI technology…. of PT1
Trust (PT) et al (2019) financial transactions
AI technology ….. secure all the time PT2
I have confidence in the privacy of AI PT3
…..banking services
Perceived Owusu et al AI technology banking services … PB1
Benefit (PB) (2020) lot of comforts
PB2

24 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

The time spent is worthwhile …. AI

technology-enabled digital banking
Customer Sikdar et al CS3
Satisfaction (2015) Overall I am fully satisfied …my
(CS) bank

Customer Alonso-Dos- CL3

Loyalty (CL) Santos et al
(2020) I intend to continue …..offered by
my bank

Assessment of Validity
The level of accuracy in research is referred to as validity. The data used in a scaling procedure
needs to be relevant to the measured construct and objective. Construct validity and content/face
measures were used to evaluate the validity of the model.
Content Validity
The content validity of a construct is determined by how well the measure encompasses the
theoretical definition of the construct (Rungtusanatham, 1998). The dimensions of online service
quality were determined through a literature review, and experts in the field of online services
and academics in marketing have verified the content validity of the instrument utilized in this
study.
Construct Validity
When a set of measured items accurately reflects the theoretical latent construct, it is said to have
construct validity, which is related to measurement accuracy (Hair, 2007). According to O'Leary-
Kelly and Vokurka (1998), construct validity can be proven by empirically evaluating constructs'
unidimensionality. Confirmatory Factor Analysis (CFA) offers better control when evaluating
unidimensionality. This research applied confirmatory factor analysis to evaluate both
discriminant and convergent validity.
Convergent Validity
Convergent validity is a characteristic that items used as indicators of a particular construct cover
or share a significant amount of variance (Hair et al., 2007). This study used the following
techniques to assess convergent validity: Construct reliability and factor loading analysis. Factor
loading greater than 0.5 indicates good convergent validity, and in the case of high convergent
validity, high loading on a factor would imply that they converge on a common point (Hair et al.,
2007). Table 3 first displays the factor loading of the 36 items in the AI technology factors,
service quality scale, customer satisfaction, and loyalty. The items' factor loading values of the
PB3, CS1, CS2, CL1, and CL2 items are close to or lesser than 0.5, indicating the scale's poor
convergent validity (see Table 3). As seen in Table 5, after removing the PB3, CS1, CS2, CL1,
and CL2 items from the questionnaire, The scale exhibits good convergent validity since all of
the item's factor loading values are either greater than or close to 0.5. High construct reliability
means that the measures consistently record the same latent construct, which is known as internal
consistency. When estimating construct reliability, 0.6 or higher is generally regarded as

25 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

indicative of good construct reliability. Table 5 shows that construct reliability values greater
than 0.6 have been identified in all customer loyalty-related constructs, indicating good construct
reliability.

Table 5: Refined FL, CA, AVE and CR scores for all items of AI technology factors, service quality
dimensions, customer satisfaction and customer loyalty.
Dimensions Statements
Code F L CA AVE CR

Reliability RL1
0.85
0.817 0.733 0.819
RL2
0.879
RL3
0.839
Assurance ASS1
0.81
0.749 0.665 0.751
ASS2
0.823
ASS3
0.814
Customization CUS1
0.807
0.773 0.689 0.778
CUS2
0.845
CUS3
0.837
Empathy EM1
0.786
0.725 0.646 0.729
EM2
0.858
EM3
0.762
Responsiveness RE1
0.843
0.807 0.726 0.811
RE2
0.855
RE3
0.853
Perceived PU1
0.782
Usefulness 0.833 0.76 0.852
PU2
0.924
PU3
0.887
Perceived Ease of PEU1
0.833
Use 0.75 0.657 0.745
PEU2
0.806
PEU3
0.794

26 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

Perceived Risk PR1

0.906
0.903 0.836 0.950
PR2
0.888
PR3
0.948
Perceived Trust PT1
0.871
0.858 0.768 0.959
PT2
0.904
PT3
0.857
Perceived Benefit PB1
0.885
0.787 0.824 0.814
PB2
0.927
Customer CS3 1
Satisfaction
Customer Loyalty CL3 1

*FL=Factor Loading, CA=Cronbach’s alpha, AVE=Average variance extracted, CR=Composite

reliability

Discriminant Validity
Discriminant validity, as per Hair et al. (2007), denotes how much a construct differs from other
constructs in reality. A construct with high discriminant validity is distinct and can capture certain
phenomena that additional measures cannot. If a process does not exhibit a significant correlation
with the other events that it aims to distinguish from, it is considered to have discriminant validity
(O'Leary-Kelly and Vokurka, 1998). According to Fornell and Larcker, discriminant validity is
established if the AVE is higher than the squared correlation coefficient of each of the constructs
(1981). Table 6 illustrates that the AVE value exceeds the squared correlation coefficient for
every construct, indicating that every construct in the digital banking customer loyalty scale
possesses discriminant validity.

Table 6: Discriminant Validity Analysis

Discriminant Validity-FLC
ASS CL CS CUS EM PB PEU PR PT PU RE RL
AS 0.81
S 5
0.44
CL 7 1
0.35 0.43
CS 8 6 1
CU 0.44 0.28 0.83
S 7 3 0.26 1
0.61 0.49 0.39 0.46 0.80
EM 5 6 5 6 2

27 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

0.34 0.32 0.62 0.22 0.32 0.90

PB 1 7 4 8 6 6
PE 0.34 0.15 0.52 0.37 0.12 0.81
U 3 0.29 8 5 8 4 2
- - - -
0.03 0.06 0.05 0.00 0.03 0.09 0.02 0.91
PR 7 4 6 5 5 6 4 6
- - - - -
0.05 0.08 0.02 0.02 0.04 0.05 0.05 0.72 0.87
PT 6 8 9 7 7 5 2 3 6
- -
0.39 0.50 0.23 0.27 0.40 0.22 0.29 0.02 0.03 0.86
PU 2 7 4 7 5 3 8 2 6 7
- -
0.53 0.43 0.35 0.48 0.48 0.28 0.40 0.04 0.06 0.31 0.85
RE 6 2 1 9 3 7 7 5 1 6 1
-
0.63 0.35 0.31 0.31 0.51 0.25 0.13 0.01 0.01 0.23 0.42 0.85
RL 8 4 7 5 1 2 5 6 5 1 6 7

Multicollinearity-Statistics
According to Hair, Ringle, and Sarstedt (2011), a VIF value of 5 or higher in the PLS_PM context
indicates a possible collinearity issue. The present model has no collinearity issue, as shown by
all of the VIF values in Table 7 that are less than 5.

Table 7: Collinearity Statistics

Collinearity Stats-
VIF

VIF
ASS1 1.505
ASS2 1.611
ASS3 1.425
CL3 1
CS3 1
CUS1 1.526
CUS2 1.738
CUS3 1.567
EM1 1.508
EM2 1.681
EM3 1.302
PB1 1.729

28 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

PB2 1.729
PEU1 1.496
PEU2 1.447
PEU3 1.468
PR1 2.980
PR2 2.546
PR3 3.517
PT1 2.721
PT2 1.798
PT3 2.489
PU1 1.582
PU2 2.798
PU3 2.397
RE1 1.735
RE2 1.846
RE3 1.719
RL1 1.868
RL2 2.007
RL3 1.677

Model Fit
Chi-square/df statistics, Tucker Lewis Index (TLI), and Comparative Fit Index (CFI) were
utilized in this study to evaluate the model fit of the customer loyalty model for digital banking
(Table 8). Chi-square/df was employed as the absolute fit index in this study. An ideal model,
according to Hu and Bentler (1999), has a Chi-square/df value of less than 2. In this study, the
Chi-square/df value for the customer loyalty model in digital banking was 1.689, indicating a
good model fit. Hair et al. (2007) state that another indicator of a good model fit is a Tucker
Lewis Index (TLI) value and a Comparative Fit Index (CFI) value that is near to 0.9. The digital
banking customer loyalty model's TLI and CFI values in this study were 0.84 and 0.876,
respectively, indicating a good model fit.
Table 8: Model fit indices
Type of Model Fit Index Model Fit Index Values

1.689
Absolute Fit Index Chi-square/df

Comparative Fit Index (CFI)

0.876
Relative Measure Indices

Tucker Lewis Index (TLI)

0.84

29 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

The conclusions drawn from the goodness-of-fit indices indicate that the factor framework of the
ten suggested constructs was well-established. The factors related to AI technology and the
quality dimensions of digital banking services include the following: (1) Reliability, (2)
Assurance, (3) Customization, (4) Empathy, (5) Responsiveness, (6) Perceived Usefulness, (7)
Perceived Ease of Use, (8) Perceived Risk, (9) Perceived Trust, and (10) Perceived Benefit.
It is clear from the explanation above that there is already a high degree of prediction for the
phenomena under study in the final research model.

Phase III: Scale validation

Unidimensionality and reliability: According to the analysis, every item on the present scale is
associated with a single, fundamental construct, indicating the measures' unidimensionality
(Gerbing & Anderson, 1988). The results also showed in Table 5 that the scores of Cronbach's
alpha, which vary between 0.72 and 0.9, and composite reliability, which range within 0.73 and
0.96 (Nunnally & Bernstein, 1994), validated the good construct reliability and high internal
consistency (refer Figure 1).

Figure 1: Validated Research Model

30 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

V. FINDINGS & DISCUSSION

This study utilized the highly valid and reliable scale development procedures proposed by
Parasuraman et al. (1988) and Davis et al. (2002). The researchers have developed a 31-item
scale covering 12 factors to measure customer loyalty to AI-powered digital banking. The study's
context is restricted to India. The outcomes of the study revealed that customer loyalty could be
measured as a three-factor construct formulating service quality dimensions (reliability,
assurance, customization, empathy, and responsiveness), AI technology acceptance factors
(perceived usefulness, perceived ease of use, perceived risk, perceived trust, perceived benefit)
and customer satisfaction. Higher internal consistency of the scale has been demonstrated across
various samples. Tests of reliability, such as discriminant validity and construct validity, have
been obtained by the study. "Perceived risk" was the most significant factor among the three
variables for evaluating customer loyalty. Perceived trust is the second most crucial component.
The other two most moderately significant factors are perceived usefulness and reliability.
Hence, this study adopts altered TAM factors (PU, PEoU, PR, PT, and PB) and modified Service
Quality dimensions (RACER) to quantify customer loyalty to digital banking services with
respect to AI.
The findings show that AI technology factors and aspects of service quality positively influence
customer satisfaction and loyalty. The results of the study resemble that the scores of Cronbach's
alpha, ranges between 0.72 and 0.9, and composite reliability, ranges within 0.73 and 0.96,
validates the good construct reliability and high internal consistency. The TLI and CFI values in
this study are 0.84 and 0.876, respectively that is close to threshold limits.
In relation to the model that is being tested, factors such as AI technology, customer loyalty,
customer satisfaction, and service quality dimensions all measure to establish a good fit model,
indicating that the developed model and the actual data used in the field are consistent.

VI. MANAGERIAL IMPLICATIONS

The present research adds to the body of knowledge on the subject by developing a valid and
reliable scale to evaluate customer loyalty regarding the AI-powered digital banking industry in
India. It has shown that customer satisfaction, service quality dimensions, and AI technology
factors can be used to measure a customer's loyalty to a digital bank.
The findings of the study have a number of significant implications for financial services
marketers. First, the study has shown that banks' banking services significantly determine
customer loyalty. Policymakers and marketers should always remember to offer the best services
at competitive prices, as it will be challenging to retain customers if the services are inadequate.
Second, the study will help judge consumer confidence in the banks' banking services. This will
make it easier to estimate the degree of confidence that customers have in their respective banks.
Third, studies found an indirect relationship between customer loyalty and service quality
inevitably point out upgrading service quality will increase customer satisfaction and loyalty.
Enhancement of internal or employee loyalty is typically the primary approach toward raising
the standard of services. Fourth, the study has shown that having a solid online presence is
crucial, even for developing nations like India. For this reason, every bank must have every
digital touch point possible to make their customers' lives easier.

31 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

VII. LIMITATIONS & SCOPE FOR FUTURE RESEARCH

Specific issues with this study need to be addressed. The results cannot be broadly applied
because the data collection was limited to India. Therefore, to understand the variations in the
results, future research should be carried out in other Asian nations. A comparative analysis can
be performed to learn more about the disparities in customer loyalty between developed and
developing countries. Secondly, the scale is specific to the banking industry. To ascertain any
differences, further research can be done in other sectors. It will be possible for future researchers
to investigate whether or not Islamic banks and conventional banks differ in terms of client
loyalty. This model could be valuable and intriguing for measuring customer loyalty to digital
service offerings.
REFERENCES
[1] Asubonteng, P., McCleary, K.J. and Swan, J.E. (1996), “SERVQUAL revisited: a critical
review of service quality”, The Journal of Services Marketing, Vol. 10 No. 6, pp. 62-81.
[2] Bagozzi. (1990). “The level of effort required for behavior as a moderator of the attitude-
behaviour relation,” European Journal of Social Psychology, Vol.20, Issue-1.
[3] Bartlett, M. S., “The effect of standardization on a Chi-square approximation in factor
analysis,” Biometrika, 38(3/4), 337-344. 1951.
[4] B. Davis and Fred D. Davis, Gordon, Michael G. Morris, Viswanath Venkatesh. (2002). User
Acceptance of Information Technology: Toward a Unified View, MIS Quarterly, Vol. 27, No. 3, pp.
425-478 (54 pages).

[5] Chien-Ta Bruce Ho and Wen-Chuan Lin. (2009). Measuring the service qualityof
internet banking: scale development and validation, European Business Review,
Emrald Group Publishing Ltd., Volume 22(1), pp 5-24.
doi.org/10.1108/09555341011008981
[6] David W. Gerbing and James C. Anderson. (1988). An Updated Paradigm for Scale
Development Incorporating Unidimensionality and Its Assessment, Journal of Marketing
Research, Volume 25, Issue 2, doi.org/10.1177/002224378802500207
[7] Davis. (1989). “Perceived Usefulness, Perceived Ease of Use, and User Acceptance of
Information Technology,” MIS Quarterly 13(3): 319-339.
[8] Davis A., Gordon L Flett, Avi Besser (2002). Validation of a new scale for measuring
problematic internet use: implications for pre-employment screening, 5(4):331-45. doi:
10.1089/109493102760275581
[9] Fornell, C., & Larker, D. (1981).Evaluating structural equation models with unobserved
variables and Measurement errors. Journal of Research Marketing, 27(3), 445-466.
[10] Gefen, D. (2002), “Customer loyalty in e-commerce”, Journal of the Association for
Information Systems, Vol. 3 No. 1, pp. 27-51.
[11] Ghosh, S., Surjadaja, H. and Antony, J. (2004), “Optimization of the determinants of e-
service operations”, Business Process Management Journal, Vol. 10 No. 6, pp. 616-36.

32 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

[12] Hardesty, David M. Bearden, William O. (2004). The use of expert judges in scale
development: Implications for improving face validity of measures of unobservable constructs,
Journal of Business Research, Volume 57, Issue 2, Pages 98-107.
[13] Hair, J., Anderson, R., Tatham, R. and Black, W. (1998) Multivariate data analysis. 5th
Edition, Prentice Hall, New Jersey.
[14] Hair,J. F., Black, W. C., Babin, B. J.,Anderson, R. E. , &Tatham, R. L. (2006). Multivariate
Data Analysis (6th Ed.). Pearson Education, India.
[15] Hair, J.F., Money, A.H., Samouel, P. and Page, M. (2007), "Research Methods for
Business", Education + Training, Vol. 49 No. 4, pp. 336-
337. doi.org/10.1108/et.2007.49.4.336.2
[16] Hair, J., Ringle, C. and Sarstedt, M. (2011) PLS-SEM: Indeed a Silver Bullet. Journal of
Marketing Theory and Practice, 19, 139-151. doi.org/10.2753/MTP1069-6679190202
[17] Hu, L. T., & Bentler, P. M. (1999). Cutoff Criteria for Fit Indexes in Covariance Structure
Analysis: Conventional Criteria versus New Alternatives. Structural Equation Modeling, 6, 1-
55. doi.org/10.1080/10705519909540118
[18] Indimas, S., & Fachira, I. (2017). Customer loyalty: The effects of service loyalty and the
mediating role of customer satisfaction study case: Pt sabda alam hotel. Journal of Business and
Management, 6(2), 250–261.
[19] Jayawardhena, C. and Foley, P. (2000), “Changes in banking sector – the case of internet
banking in UK”, Journal of Internet Research: Networking and Policy, Vol. 10 No. 1, pp. 19-
30.
[20] Jane Upton. (2013). “Encyclopedia of Behavioral Medicine,” pp.1580-1581.
[21] Jun, M. and Cai, S. (2001), “The key determinants of internet banking service quality: a
content analysis”, International Journal of Bank Marketing, Vol. 19 No. 7, pp. 276-91.
[22] Kaiser, H.F., “The varimax criterion for analytic rotation in factor analysis,” Psychometrika,
23, 187-200. 1958.
[23] Kaplan and Haenlein. (2007). “A model to determine customer lifetime value in a retail
banking context,” European Management Journal, Vol.25, Issue-3.
[24] Lee J Cronbach. (1951). Coefficent alpha and the internal structure of tests, Psychometrika,
Volume 16, pages 297-334.
[25] Liao, Z. and Cheung, M. (2002), “Internet-based e-banking and consumer attitudes: an
empirical study”, Information & Management, Vol. 39, pp. 283-95.
[26] L.Margherio. (1998). “The Emerging Digital Economy,” U.S. Department of Commerce.
[27] Menor, L.J. and Roth, A.V. (2007) New Service Development Competence in Retail
Banking: Construct Development and Measurement Validation. Journal of Operations
Management, 25, 825-846. doi.org/10.1016/j.jom.2006.07.004
[28] Nirav Halvadia and Prof. Shekha Patel. (2009). ONLINE SERVICE QUALITY: SCALE
DEVELOPMENT AND VALIDATION, GJRIM, Volume 9(1).
[29] Nunnally, J.C. and Bernstein, I.H. (1994). The Assessment of Reliability, Psychometric
Theory (3rd ed.), New York: McGraw Hill, 248-292.
[30] O’Leary-Kelly, S. W., &Vokurka, R. J. (1998).The empirical assessment of construct
validity, Journal of Operations Management, 16(1), 387-405.

33 | P a g e
 Central European Management Journal ISSN:2336-2693 | E-ISSN:2336-4890
Vol. 32 Iss. 1 (2024)

[31] Oliver, R.L. (1997). “Satisfaction: A Behavioral Perspective on the Consumer,” The
McGraw-Hill Companies, Inc., New York.
[32] Oliver, R.L. (1999). “Whence Consumer Loyalty,” Journal of Marketing 63, 33-34.
[33] 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, Vol. 49 No. 4, pp. 41-
50.
[34] Parasuraman, A., Zeithaml, V. A., & Berry, L. L. (1988). SERVQUAL: a multiple- item
scale for measuring consumer perceptions of services quality. Journal of Retailing, 64(1), 12-40.
[35] Parasuraman, A., Zeithaml, V., & Berry, L.L. (1991). Refinement and reassessment of the
SERVQUAL scale. Journal of Retailing, 67 (4), 420-450.
[36] Parasuraman et al. (2002). “Service Quality Delivery Through web Sites: A Critical Review
of tent Knowledge,” Journal of the Academy of Marketing Science, 30(4):362-375.
[37] Peha and Khamitov. (2004). “PayCash: A secure Efficient Internet Payment System,”
Electronic Commerce Research and Applications, 3, 381-388.
[38] Ranganathan, C., & Ganapathy, S. (2002). Key dimensions of business-to-consumer
websites. Information and Management, 39(1), 457-465.
[39] Rowley, J. (2006), “An analysis of the e-service literature: towards a research agenda”,
Internet Research, Vol. 16 No. 3, pp. 339-59.
[40] Rungtusanatham, M. J. (1998). Let’s not overlook content validity. Decision Line, 29(4),
10-13.
[41] Santos, J. (2003), “E-service quality: a model of virtual service quality dimensions”,
Managing Service Quality, Vol. 13 No. 3, pp. 233-46.
[42] Sheng and Liu. (2010). “An empirical study on the effect of e-service quality on online
customer satisfaction and loyalty,” Nankai Business Review International.
[43] Suleman et al. (2012). “Internet banking service quality, e-customer satisfaction and loyalty:
the modified e-SERVQUAL model,” TQM Journal.
[44] Weijters et al. (2005). “Customers’ usage of self-service technology in retail setting,”
Vlerick Leuven Gent working Paper Series 2005/19.
[45] Yang, Z., Peterson, R.T. and Huang, L. (2001), “Taking the pulse of internet
pharmacies”, Marketing Health Services, Vol. 21 No. 2, pp. 5-10.
[46] Yang L., Yinghao Zhao, Yicun Wang, Lei Liu, Xingyi Zhang, Bingjin Li, Ranji Cui.
(2015). The Effects of Psychological Stress on Depression, PMC,
DOI: 10.2174/1570159x1304150831150507
[47] Yasin, M., Correia, E. and Lisboa, J. (2004), “The profitability of customer-targeted quality
improvement efforts: an empirical examination”, The TQM Magazine, Vol. 16 No. 1, pp. 45-
9.
[48] Zeithaml, V.A., Parasuraman, A. and Malhotra, A. (2000), “e-Service quality: definition,
dimensions and conceptual model”, working paper, Marketing Science Institute, Cambridge,
MA.

34 | P a g e